867 copy The Meteoritical Society 2007 Printed in USA

Meteoritics amp Planetary Science 42 Nr 45 867ndash882 (2007)Abstract available online at httpmeteoriticsorg

The Lake Bosumtwi meteorite impact structure GhanamdashWhere is the magnetic source

Hernan UGALDE1 William A MORRIS1 Lauri J PESONEN2 and Sylvester K DANUOR3

1McMaster Applied Geophysics and Geological Imaging Consortium School of Geography and Earth Sciences McMaster University 1280 Main Street West Hamilton Ontario L8S 4K1 Canada

2Division of Geophysics University of Helsinki Helsinki Finland 3Department of Physics Kwame Nkrumah University of Science and Technology Kumasi Ghana

Corresponding author E-mail ugaldehmcmasterca

(Received 25 July 2006 revision accepted 17 January 2007)

AbstractndashThe Bosumtwi impact structure (Ghana) is a young and well-preserved structure where avast amount of information is available to constrain any geophysical model Previous analysis of theairborne magnetic data and results of numerical simulation of impact predicted a strongly magneticimpact-melt body underneath the lake Recent drilling through the structure did not penetrate such anexpected impact-melt rock magnetic source A new 3-D magnetic model for the structure wasconstructed based on a newly acquired higher-resolution marine magnetic data set with considerationof the observed gravity data on the lake previous seismic models and the magnetic properties andlithology identified in the two International Continental Scientific Drilling Program (ICDP) deepboreholes The new model contains highly magnetic bodies located in the northeast sector of thestructure not centered onto the drilling sites As in previous models higher magnetization than thatmeasured in outcropping impactites had to be assigned to the unexposed source bodies Integration ofthe new model with the borehole petrophysics and published geology indicates that these bodieslikely correspond to an extension to the south of the Kumasi batholith which outcrops to the northeastof the structure The possibility that these source bodies are related to the seismically identifiedcentral uplift or to an unmapped impact-melt sheet predicted by previous models of the structure is notsupported Detailed magnetic scanning of the Kumasi batholith to the north and the Bansu intrusionto the south would provide a test for this interpretation

INTRODUCTION

Impact cratering processes are essential for theunderstanding of the geological and biological evolution ofplanet Earth Terrestrial impact cratering research isimportant as the Earth is the only planet from which we canobtain a reasonable spatial and temporal impact recordIntegration of results from seismic surveys potential fieldstudies remote sensing drilling and numerical modeling ofimpact processes coupled with petrophysical data ofimpactites and target rocks provides constraints ondetermining crater dimensions and processes involved incrater formation (Artemieva et al 2004) Analysis of gravityand magnetic data of ancient impact structures givesestimates for the degree of the erosion and thus provides atool for estimating the original dimensions of deeply erodedstructures (Pilkington and Grieve 1992 Plado et al 2000Ugalde 2006) Such data may also be crucial to delineate

impact breccia especially the often strongly magnetic impactmelt bodies

The Bosumtwi impact structure in Ghana (6deg30primeN1deg25primeW) is a unique terrestrial case that allows us to studyhow an impact structure on Earth has been formed Thestructure is located about 30 km southeast of Kumasi (Fig 1)The crater is occupied by Lake Bosumtwi and has a rim-to-rim diameter of 105 km (Koeberl and Reimold 2005) Thelake itself has a diameter of 8 km and a maximum depth of75 m (Scholz et al 2002) The crater structure is the youngestof the moderately large and well-preserved impact structureson Earth the Bosumtwi impact structure was formed by ameteorite impact about 107 Myr ago in lower greenschistfacies supracrustal rocks of the 21ndash22 Gyr old BirimianSupergroup which are intruded by various mostly graniticand granodioritic bodies (Koeberl et al 1997 Koeberl andReimold 2005) (Fig 2) Based on recent numerical modelingand tektite distribution analysis the Bosumtwi crater was

868 H Ugalde et al

likely produced by an iron meteorite which came probablyfrom the northeast with a velocity of sim20 km sminus1 and impactedthe Birimian target rocks at an angle of 30ndash45deg (Artemievaet al 2004)

The structure is accessible and a wide variety ofgeophysical (airborne shipborne) data has been accumulatedto allow detailed geophysical modeling In addition a recentInternational Continental Scientific Drilling Program (ICDP)drilling project into Lake Bosumtwi in 2004 (Koeberl et al2007 Coney et al 2007 Ferriegravere et al 2007) has providedsignificant new ground truth with which to calibrate thegeophysical data sets for the interior of the structure Basedon a high-resolution airborne magnetic survey and thenavailable petrophysical and paleomagnetic data Plado et al(2000) presented a magnetic model for the Bosumtwistructure The model suggests the presence of a highlymagnetic impact-melt body just north of the lake center Inorder to reproduce the observed magnetic anomalies the

magnetic parameters of the model (susceptibility andintensity of remanent magnetization) had to be much higherthan those of suevites accessible in outcrop to the north andsouth of the lake (for a review of the geology see Koeberl andReimold 2005) The model was constructed from polygonalprisms located between 200 and 600 m depth The Plado et al(2000) model was subsequently shown to be consistent withnumerical models of the cratering process and the relatedestimate of impact-melt production estimation (Artemievaet al 2004)

Since 2000 seismic and gravity surveys have beencarried out across Lake Bosumtwi (Karp et al 2002 Scholzet al 2002) The results of the seismic surveys point to a four-layer structure consisting of water sediments impact brecciaand unfractured basement at the bottom In this modeling thebreccia layer resembled the magnetic melt body of Plado et al(2000) The seismic data outlined a double-peaked centraluplift below the lake sediments The location of the central

Fig 1 a) The location of Lake Bosumtwi in Africa (top) and Ghana (bottom) The bottom panel shows topography data from SRTM The citiesof Accra and Kumasi are marked for reference The rectangle in the bottom panel marks the location of the detailed image on the right b) Thetopography of the lake area and location the marine magnetic profiles The positions of the 2004 ICDP deep boreholes are shown for reference

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 869

uplift is slightly to the northwest of the center of the craterstructure A gravity-based 3-D model supports the water-sediment-breccia structure and is consistent with sedimentthickness and geometry of the central uplift as indicated bythe seismic data However this gravity model is unable tolocate the breccia-basement interface because of a lack of asufficient density contrast between the two units (Ugaldeet al 2007)

The two deep boreholes through the interior of thestructure (LB-07A LB-08A) revealed some surprisesAlthough the drilling yielded unequivocal evidence of thethree anticipated upper layers (Coney et al 2007 Ferriegravereet al 2007) it did not penetrate the hypothesized highlymagnetic body in the structure Instead drilling went througha weakly magnetic breccia with only minor amounts ofimpact-melt fragments (Coney et al 2007 Ferriegravere et al2007 Elbra et al 2007 Kontny et al 2007 Morris et al2007) The borehole magnetic logging revealed distinct short-wavelength magnetic anomalies in the bottom part (Morris

et al 2007) However those anomalies must be analyzedcarefully as they were determined from samples spaced at10 cm intervals and with a millimeter-scale source-sensorseparation Therefore the amplitude and frequencycomponents of these anomalies are not directly comparable tothe marine and airborne magnetic data sets Petrophysicalscanning of the cores is generally consistent with the boreholelogs and shows sporadic susceptibility and remanenceenhancements (Elbra et al 2007 Kontny et al 2007 Morriset al 2007) However these susceptibility values are oneorder of magnitude lower that the values required by thePlado et al (2000) model The questions therefore are whereis the magnetic source body and how was it formed Whatrock type is causing the anomalies Are the boreholeanomalies related to the hidden magnetic body or are theyrelated to more local layers within the stratigraphy

To answer these questions we re-analyzed the airborneand marine magnetic data for the Bosumtwi structure Thegoal was to obtain a new magnetic model which would be

Fig 2 A simplified geological map of the Bosumtwi impact crater area (modified after Koeberl and Reimold 2005)

870 H Ugalde et al

consistent with existing petrophysical and paleomagnetic datasets (Elbra et al 2007 Kontny et al 2007 Schleifer et al2006) as well as the gravity and seismic data

THE BOSUMTWI IMPACT STRUCTURE

Regional Geology

The regional Precambrian geology in and around LakeBosumtwi is dominated by the 22ndash21 Gyr old supracrustalrocks of the Birimian Supergroup comprising mainly meta-graywackes shales and mica schists and occurring in theform of several broad metasedimentary and metavolcanics

belts (Karikari et al 2007) A characteristic feature of thesupracrustal rocks is a northeast-southwest fabric trend withsteep dips either to the northwest or southeast (Reimold et al1998) (Fig 2) These regional belts are easily recognized inaeromagnetic maps (Fig 3b) A variety of granitoid intrusions(biotite or amphibole granites) is also present in theBosumtwi region (Junner 1937 Moon and Mason 1967Koeberl and Reimold 2005) Granite intrusions probablyconnected with the Kumasi batholith outcrop around thenorth northeast east and west sides of the lake (eg thePepiakese granite on the northeast side of the lake [Jones et al1981 Koeberl and Reimold 2005]) Biotite and hornblendegranodiorites (Bansu intrusion) (Koeberl and Reimold 2005)

Fig 3 The total magnetic field at Lake Bosumtwi a) The 1960 airborne survey as published by Jones et al (1981) b) The 1997 airborneresidual anomaly map after Plado et al (2000) Data from Pesonen et al (2003) Contour interval is 2 nT c) The 2001 marine survey Thedata collection lines are noted in black and the location of the 2004 ICDP drilling sites are marked as stars The lake outline is in black Thecontour interval is 2 nT Data from Danuor (2004) d) A south-north profile across the structure (line 4E of [c]) showing both the marine (lakelevel) and airborne (70 m altitude) data sets for comparison Note the greater amplitude of the marine data due to its location closer to themagnetic sources

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 871

outcrop to the south and west of the lake Numerous narrowdikes of biotite granitoid occur at many basement exposuresin the crater rim (Reimold et al 1998)

Recent rock formations include the lake beds and theproducts of weathering (laterites soil) which can havethicknesses up to 10 m Although no impact melt rock hasbeen found around the crater (Jones et al 1981 Reimold et al1998) numerous breccia and suevite exposures have beenmapped (Koeberl and Reimold 2005) In early 1999 ashallow drilling program to the north of the crater rim wasundertaken to determine the extent of the ejecta blanketaround the crater and to obtain subsurface core samples formineralogical petrological and geochemical studies of ejecta(Boamah and Koeberl 2003) A variety of impactite brecciaswere found including impact glassndashrich suevite and severaltypes of lithic breccias (Boamah and Koeberl 2003)

The impact origin of the structure was proven eg byevidence of shock deformation features (eg planardeformation features [PDFs] high pressure mineral phases inimpactites) For a review see Koeberl and Reimold (2005)and references therein

PREVIOUS GEOPHYSICAL STUDIES

Magnetic Data

The first airborne magnetic study of the structure wasmade in 1960 by Hunting Surveys Ltd (Jones et al 1981) at aflight altitude of 300 m and with a line spacing of 500 m Theoccurrence of a central negative magnetic anomaly of ~40 nTwith positive side (flank) anomalies of ~20 nT was reported(Fig 3a) The anomaly was modeled with a 200 m thickbreccia lens below the lake sediments occurring at a depth of500 m

In 1997 a high-resolution low-altitude (70 m) airbornegeophysical survey across the Bosumtwi structure was carriedout by the Geological Survey of Finland in cooperation withthe University of Vienna and the Ghana Geological SurveyDepartment (Pesonen et al 1997 Pesonen et al 2003) Itincluded measurements of the total magnetic fieldelectromagnetic field and gamma radiation data Linespacing was 500 m and the sampling rate 75 m (magnetics)15 m (EM) and 60 m (radiometric) respectively along north-south flight lines (Pesonen et al 2003) A series of 21geophysical maps (magnetic electromagnetic radiometric)of the structure were compiled by these authors

After eliminating the regional magnetic field Plado et al(2000) published a residual magnetic anomaly map of theBosumtwi structure which is strikingly similar to the onepublished by Jones et al (1981) but much more detailed(Fig 3b) The central part of the structure is characterized byrelatively high gradients At the center of the lake there is onelarge negative magnetic anomaly with two positive sideanomalies to the north and south of it consistent with the

expected expression of a magnetic source at near-equatorialmagnetic latitudes (see Fig 5 of Pesonen et al 2003)

Investigation of the magnetic anomaly pattern (Fig 3b)clearly illustrates the idea that the impact occurred into asouthwest-northeast trending metasedimentary-metavolcanicbelt This belt is not as strongly visible at the lake as is thesimilarly trending (southwest-northeast) belt to the southeastside of the structure which is superimposed by the impactstructure at its southwest edge There is a weak ring-shapedmagnetic halo around the lake The central negative anomalyis oblique to the overall southwest-northeast trendingmagnetic lineation and it is seemingly composed of 4 to 6patches It is flanked by northerly and somewhat weakersoutherly positive anomalies A weakly positive anomaly spotalmost at the center of the lake was interpreted by Plado et al(2000) as a possible central uplift

In 2001 the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi Ghana) collected a marinemagnetic data set to complement the existing airborne data(Danuor 2004) This new data set had a wider spacing of800 m and the magnetometer was towed behind the marineplatform used for that purpose Albeit having a wider linespacing than the airborne data set the marine data had theadvantage that it measured the magnetic field closer to themagnetic sources Also because of the slower movement ofthe platform the sampling rate was improved The smallerdistance to the sources and enhanced sampling rate allows abetter representation of the magnetic anomalies which cannot be achieved by pure downward analytical continuationfrom the airplane data collection surface to the lake levelFigure 3d shows a south-north profile across the main part ofthe magnetic anomaly with both the marine and airbornemagnetic data sets The greater amplitude of the marine dataset can be observed In order to isolate the main magneticanomalies and their probable causative bodies we computedthe amplitude of the analytical signal which is computedas (Nabighian 1972) and gives maxima over the edge of themagnetized bodies causing the observed anomalies with theadvantage of being independent of remanent magnetization(Nabighian 1972 Roest et al 1992 MacLeod et al 1993)The NE of the lake exhibits a couple of high-amplitude mid-to long-wavelength anomalies (A1 B1 of Fig 4) There is alarge anomaly that encircles the northern part of the lakewhich previously had been tentatively interpreted as a largepackage of highly magnetic melt-rich rocks However theanalytical signal map of Fig 4 permits the separation of theanomaly into at least three main high-amplitude anomalies ofabout 1000ndash2000 m wavelength (see Fig 4 anomalies A1A2 A3) plus some highly magnetized material to its sidesThe rest of the lake does not exhibit any other significantanomalies except for the previously noted B1 and anomaliesB3 and B2 (composed of two smaller magnetic sources) to thesouth of it B1 and B2 are aligned in a SW-NE trend parallel

ASIG X Y( ) partF partXfrasl( )2 partF partYfrasl( )2 partF partZfrasl( )2+ +=

872 H Ugalde et al

to the dominant strike of the metasedimentary andmetavolcanic belts observed in the region and anomalies C1and C2 at the western side of the lake both have mid- tohigh- amplitude and 700ndash900 m wavelength The smalleramplitude of C1 and C2 denotes probable deeper settings oftheir corresponding magnetic sources or poorer magneticproperties than those relating to the anomalies A and B

Magnetic Properties of the Rocks

The magnetic properties of the rocks found in and aroundthe Bosumtwi structure are of key importance in themodeling The data used by Plado et al (2000) aresummarized in Table 1 together with data used in this work

Plado et al (2000) demonstrated that there is a cleardifference between petrophysical properties of the targetrocks and the impact-derived suevites The fallout sueviteshave lower densities (higher porosities) and somewhat highermagnetizations (susceptibility remanent magnetization)compared to the target rocks The remanent magnetization ofsuevites prevails over induced magnetization (theKoenigsberger ratio of remanent over induced magnetizationis much higher than 1 ~4) The target rockshave strikingly homogeneous physical properties withnoticeable weak remanent magnetization and Q values

(Q lt005) A graphitic shale had fairly high remanentmagnetization (natural remanent magnetization [NRM] =386 Am Q = 067) whereas the granites (such as thePepiakese body) were magnetically very weak (NRM lt02 Am Q lt016) Unfortunately there are no measurementsof magnetic properties for the Kumasi batholith to the north ofthe lake and the Bansu intrusion to the south although both ofthem exhibit a direct spatial correlation with the airbornemagnetic anomalies

Petrophysical measurements were accomplished on thecore recovered from both holes LB-07A and LB-08A at thefacilities of the GeoForschungsZentrum (GFZ) in Potsdam in2004 (Milkereit et al 2006 Ugalde 2006 Morris et al 2007)See Morris et al (2007) for a detailed description of themagnetic susceptibility logging

NRM was measured in a few selected samples of thecores (Elbra et al 2007 Kontny et al 2007) In general NRMintensities on impact breccias vary between 04 and 20 Am(Kontny et al 2007) Although these values are higher thanthe NRM intensities reported for the target rocks sampledaround the lake (01ndash39 mAm) (Plado et al 2000) they werederived from small clasts or pieces of melt and therefore donot constitute a representative high-magnetization unit thatcould account for the large magnetic anomalies observed inthe lake area

Fig 4 Amplitude of the analytical signal (ASIG) computed from the marine magnetic data set The two ICDP deep boreholes are shown asstars The main anomalies are labeled See text for details

Q MR= MIfrasl

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 873

Previous Magnetic Models

Based on aeromagnetic data and petrophysical andpaleomagnetic data of oriented samples of impactites andtarget rocks outcropping around the lake Plado et al (2000)published a 3-D magnetic model of the structure The modelconsisted of polygonal bodies at depth from 200 to 600 mdiminishing downwards from their surface cross section Themodel of Plado et al (2000) is shown in Fig 5a

However in order to obtain a meaningful match of themodel with the observed magnetic data Plado et al (2000)had to assign much higher values to the magnetic properties tothe body than what was actually measured on rocks (seeTable 1) In the model water sediment and basement wereall treated as the same background unit with zeromagnetization The magnetic body constituting the source ofthe magnetic anomaly (the hypothesized impact melt) wasgiven uniform magnetic properties (susceptibility =00033 SI NRM = 0367 Amminus1 and Q ~43) The remanentmagnetization direction was set to D = 3 I = minus1 as measuredon suevites outcropping around the lake Plado et al (2000)discussed in detail the various possibilities for why themagnetic parameters within the structure could differ fromthose outside

Danuor obtained a 25-D model for one profile across thecenter of the structure (Danuor 2004) Similar to Plado et al(2000) in order to reproduce the observed anomaly he used ahigh magnetic susceptibility of 003 SI but assumed onlyinduced magnetization (no NRM) The proposed source bodyextended from about 250 m to about 610 m depth (Danuor2004) Danuorrsquos model is shown in Fig 5b

NEW MAGNETIC MODELING

Synthetic Modeling

Due to the dipolar nature of the Earthrsquos magnetic fieldobserved magnetic anomalies at low magnetic latitudes canbe rather complex and thus to find the proper location of themagnetic sources relative to the anomalies can be difficult Tobetter understand the expected pattern of magnetic anomalies

at low magnetic latitudes as at Lake Bosumtwi we firstperformed some synthetic modeling

The first model (Figs 6andashc) shows the magneticsignature of a magnetized prismatic body of 4000 times 4000 times400 m The size of the body is the same as that assumed byPlado et al (2000) This proves that a highly magneticanomalous source body will give rise to a negative magneticanomaly at the observed latitude (65degN) of Lake BosumtwiThe negative anomaly appears slightly to the south of thecenter of the body and is accompanied by positive anomalieson the northern and southern sides of the major anomaly Ofthese side anomalies the northern one has slightly higheramplitude than the southern one From this model it is alsoclear that the measured magnetization parameters of the deepICDP boreholes LB-07A and LB-08A are insufficient togenerate the observed magnetic anomaly Adding remanentmagnetization parallel to the ambient field has the same effectas increasing the magnetic susceptibility the shape of theanomaly does not change only its amplitude

The second synthetic model (Fig 6d) shows themagnetic signature of a thin sheet of 4000 times 4000 times 5 m Thepurpose of this model is to test the hypothesis of the observedmagnetic anomaly being caused by a stack of thin meltsheets This hypothesis arose at early stages of interpretationof the LB-07A and LB-08A borehole magnetic data whichshowed high-frequency and high-amplitude magneticanomalies However the borehole data were acquired at asampling rate of 10 cm and spacing of millimeters betweenthe magnetic sources and the sensors therefore the frequencycontent of these anomalies is related to the different samplingrate and closer distance to the magnetic sources and is notcomparable to the observed magnetic anomalies as detectedfrom the lake surface or from an airplane In any case thismodel is intended to verify whether the thin magnetic sheetidea is feasible or not The magnetization parameters used arethe same as for the 3-D model (Table 1) The amplitude of theanomaly created by such a thin sheet is 025 nT at a depth of200 m and 004 nT at 600 m depth Thus to reproduce theobserved plusmn30 nT anomaly we need at least 120 of these thinsheets assuming the best possible scenario of shallowsources at 200 m where the amplitude is the greatest If we

Table 1 Magnetic properties of the impact and target rocks as measured on outcrops around Lake Bosumtwi by Plado et al (2000) and on the LB-07A and LB-08A drill cores (Kontny et al 2007) and magnetization parameters used in the 3-D model

Magnetic susceptibilityPlado et al (2000)

Magnetic susceptibilityThis work

Remanent magnetizationPlado et al (2000)(Am)

Remanent magnetizationThis work(Am)

Rock typelayer Measured Model Measured Model Measured Model Measured Model

Water N A N App N A 0 N A 0 N A 0Sediments N A N App N A 0 N A 0 N A 0Magnetic bodies 000033 00033 00003 00002ndash005 00368 00368

D = 3 I = minus1

00004ndash02 0ndash15D = minus6I = minus12

Target rocks (background) 000015 0 00003 00003 0057 0N A = not available N App = not applicable

874 H Ugalde et al

consider the decreasing amplitude with depth the number ofsheets scales up to at least 200 Therefore the more realisticapproach of stacking the thin sheets one below the otherleaving some space in between them would first requiremore space (gt1000 m taking into account 1000 m for the200 thin sheets plus 60ndash100 m of host rock considering 05ndash1 m between each sheet) and secondly the computedmagnetic response of such a geometry would still be smallerthan the observed field Furthermore this alternate modelwould extend from 200 to 1200 m depth which is notfeasible according to the borehole logs (Coney et al 2007Deutsch et al 2007 Ferriegravere et al 2007) and also wouldmean a block of 4000 times 4000 times 1000 m of highly magneticmaterial which was not encountered by the deep ICDPboreholes

New 3-D Model

A new 3-D model was created from the marine magneticdata (Fig 7) The 3-D structure was built over the ~800 mspaced north-south profiles Each section is 850ndash1000 m wide(depending on the separation of the survey lines) The bodiesthat constitute each section are draped over the craterbathymetry A combination of forward modeling andinversion leads to a model that has an error with an average of084 nT and a standard deviation of 82 nT The basic modelwhich was constructed on the basis of the geometry inheritedby a similar 3-D model built with the gravity data (Ugaldeet al 2007) comprises four layers water sediment impactbreccia (or magnetic rocks underlying the sediments) andmdashfor consistency with Kontny et alrsquos (2007) petrophysical

Fig 5 Previous attempts to model the observed magnetic anomalies at Lake Bosumtwi a) A cross section through the center of the craterfrom Plado et al (2000) b) The Danuor (2004) model across the center of the structure Note the different directions and scales for these twoprofiles

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 875

measurementsmdasha magnetic background Kontny et al (2007)measured high remanent magnetization intensities on samplesfrom both ICDP deep boreholes (LB-07A and LB-08A)which vary depending on the volume percent content ofpyrrhotite However those magnetization variations on thebackground are beyond the resolution of the marine magneticdata set used here

The parameters used in this model calculation aresummarized in Table 1 Starting from the source bodygeometry derived from the gravity study and applyinginversion the magnetic properties of the bodies constitutingeach line were adjusted line by line in order to reproduce theobserved anomalies The inversion was constrained tomagnetic susceptibilities lt005 SI and NRM lt15 Am Whatmakes this model really 3-D is that all the bodies across theentire structure are active when computing a single line thus

unlike 2-D or 25-D models each line is affected by thebodies located laterally to it As seen in Fig 7 this modelallows a non-uniform magnetization distribution throughoutthe structure Magnetic susceptibilities vary from 00002 to005 SI and NRM intensity from 0 to 15 Am parallel to theambient field (D = minus6deg I = minus12deg)

The model complies to the borehole petrophysics withmagnetic susceptibilities and remanence values within therange of what was measured on the recovered core from thedeep ICDP boreholes (Elbra et al 2007 Kontny et al 2007Morris et al 2007) The susceptibility values are uniformlylow across most of the lake with the exception of thenortheast side of it There at a distance of ~1800 m from theICDP boreholes lie 4 bodies that cover an area of ~4000 times2000 m and where susceptibilities are much higher than thosemeasured on the IDCP core (Fig 7) (Figs 3 and 4 of Morris

Fig 6 Synthetic models for a source at a magnetic latitude similar to that of the Lake Bosumtwi impact structure The magnetic source is onemagnetized block with the parameters as given in Plado et al (2000) See text and Table 1 for details a) Plan view of the model (black outline)and its computed magnetic response b) 3-D perspective view of the model showing the synthetic lines every 100 m c) Cross section of themodel along the center of the synthetic structure d) Cross section of the thin sheet model Note the different vertical scales in (c) and (d)

876 H Ugalde et al

Fig 7 Plan view and cross sections of the new 3-D model generated in this work Top left plan view of the magnetized bodies color-codedby magnetic susceptibility Top right computed magnetic response of the 3-D model Middle and bottom rows four north-south cross sectionsof the model from west (line 2) to east (line 5) The bodies are color-coded by magnetic susceptibility The body at the top is water (k = 0)constrained by the lake bathymetry followed by the post-impact sediment layer (k = 0) which is constrained by the gravity model

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 877

et al 2007) NRM intensities are higher than 1 Am acrossmost of the lake However as the direction of remanence isparallel to the ambient field the real effect of increasing theintensity of NRM is to increase the effective susceptibility

(1)

where k is magnetic susceptibility keff is the effectivesusceptibility after adding remanence Q the Koenigsbergerratio and MI and MR are the intensity of the induced andremanent magnetization respectively Consequently thesame effect can be accomplished by reducing the intensity ofNRM and increasing susceptibility or vice-versa butreasonable limits on the amount of each are imposed bymagnetic mineralogy (compare Kontny et al 2007)

Analysis of the amplitude of the analytical signal map(Fig 4) also locates the main source of the observed magneticanomaly at the northeast corner of the lake (A1 A2 of Fig 4)and on a northeast to southwest line to the south of it (B1 andB2 of Fig 4) whichmdashnot surprisinglymdashaligns with thehighest magnetic susceptibilities found by the model

DISCUSSION

We succeeded in finding a model that fits the observedmagnetic anomalies and the petrophysical measurementsfrom the ICDP boreholes LB-07A and LB-08A Howeverthere are still some unresolved problems that we attempt toaddress here Due to the expected relatively highmagnetization of impact-melt rocks (Ugalde et al 2005Henkel and Reimold 2002) the amount of impact meltproduced by the Bosumtwi impact becomes an importantconstraint in understanding the source of the observedmagnetic anomalies Artemieva et al (2004) estimated thevolume of impact melt produced by the impact (projectilediameter between 076ndash104 km for an impact angle between30degndash60deg impacting at 20 kms) at 26ndash43 km3 and athickness of ~200 m for the melt sheet near the location of thetwo ICDP deep boreholes Plado et al (2000) interpreted theobserved magnetic anomalies as an approximately triangularbody of impact melt extending for 4000 m in the north-southdirection of 400 m thickness and of 4000 m width in east-west extent which would give approximately 32 km3 ofimpact melt However the ICDP deep boreholes did not finda considerable amount of impact melt just centimeter-widesuevite dikes in LB-08A and lt50 m of suevite in LB-07A(Coney et al 2007 Ferriegravere et al 2007 Deutsch et al 2007)In addition the melt proportion in suevitic breccias in theseboreholes was shown by these authors to be generally verysmall amounting to just a few volume The gravity modelof Ugalde et al (2007) also supports an impactite unit muchthinner than expected Therefore there are three options tounderstand the source of the magnetic anomalies a) the ICDP

deep boreholes missed the true source of the magneticanomalies and there is a large body of magnetic rock belowthe ICDP boreholes but at a greater depth b) there is a largebody of impact melt in the northeast sector of the impactstructure c) the source of the magnetic anomalies is notimpact melt and another source mechanism has yet to befound

The first option can be ruled out based on the magneticproperties of the impactites measured around the lake byPlado et al (2000) and the magnetic properties of the targetlithologies measured by Kontny et al (2007) Because of theexponential decay of the magnetic anomalies with depth tothe top of the associated magnetic sources (Spector and Grant1970) in order to have a deeper magnetic source below theICDP deep boreholes the magnetic properties of these bodieswould have to be at least ten times those measured around thelake or at the ICDP boreholes That is very unlikelyconsidering the magnetic properties measured in a few meltfragments found in the impact breccia and the targetlithologies at LB-07A and LB-08A (Kontny and Just 2006Kontny et al 2007) Kontny et al (2007) measured an averagesusceptibility of ~300 times 10minus6 SI and Q ~9 for the targetlithologies That would give an effective susceptibility of keff= k(1 + Q) = 3000 times 10minus5 SI for the whole stratigraphiccolumn which is still not sufficient to produce such largeanomalies Also if that were the case the borehole magneticlogs would show much higher amplitude anomalies towardsthe base of the boreholes (Morris et al 2007) Furthermorethe location of the ICDP deep boreholes does not match thelocation of any of the magnetic anomalies as indicated by theamplitude of the analytical signal (Fig 4) nor does thewavelength of the anomalies correspond to what is expectedfor a magnetic source located 500 m below the observationplane (much broader anomalies highly reduced amplitudes)Plado et al (2000) interpreted the magnetic anomalies asimpact melt based on the hypothesis that the magneticproperties of the suevites around the lake were diminished byweathering and alteration processes Thus the measuredvalues were not representative of the real magnetizationparameters of the impact-melt rocks that were assumed to beinside the impact structure Kontny et al (2007) indeedshowed that the rock magnetic parameters for all rocks at theICDP deep boreholes (impact lithologies and basement rocks)are distinctly higher than those reported by Plado et al(2000) however neither ICDP deep borehole intersected anyimpact melt (Coney et al 2007 Ferriegravere et al 2007 Deutschet al 2007) which rules out impact melt as a magnetic sourceFinally based on the location of the magnetic anomalies andtheir wavelength and intensity we cannot locate the source ofthe magnetic anomalies below the ICDP deep boreholes at agreater depth

The second and third options have similar implicationsregarding the location of the magnetic sources but differ withregard to the origin of the anomalies Considering both the

keff k 1 Q+( ) k 1 MR

MI-------------+⎝ ⎠

⎛ ⎞==

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

868 H Ugalde et al

likely produced by an iron meteorite which came probablyfrom the northeast with a velocity of sim20 km sminus1 and impactedthe Birimian target rocks at an angle of 30ndash45deg (Artemievaet al 2004)

The structure is accessible and a wide variety ofgeophysical (airborne shipborne) data has been accumulatedto allow detailed geophysical modeling In addition a recentInternational Continental Scientific Drilling Program (ICDP)drilling project into Lake Bosumtwi in 2004 (Koeberl et al2007 Coney et al 2007 Ferriegravere et al 2007) has providedsignificant new ground truth with which to calibrate thegeophysical data sets for the interior of the structure Basedon a high-resolution airborne magnetic survey and thenavailable petrophysical and paleomagnetic data Plado et al(2000) presented a magnetic model for the Bosumtwistructure The model suggests the presence of a highlymagnetic impact-melt body just north of the lake center Inorder to reproduce the observed magnetic anomalies the

magnetic parameters of the model (susceptibility andintensity of remanent magnetization) had to be much higherthan those of suevites accessible in outcrop to the north andsouth of the lake (for a review of the geology see Koeberl andReimold 2005) The model was constructed from polygonalprisms located between 200 and 600 m depth The Plado et al(2000) model was subsequently shown to be consistent withnumerical models of the cratering process and the relatedestimate of impact-melt production estimation (Artemievaet al 2004)

Since 2000 seismic and gravity surveys have beencarried out across Lake Bosumtwi (Karp et al 2002 Scholzet al 2002) The results of the seismic surveys point to a four-layer structure consisting of water sediments impact brecciaand unfractured basement at the bottom In this modeling thebreccia layer resembled the magnetic melt body of Plado et al(2000) The seismic data outlined a double-peaked centraluplift below the lake sediments The location of the central

Fig 1 a) The location of Lake Bosumtwi in Africa (top) and Ghana (bottom) The bottom panel shows topography data from SRTM The citiesof Accra and Kumasi are marked for reference The rectangle in the bottom panel marks the location of the detailed image on the right b) Thetopography of the lake area and location the marine magnetic profiles The positions of the 2004 ICDP deep boreholes are shown for reference

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 869

uplift is slightly to the northwest of the center of the craterstructure A gravity-based 3-D model supports the water-sediment-breccia structure and is consistent with sedimentthickness and geometry of the central uplift as indicated bythe seismic data However this gravity model is unable tolocate the breccia-basement interface because of a lack of asufficient density contrast between the two units (Ugaldeet al 2007)

The two deep boreholes through the interior of thestructure (LB-07A LB-08A) revealed some surprisesAlthough the drilling yielded unequivocal evidence of thethree anticipated upper layers (Coney et al 2007 Ferriegravereet al 2007) it did not penetrate the hypothesized highlymagnetic body in the structure Instead drilling went througha weakly magnetic breccia with only minor amounts ofimpact-melt fragments (Coney et al 2007 Ferriegravere et al2007 Elbra et al 2007 Kontny et al 2007 Morris et al2007) The borehole magnetic logging revealed distinct short-wavelength magnetic anomalies in the bottom part (Morris

et al 2007) However those anomalies must be analyzedcarefully as they were determined from samples spaced at10 cm intervals and with a millimeter-scale source-sensorseparation Therefore the amplitude and frequencycomponents of these anomalies are not directly comparable tothe marine and airborne magnetic data sets Petrophysicalscanning of the cores is generally consistent with the boreholelogs and shows sporadic susceptibility and remanenceenhancements (Elbra et al 2007 Kontny et al 2007 Morriset al 2007) However these susceptibility values are oneorder of magnitude lower that the values required by thePlado et al (2000) model The questions therefore are whereis the magnetic source body and how was it formed Whatrock type is causing the anomalies Are the boreholeanomalies related to the hidden magnetic body or are theyrelated to more local layers within the stratigraphy

To answer these questions we re-analyzed the airborneand marine magnetic data for the Bosumtwi structure Thegoal was to obtain a new magnetic model which would be

Fig 2 A simplified geological map of the Bosumtwi impact crater area (modified after Koeberl and Reimold 2005)

870 H Ugalde et al

consistent with existing petrophysical and paleomagnetic datasets (Elbra et al 2007 Kontny et al 2007 Schleifer et al2006) as well as the gravity and seismic data

THE BOSUMTWI IMPACT STRUCTURE

Regional Geology

The regional Precambrian geology in and around LakeBosumtwi is dominated by the 22ndash21 Gyr old supracrustalrocks of the Birimian Supergroup comprising mainly meta-graywackes shales and mica schists and occurring in theform of several broad metasedimentary and metavolcanics

belts (Karikari et al 2007) A characteristic feature of thesupracrustal rocks is a northeast-southwest fabric trend withsteep dips either to the northwest or southeast (Reimold et al1998) (Fig 2) These regional belts are easily recognized inaeromagnetic maps (Fig 3b) A variety of granitoid intrusions(biotite or amphibole granites) is also present in theBosumtwi region (Junner 1937 Moon and Mason 1967Koeberl and Reimold 2005) Granite intrusions probablyconnected with the Kumasi batholith outcrop around thenorth northeast east and west sides of the lake (eg thePepiakese granite on the northeast side of the lake [Jones et al1981 Koeberl and Reimold 2005]) Biotite and hornblendegranodiorites (Bansu intrusion) (Koeberl and Reimold 2005)

Fig 3 The total magnetic field at Lake Bosumtwi a) The 1960 airborne survey as published by Jones et al (1981) b) The 1997 airborneresidual anomaly map after Plado et al (2000) Data from Pesonen et al (2003) Contour interval is 2 nT c) The 2001 marine survey Thedata collection lines are noted in black and the location of the 2004 ICDP drilling sites are marked as stars The lake outline is in black Thecontour interval is 2 nT Data from Danuor (2004) d) A south-north profile across the structure (line 4E of [c]) showing both the marine (lakelevel) and airborne (70 m altitude) data sets for comparison Note the greater amplitude of the marine data due to its location closer to themagnetic sources

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 871

outcrop to the south and west of the lake Numerous narrowdikes of biotite granitoid occur at many basement exposuresin the crater rim (Reimold et al 1998)

Recent rock formations include the lake beds and theproducts of weathering (laterites soil) which can havethicknesses up to 10 m Although no impact melt rock hasbeen found around the crater (Jones et al 1981 Reimold et al1998) numerous breccia and suevite exposures have beenmapped (Koeberl and Reimold 2005) In early 1999 ashallow drilling program to the north of the crater rim wasundertaken to determine the extent of the ejecta blanketaround the crater and to obtain subsurface core samples formineralogical petrological and geochemical studies of ejecta(Boamah and Koeberl 2003) A variety of impactite brecciaswere found including impact glassndashrich suevite and severaltypes of lithic breccias (Boamah and Koeberl 2003)

The impact origin of the structure was proven eg byevidence of shock deformation features (eg planardeformation features [PDFs] high pressure mineral phases inimpactites) For a review see Koeberl and Reimold (2005)and references therein

PREVIOUS GEOPHYSICAL STUDIES

Magnetic Data

The first airborne magnetic study of the structure wasmade in 1960 by Hunting Surveys Ltd (Jones et al 1981) at aflight altitude of 300 m and with a line spacing of 500 m Theoccurrence of a central negative magnetic anomaly of ~40 nTwith positive side (flank) anomalies of ~20 nT was reported(Fig 3a) The anomaly was modeled with a 200 m thickbreccia lens below the lake sediments occurring at a depth of500 m

In 1997 a high-resolution low-altitude (70 m) airbornegeophysical survey across the Bosumtwi structure was carriedout by the Geological Survey of Finland in cooperation withthe University of Vienna and the Ghana Geological SurveyDepartment (Pesonen et al 1997 Pesonen et al 2003) Itincluded measurements of the total magnetic fieldelectromagnetic field and gamma radiation data Linespacing was 500 m and the sampling rate 75 m (magnetics)15 m (EM) and 60 m (radiometric) respectively along north-south flight lines (Pesonen et al 2003) A series of 21geophysical maps (magnetic electromagnetic radiometric)of the structure were compiled by these authors

After eliminating the regional magnetic field Plado et al(2000) published a residual magnetic anomaly map of theBosumtwi structure which is strikingly similar to the onepublished by Jones et al (1981) but much more detailed(Fig 3b) The central part of the structure is characterized byrelatively high gradients At the center of the lake there is onelarge negative magnetic anomaly with two positive sideanomalies to the north and south of it consistent with the

expected expression of a magnetic source at near-equatorialmagnetic latitudes (see Fig 5 of Pesonen et al 2003)

Investigation of the magnetic anomaly pattern (Fig 3b)clearly illustrates the idea that the impact occurred into asouthwest-northeast trending metasedimentary-metavolcanicbelt This belt is not as strongly visible at the lake as is thesimilarly trending (southwest-northeast) belt to the southeastside of the structure which is superimposed by the impactstructure at its southwest edge There is a weak ring-shapedmagnetic halo around the lake The central negative anomalyis oblique to the overall southwest-northeast trendingmagnetic lineation and it is seemingly composed of 4 to 6patches It is flanked by northerly and somewhat weakersoutherly positive anomalies A weakly positive anomaly spotalmost at the center of the lake was interpreted by Plado et al(2000) as a possible central uplift

In 2001 the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi Ghana) collected a marinemagnetic data set to complement the existing airborne data(Danuor 2004) This new data set had a wider spacing of800 m and the magnetometer was towed behind the marineplatform used for that purpose Albeit having a wider linespacing than the airborne data set the marine data had theadvantage that it measured the magnetic field closer to themagnetic sources Also because of the slower movement ofthe platform the sampling rate was improved The smallerdistance to the sources and enhanced sampling rate allows abetter representation of the magnetic anomalies which cannot be achieved by pure downward analytical continuationfrom the airplane data collection surface to the lake levelFigure 3d shows a south-north profile across the main part ofthe magnetic anomaly with both the marine and airbornemagnetic data sets The greater amplitude of the marine dataset can be observed In order to isolate the main magneticanomalies and their probable causative bodies we computedthe amplitude of the analytical signal which is computedas (Nabighian 1972) and gives maxima over the edge of themagnetized bodies causing the observed anomalies with theadvantage of being independent of remanent magnetization(Nabighian 1972 Roest et al 1992 MacLeod et al 1993)The NE of the lake exhibits a couple of high-amplitude mid-to long-wavelength anomalies (A1 B1 of Fig 4) There is alarge anomaly that encircles the northern part of the lakewhich previously had been tentatively interpreted as a largepackage of highly magnetic melt-rich rocks However theanalytical signal map of Fig 4 permits the separation of theanomaly into at least three main high-amplitude anomalies ofabout 1000ndash2000 m wavelength (see Fig 4 anomalies A1A2 A3) plus some highly magnetized material to its sidesThe rest of the lake does not exhibit any other significantanomalies except for the previously noted B1 and anomaliesB3 and B2 (composed of two smaller magnetic sources) to thesouth of it B1 and B2 are aligned in a SW-NE trend parallel

ASIG X Y( ) partF partXfrasl( )2 partF partYfrasl( )2 partF partZfrasl( )2+ +=

872 H Ugalde et al

to the dominant strike of the metasedimentary andmetavolcanic belts observed in the region and anomalies C1and C2 at the western side of the lake both have mid- tohigh- amplitude and 700ndash900 m wavelength The smalleramplitude of C1 and C2 denotes probable deeper settings oftheir corresponding magnetic sources or poorer magneticproperties than those relating to the anomalies A and B

Magnetic Properties of the Rocks

The magnetic properties of the rocks found in and aroundthe Bosumtwi structure are of key importance in themodeling The data used by Plado et al (2000) aresummarized in Table 1 together with data used in this work

Plado et al (2000) demonstrated that there is a cleardifference between petrophysical properties of the targetrocks and the impact-derived suevites The fallout sueviteshave lower densities (higher porosities) and somewhat highermagnetizations (susceptibility remanent magnetization)compared to the target rocks The remanent magnetization ofsuevites prevails over induced magnetization (theKoenigsberger ratio of remanent over induced magnetizationis much higher than 1 ~4) The target rockshave strikingly homogeneous physical properties withnoticeable weak remanent magnetization and Q values

(Q lt005) A graphitic shale had fairly high remanentmagnetization (natural remanent magnetization [NRM] =386 Am Q = 067) whereas the granites (such as thePepiakese body) were magnetically very weak (NRM lt02 Am Q lt016) Unfortunately there are no measurementsof magnetic properties for the Kumasi batholith to the north ofthe lake and the Bansu intrusion to the south although both ofthem exhibit a direct spatial correlation with the airbornemagnetic anomalies

Petrophysical measurements were accomplished on thecore recovered from both holes LB-07A and LB-08A at thefacilities of the GeoForschungsZentrum (GFZ) in Potsdam in2004 (Milkereit et al 2006 Ugalde 2006 Morris et al 2007)See Morris et al (2007) for a detailed description of themagnetic susceptibility logging

NRM was measured in a few selected samples of thecores (Elbra et al 2007 Kontny et al 2007) In general NRMintensities on impact breccias vary between 04 and 20 Am(Kontny et al 2007) Although these values are higher thanthe NRM intensities reported for the target rocks sampledaround the lake (01ndash39 mAm) (Plado et al 2000) they werederived from small clasts or pieces of melt and therefore donot constitute a representative high-magnetization unit thatcould account for the large magnetic anomalies observed inthe lake area

Fig 4 Amplitude of the analytical signal (ASIG) computed from the marine magnetic data set The two ICDP deep boreholes are shown asstars The main anomalies are labeled See text for details

Q MR= MIfrasl

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 873

Previous Magnetic Models

Based on aeromagnetic data and petrophysical andpaleomagnetic data of oriented samples of impactites andtarget rocks outcropping around the lake Plado et al (2000)published a 3-D magnetic model of the structure The modelconsisted of polygonal bodies at depth from 200 to 600 mdiminishing downwards from their surface cross section Themodel of Plado et al (2000) is shown in Fig 5a

However in order to obtain a meaningful match of themodel with the observed magnetic data Plado et al (2000)had to assign much higher values to the magnetic properties tothe body than what was actually measured on rocks (seeTable 1) In the model water sediment and basement wereall treated as the same background unit with zeromagnetization The magnetic body constituting the source ofthe magnetic anomaly (the hypothesized impact melt) wasgiven uniform magnetic properties (susceptibility =00033 SI NRM = 0367 Amminus1 and Q ~43) The remanentmagnetization direction was set to D = 3 I = minus1 as measuredon suevites outcropping around the lake Plado et al (2000)discussed in detail the various possibilities for why themagnetic parameters within the structure could differ fromthose outside

Danuor obtained a 25-D model for one profile across thecenter of the structure (Danuor 2004) Similar to Plado et al(2000) in order to reproduce the observed anomaly he used ahigh magnetic susceptibility of 003 SI but assumed onlyinduced magnetization (no NRM) The proposed source bodyextended from about 250 m to about 610 m depth (Danuor2004) Danuorrsquos model is shown in Fig 5b

NEW MAGNETIC MODELING

Synthetic Modeling

Due to the dipolar nature of the Earthrsquos magnetic fieldobserved magnetic anomalies at low magnetic latitudes canbe rather complex and thus to find the proper location of themagnetic sources relative to the anomalies can be difficult Tobetter understand the expected pattern of magnetic anomalies

at low magnetic latitudes as at Lake Bosumtwi we firstperformed some synthetic modeling

The first model (Figs 6andashc) shows the magneticsignature of a magnetized prismatic body of 4000 times 4000 times400 m The size of the body is the same as that assumed byPlado et al (2000) This proves that a highly magneticanomalous source body will give rise to a negative magneticanomaly at the observed latitude (65degN) of Lake BosumtwiThe negative anomaly appears slightly to the south of thecenter of the body and is accompanied by positive anomalieson the northern and southern sides of the major anomaly Ofthese side anomalies the northern one has slightly higheramplitude than the southern one From this model it is alsoclear that the measured magnetization parameters of the deepICDP boreholes LB-07A and LB-08A are insufficient togenerate the observed magnetic anomaly Adding remanentmagnetization parallel to the ambient field has the same effectas increasing the magnetic susceptibility the shape of theanomaly does not change only its amplitude

The second synthetic model (Fig 6d) shows themagnetic signature of a thin sheet of 4000 times 4000 times 5 m Thepurpose of this model is to test the hypothesis of the observedmagnetic anomaly being caused by a stack of thin meltsheets This hypothesis arose at early stages of interpretationof the LB-07A and LB-08A borehole magnetic data whichshowed high-frequency and high-amplitude magneticanomalies However the borehole data were acquired at asampling rate of 10 cm and spacing of millimeters betweenthe magnetic sources and the sensors therefore the frequencycontent of these anomalies is related to the different samplingrate and closer distance to the magnetic sources and is notcomparable to the observed magnetic anomalies as detectedfrom the lake surface or from an airplane In any case thismodel is intended to verify whether the thin magnetic sheetidea is feasible or not The magnetization parameters used arethe same as for the 3-D model (Table 1) The amplitude of theanomaly created by such a thin sheet is 025 nT at a depth of200 m and 004 nT at 600 m depth Thus to reproduce theobserved plusmn30 nT anomaly we need at least 120 of these thinsheets assuming the best possible scenario of shallowsources at 200 m where the amplitude is the greatest If we

Table 1 Magnetic properties of the impact and target rocks as measured on outcrops around Lake Bosumtwi by Plado et al (2000) and on the LB-07A and LB-08A drill cores (Kontny et al 2007) and magnetization parameters used in the 3-D model

Magnetic susceptibilityPlado et al (2000)

Magnetic susceptibilityThis work

Remanent magnetizationPlado et al (2000)(Am)

Remanent magnetizationThis work(Am)

Rock typelayer Measured Model Measured Model Measured Model Measured Model

Water N A N App N A 0 N A 0 N A 0Sediments N A N App N A 0 N A 0 N A 0Magnetic bodies 000033 00033 00003 00002ndash005 00368 00368

D = 3 I = minus1

00004ndash02 0ndash15D = minus6I = minus12

Target rocks (background) 000015 0 00003 00003 0057 0N A = not available N App = not applicable

874 H Ugalde et al

consider the decreasing amplitude with depth the number ofsheets scales up to at least 200 Therefore the more realisticapproach of stacking the thin sheets one below the otherleaving some space in between them would first requiremore space (gt1000 m taking into account 1000 m for the200 thin sheets plus 60ndash100 m of host rock considering 05ndash1 m between each sheet) and secondly the computedmagnetic response of such a geometry would still be smallerthan the observed field Furthermore this alternate modelwould extend from 200 to 1200 m depth which is notfeasible according to the borehole logs (Coney et al 2007Deutsch et al 2007 Ferriegravere et al 2007) and also wouldmean a block of 4000 times 4000 times 1000 m of highly magneticmaterial which was not encountered by the deep ICDPboreholes

New 3-D Model

A new 3-D model was created from the marine magneticdata (Fig 7) The 3-D structure was built over the ~800 mspaced north-south profiles Each section is 850ndash1000 m wide(depending on the separation of the survey lines) The bodiesthat constitute each section are draped over the craterbathymetry A combination of forward modeling andinversion leads to a model that has an error with an average of084 nT and a standard deviation of 82 nT The basic modelwhich was constructed on the basis of the geometry inheritedby a similar 3-D model built with the gravity data (Ugaldeet al 2007) comprises four layers water sediment impactbreccia (or magnetic rocks underlying the sediments) andmdashfor consistency with Kontny et alrsquos (2007) petrophysical

Fig 5 Previous attempts to model the observed magnetic anomalies at Lake Bosumtwi a) A cross section through the center of the craterfrom Plado et al (2000) b) The Danuor (2004) model across the center of the structure Note the different directions and scales for these twoprofiles

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 875

measurementsmdasha magnetic background Kontny et al (2007)measured high remanent magnetization intensities on samplesfrom both ICDP deep boreholes (LB-07A and LB-08A)which vary depending on the volume percent content ofpyrrhotite However those magnetization variations on thebackground are beyond the resolution of the marine magneticdata set used here

The parameters used in this model calculation aresummarized in Table 1 Starting from the source bodygeometry derived from the gravity study and applyinginversion the magnetic properties of the bodies constitutingeach line were adjusted line by line in order to reproduce theobserved anomalies The inversion was constrained tomagnetic susceptibilities lt005 SI and NRM lt15 Am Whatmakes this model really 3-D is that all the bodies across theentire structure are active when computing a single line thus

unlike 2-D or 25-D models each line is affected by thebodies located laterally to it As seen in Fig 7 this modelallows a non-uniform magnetization distribution throughoutthe structure Magnetic susceptibilities vary from 00002 to005 SI and NRM intensity from 0 to 15 Am parallel to theambient field (D = minus6deg I = minus12deg)

The model complies to the borehole petrophysics withmagnetic susceptibilities and remanence values within therange of what was measured on the recovered core from thedeep ICDP boreholes (Elbra et al 2007 Kontny et al 2007Morris et al 2007) The susceptibility values are uniformlylow across most of the lake with the exception of thenortheast side of it There at a distance of ~1800 m from theICDP boreholes lie 4 bodies that cover an area of ~4000 times2000 m and where susceptibilities are much higher than thosemeasured on the IDCP core (Fig 7) (Figs 3 and 4 of Morris

Fig 6 Synthetic models for a source at a magnetic latitude similar to that of the Lake Bosumtwi impact structure The magnetic source is onemagnetized block with the parameters as given in Plado et al (2000) See text and Table 1 for details a) Plan view of the model (black outline)and its computed magnetic response b) 3-D perspective view of the model showing the synthetic lines every 100 m c) Cross section of themodel along the center of the synthetic structure d) Cross section of the thin sheet model Note the different vertical scales in (c) and (d)

876 H Ugalde et al

Fig 7 Plan view and cross sections of the new 3-D model generated in this work Top left plan view of the magnetized bodies color-codedby magnetic susceptibility Top right computed magnetic response of the 3-D model Middle and bottom rows four north-south cross sectionsof the model from west (line 2) to east (line 5) The bodies are color-coded by magnetic susceptibility The body at the top is water (k = 0)constrained by the lake bathymetry followed by the post-impact sediment layer (k = 0) which is constrained by the gravity model

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 877

et al 2007) NRM intensities are higher than 1 Am acrossmost of the lake However as the direction of remanence isparallel to the ambient field the real effect of increasing theintensity of NRM is to increase the effective susceptibility

(1)

where k is magnetic susceptibility keff is the effectivesusceptibility after adding remanence Q the Koenigsbergerratio and MI and MR are the intensity of the induced andremanent magnetization respectively Consequently thesame effect can be accomplished by reducing the intensity ofNRM and increasing susceptibility or vice-versa butreasonable limits on the amount of each are imposed bymagnetic mineralogy (compare Kontny et al 2007)

Analysis of the amplitude of the analytical signal map(Fig 4) also locates the main source of the observed magneticanomaly at the northeast corner of the lake (A1 A2 of Fig 4)and on a northeast to southwest line to the south of it (B1 andB2 of Fig 4) whichmdashnot surprisinglymdashaligns with thehighest magnetic susceptibilities found by the model

DISCUSSION

We succeeded in finding a model that fits the observedmagnetic anomalies and the petrophysical measurementsfrom the ICDP boreholes LB-07A and LB-08A Howeverthere are still some unresolved problems that we attempt toaddress here Due to the expected relatively highmagnetization of impact-melt rocks (Ugalde et al 2005Henkel and Reimold 2002) the amount of impact meltproduced by the Bosumtwi impact becomes an importantconstraint in understanding the source of the observedmagnetic anomalies Artemieva et al (2004) estimated thevolume of impact melt produced by the impact (projectilediameter between 076ndash104 km for an impact angle between30degndash60deg impacting at 20 kms) at 26ndash43 km3 and athickness of ~200 m for the melt sheet near the location of thetwo ICDP deep boreholes Plado et al (2000) interpreted theobserved magnetic anomalies as an approximately triangularbody of impact melt extending for 4000 m in the north-southdirection of 400 m thickness and of 4000 m width in east-west extent which would give approximately 32 km3 ofimpact melt However the ICDP deep boreholes did not finda considerable amount of impact melt just centimeter-widesuevite dikes in LB-08A and lt50 m of suevite in LB-07A(Coney et al 2007 Ferriegravere et al 2007 Deutsch et al 2007)In addition the melt proportion in suevitic breccias in theseboreholes was shown by these authors to be generally verysmall amounting to just a few volume The gravity modelof Ugalde et al (2007) also supports an impactite unit muchthinner than expected Therefore there are three options tounderstand the source of the magnetic anomalies a) the ICDP

deep boreholes missed the true source of the magneticanomalies and there is a large body of magnetic rock belowthe ICDP boreholes but at a greater depth b) there is a largebody of impact melt in the northeast sector of the impactstructure c) the source of the magnetic anomalies is notimpact melt and another source mechanism has yet to befound

The first option can be ruled out based on the magneticproperties of the impactites measured around the lake byPlado et al (2000) and the magnetic properties of the targetlithologies measured by Kontny et al (2007) Because of theexponential decay of the magnetic anomalies with depth tothe top of the associated magnetic sources (Spector and Grant1970) in order to have a deeper magnetic source below theICDP deep boreholes the magnetic properties of these bodieswould have to be at least ten times those measured around thelake or at the ICDP boreholes That is very unlikelyconsidering the magnetic properties measured in a few meltfragments found in the impact breccia and the targetlithologies at LB-07A and LB-08A (Kontny and Just 2006Kontny et al 2007) Kontny et al (2007) measured an averagesusceptibility of ~300 times 10minus6 SI and Q ~9 for the targetlithologies That would give an effective susceptibility of keff= k(1 + Q) = 3000 times 10minus5 SI for the whole stratigraphiccolumn which is still not sufficient to produce such largeanomalies Also if that were the case the borehole magneticlogs would show much higher amplitude anomalies towardsthe base of the boreholes (Morris et al 2007) Furthermorethe location of the ICDP deep boreholes does not match thelocation of any of the magnetic anomalies as indicated by theamplitude of the analytical signal (Fig 4) nor does thewavelength of the anomalies correspond to what is expectedfor a magnetic source located 500 m below the observationplane (much broader anomalies highly reduced amplitudes)Plado et al (2000) interpreted the magnetic anomalies asimpact melt based on the hypothesis that the magneticproperties of the suevites around the lake were diminished byweathering and alteration processes Thus the measuredvalues were not representative of the real magnetizationparameters of the impact-melt rocks that were assumed to beinside the impact structure Kontny et al (2007) indeedshowed that the rock magnetic parameters for all rocks at theICDP deep boreholes (impact lithologies and basement rocks)are distinctly higher than those reported by Plado et al(2000) however neither ICDP deep borehole intersected anyimpact melt (Coney et al 2007 Ferriegravere et al 2007 Deutschet al 2007) which rules out impact melt as a magnetic sourceFinally based on the location of the magnetic anomalies andtheir wavelength and intensity we cannot locate the source ofthe magnetic anomalies below the ICDP deep boreholes at agreater depth

The second and third options have similar implicationsregarding the location of the magnetic sources but differ withregard to the origin of the anomalies Considering both the

keff k 1 Q+( ) k 1 MR

MI-------------+⎝ ⎠

⎛ ⎞==

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 869

uplift is slightly to the northwest of the center of the craterstructure A gravity-based 3-D model supports the water-sediment-breccia structure and is consistent with sedimentthickness and geometry of the central uplift as indicated bythe seismic data However this gravity model is unable tolocate the breccia-basement interface because of a lack of asufficient density contrast between the two units (Ugaldeet al 2007)

The two deep boreholes through the interior of thestructure (LB-07A LB-08A) revealed some surprisesAlthough the drilling yielded unequivocal evidence of thethree anticipated upper layers (Coney et al 2007 Ferriegravereet al 2007) it did not penetrate the hypothesized highlymagnetic body in the structure Instead drilling went througha weakly magnetic breccia with only minor amounts ofimpact-melt fragments (Coney et al 2007 Ferriegravere et al2007 Elbra et al 2007 Kontny et al 2007 Morris et al2007) The borehole magnetic logging revealed distinct short-wavelength magnetic anomalies in the bottom part (Morris

et al 2007) However those anomalies must be analyzedcarefully as they were determined from samples spaced at10 cm intervals and with a millimeter-scale source-sensorseparation Therefore the amplitude and frequencycomponents of these anomalies are not directly comparable tothe marine and airborne magnetic data sets Petrophysicalscanning of the cores is generally consistent with the boreholelogs and shows sporadic susceptibility and remanenceenhancements (Elbra et al 2007 Kontny et al 2007 Morriset al 2007) However these susceptibility values are oneorder of magnitude lower that the values required by thePlado et al (2000) model The questions therefore are whereis the magnetic source body and how was it formed Whatrock type is causing the anomalies Are the boreholeanomalies related to the hidden magnetic body or are theyrelated to more local layers within the stratigraphy

To answer these questions we re-analyzed the airborneand marine magnetic data for the Bosumtwi structure Thegoal was to obtain a new magnetic model which would be

Fig 2 A simplified geological map of the Bosumtwi impact crater area (modified after Koeberl and Reimold 2005)

870 H Ugalde et al

consistent with existing petrophysical and paleomagnetic datasets (Elbra et al 2007 Kontny et al 2007 Schleifer et al2006) as well as the gravity and seismic data

THE BOSUMTWI IMPACT STRUCTURE

Regional Geology

The regional Precambrian geology in and around LakeBosumtwi is dominated by the 22ndash21 Gyr old supracrustalrocks of the Birimian Supergroup comprising mainly meta-graywackes shales and mica schists and occurring in theform of several broad metasedimentary and metavolcanics

belts (Karikari et al 2007) A characteristic feature of thesupracrustal rocks is a northeast-southwest fabric trend withsteep dips either to the northwest or southeast (Reimold et al1998) (Fig 2) These regional belts are easily recognized inaeromagnetic maps (Fig 3b) A variety of granitoid intrusions(biotite or amphibole granites) is also present in theBosumtwi region (Junner 1937 Moon and Mason 1967Koeberl and Reimold 2005) Granite intrusions probablyconnected with the Kumasi batholith outcrop around thenorth northeast east and west sides of the lake (eg thePepiakese granite on the northeast side of the lake [Jones et al1981 Koeberl and Reimold 2005]) Biotite and hornblendegranodiorites (Bansu intrusion) (Koeberl and Reimold 2005)

Fig 3 The total magnetic field at Lake Bosumtwi a) The 1960 airborne survey as published by Jones et al (1981) b) The 1997 airborneresidual anomaly map after Plado et al (2000) Data from Pesonen et al (2003) Contour interval is 2 nT c) The 2001 marine survey Thedata collection lines are noted in black and the location of the 2004 ICDP drilling sites are marked as stars The lake outline is in black Thecontour interval is 2 nT Data from Danuor (2004) d) A south-north profile across the structure (line 4E of [c]) showing both the marine (lakelevel) and airborne (70 m altitude) data sets for comparison Note the greater amplitude of the marine data due to its location closer to themagnetic sources

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 871

outcrop to the south and west of the lake Numerous narrowdikes of biotite granitoid occur at many basement exposuresin the crater rim (Reimold et al 1998)

Recent rock formations include the lake beds and theproducts of weathering (laterites soil) which can havethicknesses up to 10 m Although no impact melt rock hasbeen found around the crater (Jones et al 1981 Reimold et al1998) numerous breccia and suevite exposures have beenmapped (Koeberl and Reimold 2005) In early 1999 ashallow drilling program to the north of the crater rim wasundertaken to determine the extent of the ejecta blanketaround the crater and to obtain subsurface core samples formineralogical petrological and geochemical studies of ejecta(Boamah and Koeberl 2003) A variety of impactite brecciaswere found including impact glassndashrich suevite and severaltypes of lithic breccias (Boamah and Koeberl 2003)

The impact origin of the structure was proven eg byevidence of shock deformation features (eg planardeformation features [PDFs] high pressure mineral phases inimpactites) For a review see Koeberl and Reimold (2005)and references therein

PREVIOUS GEOPHYSICAL STUDIES

Magnetic Data

The first airborne magnetic study of the structure wasmade in 1960 by Hunting Surveys Ltd (Jones et al 1981) at aflight altitude of 300 m and with a line spacing of 500 m Theoccurrence of a central negative magnetic anomaly of ~40 nTwith positive side (flank) anomalies of ~20 nT was reported(Fig 3a) The anomaly was modeled with a 200 m thickbreccia lens below the lake sediments occurring at a depth of500 m

In 1997 a high-resolution low-altitude (70 m) airbornegeophysical survey across the Bosumtwi structure was carriedout by the Geological Survey of Finland in cooperation withthe University of Vienna and the Ghana Geological SurveyDepartment (Pesonen et al 1997 Pesonen et al 2003) Itincluded measurements of the total magnetic fieldelectromagnetic field and gamma radiation data Linespacing was 500 m and the sampling rate 75 m (magnetics)15 m (EM) and 60 m (radiometric) respectively along north-south flight lines (Pesonen et al 2003) A series of 21geophysical maps (magnetic electromagnetic radiometric)of the structure were compiled by these authors

After eliminating the regional magnetic field Plado et al(2000) published a residual magnetic anomaly map of theBosumtwi structure which is strikingly similar to the onepublished by Jones et al (1981) but much more detailed(Fig 3b) The central part of the structure is characterized byrelatively high gradients At the center of the lake there is onelarge negative magnetic anomaly with two positive sideanomalies to the north and south of it consistent with the

expected expression of a magnetic source at near-equatorialmagnetic latitudes (see Fig 5 of Pesonen et al 2003)

Investigation of the magnetic anomaly pattern (Fig 3b)clearly illustrates the idea that the impact occurred into asouthwest-northeast trending metasedimentary-metavolcanicbelt This belt is not as strongly visible at the lake as is thesimilarly trending (southwest-northeast) belt to the southeastside of the structure which is superimposed by the impactstructure at its southwest edge There is a weak ring-shapedmagnetic halo around the lake The central negative anomalyis oblique to the overall southwest-northeast trendingmagnetic lineation and it is seemingly composed of 4 to 6patches It is flanked by northerly and somewhat weakersoutherly positive anomalies A weakly positive anomaly spotalmost at the center of the lake was interpreted by Plado et al(2000) as a possible central uplift

In 2001 the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi Ghana) collected a marinemagnetic data set to complement the existing airborne data(Danuor 2004) This new data set had a wider spacing of800 m and the magnetometer was towed behind the marineplatform used for that purpose Albeit having a wider linespacing than the airborne data set the marine data had theadvantage that it measured the magnetic field closer to themagnetic sources Also because of the slower movement ofthe platform the sampling rate was improved The smallerdistance to the sources and enhanced sampling rate allows abetter representation of the magnetic anomalies which cannot be achieved by pure downward analytical continuationfrom the airplane data collection surface to the lake levelFigure 3d shows a south-north profile across the main part ofthe magnetic anomaly with both the marine and airbornemagnetic data sets The greater amplitude of the marine dataset can be observed In order to isolate the main magneticanomalies and their probable causative bodies we computedthe amplitude of the analytical signal which is computedas (Nabighian 1972) and gives maxima over the edge of themagnetized bodies causing the observed anomalies with theadvantage of being independent of remanent magnetization(Nabighian 1972 Roest et al 1992 MacLeod et al 1993)The NE of the lake exhibits a couple of high-amplitude mid-to long-wavelength anomalies (A1 B1 of Fig 4) There is alarge anomaly that encircles the northern part of the lakewhich previously had been tentatively interpreted as a largepackage of highly magnetic melt-rich rocks However theanalytical signal map of Fig 4 permits the separation of theanomaly into at least three main high-amplitude anomalies ofabout 1000ndash2000 m wavelength (see Fig 4 anomalies A1A2 A3) plus some highly magnetized material to its sidesThe rest of the lake does not exhibit any other significantanomalies except for the previously noted B1 and anomaliesB3 and B2 (composed of two smaller magnetic sources) to thesouth of it B1 and B2 are aligned in a SW-NE trend parallel

ASIG X Y( ) partF partXfrasl( )2 partF partYfrasl( )2 partF partZfrasl( )2+ +=

872 H Ugalde et al

to the dominant strike of the metasedimentary andmetavolcanic belts observed in the region and anomalies C1and C2 at the western side of the lake both have mid- tohigh- amplitude and 700ndash900 m wavelength The smalleramplitude of C1 and C2 denotes probable deeper settings oftheir corresponding magnetic sources or poorer magneticproperties than those relating to the anomalies A and B

Magnetic Properties of the Rocks

The magnetic properties of the rocks found in and aroundthe Bosumtwi structure are of key importance in themodeling The data used by Plado et al (2000) aresummarized in Table 1 together with data used in this work

Plado et al (2000) demonstrated that there is a cleardifference between petrophysical properties of the targetrocks and the impact-derived suevites The fallout sueviteshave lower densities (higher porosities) and somewhat highermagnetizations (susceptibility remanent magnetization)compared to the target rocks The remanent magnetization ofsuevites prevails over induced magnetization (theKoenigsberger ratio of remanent over induced magnetizationis much higher than 1 ~4) The target rockshave strikingly homogeneous physical properties withnoticeable weak remanent magnetization and Q values

(Q lt005) A graphitic shale had fairly high remanentmagnetization (natural remanent magnetization [NRM] =386 Am Q = 067) whereas the granites (such as thePepiakese body) were magnetically very weak (NRM lt02 Am Q lt016) Unfortunately there are no measurementsof magnetic properties for the Kumasi batholith to the north ofthe lake and the Bansu intrusion to the south although both ofthem exhibit a direct spatial correlation with the airbornemagnetic anomalies

Petrophysical measurements were accomplished on thecore recovered from both holes LB-07A and LB-08A at thefacilities of the GeoForschungsZentrum (GFZ) in Potsdam in2004 (Milkereit et al 2006 Ugalde 2006 Morris et al 2007)See Morris et al (2007) for a detailed description of themagnetic susceptibility logging

NRM was measured in a few selected samples of thecores (Elbra et al 2007 Kontny et al 2007) In general NRMintensities on impact breccias vary between 04 and 20 Am(Kontny et al 2007) Although these values are higher thanthe NRM intensities reported for the target rocks sampledaround the lake (01ndash39 mAm) (Plado et al 2000) they werederived from small clasts or pieces of melt and therefore donot constitute a representative high-magnetization unit thatcould account for the large magnetic anomalies observed inthe lake area

Fig 4 Amplitude of the analytical signal (ASIG) computed from the marine magnetic data set The two ICDP deep boreholes are shown asstars The main anomalies are labeled See text for details

Q MR= MIfrasl

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 873

Previous Magnetic Models

Based on aeromagnetic data and petrophysical andpaleomagnetic data of oriented samples of impactites andtarget rocks outcropping around the lake Plado et al (2000)published a 3-D magnetic model of the structure The modelconsisted of polygonal bodies at depth from 200 to 600 mdiminishing downwards from their surface cross section Themodel of Plado et al (2000) is shown in Fig 5a

However in order to obtain a meaningful match of themodel with the observed magnetic data Plado et al (2000)had to assign much higher values to the magnetic properties tothe body than what was actually measured on rocks (seeTable 1) In the model water sediment and basement wereall treated as the same background unit with zeromagnetization The magnetic body constituting the source ofthe magnetic anomaly (the hypothesized impact melt) wasgiven uniform magnetic properties (susceptibility =00033 SI NRM = 0367 Amminus1 and Q ~43) The remanentmagnetization direction was set to D = 3 I = minus1 as measuredon suevites outcropping around the lake Plado et al (2000)discussed in detail the various possibilities for why themagnetic parameters within the structure could differ fromthose outside

Danuor obtained a 25-D model for one profile across thecenter of the structure (Danuor 2004) Similar to Plado et al(2000) in order to reproduce the observed anomaly he used ahigh magnetic susceptibility of 003 SI but assumed onlyinduced magnetization (no NRM) The proposed source bodyextended from about 250 m to about 610 m depth (Danuor2004) Danuorrsquos model is shown in Fig 5b

NEW MAGNETIC MODELING

Synthetic Modeling

Due to the dipolar nature of the Earthrsquos magnetic fieldobserved magnetic anomalies at low magnetic latitudes canbe rather complex and thus to find the proper location of themagnetic sources relative to the anomalies can be difficult Tobetter understand the expected pattern of magnetic anomalies

at low magnetic latitudes as at Lake Bosumtwi we firstperformed some synthetic modeling

The first model (Figs 6andashc) shows the magneticsignature of a magnetized prismatic body of 4000 times 4000 times400 m The size of the body is the same as that assumed byPlado et al (2000) This proves that a highly magneticanomalous source body will give rise to a negative magneticanomaly at the observed latitude (65degN) of Lake BosumtwiThe negative anomaly appears slightly to the south of thecenter of the body and is accompanied by positive anomalieson the northern and southern sides of the major anomaly Ofthese side anomalies the northern one has slightly higheramplitude than the southern one From this model it is alsoclear that the measured magnetization parameters of the deepICDP boreholes LB-07A and LB-08A are insufficient togenerate the observed magnetic anomaly Adding remanentmagnetization parallel to the ambient field has the same effectas increasing the magnetic susceptibility the shape of theanomaly does not change only its amplitude

The second synthetic model (Fig 6d) shows themagnetic signature of a thin sheet of 4000 times 4000 times 5 m Thepurpose of this model is to test the hypothesis of the observedmagnetic anomaly being caused by a stack of thin meltsheets This hypothesis arose at early stages of interpretationof the LB-07A and LB-08A borehole magnetic data whichshowed high-frequency and high-amplitude magneticanomalies However the borehole data were acquired at asampling rate of 10 cm and spacing of millimeters betweenthe magnetic sources and the sensors therefore the frequencycontent of these anomalies is related to the different samplingrate and closer distance to the magnetic sources and is notcomparable to the observed magnetic anomalies as detectedfrom the lake surface or from an airplane In any case thismodel is intended to verify whether the thin magnetic sheetidea is feasible or not The magnetization parameters used arethe same as for the 3-D model (Table 1) The amplitude of theanomaly created by such a thin sheet is 025 nT at a depth of200 m and 004 nT at 600 m depth Thus to reproduce theobserved plusmn30 nT anomaly we need at least 120 of these thinsheets assuming the best possible scenario of shallowsources at 200 m where the amplitude is the greatest If we

Table 1 Magnetic properties of the impact and target rocks as measured on outcrops around Lake Bosumtwi by Plado et al (2000) and on the LB-07A and LB-08A drill cores (Kontny et al 2007) and magnetization parameters used in the 3-D model

Magnetic susceptibilityPlado et al (2000)

Magnetic susceptibilityThis work

Remanent magnetizationPlado et al (2000)(Am)

Remanent magnetizationThis work(Am)

Rock typelayer Measured Model Measured Model Measured Model Measured Model

Water N A N App N A 0 N A 0 N A 0Sediments N A N App N A 0 N A 0 N A 0Magnetic bodies 000033 00033 00003 00002ndash005 00368 00368

D = 3 I = minus1

00004ndash02 0ndash15D = minus6I = minus12

Target rocks (background) 000015 0 00003 00003 0057 0N A = not available N App = not applicable

874 H Ugalde et al

consider the decreasing amplitude with depth the number ofsheets scales up to at least 200 Therefore the more realisticapproach of stacking the thin sheets one below the otherleaving some space in between them would first requiremore space (gt1000 m taking into account 1000 m for the200 thin sheets plus 60ndash100 m of host rock considering 05ndash1 m between each sheet) and secondly the computedmagnetic response of such a geometry would still be smallerthan the observed field Furthermore this alternate modelwould extend from 200 to 1200 m depth which is notfeasible according to the borehole logs (Coney et al 2007Deutsch et al 2007 Ferriegravere et al 2007) and also wouldmean a block of 4000 times 4000 times 1000 m of highly magneticmaterial which was not encountered by the deep ICDPboreholes

New 3-D Model

A new 3-D model was created from the marine magneticdata (Fig 7) The 3-D structure was built over the ~800 mspaced north-south profiles Each section is 850ndash1000 m wide(depending on the separation of the survey lines) The bodiesthat constitute each section are draped over the craterbathymetry A combination of forward modeling andinversion leads to a model that has an error with an average of084 nT and a standard deviation of 82 nT The basic modelwhich was constructed on the basis of the geometry inheritedby a similar 3-D model built with the gravity data (Ugaldeet al 2007) comprises four layers water sediment impactbreccia (or magnetic rocks underlying the sediments) andmdashfor consistency with Kontny et alrsquos (2007) petrophysical

Fig 5 Previous attempts to model the observed magnetic anomalies at Lake Bosumtwi a) A cross section through the center of the craterfrom Plado et al (2000) b) The Danuor (2004) model across the center of the structure Note the different directions and scales for these twoprofiles

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 875

measurementsmdasha magnetic background Kontny et al (2007)measured high remanent magnetization intensities on samplesfrom both ICDP deep boreholes (LB-07A and LB-08A)which vary depending on the volume percent content ofpyrrhotite However those magnetization variations on thebackground are beyond the resolution of the marine magneticdata set used here

The parameters used in this model calculation aresummarized in Table 1 Starting from the source bodygeometry derived from the gravity study and applyinginversion the magnetic properties of the bodies constitutingeach line were adjusted line by line in order to reproduce theobserved anomalies The inversion was constrained tomagnetic susceptibilities lt005 SI and NRM lt15 Am Whatmakes this model really 3-D is that all the bodies across theentire structure are active when computing a single line thus

unlike 2-D or 25-D models each line is affected by thebodies located laterally to it As seen in Fig 7 this modelallows a non-uniform magnetization distribution throughoutthe structure Magnetic susceptibilities vary from 00002 to005 SI and NRM intensity from 0 to 15 Am parallel to theambient field (D = minus6deg I = minus12deg)

The model complies to the borehole petrophysics withmagnetic susceptibilities and remanence values within therange of what was measured on the recovered core from thedeep ICDP boreholes (Elbra et al 2007 Kontny et al 2007Morris et al 2007) The susceptibility values are uniformlylow across most of the lake with the exception of thenortheast side of it There at a distance of ~1800 m from theICDP boreholes lie 4 bodies that cover an area of ~4000 times2000 m and where susceptibilities are much higher than thosemeasured on the IDCP core (Fig 7) (Figs 3 and 4 of Morris

Fig 6 Synthetic models for a source at a magnetic latitude similar to that of the Lake Bosumtwi impact structure The magnetic source is onemagnetized block with the parameters as given in Plado et al (2000) See text and Table 1 for details a) Plan view of the model (black outline)and its computed magnetic response b) 3-D perspective view of the model showing the synthetic lines every 100 m c) Cross section of themodel along the center of the synthetic structure d) Cross section of the thin sheet model Note the different vertical scales in (c) and (d)

876 H Ugalde et al

Fig 7 Plan view and cross sections of the new 3-D model generated in this work Top left plan view of the magnetized bodies color-codedby magnetic susceptibility Top right computed magnetic response of the 3-D model Middle and bottom rows four north-south cross sectionsof the model from west (line 2) to east (line 5) The bodies are color-coded by magnetic susceptibility The body at the top is water (k = 0)constrained by the lake bathymetry followed by the post-impact sediment layer (k = 0) which is constrained by the gravity model

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 877

et al 2007) NRM intensities are higher than 1 Am acrossmost of the lake However as the direction of remanence isparallel to the ambient field the real effect of increasing theintensity of NRM is to increase the effective susceptibility

(1)

where k is magnetic susceptibility keff is the effectivesusceptibility after adding remanence Q the Koenigsbergerratio and MI and MR are the intensity of the induced andremanent magnetization respectively Consequently thesame effect can be accomplished by reducing the intensity ofNRM and increasing susceptibility or vice-versa butreasonable limits on the amount of each are imposed bymagnetic mineralogy (compare Kontny et al 2007)

Analysis of the amplitude of the analytical signal map(Fig 4) also locates the main source of the observed magneticanomaly at the northeast corner of the lake (A1 A2 of Fig 4)and on a northeast to southwest line to the south of it (B1 andB2 of Fig 4) whichmdashnot surprisinglymdashaligns with thehighest magnetic susceptibilities found by the model

DISCUSSION

We succeeded in finding a model that fits the observedmagnetic anomalies and the petrophysical measurementsfrom the ICDP boreholes LB-07A and LB-08A Howeverthere are still some unresolved problems that we attempt toaddress here Due to the expected relatively highmagnetization of impact-melt rocks (Ugalde et al 2005Henkel and Reimold 2002) the amount of impact meltproduced by the Bosumtwi impact becomes an importantconstraint in understanding the source of the observedmagnetic anomalies Artemieva et al (2004) estimated thevolume of impact melt produced by the impact (projectilediameter between 076ndash104 km for an impact angle between30degndash60deg impacting at 20 kms) at 26ndash43 km3 and athickness of ~200 m for the melt sheet near the location of thetwo ICDP deep boreholes Plado et al (2000) interpreted theobserved magnetic anomalies as an approximately triangularbody of impact melt extending for 4000 m in the north-southdirection of 400 m thickness and of 4000 m width in east-west extent which would give approximately 32 km3 ofimpact melt However the ICDP deep boreholes did not finda considerable amount of impact melt just centimeter-widesuevite dikes in LB-08A and lt50 m of suevite in LB-07A(Coney et al 2007 Ferriegravere et al 2007 Deutsch et al 2007)In addition the melt proportion in suevitic breccias in theseboreholes was shown by these authors to be generally verysmall amounting to just a few volume The gravity modelof Ugalde et al (2007) also supports an impactite unit muchthinner than expected Therefore there are three options tounderstand the source of the magnetic anomalies a) the ICDP

deep boreholes missed the true source of the magneticanomalies and there is a large body of magnetic rock belowthe ICDP boreholes but at a greater depth b) there is a largebody of impact melt in the northeast sector of the impactstructure c) the source of the magnetic anomalies is notimpact melt and another source mechanism has yet to befound

The first option can be ruled out based on the magneticproperties of the impactites measured around the lake byPlado et al (2000) and the magnetic properties of the targetlithologies measured by Kontny et al (2007) Because of theexponential decay of the magnetic anomalies with depth tothe top of the associated magnetic sources (Spector and Grant1970) in order to have a deeper magnetic source below theICDP deep boreholes the magnetic properties of these bodieswould have to be at least ten times those measured around thelake or at the ICDP boreholes That is very unlikelyconsidering the magnetic properties measured in a few meltfragments found in the impact breccia and the targetlithologies at LB-07A and LB-08A (Kontny and Just 2006Kontny et al 2007) Kontny et al (2007) measured an averagesusceptibility of ~300 times 10minus6 SI and Q ~9 for the targetlithologies That would give an effective susceptibility of keff= k(1 + Q) = 3000 times 10minus5 SI for the whole stratigraphiccolumn which is still not sufficient to produce such largeanomalies Also if that were the case the borehole magneticlogs would show much higher amplitude anomalies towardsthe base of the boreholes (Morris et al 2007) Furthermorethe location of the ICDP deep boreholes does not match thelocation of any of the magnetic anomalies as indicated by theamplitude of the analytical signal (Fig 4) nor does thewavelength of the anomalies correspond to what is expectedfor a magnetic source located 500 m below the observationplane (much broader anomalies highly reduced amplitudes)Plado et al (2000) interpreted the magnetic anomalies asimpact melt based on the hypothesis that the magneticproperties of the suevites around the lake were diminished byweathering and alteration processes Thus the measuredvalues were not representative of the real magnetizationparameters of the impact-melt rocks that were assumed to beinside the impact structure Kontny et al (2007) indeedshowed that the rock magnetic parameters for all rocks at theICDP deep boreholes (impact lithologies and basement rocks)are distinctly higher than those reported by Plado et al(2000) however neither ICDP deep borehole intersected anyimpact melt (Coney et al 2007 Ferriegravere et al 2007 Deutschet al 2007) which rules out impact melt as a magnetic sourceFinally based on the location of the magnetic anomalies andtheir wavelength and intensity we cannot locate the source ofthe magnetic anomalies below the ICDP deep boreholes at agreater depth

The second and third options have similar implicationsregarding the location of the magnetic sources but differ withregard to the origin of the anomalies Considering both the

keff k 1 Q+( ) k 1 MR

MI-------------+⎝ ⎠

⎛ ⎞==

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

870 H Ugalde et al

consistent with existing petrophysical and paleomagnetic datasets (Elbra et al 2007 Kontny et al 2007 Schleifer et al2006) as well as the gravity and seismic data

THE BOSUMTWI IMPACT STRUCTURE

Regional Geology

The regional Precambrian geology in and around LakeBosumtwi is dominated by the 22ndash21 Gyr old supracrustalrocks of the Birimian Supergroup comprising mainly meta-graywackes shales and mica schists and occurring in theform of several broad metasedimentary and metavolcanics

belts (Karikari et al 2007) A characteristic feature of thesupracrustal rocks is a northeast-southwest fabric trend withsteep dips either to the northwest or southeast (Reimold et al1998) (Fig 2) These regional belts are easily recognized inaeromagnetic maps (Fig 3b) A variety of granitoid intrusions(biotite or amphibole granites) is also present in theBosumtwi region (Junner 1937 Moon and Mason 1967Koeberl and Reimold 2005) Granite intrusions probablyconnected with the Kumasi batholith outcrop around thenorth northeast east and west sides of the lake (eg thePepiakese granite on the northeast side of the lake [Jones et al1981 Koeberl and Reimold 2005]) Biotite and hornblendegranodiorites (Bansu intrusion) (Koeberl and Reimold 2005)

Fig 3 The total magnetic field at Lake Bosumtwi a) The 1960 airborne survey as published by Jones et al (1981) b) The 1997 airborneresidual anomaly map after Plado et al (2000) Data from Pesonen et al (2003) Contour interval is 2 nT c) The 2001 marine survey Thedata collection lines are noted in black and the location of the 2004 ICDP drilling sites are marked as stars The lake outline is in black Thecontour interval is 2 nT Data from Danuor (2004) d) A south-north profile across the structure (line 4E of [c]) showing both the marine (lakelevel) and airborne (70 m altitude) data sets for comparison Note the greater amplitude of the marine data due to its location closer to themagnetic sources

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 871

outcrop to the south and west of the lake Numerous narrowdikes of biotite granitoid occur at many basement exposuresin the crater rim (Reimold et al 1998)

Recent rock formations include the lake beds and theproducts of weathering (laterites soil) which can havethicknesses up to 10 m Although no impact melt rock hasbeen found around the crater (Jones et al 1981 Reimold et al1998) numerous breccia and suevite exposures have beenmapped (Koeberl and Reimold 2005) In early 1999 ashallow drilling program to the north of the crater rim wasundertaken to determine the extent of the ejecta blanketaround the crater and to obtain subsurface core samples formineralogical petrological and geochemical studies of ejecta(Boamah and Koeberl 2003) A variety of impactite brecciaswere found including impact glassndashrich suevite and severaltypes of lithic breccias (Boamah and Koeberl 2003)

The impact origin of the structure was proven eg byevidence of shock deformation features (eg planardeformation features [PDFs] high pressure mineral phases inimpactites) For a review see Koeberl and Reimold (2005)and references therein

PREVIOUS GEOPHYSICAL STUDIES

Magnetic Data

The first airborne magnetic study of the structure wasmade in 1960 by Hunting Surveys Ltd (Jones et al 1981) at aflight altitude of 300 m and with a line spacing of 500 m Theoccurrence of a central negative magnetic anomaly of ~40 nTwith positive side (flank) anomalies of ~20 nT was reported(Fig 3a) The anomaly was modeled with a 200 m thickbreccia lens below the lake sediments occurring at a depth of500 m

In 1997 a high-resolution low-altitude (70 m) airbornegeophysical survey across the Bosumtwi structure was carriedout by the Geological Survey of Finland in cooperation withthe University of Vienna and the Ghana Geological SurveyDepartment (Pesonen et al 1997 Pesonen et al 2003) Itincluded measurements of the total magnetic fieldelectromagnetic field and gamma radiation data Linespacing was 500 m and the sampling rate 75 m (magnetics)15 m (EM) and 60 m (radiometric) respectively along north-south flight lines (Pesonen et al 2003) A series of 21geophysical maps (magnetic electromagnetic radiometric)of the structure were compiled by these authors

After eliminating the regional magnetic field Plado et al(2000) published a residual magnetic anomaly map of theBosumtwi structure which is strikingly similar to the onepublished by Jones et al (1981) but much more detailed(Fig 3b) The central part of the structure is characterized byrelatively high gradients At the center of the lake there is onelarge negative magnetic anomaly with two positive sideanomalies to the north and south of it consistent with the

expected expression of a magnetic source at near-equatorialmagnetic latitudes (see Fig 5 of Pesonen et al 2003)

Investigation of the magnetic anomaly pattern (Fig 3b)clearly illustrates the idea that the impact occurred into asouthwest-northeast trending metasedimentary-metavolcanicbelt This belt is not as strongly visible at the lake as is thesimilarly trending (southwest-northeast) belt to the southeastside of the structure which is superimposed by the impactstructure at its southwest edge There is a weak ring-shapedmagnetic halo around the lake The central negative anomalyis oblique to the overall southwest-northeast trendingmagnetic lineation and it is seemingly composed of 4 to 6patches It is flanked by northerly and somewhat weakersoutherly positive anomalies A weakly positive anomaly spotalmost at the center of the lake was interpreted by Plado et al(2000) as a possible central uplift

In 2001 the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi Ghana) collected a marinemagnetic data set to complement the existing airborne data(Danuor 2004) This new data set had a wider spacing of800 m and the magnetometer was towed behind the marineplatform used for that purpose Albeit having a wider linespacing than the airborne data set the marine data had theadvantage that it measured the magnetic field closer to themagnetic sources Also because of the slower movement ofthe platform the sampling rate was improved The smallerdistance to the sources and enhanced sampling rate allows abetter representation of the magnetic anomalies which cannot be achieved by pure downward analytical continuationfrom the airplane data collection surface to the lake levelFigure 3d shows a south-north profile across the main part ofthe magnetic anomaly with both the marine and airbornemagnetic data sets The greater amplitude of the marine dataset can be observed In order to isolate the main magneticanomalies and their probable causative bodies we computedthe amplitude of the analytical signal which is computedas (Nabighian 1972) and gives maxima over the edge of themagnetized bodies causing the observed anomalies with theadvantage of being independent of remanent magnetization(Nabighian 1972 Roest et al 1992 MacLeod et al 1993)The NE of the lake exhibits a couple of high-amplitude mid-to long-wavelength anomalies (A1 B1 of Fig 4) There is alarge anomaly that encircles the northern part of the lakewhich previously had been tentatively interpreted as a largepackage of highly magnetic melt-rich rocks However theanalytical signal map of Fig 4 permits the separation of theanomaly into at least three main high-amplitude anomalies ofabout 1000ndash2000 m wavelength (see Fig 4 anomalies A1A2 A3) plus some highly magnetized material to its sidesThe rest of the lake does not exhibit any other significantanomalies except for the previously noted B1 and anomaliesB3 and B2 (composed of two smaller magnetic sources) to thesouth of it B1 and B2 are aligned in a SW-NE trend parallel

ASIG X Y( ) partF partXfrasl( )2 partF partYfrasl( )2 partF partZfrasl( )2+ +=

872 H Ugalde et al

to the dominant strike of the metasedimentary andmetavolcanic belts observed in the region and anomalies C1and C2 at the western side of the lake both have mid- tohigh- amplitude and 700ndash900 m wavelength The smalleramplitude of C1 and C2 denotes probable deeper settings oftheir corresponding magnetic sources or poorer magneticproperties than those relating to the anomalies A and B

Magnetic Properties of the Rocks

The magnetic properties of the rocks found in and aroundthe Bosumtwi structure are of key importance in themodeling The data used by Plado et al (2000) aresummarized in Table 1 together with data used in this work

Plado et al (2000) demonstrated that there is a cleardifference between petrophysical properties of the targetrocks and the impact-derived suevites The fallout sueviteshave lower densities (higher porosities) and somewhat highermagnetizations (susceptibility remanent magnetization)compared to the target rocks The remanent magnetization ofsuevites prevails over induced magnetization (theKoenigsberger ratio of remanent over induced magnetizationis much higher than 1 ~4) The target rockshave strikingly homogeneous physical properties withnoticeable weak remanent magnetization and Q values

(Q lt005) A graphitic shale had fairly high remanentmagnetization (natural remanent magnetization [NRM] =386 Am Q = 067) whereas the granites (such as thePepiakese body) were magnetically very weak (NRM lt02 Am Q lt016) Unfortunately there are no measurementsof magnetic properties for the Kumasi batholith to the north ofthe lake and the Bansu intrusion to the south although both ofthem exhibit a direct spatial correlation with the airbornemagnetic anomalies

Petrophysical measurements were accomplished on thecore recovered from both holes LB-07A and LB-08A at thefacilities of the GeoForschungsZentrum (GFZ) in Potsdam in2004 (Milkereit et al 2006 Ugalde 2006 Morris et al 2007)See Morris et al (2007) for a detailed description of themagnetic susceptibility logging

NRM was measured in a few selected samples of thecores (Elbra et al 2007 Kontny et al 2007) In general NRMintensities on impact breccias vary between 04 and 20 Am(Kontny et al 2007) Although these values are higher thanthe NRM intensities reported for the target rocks sampledaround the lake (01ndash39 mAm) (Plado et al 2000) they werederived from small clasts or pieces of melt and therefore donot constitute a representative high-magnetization unit thatcould account for the large magnetic anomalies observed inthe lake area

Fig 4 Amplitude of the analytical signal (ASIG) computed from the marine magnetic data set The two ICDP deep boreholes are shown asstars The main anomalies are labeled See text for details

Q MR= MIfrasl

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 873

Previous Magnetic Models

Based on aeromagnetic data and petrophysical andpaleomagnetic data of oriented samples of impactites andtarget rocks outcropping around the lake Plado et al (2000)published a 3-D magnetic model of the structure The modelconsisted of polygonal bodies at depth from 200 to 600 mdiminishing downwards from their surface cross section Themodel of Plado et al (2000) is shown in Fig 5a

However in order to obtain a meaningful match of themodel with the observed magnetic data Plado et al (2000)had to assign much higher values to the magnetic properties tothe body than what was actually measured on rocks (seeTable 1) In the model water sediment and basement wereall treated as the same background unit with zeromagnetization The magnetic body constituting the source ofthe magnetic anomaly (the hypothesized impact melt) wasgiven uniform magnetic properties (susceptibility =00033 SI NRM = 0367 Amminus1 and Q ~43) The remanentmagnetization direction was set to D = 3 I = minus1 as measuredon suevites outcropping around the lake Plado et al (2000)discussed in detail the various possibilities for why themagnetic parameters within the structure could differ fromthose outside

Danuor obtained a 25-D model for one profile across thecenter of the structure (Danuor 2004) Similar to Plado et al(2000) in order to reproduce the observed anomaly he used ahigh magnetic susceptibility of 003 SI but assumed onlyinduced magnetization (no NRM) The proposed source bodyextended from about 250 m to about 610 m depth (Danuor2004) Danuorrsquos model is shown in Fig 5b

NEW MAGNETIC MODELING

Synthetic Modeling

Due to the dipolar nature of the Earthrsquos magnetic fieldobserved magnetic anomalies at low magnetic latitudes canbe rather complex and thus to find the proper location of themagnetic sources relative to the anomalies can be difficult Tobetter understand the expected pattern of magnetic anomalies

at low magnetic latitudes as at Lake Bosumtwi we firstperformed some synthetic modeling

The first model (Figs 6andashc) shows the magneticsignature of a magnetized prismatic body of 4000 times 4000 times400 m The size of the body is the same as that assumed byPlado et al (2000) This proves that a highly magneticanomalous source body will give rise to a negative magneticanomaly at the observed latitude (65degN) of Lake BosumtwiThe negative anomaly appears slightly to the south of thecenter of the body and is accompanied by positive anomalieson the northern and southern sides of the major anomaly Ofthese side anomalies the northern one has slightly higheramplitude than the southern one From this model it is alsoclear that the measured magnetization parameters of the deepICDP boreholes LB-07A and LB-08A are insufficient togenerate the observed magnetic anomaly Adding remanentmagnetization parallel to the ambient field has the same effectas increasing the magnetic susceptibility the shape of theanomaly does not change only its amplitude

The second synthetic model (Fig 6d) shows themagnetic signature of a thin sheet of 4000 times 4000 times 5 m Thepurpose of this model is to test the hypothesis of the observedmagnetic anomaly being caused by a stack of thin meltsheets This hypothesis arose at early stages of interpretationof the LB-07A and LB-08A borehole magnetic data whichshowed high-frequency and high-amplitude magneticanomalies However the borehole data were acquired at asampling rate of 10 cm and spacing of millimeters betweenthe magnetic sources and the sensors therefore the frequencycontent of these anomalies is related to the different samplingrate and closer distance to the magnetic sources and is notcomparable to the observed magnetic anomalies as detectedfrom the lake surface or from an airplane In any case thismodel is intended to verify whether the thin magnetic sheetidea is feasible or not The magnetization parameters used arethe same as for the 3-D model (Table 1) The amplitude of theanomaly created by such a thin sheet is 025 nT at a depth of200 m and 004 nT at 600 m depth Thus to reproduce theobserved plusmn30 nT anomaly we need at least 120 of these thinsheets assuming the best possible scenario of shallowsources at 200 m where the amplitude is the greatest If we

Table 1 Magnetic properties of the impact and target rocks as measured on outcrops around Lake Bosumtwi by Plado et al (2000) and on the LB-07A and LB-08A drill cores (Kontny et al 2007) and magnetization parameters used in the 3-D model

Magnetic susceptibilityPlado et al (2000)

Magnetic susceptibilityThis work

Remanent magnetizationPlado et al (2000)(Am)

Remanent magnetizationThis work(Am)

Rock typelayer Measured Model Measured Model Measured Model Measured Model

Water N A N App N A 0 N A 0 N A 0Sediments N A N App N A 0 N A 0 N A 0Magnetic bodies 000033 00033 00003 00002ndash005 00368 00368

D = 3 I = minus1

00004ndash02 0ndash15D = minus6I = minus12

Target rocks (background) 000015 0 00003 00003 0057 0N A = not available N App = not applicable

874 H Ugalde et al

consider the decreasing amplitude with depth the number ofsheets scales up to at least 200 Therefore the more realisticapproach of stacking the thin sheets one below the otherleaving some space in between them would first requiremore space (gt1000 m taking into account 1000 m for the200 thin sheets plus 60ndash100 m of host rock considering 05ndash1 m between each sheet) and secondly the computedmagnetic response of such a geometry would still be smallerthan the observed field Furthermore this alternate modelwould extend from 200 to 1200 m depth which is notfeasible according to the borehole logs (Coney et al 2007Deutsch et al 2007 Ferriegravere et al 2007) and also wouldmean a block of 4000 times 4000 times 1000 m of highly magneticmaterial which was not encountered by the deep ICDPboreholes

New 3-D Model

A new 3-D model was created from the marine magneticdata (Fig 7) The 3-D structure was built over the ~800 mspaced north-south profiles Each section is 850ndash1000 m wide(depending on the separation of the survey lines) The bodiesthat constitute each section are draped over the craterbathymetry A combination of forward modeling andinversion leads to a model that has an error with an average of084 nT and a standard deviation of 82 nT The basic modelwhich was constructed on the basis of the geometry inheritedby a similar 3-D model built with the gravity data (Ugaldeet al 2007) comprises four layers water sediment impactbreccia (or magnetic rocks underlying the sediments) andmdashfor consistency with Kontny et alrsquos (2007) petrophysical

Fig 5 Previous attempts to model the observed magnetic anomalies at Lake Bosumtwi a) A cross section through the center of the craterfrom Plado et al (2000) b) The Danuor (2004) model across the center of the structure Note the different directions and scales for these twoprofiles

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 875

measurementsmdasha magnetic background Kontny et al (2007)measured high remanent magnetization intensities on samplesfrom both ICDP deep boreholes (LB-07A and LB-08A)which vary depending on the volume percent content ofpyrrhotite However those magnetization variations on thebackground are beyond the resolution of the marine magneticdata set used here

The parameters used in this model calculation aresummarized in Table 1 Starting from the source bodygeometry derived from the gravity study and applyinginversion the magnetic properties of the bodies constitutingeach line were adjusted line by line in order to reproduce theobserved anomalies The inversion was constrained tomagnetic susceptibilities lt005 SI and NRM lt15 Am Whatmakes this model really 3-D is that all the bodies across theentire structure are active when computing a single line thus

unlike 2-D or 25-D models each line is affected by thebodies located laterally to it As seen in Fig 7 this modelallows a non-uniform magnetization distribution throughoutthe structure Magnetic susceptibilities vary from 00002 to005 SI and NRM intensity from 0 to 15 Am parallel to theambient field (D = minus6deg I = minus12deg)

The model complies to the borehole petrophysics withmagnetic susceptibilities and remanence values within therange of what was measured on the recovered core from thedeep ICDP boreholes (Elbra et al 2007 Kontny et al 2007Morris et al 2007) The susceptibility values are uniformlylow across most of the lake with the exception of thenortheast side of it There at a distance of ~1800 m from theICDP boreholes lie 4 bodies that cover an area of ~4000 times2000 m and where susceptibilities are much higher than thosemeasured on the IDCP core (Fig 7) (Figs 3 and 4 of Morris

Fig 6 Synthetic models for a source at a magnetic latitude similar to that of the Lake Bosumtwi impact structure The magnetic source is onemagnetized block with the parameters as given in Plado et al (2000) See text and Table 1 for details a) Plan view of the model (black outline)and its computed magnetic response b) 3-D perspective view of the model showing the synthetic lines every 100 m c) Cross section of themodel along the center of the synthetic structure d) Cross section of the thin sheet model Note the different vertical scales in (c) and (d)

876 H Ugalde et al

Fig 7 Plan view and cross sections of the new 3-D model generated in this work Top left plan view of the magnetized bodies color-codedby magnetic susceptibility Top right computed magnetic response of the 3-D model Middle and bottom rows four north-south cross sectionsof the model from west (line 2) to east (line 5) The bodies are color-coded by magnetic susceptibility The body at the top is water (k = 0)constrained by the lake bathymetry followed by the post-impact sediment layer (k = 0) which is constrained by the gravity model

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 877

et al 2007) NRM intensities are higher than 1 Am acrossmost of the lake However as the direction of remanence isparallel to the ambient field the real effect of increasing theintensity of NRM is to increase the effective susceptibility

(1)

where k is magnetic susceptibility keff is the effectivesusceptibility after adding remanence Q the Koenigsbergerratio and MI and MR are the intensity of the induced andremanent magnetization respectively Consequently thesame effect can be accomplished by reducing the intensity ofNRM and increasing susceptibility or vice-versa butreasonable limits on the amount of each are imposed bymagnetic mineralogy (compare Kontny et al 2007)

Analysis of the amplitude of the analytical signal map(Fig 4) also locates the main source of the observed magneticanomaly at the northeast corner of the lake (A1 A2 of Fig 4)and on a northeast to southwest line to the south of it (B1 andB2 of Fig 4) whichmdashnot surprisinglymdashaligns with thehighest magnetic susceptibilities found by the model

DISCUSSION

We succeeded in finding a model that fits the observedmagnetic anomalies and the petrophysical measurementsfrom the ICDP boreholes LB-07A and LB-08A Howeverthere are still some unresolved problems that we attempt toaddress here Due to the expected relatively highmagnetization of impact-melt rocks (Ugalde et al 2005Henkel and Reimold 2002) the amount of impact meltproduced by the Bosumtwi impact becomes an importantconstraint in understanding the source of the observedmagnetic anomalies Artemieva et al (2004) estimated thevolume of impact melt produced by the impact (projectilediameter between 076ndash104 km for an impact angle between30degndash60deg impacting at 20 kms) at 26ndash43 km3 and athickness of ~200 m for the melt sheet near the location of thetwo ICDP deep boreholes Plado et al (2000) interpreted theobserved magnetic anomalies as an approximately triangularbody of impact melt extending for 4000 m in the north-southdirection of 400 m thickness and of 4000 m width in east-west extent which would give approximately 32 km3 ofimpact melt However the ICDP deep boreholes did not finda considerable amount of impact melt just centimeter-widesuevite dikes in LB-08A and lt50 m of suevite in LB-07A(Coney et al 2007 Ferriegravere et al 2007 Deutsch et al 2007)In addition the melt proportion in suevitic breccias in theseboreholes was shown by these authors to be generally verysmall amounting to just a few volume The gravity modelof Ugalde et al (2007) also supports an impactite unit muchthinner than expected Therefore there are three options tounderstand the source of the magnetic anomalies a) the ICDP

deep boreholes missed the true source of the magneticanomalies and there is a large body of magnetic rock belowthe ICDP boreholes but at a greater depth b) there is a largebody of impact melt in the northeast sector of the impactstructure c) the source of the magnetic anomalies is notimpact melt and another source mechanism has yet to befound

The first option can be ruled out based on the magneticproperties of the impactites measured around the lake byPlado et al (2000) and the magnetic properties of the targetlithologies measured by Kontny et al (2007) Because of theexponential decay of the magnetic anomalies with depth tothe top of the associated magnetic sources (Spector and Grant1970) in order to have a deeper magnetic source below theICDP deep boreholes the magnetic properties of these bodieswould have to be at least ten times those measured around thelake or at the ICDP boreholes That is very unlikelyconsidering the magnetic properties measured in a few meltfragments found in the impact breccia and the targetlithologies at LB-07A and LB-08A (Kontny and Just 2006Kontny et al 2007) Kontny et al (2007) measured an averagesusceptibility of ~300 times 10minus6 SI and Q ~9 for the targetlithologies That would give an effective susceptibility of keff= k(1 + Q) = 3000 times 10minus5 SI for the whole stratigraphiccolumn which is still not sufficient to produce such largeanomalies Also if that were the case the borehole magneticlogs would show much higher amplitude anomalies towardsthe base of the boreholes (Morris et al 2007) Furthermorethe location of the ICDP deep boreholes does not match thelocation of any of the magnetic anomalies as indicated by theamplitude of the analytical signal (Fig 4) nor does thewavelength of the anomalies correspond to what is expectedfor a magnetic source located 500 m below the observationplane (much broader anomalies highly reduced amplitudes)Plado et al (2000) interpreted the magnetic anomalies asimpact melt based on the hypothesis that the magneticproperties of the suevites around the lake were diminished byweathering and alteration processes Thus the measuredvalues were not representative of the real magnetizationparameters of the impact-melt rocks that were assumed to beinside the impact structure Kontny et al (2007) indeedshowed that the rock magnetic parameters for all rocks at theICDP deep boreholes (impact lithologies and basement rocks)are distinctly higher than those reported by Plado et al(2000) however neither ICDP deep borehole intersected anyimpact melt (Coney et al 2007 Ferriegravere et al 2007 Deutschet al 2007) which rules out impact melt as a magnetic sourceFinally based on the location of the magnetic anomalies andtheir wavelength and intensity we cannot locate the source ofthe magnetic anomalies below the ICDP deep boreholes at agreater depth

The second and third options have similar implicationsregarding the location of the magnetic sources but differ withregard to the origin of the anomalies Considering both the

keff k 1 Q+( ) k 1 MR

MI-------------+⎝ ⎠

⎛ ⎞==

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 871

outcrop to the south and west of the lake Numerous narrowdikes of biotite granitoid occur at many basement exposuresin the crater rim (Reimold et al 1998)

Recent rock formations include the lake beds and theproducts of weathering (laterites soil) which can havethicknesses up to 10 m Although no impact melt rock hasbeen found around the crater (Jones et al 1981 Reimold et al1998) numerous breccia and suevite exposures have beenmapped (Koeberl and Reimold 2005) In early 1999 ashallow drilling program to the north of the crater rim wasundertaken to determine the extent of the ejecta blanketaround the crater and to obtain subsurface core samples formineralogical petrological and geochemical studies of ejecta(Boamah and Koeberl 2003) A variety of impactite brecciaswere found including impact glassndashrich suevite and severaltypes of lithic breccias (Boamah and Koeberl 2003)

The impact origin of the structure was proven eg byevidence of shock deformation features (eg planardeformation features [PDFs] high pressure mineral phases inimpactites) For a review see Koeberl and Reimold (2005)and references therein

PREVIOUS GEOPHYSICAL STUDIES

Magnetic Data

The first airborne magnetic study of the structure wasmade in 1960 by Hunting Surveys Ltd (Jones et al 1981) at aflight altitude of 300 m and with a line spacing of 500 m Theoccurrence of a central negative magnetic anomaly of ~40 nTwith positive side (flank) anomalies of ~20 nT was reported(Fig 3a) The anomaly was modeled with a 200 m thickbreccia lens below the lake sediments occurring at a depth of500 m

In 1997 a high-resolution low-altitude (70 m) airbornegeophysical survey across the Bosumtwi structure was carriedout by the Geological Survey of Finland in cooperation withthe University of Vienna and the Ghana Geological SurveyDepartment (Pesonen et al 1997 Pesonen et al 2003) Itincluded measurements of the total magnetic fieldelectromagnetic field and gamma radiation data Linespacing was 500 m and the sampling rate 75 m (magnetics)15 m (EM) and 60 m (radiometric) respectively along north-south flight lines (Pesonen et al 2003) A series of 21geophysical maps (magnetic electromagnetic radiometric)of the structure were compiled by these authors

After eliminating the regional magnetic field Plado et al(2000) published a residual magnetic anomaly map of theBosumtwi structure which is strikingly similar to the onepublished by Jones et al (1981) but much more detailed(Fig 3b) The central part of the structure is characterized byrelatively high gradients At the center of the lake there is onelarge negative magnetic anomaly with two positive sideanomalies to the north and south of it consistent with the

expected expression of a magnetic source at near-equatorialmagnetic latitudes (see Fig 5 of Pesonen et al 2003)

Investigation of the magnetic anomaly pattern (Fig 3b)clearly illustrates the idea that the impact occurred into asouthwest-northeast trending metasedimentary-metavolcanicbelt This belt is not as strongly visible at the lake as is thesimilarly trending (southwest-northeast) belt to the southeastside of the structure which is superimposed by the impactstructure at its southwest edge There is a weak ring-shapedmagnetic halo around the lake The central negative anomalyis oblique to the overall southwest-northeast trendingmagnetic lineation and it is seemingly composed of 4 to 6patches It is flanked by northerly and somewhat weakersoutherly positive anomalies A weakly positive anomaly spotalmost at the center of the lake was interpreted by Plado et al(2000) as a possible central uplift

In 2001 the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi Ghana) collected a marinemagnetic data set to complement the existing airborne data(Danuor 2004) This new data set had a wider spacing of800 m and the magnetometer was towed behind the marineplatform used for that purpose Albeit having a wider linespacing than the airborne data set the marine data had theadvantage that it measured the magnetic field closer to themagnetic sources Also because of the slower movement ofthe platform the sampling rate was improved The smallerdistance to the sources and enhanced sampling rate allows abetter representation of the magnetic anomalies which cannot be achieved by pure downward analytical continuationfrom the airplane data collection surface to the lake levelFigure 3d shows a south-north profile across the main part ofthe magnetic anomaly with both the marine and airbornemagnetic data sets The greater amplitude of the marine dataset can be observed In order to isolate the main magneticanomalies and their probable causative bodies we computedthe amplitude of the analytical signal which is computedas (Nabighian 1972) and gives maxima over the edge of themagnetized bodies causing the observed anomalies with theadvantage of being independent of remanent magnetization(Nabighian 1972 Roest et al 1992 MacLeod et al 1993)The NE of the lake exhibits a couple of high-amplitude mid-to long-wavelength anomalies (A1 B1 of Fig 4) There is alarge anomaly that encircles the northern part of the lakewhich previously had been tentatively interpreted as a largepackage of highly magnetic melt-rich rocks However theanalytical signal map of Fig 4 permits the separation of theanomaly into at least three main high-amplitude anomalies ofabout 1000ndash2000 m wavelength (see Fig 4 anomalies A1A2 A3) plus some highly magnetized material to its sidesThe rest of the lake does not exhibit any other significantanomalies except for the previously noted B1 and anomaliesB3 and B2 (composed of two smaller magnetic sources) to thesouth of it B1 and B2 are aligned in a SW-NE trend parallel

ASIG X Y( ) partF partXfrasl( )2 partF partYfrasl( )2 partF partZfrasl( )2+ +=

872 H Ugalde et al

to the dominant strike of the metasedimentary andmetavolcanic belts observed in the region and anomalies C1and C2 at the western side of the lake both have mid- tohigh- amplitude and 700ndash900 m wavelength The smalleramplitude of C1 and C2 denotes probable deeper settings oftheir corresponding magnetic sources or poorer magneticproperties than those relating to the anomalies A and B

Magnetic Properties of the Rocks

The magnetic properties of the rocks found in and aroundthe Bosumtwi structure are of key importance in themodeling The data used by Plado et al (2000) aresummarized in Table 1 together with data used in this work

Plado et al (2000) demonstrated that there is a cleardifference between petrophysical properties of the targetrocks and the impact-derived suevites The fallout sueviteshave lower densities (higher porosities) and somewhat highermagnetizations (susceptibility remanent magnetization)compared to the target rocks The remanent magnetization ofsuevites prevails over induced magnetization (theKoenigsberger ratio of remanent over induced magnetizationis much higher than 1 ~4) The target rockshave strikingly homogeneous physical properties withnoticeable weak remanent magnetization and Q values

(Q lt005) A graphitic shale had fairly high remanentmagnetization (natural remanent magnetization [NRM] =386 Am Q = 067) whereas the granites (such as thePepiakese body) were magnetically very weak (NRM lt02 Am Q lt016) Unfortunately there are no measurementsof magnetic properties for the Kumasi batholith to the north ofthe lake and the Bansu intrusion to the south although both ofthem exhibit a direct spatial correlation with the airbornemagnetic anomalies

Petrophysical measurements were accomplished on thecore recovered from both holes LB-07A and LB-08A at thefacilities of the GeoForschungsZentrum (GFZ) in Potsdam in2004 (Milkereit et al 2006 Ugalde 2006 Morris et al 2007)See Morris et al (2007) for a detailed description of themagnetic susceptibility logging

NRM was measured in a few selected samples of thecores (Elbra et al 2007 Kontny et al 2007) In general NRMintensities on impact breccias vary between 04 and 20 Am(Kontny et al 2007) Although these values are higher thanthe NRM intensities reported for the target rocks sampledaround the lake (01ndash39 mAm) (Plado et al 2000) they werederived from small clasts or pieces of melt and therefore donot constitute a representative high-magnetization unit thatcould account for the large magnetic anomalies observed inthe lake area

Fig 4 Amplitude of the analytical signal (ASIG) computed from the marine magnetic data set The two ICDP deep boreholes are shown asstars The main anomalies are labeled See text for details

Q MR= MIfrasl

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 873

Previous Magnetic Models

Based on aeromagnetic data and petrophysical andpaleomagnetic data of oriented samples of impactites andtarget rocks outcropping around the lake Plado et al (2000)published a 3-D magnetic model of the structure The modelconsisted of polygonal bodies at depth from 200 to 600 mdiminishing downwards from their surface cross section Themodel of Plado et al (2000) is shown in Fig 5a

However in order to obtain a meaningful match of themodel with the observed magnetic data Plado et al (2000)had to assign much higher values to the magnetic properties tothe body than what was actually measured on rocks (seeTable 1) In the model water sediment and basement wereall treated as the same background unit with zeromagnetization The magnetic body constituting the source ofthe magnetic anomaly (the hypothesized impact melt) wasgiven uniform magnetic properties (susceptibility =00033 SI NRM = 0367 Amminus1 and Q ~43) The remanentmagnetization direction was set to D = 3 I = minus1 as measuredon suevites outcropping around the lake Plado et al (2000)discussed in detail the various possibilities for why themagnetic parameters within the structure could differ fromthose outside

Danuor obtained a 25-D model for one profile across thecenter of the structure (Danuor 2004) Similar to Plado et al(2000) in order to reproduce the observed anomaly he used ahigh magnetic susceptibility of 003 SI but assumed onlyinduced magnetization (no NRM) The proposed source bodyextended from about 250 m to about 610 m depth (Danuor2004) Danuorrsquos model is shown in Fig 5b

NEW MAGNETIC MODELING

Synthetic Modeling

Due to the dipolar nature of the Earthrsquos magnetic fieldobserved magnetic anomalies at low magnetic latitudes canbe rather complex and thus to find the proper location of themagnetic sources relative to the anomalies can be difficult Tobetter understand the expected pattern of magnetic anomalies

at low magnetic latitudes as at Lake Bosumtwi we firstperformed some synthetic modeling

The first model (Figs 6andashc) shows the magneticsignature of a magnetized prismatic body of 4000 times 4000 times400 m The size of the body is the same as that assumed byPlado et al (2000) This proves that a highly magneticanomalous source body will give rise to a negative magneticanomaly at the observed latitude (65degN) of Lake BosumtwiThe negative anomaly appears slightly to the south of thecenter of the body and is accompanied by positive anomalieson the northern and southern sides of the major anomaly Ofthese side anomalies the northern one has slightly higheramplitude than the southern one From this model it is alsoclear that the measured magnetization parameters of the deepICDP boreholes LB-07A and LB-08A are insufficient togenerate the observed magnetic anomaly Adding remanentmagnetization parallel to the ambient field has the same effectas increasing the magnetic susceptibility the shape of theanomaly does not change only its amplitude

The second synthetic model (Fig 6d) shows themagnetic signature of a thin sheet of 4000 times 4000 times 5 m Thepurpose of this model is to test the hypothesis of the observedmagnetic anomaly being caused by a stack of thin meltsheets This hypothesis arose at early stages of interpretationof the LB-07A and LB-08A borehole magnetic data whichshowed high-frequency and high-amplitude magneticanomalies However the borehole data were acquired at asampling rate of 10 cm and spacing of millimeters betweenthe magnetic sources and the sensors therefore the frequencycontent of these anomalies is related to the different samplingrate and closer distance to the magnetic sources and is notcomparable to the observed magnetic anomalies as detectedfrom the lake surface or from an airplane In any case thismodel is intended to verify whether the thin magnetic sheetidea is feasible or not The magnetization parameters used arethe same as for the 3-D model (Table 1) The amplitude of theanomaly created by such a thin sheet is 025 nT at a depth of200 m and 004 nT at 600 m depth Thus to reproduce theobserved plusmn30 nT anomaly we need at least 120 of these thinsheets assuming the best possible scenario of shallowsources at 200 m where the amplitude is the greatest If we

Table 1 Magnetic properties of the impact and target rocks as measured on outcrops around Lake Bosumtwi by Plado et al (2000) and on the LB-07A and LB-08A drill cores (Kontny et al 2007) and magnetization parameters used in the 3-D model

Magnetic susceptibilityPlado et al (2000)

Magnetic susceptibilityThis work

Remanent magnetizationPlado et al (2000)(Am)

Remanent magnetizationThis work(Am)

Rock typelayer Measured Model Measured Model Measured Model Measured Model

Water N A N App N A 0 N A 0 N A 0Sediments N A N App N A 0 N A 0 N A 0Magnetic bodies 000033 00033 00003 00002ndash005 00368 00368

D = 3 I = minus1

00004ndash02 0ndash15D = minus6I = minus12

Target rocks (background) 000015 0 00003 00003 0057 0N A = not available N App = not applicable

874 H Ugalde et al

consider the decreasing amplitude with depth the number ofsheets scales up to at least 200 Therefore the more realisticapproach of stacking the thin sheets one below the otherleaving some space in between them would first requiremore space (gt1000 m taking into account 1000 m for the200 thin sheets plus 60ndash100 m of host rock considering 05ndash1 m between each sheet) and secondly the computedmagnetic response of such a geometry would still be smallerthan the observed field Furthermore this alternate modelwould extend from 200 to 1200 m depth which is notfeasible according to the borehole logs (Coney et al 2007Deutsch et al 2007 Ferriegravere et al 2007) and also wouldmean a block of 4000 times 4000 times 1000 m of highly magneticmaterial which was not encountered by the deep ICDPboreholes

New 3-D Model

A new 3-D model was created from the marine magneticdata (Fig 7) The 3-D structure was built over the ~800 mspaced north-south profiles Each section is 850ndash1000 m wide(depending on the separation of the survey lines) The bodiesthat constitute each section are draped over the craterbathymetry A combination of forward modeling andinversion leads to a model that has an error with an average of084 nT and a standard deviation of 82 nT The basic modelwhich was constructed on the basis of the geometry inheritedby a similar 3-D model built with the gravity data (Ugaldeet al 2007) comprises four layers water sediment impactbreccia (or magnetic rocks underlying the sediments) andmdashfor consistency with Kontny et alrsquos (2007) petrophysical

Fig 5 Previous attempts to model the observed magnetic anomalies at Lake Bosumtwi a) A cross section through the center of the craterfrom Plado et al (2000) b) The Danuor (2004) model across the center of the structure Note the different directions and scales for these twoprofiles

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 875

measurementsmdasha magnetic background Kontny et al (2007)measured high remanent magnetization intensities on samplesfrom both ICDP deep boreholes (LB-07A and LB-08A)which vary depending on the volume percent content ofpyrrhotite However those magnetization variations on thebackground are beyond the resolution of the marine magneticdata set used here

The parameters used in this model calculation aresummarized in Table 1 Starting from the source bodygeometry derived from the gravity study and applyinginversion the magnetic properties of the bodies constitutingeach line were adjusted line by line in order to reproduce theobserved anomalies The inversion was constrained tomagnetic susceptibilities lt005 SI and NRM lt15 Am Whatmakes this model really 3-D is that all the bodies across theentire structure are active when computing a single line thus

unlike 2-D or 25-D models each line is affected by thebodies located laterally to it As seen in Fig 7 this modelallows a non-uniform magnetization distribution throughoutthe structure Magnetic susceptibilities vary from 00002 to005 SI and NRM intensity from 0 to 15 Am parallel to theambient field (D = minus6deg I = minus12deg)

The model complies to the borehole petrophysics withmagnetic susceptibilities and remanence values within therange of what was measured on the recovered core from thedeep ICDP boreholes (Elbra et al 2007 Kontny et al 2007Morris et al 2007) The susceptibility values are uniformlylow across most of the lake with the exception of thenortheast side of it There at a distance of ~1800 m from theICDP boreholes lie 4 bodies that cover an area of ~4000 times2000 m and where susceptibilities are much higher than thosemeasured on the IDCP core (Fig 7) (Figs 3 and 4 of Morris

Fig 6 Synthetic models for a source at a magnetic latitude similar to that of the Lake Bosumtwi impact structure The magnetic source is onemagnetized block with the parameters as given in Plado et al (2000) See text and Table 1 for details a) Plan view of the model (black outline)and its computed magnetic response b) 3-D perspective view of the model showing the synthetic lines every 100 m c) Cross section of themodel along the center of the synthetic structure d) Cross section of the thin sheet model Note the different vertical scales in (c) and (d)

876 H Ugalde et al

Fig 7 Plan view and cross sections of the new 3-D model generated in this work Top left plan view of the magnetized bodies color-codedby magnetic susceptibility Top right computed magnetic response of the 3-D model Middle and bottom rows four north-south cross sectionsof the model from west (line 2) to east (line 5) The bodies are color-coded by magnetic susceptibility The body at the top is water (k = 0)constrained by the lake bathymetry followed by the post-impact sediment layer (k = 0) which is constrained by the gravity model

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 877

et al 2007) NRM intensities are higher than 1 Am acrossmost of the lake However as the direction of remanence isparallel to the ambient field the real effect of increasing theintensity of NRM is to increase the effective susceptibility

(1)

where k is magnetic susceptibility keff is the effectivesusceptibility after adding remanence Q the Koenigsbergerratio and MI and MR are the intensity of the induced andremanent magnetization respectively Consequently thesame effect can be accomplished by reducing the intensity ofNRM and increasing susceptibility or vice-versa butreasonable limits on the amount of each are imposed bymagnetic mineralogy (compare Kontny et al 2007)

Analysis of the amplitude of the analytical signal map(Fig 4) also locates the main source of the observed magneticanomaly at the northeast corner of the lake (A1 A2 of Fig 4)and on a northeast to southwest line to the south of it (B1 andB2 of Fig 4) whichmdashnot surprisinglymdashaligns with thehighest magnetic susceptibilities found by the model

DISCUSSION

We succeeded in finding a model that fits the observedmagnetic anomalies and the petrophysical measurementsfrom the ICDP boreholes LB-07A and LB-08A Howeverthere are still some unresolved problems that we attempt toaddress here Due to the expected relatively highmagnetization of impact-melt rocks (Ugalde et al 2005Henkel and Reimold 2002) the amount of impact meltproduced by the Bosumtwi impact becomes an importantconstraint in understanding the source of the observedmagnetic anomalies Artemieva et al (2004) estimated thevolume of impact melt produced by the impact (projectilediameter between 076ndash104 km for an impact angle between30degndash60deg impacting at 20 kms) at 26ndash43 km3 and athickness of ~200 m for the melt sheet near the location of thetwo ICDP deep boreholes Plado et al (2000) interpreted theobserved magnetic anomalies as an approximately triangularbody of impact melt extending for 4000 m in the north-southdirection of 400 m thickness and of 4000 m width in east-west extent which would give approximately 32 km3 ofimpact melt However the ICDP deep boreholes did not finda considerable amount of impact melt just centimeter-widesuevite dikes in LB-08A and lt50 m of suevite in LB-07A(Coney et al 2007 Ferriegravere et al 2007 Deutsch et al 2007)In addition the melt proportion in suevitic breccias in theseboreholes was shown by these authors to be generally verysmall amounting to just a few volume The gravity modelof Ugalde et al (2007) also supports an impactite unit muchthinner than expected Therefore there are three options tounderstand the source of the magnetic anomalies a) the ICDP

deep boreholes missed the true source of the magneticanomalies and there is a large body of magnetic rock belowthe ICDP boreholes but at a greater depth b) there is a largebody of impact melt in the northeast sector of the impactstructure c) the source of the magnetic anomalies is notimpact melt and another source mechanism has yet to befound

The first option can be ruled out based on the magneticproperties of the impactites measured around the lake byPlado et al (2000) and the magnetic properties of the targetlithologies measured by Kontny et al (2007) Because of theexponential decay of the magnetic anomalies with depth tothe top of the associated magnetic sources (Spector and Grant1970) in order to have a deeper magnetic source below theICDP deep boreholes the magnetic properties of these bodieswould have to be at least ten times those measured around thelake or at the ICDP boreholes That is very unlikelyconsidering the magnetic properties measured in a few meltfragments found in the impact breccia and the targetlithologies at LB-07A and LB-08A (Kontny and Just 2006Kontny et al 2007) Kontny et al (2007) measured an averagesusceptibility of ~300 times 10minus6 SI and Q ~9 for the targetlithologies That would give an effective susceptibility of keff= k(1 + Q) = 3000 times 10minus5 SI for the whole stratigraphiccolumn which is still not sufficient to produce such largeanomalies Also if that were the case the borehole magneticlogs would show much higher amplitude anomalies towardsthe base of the boreholes (Morris et al 2007) Furthermorethe location of the ICDP deep boreholes does not match thelocation of any of the magnetic anomalies as indicated by theamplitude of the analytical signal (Fig 4) nor does thewavelength of the anomalies correspond to what is expectedfor a magnetic source located 500 m below the observationplane (much broader anomalies highly reduced amplitudes)Plado et al (2000) interpreted the magnetic anomalies asimpact melt based on the hypothesis that the magneticproperties of the suevites around the lake were diminished byweathering and alteration processes Thus the measuredvalues were not representative of the real magnetizationparameters of the impact-melt rocks that were assumed to beinside the impact structure Kontny et al (2007) indeedshowed that the rock magnetic parameters for all rocks at theICDP deep boreholes (impact lithologies and basement rocks)are distinctly higher than those reported by Plado et al(2000) however neither ICDP deep borehole intersected anyimpact melt (Coney et al 2007 Ferriegravere et al 2007 Deutschet al 2007) which rules out impact melt as a magnetic sourceFinally based on the location of the magnetic anomalies andtheir wavelength and intensity we cannot locate the source ofthe magnetic anomalies below the ICDP deep boreholes at agreater depth

The second and third options have similar implicationsregarding the location of the magnetic sources but differ withregard to the origin of the anomalies Considering both the

keff k 1 Q+( ) k 1 MR

MI-------------+⎝ ⎠

⎛ ⎞==

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

872 H Ugalde et al

to the dominant strike of the metasedimentary andmetavolcanic belts observed in the region and anomalies C1and C2 at the western side of the lake both have mid- tohigh- amplitude and 700ndash900 m wavelength The smalleramplitude of C1 and C2 denotes probable deeper settings oftheir corresponding magnetic sources or poorer magneticproperties than those relating to the anomalies A and B

Magnetic Properties of the Rocks

The magnetic properties of the rocks found in and aroundthe Bosumtwi structure are of key importance in themodeling The data used by Plado et al (2000) aresummarized in Table 1 together with data used in this work

Plado et al (2000) demonstrated that there is a cleardifference between petrophysical properties of the targetrocks and the impact-derived suevites The fallout sueviteshave lower densities (higher porosities) and somewhat highermagnetizations (susceptibility remanent magnetization)compared to the target rocks The remanent magnetization ofsuevites prevails over induced magnetization (theKoenigsberger ratio of remanent over induced magnetizationis much higher than 1 ~4) The target rockshave strikingly homogeneous physical properties withnoticeable weak remanent magnetization and Q values

(Q lt005) A graphitic shale had fairly high remanentmagnetization (natural remanent magnetization [NRM] =386 Am Q = 067) whereas the granites (such as thePepiakese body) were magnetically very weak (NRM lt02 Am Q lt016) Unfortunately there are no measurementsof magnetic properties for the Kumasi batholith to the north ofthe lake and the Bansu intrusion to the south although both ofthem exhibit a direct spatial correlation with the airbornemagnetic anomalies

Petrophysical measurements were accomplished on thecore recovered from both holes LB-07A and LB-08A at thefacilities of the GeoForschungsZentrum (GFZ) in Potsdam in2004 (Milkereit et al 2006 Ugalde 2006 Morris et al 2007)See Morris et al (2007) for a detailed description of themagnetic susceptibility logging

NRM was measured in a few selected samples of thecores (Elbra et al 2007 Kontny et al 2007) In general NRMintensities on impact breccias vary between 04 and 20 Am(Kontny et al 2007) Although these values are higher thanthe NRM intensities reported for the target rocks sampledaround the lake (01ndash39 mAm) (Plado et al 2000) they werederived from small clasts or pieces of melt and therefore donot constitute a representative high-magnetization unit thatcould account for the large magnetic anomalies observed inthe lake area

Fig 4 Amplitude of the analytical signal (ASIG) computed from the marine magnetic data set The two ICDP deep boreholes are shown asstars The main anomalies are labeled See text for details

Q MR= MIfrasl

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 873

Previous Magnetic Models

Based on aeromagnetic data and petrophysical andpaleomagnetic data of oriented samples of impactites andtarget rocks outcropping around the lake Plado et al (2000)published a 3-D magnetic model of the structure The modelconsisted of polygonal bodies at depth from 200 to 600 mdiminishing downwards from their surface cross section Themodel of Plado et al (2000) is shown in Fig 5a

However in order to obtain a meaningful match of themodel with the observed magnetic data Plado et al (2000)had to assign much higher values to the magnetic properties tothe body than what was actually measured on rocks (seeTable 1) In the model water sediment and basement wereall treated as the same background unit with zeromagnetization The magnetic body constituting the source ofthe magnetic anomaly (the hypothesized impact melt) wasgiven uniform magnetic properties (susceptibility =00033 SI NRM = 0367 Amminus1 and Q ~43) The remanentmagnetization direction was set to D = 3 I = minus1 as measuredon suevites outcropping around the lake Plado et al (2000)discussed in detail the various possibilities for why themagnetic parameters within the structure could differ fromthose outside

Danuor obtained a 25-D model for one profile across thecenter of the structure (Danuor 2004) Similar to Plado et al(2000) in order to reproduce the observed anomaly he used ahigh magnetic susceptibility of 003 SI but assumed onlyinduced magnetization (no NRM) The proposed source bodyextended from about 250 m to about 610 m depth (Danuor2004) Danuorrsquos model is shown in Fig 5b

NEW MAGNETIC MODELING

Synthetic Modeling

Due to the dipolar nature of the Earthrsquos magnetic fieldobserved magnetic anomalies at low magnetic latitudes canbe rather complex and thus to find the proper location of themagnetic sources relative to the anomalies can be difficult Tobetter understand the expected pattern of magnetic anomalies

at low magnetic latitudes as at Lake Bosumtwi we firstperformed some synthetic modeling

The first model (Figs 6andashc) shows the magneticsignature of a magnetized prismatic body of 4000 times 4000 times400 m The size of the body is the same as that assumed byPlado et al (2000) This proves that a highly magneticanomalous source body will give rise to a negative magneticanomaly at the observed latitude (65degN) of Lake BosumtwiThe negative anomaly appears slightly to the south of thecenter of the body and is accompanied by positive anomalieson the northern and southern sides of the major anomaly Ofthese side anomalies the northern one has slightly higheramplitude than the southern one From this model it is alsoclear that the measured magnetization parameters of the deepICDP boreholes LB-07A and LB-08A are insufficient togenerate the observed magnetic anomaly Adding remanentmagnetization parallel to the ambient field has the same effectas increasing the magnetic susceptibility the shape of theanomaly does not change only its amplitude

The second synthetic model (Fig 6d) shows themagnetic signature of a thin sheet of 4000 times 4000 times 5 m Thepurpose of this model is to test the hypothesis of the observedmagnetic anomaly being caused by a stack of thin meltsheets This hypothesis arose at early stages of interpretationof the LB-07A and LB-08A borehole magnetic data whichshowed high-frequency and high-amplitude magneticanomalies However the borehole data were acquired at asampling rate of 10 cm and spacing of millimeters betweenthe magnetic sources and the sensors therefore the frequencycontent of these anomalies is related to the different samplingrate and closer distance to the magnetic sources and is notcomparable to the observed magnetic anomalies as detectedfrom the lake surface or from an airplane In any case thismodel is intended to verify whether the thin magnetic sheetidea is feasible or not The magnetization parameters used arethe same as for the 3-D model (Table 1) The amplitude of theanomaly created by such a thin sheet is 025 nT at a depth of200 m and 004 nT at 600 m depth Thus to reproduce theobserved plusmn30 nT anomaly we need at least 120 of these thinsheets assuming the best possible scenario of shallowsources at 200 m where the amplitude is the greatest If we

Table 1 Magnetic properties of the impact and target rocks as measured on outcrops around Lake Bosumtwi by Plado et al (2000) and on the LB-07A and LB-08A drill cores (Kontny et al 2007) and magnetization parameters used in the 3-D model

Magnetic susceptibilityPlado et al (2000)

Magnetic susceptibilityThis work

Remanent magnetizationPlado et al (2000)(Am)

Remanent magnetizationThis work(Am)

Rock typelayer Measured Model Measured Model Measured Model Measured Model

Water N A N App N A 0 N A 0 N A 0Sediments N A N App N A 0 N A 0 N A 0Magnetic bodies 000033 00033 00003 00002ndash005 00368 00368

D = 3 I = minus1

00004ndash02 0ndash15D = minus6I = minus12

Target rocks (background) 000015 0 00003 00003 0057 0N A = not available N App = not applicable

874 H Ugalde et al

consider the decreasing amplitude with depth the number ofsheets scales up to at least 200 Therefore the more realisticapproach of stacking the thin sheets one below the otherleaving some space in between them would first requiremore space (gt1000 m taking into account 1000 m for the200 thin sheets plus 60ndash100 m of host rock considering 05ndash1 m between each sheet) and secondly the computedmagnetic response of such a geometry would still be smallerthan the observed field Furthermore this alternate modelwould extend from 200 to 1200 m depth which is notfeasible according to the borehole logs (Coney et al 2007Deutsch et al 2007 Ferriegravere et al 2007) and also wouldmean a block of 4000 times 4000 times 1000 m of highly magneticmaterial which was not encountered by the deep ICDPboreholes

New 3-D Model

A new 3-D model was created from the marine magneticdata (Fig 7) The 3-D structure was built over the ~800 mspaced north-south profiles Each section is 850ndash1000 m wide(depending on the separation of the survey lines) The bodiesthat constitute each section are draped over the craterbathymetry A combination of forward modeling andinversion leads to a model that has an error with an average of084 nT and a standard deviation of 82 nT The basic modelwhich was constructed on the basis of the geometry inheritedby a similar 3-D model built with the gravity data (Ugaldeet al 2007) comprises four layers water sediment impactbreccia (or magnetic rocks underlying the sediments) andmdashfor consistency with Kontny et alrsquos (2007) petrophysical

Fig 5 Previous attempts to model the observed magnetic anomalies at Lake Bosumtwi a) A cross section through the center of the craterfrom Plado et al (2000) b) The Danuor (2004) model across the center of the structure Note the different directions and scales for these twoprofiles

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 875

measurementsmdasha magnetic background Kontny et al (2007)measured high remanent magnetization intensities on samplesfrom both ICDP deep boreholes (LB-07A and LB-08A)which vary depending on the volume percent content ofpyrrhotite However those magnetization variations on thebackground are beyond the resolution of the marine magneticdata set used here

The parameters used in this model calculation aresummarized in Table 1 Starting from the source bodygeometry derived from the gravity study and applyinginversion the magnetic properties of the bodies constitutingeach line were adjusted line by line in order to reproduce theobserved anomalies The inversion was constrained tomagnetic susceptibilities lt005 SI and NRM lt15 Am Whatmakes this model really 3-D is that all the bodies across theentire structure are active when computing a single line thus

unlike 2-D or 25-D models each line is affected by thebodies located laterally to it As seen in Fig 7 this modelallows a non-uniform magnetization distribution throughoutthe structure Magnetic susceptibilities vary from 00002 to005 SI and NRM intensity from 0 to 15 Am parallel to theambient field (D = minus6deg I = minus12deg)

The model complies to the borehole petrophysics withmagnetic susceptibilities and remanence values within therange of what was measured on the recovered core from thedeep ICDP boreholes (Elbra et al 2007 Kontny et al 2007Morris et al 2007) The susceptibility values are uniformlylow across most of the lake with the exception of thenortheast side of it There at a distance of ~1800 m from theICDP boreholes lie 4 bodies that cover an area of ~4000 times2000 m and where susceptibilities are much higher than thosemeasured on the IDCP core (Fig 7) (Figs 3 and 4 of Morris

Fig 6 Synthetic models for a source at a magnetic latitude similar to that of the Lake Bosumtwi impact structure The magnetic source is onemagnetized block with the parameters as given in Plado et al (2000) See text and Table 1 for details a) Plan view of the model (black outline)and its computed magnetic response b) 3-D perspective view of the model showing the synthetic lines every 100 m c) Cross section of themodel along the center of the synthetic structure d) Cross section of the thin sheet model Note the different vertical scales in (c) and (d)

876 H Ugalde et al

Fig 7 Plan view and cross sections of the new 3-D model generated in this work Top left plan view of the magnetized bodies color-codedby magnetic susceptibility Top right computed magnetic response of the 3-D model Middle and bottom rows four north-south cross sectionsof the model from west (line 2) to east (line 5) The bodies are color-coded by magnetic susceptibility The body at the top is water (k = 0)constrained by the lake bathymetry followed by the post-impact sediment layer (k = 0) which is constrained by the gravity model

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 877

et al 2007) NRM intensities are higher than 1 Am acrossmost of the lake However as the direction of remanence isparallel to the ambient field the real effect of increasing theintensity of NRM is to increase the effective susceptibility

(1)

where k is magnetic susceptibility keff is the effectivesusceptibility after adding remanence Q the Koenigsbergerratio and MI and MR are the intensity of the induced andremanent magnetization respectively Consequently thesame effect can be accomplished by reducing the intensity ofNRM and increasing susceptibility or vice-versa butreasonable limits on the amount of each are imposed bymagnetic mineralogy (compare Kontny et al 2007)

Analysis of the amplitude of the analytical signal map(Fig 4) also locates the main source of the observed magneticanomaly at the northeast corner of the lake (A1 A2 of Fig 4)and on a northeast to southwest line to the south of it (B1 andB2 of Fig 4) whichmdashnot surprisinglymdashaligns with thehighest magnetic susceptibilities found by the model

DISCUSSION

We succeeded in finding a model that fits the observedmagnetic anomalies and the petrophysical measurementsfrom the ICDP boreholes LB-07A and LB-08A Howeverthere are still some unresolved problems that we attempt toaddress here Due to the expected relatively highmagnetization of impact-melt rocks (Ugalde et al 2005Henkel and Reimold 2002) the amount of impact meltproduced by the Bosumtwi impact becomes an importantconstraint in understanding the source of the observedmagnetic anomalies Artemieva et al (2004) estimated thevolume of impact melt produced by the impact (projectilediameter between 076ndash104 km for an impact angle between30degndash60deg impacting at 20 kms) at 26ndash43 km3 and athickness of ~200 m for the melt sheet near the location of thetwo ICDP deep boreholes Plado et al (2000) interpreted theobserved magnetic anomalies as an approximately triangularbody of impact melt extending for 4000 m in the north-southdirection of 400 m thickness and of 4000 m width in east-west extent which would give approximately 32 km3 ofimpact melt However the ICDP deep boreholes did not finda considerable amount of impact melt just centimeter-widesuevite dikes in LB-08A and lt50 m of suevite in LB-07A(Coney et al 2007 Ferriegravere et al 2007 Deutsch et al 2007)In addition the melt proportion in suevitic breccias in theseboreholes was shown by these authors to be generally verysmall amounting to just a few volume The gravity modelof Ugalde et al (2007) also supports an impactite unit muchthinner than expected Therefore there are three options tounderstand the source of the magnetic anomalies a) the ICDP

deep boreholes missed the true source of the magneticanomalies and there is a large body of magnetic rock belowthe ICDP boreholes but at a greater depth b) there is a largebody of impact melt in the northeast sector of the impactstructure c) the source of the magnetic anomalies is notimpact melt and another source mechanism has yet to befound

The first option can be ruled out based on the magneticproperties of the impactites measured around the lake byPlado et al (2000) and the magnetic properties of the targetlithologies measured by Kontny et al (2007) Because of theexponential decay of the magnetic anomalies with depth tothe top of the associated magnetic sources (Spector and Grant1970) in order to have a deeper magnetic source below theICDP deep boreholes the magnetic properties of these bodieswould have to be at least ten times those measured around thelake or at the ICDP boreholes That is very unlikelyconsidering the magnetic properties measured in a few meltfragments found in the impact breccia and the targetlithologies at LB-07A and LB-08A (Kontny and Just 2006Kontny et al 2007) Kontny et al (2007) measured an averagesusceptibility of ~300 times 10minus6 SI and Q ~9 for the targetlithologies That would give an effective susceptibility of keff= k(1 + Q) = 3000 times 10minus5 SI for the whole stratigraphiccolumn which is still not sufficient to produce such largeanomalies Also if that were the case the borehole magneticlogs would show much higher amplitude anomalies towardsthe base of the boreholes (Morris et al 2007) Furthermorethe location of the ICDP deep boreholes does not match thelocation of any of the magnetic anomalies as indicated by theamplitude of the analytical signal (Fig 4) nor does thewavelength of the anomalies correspond to what is expectedfor a magnetic source located 500 m below the observationplane (much broader anomalies highly reduced amplitudes)Plado et al (2000) interpreted the magnetic anomalies asimpact melt based on the hypothesis that the magneticproperties of the suevites around the lake were diminished byweathering and alteration processes Thus the measuredvalues were not representative of the real magnetizationparameters of the impact-melt rocks that were assumed to beinside the impact structure Kontny et al (2007) indeedshowed that the rock magnetic parameters for all rocks at theICDP deep boreholes (impact lithologies and basement rocks)are distinctly higher than those reported by Plado et al(2000) however neither ICDP deep borehole intersected anyimpact melt (Coney et al 2007 Ferriegravere et al 2007 Deutschet al 2007) which rules out impact melt as a magnetic sourceFinally based on the location of the magnetic anomalies andtheir wavelength and intensity we cannot locate the source ofthe magnetic anomalies below the ICDP deep boreholes at agreater depth

The second and third options have similar implicationsregarding the location of the magnetic sources but differ withregard to the origin of the anomalies Considering both the

keff k 1 Q+( ) k 1 MR

MI-------------+⎝ ⎠

⎛ ⎞==

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 873

Previous Magnetic Models

Based on aeromagnetic data and petrophysical andpaleomagnetic data of oriented samples of impactites andtarget rocks outcropping around the lake Plado et al (2000)published a 3-D magnetic model of the structure The modelconsisted of polygonal bodies at depth from 200 to 600 mdiminishing downwards from their surface cross section Themodel of Plado et al (2000) is shown in Fig 5a

However in order to obtain a meaningful match of themodel with the observed magnetic data Plado et al (2000)had to assign much higher values to the magnetic properties tothe body than what was actually measured on rocks (seeTable 1) In the model water sediment and basement wereall treated as the same background unit with zeromagnetization The magnetic body constituting the source ofthe magnetic anomaly (the hypothesized impact melt) wasgiven uniform magnetic properties (susceptibility =00033 SI NRM = 0367 Amminus1 and Q ~43) The remanentmagnetization direction was set to D = 3 I = minus1 as measuredon suevites outcropping around the lake Plado et al (2000)discussed in detail the various possibilities for why themagnetic parameters within the structure could differ fromthose outside

Danuor obtained a 25-D model for one profile across thecenter of the structure (Danuor 2004) Similar to Plado et al(2000) in order to reproduce the observed anomaly he used ahigh magnetic susceptibility of 003 SI but assumed onlyinduced magnetization (no NRM) The proposed source bodyextended from about 250 m to about 610 m depth (Danuor2004) Danuorrsquos model is shown in Fig 5b

NEW MAGNETIC MODELING

Synthetic Modeling

Due to the dipolar nature of the Earthrsquos magnetic fieldobserved magnetic anomalies at low magnetic latitudes canbe rather complex and thus to find the proper location of themagnetic sources relative to the anomalies can be difficult Tobetter understand the expected pattern of magnetic anomalies

at low magnetic latitudes as at Lake Bosumtwi we firstperformed some synthetic modeling

The first model (Figs 6andashc) shows the magneticsignature of a magnetized prismatic body of 4000 times 4000 times400 m The size of the body is the same as that assumed byPlado et al (2000) This proves that a highly magneticanomalous source body will give rise to a negative magneticanomaly at the observed latitude (65degN) of Lake BosumtwiThe negative anomaly appears slightly to the south of thecenter of the body and is accompanied by positive anomalieson the northern and southern sides of the major anomaly Ofthese side anomalies the northern one has slightly higheramplitude than the southern one From this model it is alsoclear that the measured magnetization parameters of the deepICDP boreholes LB-07A and LB-08A are insufficient togenerate the observed magnetic anomaly Adding remanentmagnetization parallel to the ambient field has the same effectas increasing the magnetic susceptibility the shape of theanomaly does not change only its amplitude

The second synthetic model (Fig 6d) shows themagnetic signature of a thin sheet of 4000 times 4000 times 5 m Thepurpose of this model is to test the hypothesis of the observedmagnetic anomaly being caused by a stack of thin meltsheets This hypothesis arose at early stages of interpretationof the LB-07A and LB-08A borehole magnetic data whichshowed high-frequency and high-amplitude magneticanomalies However the borehole data were acquired at asampling rate of 10 cm and spacing of millimeters betweenthe magnetic sources and the sensors therefore the frequencycontent of these anomalies is related to the different samplingrate and closer distance to the magnetic sources and is notcomparable to the observed magnetic anomalies as detectedfrom the lake surface or from an airplane In any case thismodel is intended to verify whether the thin magnetic sheetidea is feasible or not The magnetization parameters used arethe same as for the 3-D model (Table 1) The amplitude of theanomaly created by such a thin sheet is 025 nT at a depth of200 m and 004 nT at 600 m depth Thus to reproduce theobserved plusmn30 nT anomaly we need at least 120 of these thinsheets assuming the best possible scenario of shallowsources at 200 m where the amplitude is the greatest If we

Table 1 Magnetic properties of the impact and target rocks as measured on outcrops around Lake Bosumtwi by Plado et al (2000) and on the LB-07A and LB-08A drill cores (Kontny et al 2007) and magnetization parameters used in the 3-D model

Magnetic susceptibilityPlado et al (2000)

Magnetic susceptibilityThis work

Remanent magnetizationPlado et al (2000)(Am)

Remanent magnetizationThis work(Am)

Rock typelayer Measured Model Measured Model Measured Model Measured Model

Water N A N App N A 0 N A 0 N A 0Sediments N A N App N A 0 N A 0 N A 0Magnetic bodies 000033 00033 00003 00002ndash005 00368 00368

D = 3 I = minus1

00004ndash02 0ndash15D = minus6I = minus12

Target rocks (background) 000015 0 00003 00003 0057 0N A = not available N App = not applicable

874 H Ugalde et al

consider the decreasing amplitude with depth the number ofsheets scales up to at least 200 Therefore the more realisticapproach of stacking the thin sheets one below the otherleaving some space in between them would first requiremore space (gt1000 m taking into account 1000 m for the200 thin sheets plus 60ndash100 m of host rock considering 05ndash1 m between each sheet) and secondly the computedmagnetic response of such a geometry would still be smallerthan the observed field Furthermore this alternate modelwould extend from 200 to 1200 m depth which is notfeasible according to the borehole logs (Coney et al 2007Deutsch et al 2007 Ferriegravere et al 2007) and also wouldmean a block of 4000 times 4000 times 1000 m of highly magneticmaterial which was not encountered by the deep ICDPboreholes

New 3-D Model

A new 3-D model was created from the marine magneticdata (Fig 7) The 3-D structure was built over the ~800 mspaced north-south profiles Each section is 850ndash1000 m wide(depending on the separation of the survey lines) The bodiesthat constitute each section are draped over the craterbathymetry A combination of forward modeling andinversion leads to a model that has an error with an average of084 nT and a standard deviation of 82 nT The basic modelwhich was constructed on the basis of the geometry inheritedby a similar 3-D model built with the gravity data (Ugaldeet al 2007) comprises four layers water sediment impactbreccia (or magnetic rocks underlying the sediments) andmdashfor consistency with Kontny et alrsquos (2007) petrophysical

Fig 5 Previous attempts to model the observed magnetic anomalies at Lake Bosumtwi a) A cross section through the center of the craterfrom Plado et al (2000) b) The Danuor (2004) model across the center of the structure Note the different directions and scales for these twoprofiles

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 875

measurementsmdasha magnetic background Kontny et al (2007)measured high remanent magnetization intensities on samplesfrom both ICDP deep boreholes (LB-07A and LB-08A)which vary depending on the volume percent content ofpyrrhotite However those magnetization variations on thebackground are beyond the resolution of the marine magneticdata set used here

The parameters used in this model calculation aresummarized in Table 1 Starting from the source bodygeometry derived from the gravity study and applyinginversion the magnetic properties of the bodies constitutingeach line were adjusted line by line in order to reproduce theobserved anomalies The inversion was constrained tomagnetic susceptibilities lt005 SI and NRM lt15 Am Whatmakes this model really 3-D is that all the bodies across theentire structure are active when computing a single line thus

unlike 2-D or 25-D models each line is affected by thebodies located laterally to it As seen in Fig 7 this modelallows a non-uniform magnetization distribution throughoutthe structure Magnetic susceptibilities vary from 00002 to005 SI and NRM intensity from 0 to 15 Am parallel to theambient field (D = minus6deg I = minus12deg)

The model complies to the borehole petrophysics withmagnetic susceptibilities and remanence values within therange of what was measured on the recovered core from thedeep ICDP boreholes (Elbra et al 2007 Kontny et al 2007Morris et al 2007) The susceptibility values are uniformlylow across most of the lake with the exception of thenortheast side of it There at a distance of ~1800 m from theICDP boreholes lie 4 bodies that cover an area of ~4000 times2000 m and where susceptibilities are much higher than thosemeasured on the IDCP core (Fig 7) (Figs 3 and 4 of Morris

Fig 6 Synthetic models for a source at a magnetic latitude similar to that of the Lake Bosumtwi impact structure The magnetic source is onemagnetized block with the parameters as given in Plado et al (2000) See text and Table 1 for details a) Plan view of the model (black outline)and its computed magnetic response b) 3-D perspective view of the model showing the synthetic lines every 100 m c) Cross section of themodel along the center of the synthetic structure d) Cross section of the thin sheet model Note the different vertical scales in (c) and (d)

876 H Ugalde et al

Fig 7 Plan view and cross sections of the new 3-D model generated in this work Top left plan view of the magnetized bodies color-codedby magnetic susceptibility Top right computed magnetic response of the 3-D model Middle and bottom rows four north-south cross sectionsof the model from west (line 2) to east (line 5) The bodies are color-coded by magnetic susceptibility The body at the top is water (k = 0)constrained by the lake bathymetry followed by the post-impact sediment layer (k = 0) which is constrained by the gravity model

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 877

et al 2007) NRM intensities are higher than 1 Am acrossmost of the lake However as the direction of remanence isparallel to the ambient field the real effect of increasing theintensity of NRM is to increase the effective susceptibility

(1)

where k is magnetic susceptibility keff is the effectivesusceptibility after adding remanence Q the Koenigsbergerratio and MI and MR are the intensity of the induced andremanent magnetization respectively Consequently thesame effect can be accomplished by reducing the intensity ofNRM and increasing susceptibility or vice-versa butreasonable limits on the amount of each are imposed bymagnetic mineralogy (compare Kontny et al 2007)

Analysis of the amplitude of the analytical signal map(Fig 4) also locates the main source of the observed magneticanomaly at the northeast corner of the lake (A1 A2 of Fig 4)and on a northeast to southwest line to the south of it (B1 andB2 of Fig 4) whichmdashnot surprisinglymdashaligns with thehighest magnetic susceptibilities found by the model

DISCUSSION

We succeeded in finding a model that fits the observedmagnetic anomalies and the petrophysical measurementsfrom the ICDP boreholes LB-07A and LB-08A Howeverthere are still some unresolved problems that we attempt toaddress here Due to the expected relatively highmagnetization of impact-melt rocks (Ugalde et al 2005Henkel and Reimold 2002) the amount of impact meltproduced by the Bosumtwi impact becomes an importantconstraint in understanding the source of the observedmagnetic anomalies Artemieva et al (2004) estimated thevolume of impact melt produced by the impact (projectilediameter between 076ndash104 km for an impact angle between30degndash60deg impacting at 20 kms) at 26ndash43 km3 and athickness of ~200 m for the melt sheet near the location of thetwo ICDP deep boreholes Plado et al (2000) interpreted theobserved magnetic anomalies as an approximately triangularbody of impact melt extending for 4000 m in the north-southdirection of 400 m thickness and of 4000 m width in east-west extent which would give approximately 32 km3 ofimpact melt However the ICDP deep boreholes did not finda considerable amount of impact melt just centimeter-widesuevite dikes in LB-08A and lt50 m of suevite in LB-07A(Coney et al 2007 Ferriegravere et al 2007 Deutsch et al 2007)In addition the melt proportion in suevitic breccias in theseboreholes was shown by these authors to be generally verysmall amounting to just a few volume The gravity modelof Ugalde et al (2007) also supports an impactite unit muchthinner than expected Therefore there are three options tounderstand the source of the magnetic anomalies a) the ICDP

deep boreholes missed the true source of the magneticanomalies and there is a large body of magnetic rock belowthe ICDP boreholes but at a greater depth b) there is a largebody of impact melt in the northeast sector of the impactstructure c) the source of the magnetic anomalies is notimpact melt and another source mechanism has yet to befound

The first option can be ruled out based on the magneticproperties of the impactites measured around the lake byPlado et al (2000) and the magnetic properties of the targetlithologies measured by Kontny et al (2007) Because of theexponential decay of the magnetic anomalies with depth tothe top of the associated magnetic sources (Spector and Grant1970) in order to have a deeper magnetic source below theICDP deep boreholes the magnetic properties of these bodieswould have to be at least ten times those measured around thelake or at the ICDP boreholes That is very unlikelyconsidering the magnetic properties measured in a few meltfragments found in the impact breccia and the targetlithologies at LB-07A and LB-08A (Kontny and Just 2006Kontny et al 2007) Kontny et al (2007) measured an averagesusceptibility of ~300 times 10minus6 SI and Q ~9 for the targetlithologies That would give an effective susceptibility of keff= k(1 + Q) = 3000 times 10minus5 SI for the whole stratigraphiccolumn which is still not sufficient to produce such largeanomalies Also if that were the case the borehole magneticlogs would show much higher amplitude anomalies towardsthe base of the boreholes (Morris et al 2007) Furthermorethe location of the ICDP deep boreholes does not match thelocation of any of the magnetic anomalies as indicated by theamplitude of the analytical signal (Fig 4) nor does thewavelength of the anomalies correspond to what is expectedfor a magnetic source located 500 m below the observationplane (much broader anomalies highly reduced amplitudes)Plado et al (2000) interpreted the magnetic anomalies asimpact melt based on the hypothesis that the magneticproperties of the suevites around the lake were diminished byweathering and alteration processes Thus the measuredvalues were not representative of the real magnetizationparameters of the impact-melt rocks that were assumed to beinside the impact structure Kontny et al (2007) indeedshowed that the rock magnetic parameters for all rocks at theICDP deep boreholes (impact lithologies and basement rocks)are distinctly higher than those reported by Plado et al(2000) however neither ICDP deep borehole intersected anyimpact melt (Coney et al 2007 Ferriegravere et al 2007 Deutschet al 2007) which rules out impact melt as a magnetic sourceFinally based on the location of the magnetic anomalies andtheir wavelength and intensity we cannot locate the source ofthe magnetic anomalies below the ICDP deep boreholes at agreater depth

The second and third options have similar implicationsregarding the location of the magnetic sources but differ withregard to the origin of the anomalies Considering both the

keff k 1 Q+( ) k 1 MR

MI-------------+⎝ ⎠

⎛ ⎞==

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

874 H Ugalde et al

consider the decreasing amplitude with depth the number ofsheets scales up to at least 200 Therefore the more realisticapproach of stacking the thin sheets one below the otherleaving some space in between them would first requiremore space (gt1000 m taking into account 1000 m for the200 thin sheets plus 60ndash100 m of host rock considering 05ndash1 m between each sheet) and secondly the computedmagnetic response of such a geometry would still be smallerthan the observed field Furthermore this alternate modelwould extend from 200 to 1200 m depth which is notfeasible according to the borehole logs (Coney et al 2007Deutsch et al 2007 Ferriegravere et al 2007) and also wouldmean a block of 4000 times 4000 times 1000 m of highly magneticmaterial which was not encountered by the deep ICDPboreholes

New 3-D Model

A new 3-D model was created from the marine magneticdata (Fig 7) The 3-D structure was built over the ~800 mspaced north-south profiles Each section is 850ndash1000 m wide(depending on the separation of the survey lines) The bodiesthat constitute each section are draped over the craterbathymetry A combination of forward modeling andinversion leads to a model that has an error with an average of084 nT and a standard deviation of 82 nT The basic modelwhich was constructed on the basis of the geometry inheritedby a similar 3-D model built with the gravity data (Ugaldeet al 2007) comprises four layers water sediment impactbreccia (or magnetic rocks underlying the sediments) andmdashfor consistency with Kontny et alrsquos (2007) petrophysical

Fig 5 Previous attempts to model the observed magnetic anomalies at Lake Bosumtwi a) A cross section through the center of the craterfrom Plado et al (2000) b) The Danuor (2004) model across the center of the structure Note the different directions and scales for these twoprofiles

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 875

measurementsmdasha magnetic background Kontny et al (2007)measured high remanent magnetization intensities on samplesfrom both ICDP deep boreholes (LB-07A and LB-08A)which vary depending on the volume percent content ofpyrrhotite However those magnetization variations on thebackground are beyond the resolution of the marine magneticdata set used here

The parameters used in this model calculation aresummarized in Table 1 Starting from the source bodygeometry derived from the gravity study and applyinginversion the magnetic properties of the bodies constitutingeach line were adjusted line by line in order to reproduce theobserved anomalies The inversion was constrained tomagnetic susceptibilities lt005 SI and NRM lt15 Am Whatmakes this model really 3-D is that all the bodies across theentire structure are active when computing a single line thus

unlike 2-D or 25-D models each line is affected by thebodies located laterally to it As seen in Fig 7 this modelallows a non-uniform magnetization distribution throughoutthe structure Magnetic susceptibilities vary from 00002 to005 SI and NRM intensity from 0 to 15 Am parallel to theambient field (D = minus6deg I = minus12deg)

The model complies to the borehole petrophysics withmagnetic susceptibilities and remanence values within therange of what was measured on the recovered core from thedeep ICDP boreholes (Elbra et al 2007 Kontny et al 2007Morris et al 2007) The susceptibility values are uniformlylow across most of the lake with the exception of thenortheast side of it There at a distance of ~1800 m from theICDP boreholes lie 4 bodies that cover an area of ~4000 times2000 m and where susceptibilities are much higher than thosemeasured on the IDCP core (Fig 7) (Figs 3 and 4 of Morris

Fig 6 Synthetic models for a source at a magnetic latitude similar to that of the Lake Bosumtwi impact structure The magnetic source is onemagnetized block with the parameters as given in Plado et al (2000) See text and Table 1 for details a) Plan view of the model (black outline)and its computed magnetic response b) 3-D perspective view of the model showing the synthetic lines every 100 m c) Cross section of themodel along the center of the synthetic structure d) Cross section of the thin sheet model Note the different vertical scales in (c) and (d)

876 H Ugalde et al

Fig 7 Plan view and cross sections of the new 3-D model generated in this work Top left plan view of the magnetized bodies color-codedby magnetic susceptibility Top right computed magnetic response of the 3-D model Middle and bottom rows four north-south cross sectionsof the model from west (line 2) to east (line 5) The bodies are color-coded by magnetic susceptibility The body at the top is water (k = 0)constrained by the lake bathymetry followed by the post-impact sediment layer (k = 0) which is constrained by the gravity model

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 877

et al 2007) NRM intensities are higher than 1 Am acrossmost of the lake However as the direction of remanence isparallel to the ambient field the real effect of increasing theintensity of NRM is to increase the effective susceptibility

(1)

where k is magnetic susceptibility keff is the effectivesusceptibility after adding remanence Q the Koenigsbergerratio and MI and MR are the intensity of the induced andremanent magnetization respectively Consequently thesame effect can be accomplished by reducing the intensity ofNRM and increasing susceptibility or vice-versa butreasonable limits on the amount of each are imposed bymagnetic mineralogy (compare Kontny et al 2007)

Analysis of the amplitude of the analytical signal map(Fig 4) also locates the main source of the observed magneticanomaly at the northeast corner of the lake (A1 A2 of Fig 4)and on a northeast to southwest line to the south of it (B1 andB2 of Fig 4) whichmdashnot surprisinglymdashaligns with thehighest magnetic susceptibilities found by the model

DISCUSSION

We succeeded in finding a model that fits the observedmagnetic anomalies and the petrophysical measurementsfrom the ICDP boreholes LB-07A and LB-08A Howeverthere are still some unresolved problems that we attempt toaddress here Due to the expected relatively highmagnetization of impact-melt rocks (Ugalde et al 2005Henkel and Reimold 2002) the amount of impact meltproduced by the Bosumtwi impact becomes an importantconstraint in understanding the source of the observedmagnetic anomalies Artemieva et al (2004) estimated thevolume of impact melt produced by the impact (projectilediameter between 076ndash104 km for an impact angle between30degndash60deg impacting at 20 kms) at 26ndash43 km3 and athickness of ~200 m for the melt sheet near the location of thetwo ICDP deep boreholes Plado et al (2000) interpreted theobserved magnetic anomalies as an approximately triangularbody of impact melt extending for 4000 m in the north-southdirection of 400 m thickness and of 4000 m width in east-west extent which would give approximately 32 km3 ofimpact melt However the ICDP deep boreholes did not finda considerable amount of impact melt just centimeter-widesuevite dikes in LB-08A and lt50 m of suevite in LB-07A(Coney et al 2007 Ferriegravere et al 2007 Deutsch et al 2007)In addition the melt proportion in suevitic breccias in theseboreholes was shown by these authors to be generally verysmall amounting to just a few volume The gravity modelof Ugalde et al (2007) also supports an impactite unit muchthinner than expected Therefore there are three options tounderstand the source of the magnetic anomalies a) the ICDP

deep boreholes missed the true source of the magneticanomalies and there is a large body of magnetic rock belowthe ICDP boreholes but at a greater depth b) there is a largebody of impact melt in the northeast sector of the impactstructure c) the source of the magnetic anomalies is notimpact melt and another source mechanism has yet to befound

The first option can be ruled out based on the magneticproperties of the impactites measured around the lake byPlado et al (2000) and the magnetic properties of the targetlithologies measured by Kontny et al (2007) Because of theexponential decay of the magnetic anomalies with depth tothe top of the associated magnetic sources (Spector and Grant1970) in order to have a deeper magnetic source below theICDP deep boreholes the magnetic properties of these bodieswould have to be at least ten times those measured around thelake or at the ICDP boreholes That is very unlikelyconsidering the magnetic properties measured in a few meltfragments found in the impact breccia and the targetlithologies at LB-07A and LB-08A (Kontny and Just 2006Kontny et al 2007) Kontny et al (2007) measured an averagesusceptibility of ~300 times 10minus6 SI and Q ~9 for the targetlithologies That would give an effective susceptibility of keff= k(1 + Q) = 3000 times 10minus5 SI for the whole stratigraphiccolumn which is still not sufficient to produce such largeanomalies Also if that were the case the borehole magneticlogs would show much higher amplitude anomalies towardsthe base of the boreholes (Morris et al 2007) Furthermorethe location of the ICDP deep boreholes does not match thelocation of any of the magnetic anomalies as indicated by theamplitude of the analytical signal (Fig 4) nor does thewavelength of the anomalies correspond to what is expectedfor a magnetic source located 500 m below the observationplane (much broader anomalies highly reduced amplitudes)Plado et al (2000) interpreted the magnetic anomalies asimpact melt based on the hypothesis that the magneticproperties of the suevites around the lake were diminished byweathering and alteration processes Thus the measuredvalues were not representative of the real magnetizationparameters of the impact-melt rocks that were assumed to beinside the impact structure Kontny et al (2007) indeedshowed that the rock magnetic parameters for all rocks at theICDP deep boreholes (impact lithologies and basement rocks)are distinctly higher than those reported by Plado et al(2000) however neither ICDP deep borehole intersected anyimpact melt (Coney et al 2007 Ferriegravere et al 2007 Deutschet al 2007) which rules out impact melt as a magnetic sourceFinally based on the location of the magnetic anomalies andtheir wavelength and intensity we cannot locate the source ofthe magnetic anomalies below the ICDP deep boreholes at agreater depth

The second and third options have similar implicationsregarding the location of the magnetic sources but differ withregard to the origin of the anomalies Considering both the

keff k 1 Q+( ) k 1 MR

MI-------------+⎝ ⎠

⎛ ⎞==

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 875

measurementsmdasha magnetic background Kontny et al (2007)measured high remanent magnetization intensities on samplesfrom both ICDP deep boreholes (LB-07A and LB-08A)which vary depending on the volume percent content ofpyrrhotite However those magnetization variations on thebackground are beyond the resolution of the marine magneticdata set used here

The parameters used in this model calculation aresummarized in Table 1 Starting from the source bodygeometry derived from the gravity study and applyinginversion the magnetic properties of the bodies constitutingeach line were adjusted line by line in order to reproduce theobserved anomalies The inversion was constrained tomagnetic susceptibilities lt005 SI and NRM lt15 Am Whatmakes this model really 3-D is that all the bodies across theentire structure are active when computing a single line thus

unlike 2-D or 25-D models each line is affected by thebodies located laterally to it As seen in Fig 7 this modelallows a non-uniform magnetization distribution throughoutthe structure Magnetic susceptibilities vary from 00002 to005 SI and NRM intensity from 0 to 15 Am parallel to theambient field (D = minus6deg I = minus12deg)

The model complies to the borehole petrophysics withmagnetic susceptibilities and remanence values within therange of what was measured on the recovered core from thedeep ICDP boreholes (Elbra et al 2007 Kontny et al 2007Morris et al 2007) The susceptibility values are uniformlylow across most of the lake with the exception of thenortheast side of it There at a distance of ~1800 m from theICDP boreholes lie 4 bodies that cover an area of ~4000 times2000 m and where susceptibilities are much higher than thosemeasured on the IDCP core (Fig 7) (Figs 3 and 4 of Morris

Fig 6 Synthetic models for a source at a magnetic latitude similar to that of the Lake Bosumtwi impact structure The magnetic source is onemagnetized block with the parameters as given in Plado et al (2000) See text and Table 1 for details a) Plan view of the model (black outline)and its computed magnetic response b) 3-D perspective view of the model showing the synthetic lines every 100 m c) Cross section of themodel along the center of the synthetic structure d) Cross section of the thin sheet model Note the different vertical scales in (c) and (d)

876 H Ugalde et al

Fig 7 Plan view and cross sections of the new 3-D model generated in this work Top left plan view of the magnetized bodies color-codedby magnetic susceptibility Top right computed magnetic response of the 3-D model Middle and bottom rows four north-south cross sectionsof the model from west (line 2) to east (line 5) The bodies are color-coded by magnetic susceptibility The body at the top is water (k = 0)constrained by the lake bathymetry followed by the post-impact sediment layer (k = 0) which is constrained by the gravity model

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 877

et al 2007) NRM intensities are higher than 1 Am acrossmost of the lake However as the direction of remanence isparallel to the ambient field the real effect of increasing theintensity of NRM is to increase the effective susceptibility

(1)

where k is magnetic susceptibility keff is the effectivesusceptibility after adding remanence Q the Koenigsbergerratio and MI and MR are the intensity of the induced andremanent magnetization respectively Consequently thesame effect can be accomplished by reducing the intensity ofNRM and increasing susceptibility or vice-versa butreasonable limits on the amount of each are imposed bymagnetic mineralogy (compare Kontny et al 2007)

Analysis of the amplitude of the analytical signal map(Fig 4) also locates the main source of the observed magneticanomaly at the northeast corner of the lake (A1 A2 of Fig 4)and on a northeast to southwest line to the south of it (B1 andB2 of Fig 4) whichmdashnot surprisinglymdashaligns with thehighest magnetic susceptibilities found by the model

DISCUSSION

We succeeded in finding a model that fits the observedmagnetic anomalies and the petrophysical measurementsfrom the ICDP boreholes LB-07A and LB-08A Howeverthere are still some unresolved problems that we attempt toaddress here Due to the expected relatively highmagnetization of impact-melt rocks (Ugalde et al 2005Henkel and Reimold 2002) the amount of impact meltproduced by the Bosumtwi impact becomes an importantconstraint in understanding the source of the observedmagnetic anomalies Artemieva et al (2004) estimated thevolume of impact melt produced by the impact (projectilediameter between 076ndash104 km for an impact angle between30degndash60deg impacting at 20 kms) at 26ndash43 km3 and athickness of ~200 m for the melt sheet near the location of thetwo ICDP deep boreholes Plado et al (2000) interpreted theobserved magnetic anomalies as an approximately triangularbody of impact melt extending for 4000 m in the north-southdirection of 400 m thickness and of 4000 m width in east-west extent which would give approximately 32 km3 ofimpact melt However the ICDP deep boreholes did not finda considerable amount of impact melt just centimeter-widesuevite dikes in LB-08A and lt50 m of suevite in LB-07A(Coney et al 2007 Ferriegravere et al 2007 Deutsch et al 2007)In addition the melt proportion in suevitic breccias in theseboreholes was shown by these authors to be generally verysmall amounting to just a few volume The gravity modelof Ugalde et al (2007) also supports an impactite unit muchthinner than expected Therefore there are three options tounderstand the source of the magnetic anomalies a) the ICDP

deep boreholes missed the true source of the magneticanomalies and there is a large body of magnetic rock belowthe ICDP boreholes but at a greater depth b) there is a largebody of impact melt in the northeast sector of the impactstructure c) the source of the magnetic anomalies is notimpact melt and another source mechanism has yet to befound

The first option can be ruled out based on the magneticproperties of the impactites measured around the lake byPlado et al (2000) and the magnetic properties of the targetlithologies measured by Kontny et al (2007) Because of theexponential decay of the magnetic anomalies with depth tothe top of the associated magnetic sources (Spector and Grant1970) in order to have a deeper magnetic source below theICDP deep boreholes the magnetic properties of these bodieswould have to be at least ten times those measured around thelake or at the ICDP boreholes That is very unlikelyconsidering the magnetic properties measured in a few meltfragments found in the impact breccia and the targetlithologies at LB-07A and LB-08A (Kontny and Just 2006Kontny et al 2007) Kontny et al (2007) measured an averagesusceptibility of ~300 times 10minus6 SI and Q ~9 for the targetlithologies That would give an effective susceptibility of keff= k(1 + Q) = 3000 times 10minus5 SI for the whole stratigraphiccolumn which is still not sufficient to produce such largeanomalies Also if that were the case the borehole magneticlogs would show much higher amplitude anomalies towardsthe base of the boreholes (Morris et al 2007) Furthermorethe location of the ICDP deep boreholes does not match thelocation of any of the magnetic anomalies as indicated by theamplitude of the analytical signal (Fig 4) nor does thewavelength of the anomalies correspond to what is expectedfor a magnetic source located 500 m below the observationplane (much broader anomalies highly reduced amplitudes)Plado et al (2000) interpreted the magnetic anomalies asimpact melt based on the hypothesis that the magneticproperties of the suevites around the lake were diminished byweathering and alteration processes Thus the measuredvalues were not representative of the real magnetizationparameters of the impact-melt rocks that were assumed to beinside the impact structure Kontny et al (2007) indeedshowed that the rock magnetic parameters for all rocks at theICDP deep boreholes (impact lithologies and basement rocks)are distinctly higher than those reported by Plado et al(2000) however neither ICDP deep borehole intersected anyimpact melt (Coney et al 2007 Ferriegravere et al 2007 Deutschet al 2007) which rules out impact melt as a magnetic sourceFinally based on the location of the magnetic anomalies andtheir wavelength and intensity we cannot locate the source ofthe magnetic anomalies below the ICDP deep boreholes at agreater depth

The second and third options have similar implicationsregarding the location of the magnetic sources but differ withregard to the origin of the anomalies Considering both the

keff k 1 Q+( ) k 1 MR

MI-------------+⎝ ⎠

⎛ ⎞==

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

876 H Ugalde et al

Fig 7 Plan view and cross sections of the new 3-D model generated in this work Top left plan view of the magnetized bodies color-codedby magnetic susceptibility Top right computed magnetic response of the 3-D model Middle and bottom rows four north-south cross sectionsof the model from west (line 2) to east (line 5) The bodies are color-coded by magnetic susceptibility The body at the top is water (k = 0)constrained by the lake bathymetry followed by the post-impact sediment layer (k = 0) which is constrained by the gravity model

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 877

et al 2007) NRM intensities are higher than 1 Am acrossmost of the lake However as the direction of remanence isparallel to the ambient field the real effect of increasing theintensity of NRM is to increase the effective susceptibility

(1)

where k is magnetic susceptibility keff is the effectivesusceptibility after adding remanence Q the Koenigsbergerratio and MI and MR are the intensity of the induced andremanent magnetization respectively Consequently thesame effect can be accomplished by reducing the intensity ofNRM and increasing susceptibility or vice-versa butreasonable limits on the amount of each are imposed bymagnetic mineralogy (compare Kontny et al 2007)

Analysis of the amplitude of the analytical signal map(Fig 4) also locates the main source of the observed magneticanomaly at the northeast corner of the lake (A1 A2 of Fig 4)and on a northeast to southwest line to the south of it (B1 andB2 of Fig 4) whichmdashnot surprisinglymdashaligns with thehighest magnetic susceptibilities found by the model

DISCUSSION

We succeeded in finding a model that fits the observedmagnetic anomalies and the petrophysical measurementsfrom the ICDP boreholes LB-07A and LB-08A Howeverthere are still some unresolved problems that we attempt toaddress here Due to the expected relatively highmagnetization of impact-melt rocks (Ugalde et al 2005Henkel and Reimold 2002) the amount of impact meltproduced by the Bosumtwi impact becomes an importantconstraint in understanding the source of the observedmagnetic anomalies Artemieva et al (2004) estimated thevolume of impact melt produced by the impact (projectilediameter between 076ndash104 km for an impact angle between30degndash60deg impacting at 20 kms) at 26ndash43 km3 and athickness of ~200 m for the melt sheet near the location of thetwo ICDP deep boreholes Plado et al (2000) interpreted theobserved magnetic anomalies as an approximately triangularbody of impact melt extending for 4000 m in the north-southdirection of 400 m thickness and of 4000 m width in east-west extent which would give approximately 32 km3 ofimpact melt However the ICDP deep boreholes did not finda considerable amount of impact melt just centimeter-widesuevite dikes in LB-08A and lt50 m of suevite in LB-07A(Coney et al 2007 Ferriegravere et al 2007 Deutsch et al 2007)In addition the melt proportion in suevitic breccias in theseboreholes was shown by these authors to be generally verysmall amounting to just a few volume The gravity modelof Ugalde et al (2007) also supports an impactite unit muchthinner than expected Therefore there are three options tounderstand the source of the magnetic anomalies a) the ICDP

deep boreholes missed the true source of the magneticanomalies and there is a large body of magnetic rock belowthe ICDP boreholes but at a greater depth b) there is a largebody of impact melt in the northeast sector of the impactstructure c) the source of the magnetic anomalies is notimpact melt and another source mechanism has yet to befound

The first option can be ruled out based on the magneticproperties of the impactites measured around the lake byPlado et al (2000) and the magnetic properties of the targetlithologies measured by Kontny et al (2007) Because of theexponential decay of the magnetic anomalies with depth tothe top of the associated magnetic sources (Spector and Grant1970) in order to have a deeper magnetic source below theICDP deep boreholes the magnetic properties of these bodieswould have to be at least ten times those measured around thelake or at the ICDP boreholes That is very unlikelyconsidering the magnetic properties measured in a few meltfragments found in the impact breccia and the targetlithologies at LB-07A and LB-08A (Kontny and Just 2006Kontny et al 2007) Kontny et al (2007) measured an averagesusceptibility of ~300 times 10minus6 SI and Q ~9 for the targetlithologies That would give an effective susceptibility of keff= k(1 + Q) = 3000 times 10minus5 SI for the whole stratigraphiccolumn which is still not sufficient to produce such largeanomalies Also if that were the case the borehole magneticlogs would show much higher amplitude anomalies towardsthe base of the boreholes (Morris et al 2007) Furthermorethe location of the ICDP deep boreholes does not match thelocation of any of the magnetic anomalies as indicated by theamplitude of the analytical signal (Fig 4) nor does thewavelength of the anomalies correspond to what is expectedfor a magnetic source located 500 m below the observationplane (much broader anomalies highly reduced amplitudes)Plado et al (2000) interpreted the magnetic anomalies asimpact melt based on the hypothesis that the magneticproperties of the suevites around the lake were diminished byweathering and alteration processes Thus the measuredvalues were not representative of the real magnetizationparameters of the impact-melt rocks that were assumed to beinside the impact structure Kontny et al (2007) indeedshowed that the rock magnetic parameters for all rocks at theICDP deep boreholes (impact lithologies and basement rocks)are distinctly higher than those reported by Plado et al(2000) however neither ICDP deep borehole intersected anyimpact melt (Coney et al 2007 Ferriegravere et al 2007 Deutschet al 2007) which rules out impact melt as a magnetic sourceFinally based on the location of the magnetic anomalies andtheir wavelength and intensity we cannot locate the source ofthe magnetic anomalies below the ICDP deep boreholes at agreater depth

The second and third options have similar implicationsregarding the location of the magnetic sources but differ withregard to the origin of the anomalies Considering both the

keff k 1 Q+( ) k 1 MR

MI-------------+⎝ ⎠

⎛ ⎞==

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 877

et al 2007) NRM intensities are higher than 1 Am acrossmost of the lake However as the direction of remanence isparallel to the ambient field the real effect of increasing theintensity of NRM is to increase the effective susceptibility

(1)

where k is magnetic susceptibility keff is the effectivesusceptibility after adding remanence Q the Koenigsbergerratio and MI and MR are the intensity of the induced andremanent magnetization respectively Consequently thesame effect can be accomplished by reducing the intensity ofNRM and increasing susceptibility or vice-versa butreasonable limits on the amount of each are imposed bymagnetic mineralogy (compare Kontny et al 2007)

Analysis of the amplitude of the analytical signal map(Fig 4) also locates the main source of the observed magneticanomaly at the northeast corner of the lake (A1 A2 of Fig 4)and on a northeast to southwest line to the south of it (B1 andB2 of Fig 4) whichmdashnot surprisinglymdashaligns with thehighest magnetic susceptibilities found by the model

DISCUSSION

We succeeded in finding a model that fits the observedmagnetic anomalies and the petrophysical measurementsfrom the ICDP boreholes LB-07A and LB-08A Howeverthere are still some unresolved problems that we attempt toaddress here Due to the expected relatively highmagnetization of impact-melt rocks (Ugalde et al 2005Henkel and Reimold 2002) the amount of impact meltproduced by the Bosumtwi impact becomes an importantconstraint in understanding the source of the observedmagnetic anomalies Artemieva et al (2004) estimated thevolume of impact melt produced by the impact (projectilediameter between 076ndash104 km for an impact angle between30degndash60deg impacting at 20 kms) at 26ndash43 km3 and athickness of ~200 m for the melt sheet near the location of thetwo ICDP deep boreholes Plado et al (2000) interpreted theobserved magnetic anomalies as an approximately triangularbody of impact melt extending for 4000 m in the north-southdirection of 400 m thickness and of 4000 m width in east-west extent which would give approximately 32 km3 ofimpact melt However the ICDP deep boreholes did not finda considerable amount of impact melt just centimeter-widesuevite dikes in LB-08A and lt50 m of suevite in LB-07A(Coney et al 2007 Ferriegravere et al 2007 Deutsch et al 2007)In addition the melt proportion in suevitic breccias in theseboreholes was shown by these authors to be generally verysmall amounting to just a few volume The gravity modelof Ugalde et al (2007) also supports an impactite unit muchthinner than expected Therefore there are three options tounderstand the source of the magnetic anomalies a) the ICDP

deep boreholes missed the true source of the magneticanomalies and there is a large body of magnetic rock belowthe ICDP boreholes but at a greater depth b) there is a largebody of impact melt in the northeast sector of the impactstructure c) the source of the magnetic anomalies is notimpact melt and another source mechanism has yet to befound

The first option can be ruled out based on the magneticproperties of the impactites measured around the lake byPlado et al (2000) and the magnetic properties of the targetlithologies measured by Kontny et al (2007) Because of theexponential decay of the magnetic anomalies with depth tothe top of the associated magnetic sources (Spector and Grant1970) in order to have a deeper magnetic source below theICDP deep boreholes the magnetic properties of these bodieswould have to be at least ten times those measured around thelake or at the ICDP boreholes That is very unlikelyconsidering the magnetic properties measured in a few meltfragments found in the impact breccia and the targetlithologies at LB-07A and LB-08A (Kontny and Just 2006Kontny et al 2007) Kontny et al (2007) measured an averagesusceptibility of ~300 times 10minus6 SI and Q ~9 for the targetlithologies That would give an effective susceptibility of keff= k(1 + Q) = 3000 times 10minus5 SI for the whole stratigraphiccolumn which is still not sufficient to produce such largeanomalies Also if that were the case the borehole magneticlogs would show much higher amplitude anomalies towardsthe base of the boreholes (Morris et al 2007) Furthermorethe location of the ICDP deep boreholes does not match thelocation of any of the magnetic anomalies as indicated by theamplitude of the analytical signal (Fig 4) nor does thewavelength of the anomalies correspond to what is expectedfor a magnetic source located 500 m below the observationplane (much broader anomalies highly reduced amplitudes)Plado et al (2000) interpreted the magnetic anomalies asimpact melt based on the hypothesis that the magneticproperties of the suevites around the lake were diminished byweathering and alteration processes Thus the measuredvalues were not representative of the real magnetizationparameters of the impact-melt rocks that were assumed to beinside the impact structure Kontny et al (2007) indeedshowed that the rock magnetic parameters for all rocks at theICDP deep boreholes (impact lithologies and basement rocks)are distinctly higher than those reported by Plado et al(2000) however neither ICDP deep borehole intersected anyimpact melt (Coney et al 2007 Ferriegravere et al 2007 Deutschet al 2007) which rules out impact melt as a magnetic sourceFinally based on the location of the magnetic anomalies andtheir wavelength and intensity we cannot locate the source ofthe magnetic anomalies below the ICDP deep boreholes at agreater depth

The second and third options have similar implicationsregarding the location of the magnetic sources but differ withregard to the origin of the anomalies Considering both the

keff k 1 Q+( ) k 1 MR

MI-------------+⎝ ⎠

⎛ ⎞==

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

878 H Ugalde et al

3-D model constructed here and the amplitude of theanalytical signal map (Figs 4 and 7) the magnetic sources arelocated to the northeast of the lake However the genesis ofthese magnetic anomalies has yet to be explained Theimpact-melt alternative as a source of the northeast magneticanomalies is difficult to conceive because of a) the lowmagnetic properties of the impact melt found in LB-08A andLB-07A and b) the expected symmetry of a crater structuresuch as this one Slumping andor movement of basementmega-blocks that could have removed the expected impactmelt from the LB-08A and LB-07A area is an option incomparison with the observations at the Chicxulub ICDPborehole YAX-1 (Dressler et al 2003) however the marinemagnetic data do not show any evidence of faulting thatwould support this hypothesis The numerical modeling byArtemieva (2004) suggested that the impactor came from thenortheast at an angle of 30ndash45deg In this scenario it is expectedto find a higher degree of deformation in the northeast sectorof the crater structure Therefore the magnetic anomalies onthis side of the crater (A1ndashA3) could also represent stronglydeformed areas where the temperatures were sufficiently highto effect pyrrhotite remagnetization (Kontny et al 2007)However this idea relies on larger amounts of volpyrrhotite in these rocks whereas the observed abundance ofthis mineral is far less than 1 vol in the samples from theICDP boreholes (W U Reimold personal communication)to the southwest of these anomalies Finally bysuperimposing the geology of the lake surroundings (Koeberland Reimold 2005) and the airborne magnetic data set(Pesonen et al 2003) there is a clear relation between theintrusives of the Kumasi batholith (KB) and the magneticanomalies observed to the northeast of the crater and possiblyto the southwest as well although intrusives have not beenmapped in that area Figure 8 shows a composite map of theamplitude of the analytical signal from the airborne magneticdata set (Pesonen et al 2003) and the mapped geology(Koeberl and Reimold 2005) For simplicity the mainmagnetic trends were marked with dotted lines and the mainmagnetic anomalies were numbered The magnetic anomaliesto the northeast of the crater (1 of Fig 8) are located in a NE-SW trend that is parallel to two outcrops of KB to thenortheast of the trend Anomalies 2 and 3 are located onanother NE-SW trend to the south of the previous one but asthere are no outcrops of KB to the southwest of the cratertheir magnetic source is uncertain They could be related tothe Bansu Intrusion (BI) outcropping to the east and west(like anomaly 4 located on a NE-SW trend that starts at theBI outcrop on the lake shore) or to an unmapped or not-outcropping occurrence of granitoid Anomaly 5 is locatedjust above an occurrence of BI so we interpret this as themagnetic source Anomaly 6 corresponds to two sets ofanomalies on another NE-SW trend in between two KBoutcrops so we interpret them as extensions of KB toward thesouthwest Anomaly 7 is located in between KB outcrops

that plus the similar frequency content and intensity of theanomalies to those of anomaly 6 allow us to interpretanomaly 7 as KB Anomalies 8 9 and 10 are located on theedge of the metavolcanic belt of the Birimian Supergroup(Mv of Fig 2) their similar intensity to anomaly 11(widespread anomaly located on top of Mv) and smallerintensity than anomalies 3 6 7 (associated to KB) permit usto interpret them as Mv as well

Therefore the main magnetic anomalies within the lakeregion (1 and 2 of Fig 8) can be interpreted without anyldquohighly magnetic impact meltrdquo as demanded by previousworkers (Danuor 2004 Plado et al 2000) This hypothesis issupported by a forward model carried out on the totalmagnetic intensity (TMI) data for borehole LB-08A byMorris et al (2007) The borehole TMI data can be replicatedby placing a prismatic magnetic source body of 3000 times 3000times 3000 m top at 300 m depth strike = 40deg dip = 15deg k =02 SI to the northwest of borehole LB-08A With a verticalborehole and an inclined source contact the observed sharpmagnetic boundary is replicated by the borehole passingthrough an edge of the source body (Morris et al 2007)

By putting together the amplitude of the analytical signalfrom the marine data set (Fig 4) and the geology of the lake(Koeberl and Reimold 2005) the relationship between themain anomalies (A1 A2 A3) and the Kumasi batholithintrusives is even more evident (Fig 9) The anomalies southof B1 and B2 might be related to olivine pyroxenite doleriteand gabbro which were mapped as minor intrusives at theeastern side of the Bosumtwi structure (Koeberl and Reimold2005 and references therein) (OI of Fig 9) The anomaliesC2 and B3 exhibit weaker amplitudes than B1 and B2probably related to a deeper location of the intrusivesassuming that they represent a continuation of the Kumasibatholith that outcrops to the north or different and minormagnetic properties if these anomalies are really caused bythe granodiorites of the Bansu intrusion which outcrops in awide area to the southwest of the lake and on its shore atBansu (Koeberl and Reimold 2005)

Kontny et al (2007) performed rock magnetic andmagnetic mineralogy analysis on rock samples from LB-07Aand LB-08A that established ferrimagnetic pyrrhotite as themain magnetic carrier for both target and impact lithologies atBosumtwi Shock-induced remagnetization of the pre-impactpyrrhotite was presented as a mechanism to enhance themagnetization parameters of target and impact lithologiesfrom the values measured on material from the environs of theimpact structure (Plado et al 2000) to the values required bymagnetic modeling by Plado et al (2000) and this work Thismechanism is feasible considering the low Curie temperatureof pyrrhotite (~320deg C) However pyrrhotite alsodemagnetizes when subject to modest shock pressures (P lt4 GPa) (Louzada et al 2005) Thus for this mechanism to beeffective the area of thermal-remagnetization would have tobe larger than the area of shock-demagnetization In any case

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 879

the Q-values measured by Kontny et al (2007) are muchhigher than 1 indicating that remanent magnetizationdominates over induced magnetization which could alsoexplain why the magnetic susceptibility logs on LB-08A andLB-07A do not show values as high as the magnetic modelsrequire Considering the age of the structure the remanentmagnetization vector has to be subparallel to the inducing

field therefore higher Q-values translate into really higheffective susceptibilities However if pyrrhotite was thesource of the observed marine and airborne magneticanomalies then consistent magnetic anomalies would beobserved in the borehole magnetic logs from LB-08A Morriset al (2007) did not find any significant magnetic anomalieson the borehole magnetic data and furthermore explain the

Fig 8 Amplitude of the analytical signal from the airborne magnetic data set superimposed onto the regional geology around the Bosumtwiimpact structure The granitoids of the Kumasi batholith are marked KB biotite and hornblende granodiorites of the Bansu intrusion aremarked BI The main trends in the magnetic data are marked as dotted lines and the main magnetic anomalies are numbered for easieridentification See text for details

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

880 H Ugalde et al

observed anomalies with a model consisting of onemagnetized body towards the northeast of the lake and thatbarely intersects LB-08A at ~300 m depth to continue as anoff-hole source downwards

Consequently to obtain a definite explanation for thesource of the observed magnetic anomalies in thenortheastern sector of the crater more information is required

on the magnetic mineralogy and properties of the rocksoccurring to the northeast of the lake and on the Kumasibatholith and Bansu intrusion located to the northeast andsouthwest of the lake respectively

Meteorite impacts can greatly modify the magneticproperties of the target rocks (Ugalde et al 2005 Henkel andReimold 2002) either enhancing them (eg higher magnetic

Fig 9 The amplitude of the analytical signal from the marine magnetic superimposed onto the geology on the environs of the Bosumtwiimpact crater KB = granitoids of the Kumasi batholith BI = biotite and hornblende granodiorites of the Bansu intrusion PG = Pepiakesegranite and diorite OI = olivine pyroxenite dolerite and gabbro The magnetic anomalies interpreted on Fig 4 are marked A1ndashA3 B1ndashB3and C1ndashC2 The ICDP deep borehole sites are marked by red crosses The main magnetic trends interpreted from the airborne magnetic dataset are marked as dotted lines See text for details

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

The Lake Bosumtwi meteorite impact structuremdashWhere is the magnetic source 881

susceptibility because of shock decomposition of originalminerals into more magnetic phases like the shockdecomposition of biotite into magnetite at 40 GPa [Chao1968] increase of NRM by adding a secondary componentlike shock remanent magnetization [SRM] at 30 GPa [Halls1979] increase of NRM by melting and subsequent coolingunder the presence of a magnetic field) or reducing them(eg magnetic susceptibility reduction at P lt 10 GPa[Pilkington and Grieve 1992] shock demagnetization at P lt1 GPa [Cisowski and Fuller 1978]) In the case of LakeBosumtwi the maximum shock pressures induced by theimpact apparently did not reach the high values predicted bynumerical modeling (Artemieva et al 2004) and indeedreached maximum values of P lt 29 GPa in the ICDP boreholeregion (Deutsch 2006) Considering the spherical propagationof the shock wave its decay as rminus3 and the location of theICDP boreholes relative to the impact structure (near center)the maximum shock pressures reached by the impact shouldnot be higher than that measured in the ICDP borehole regionIndeed analysis of shock metamorphism on samples from thegranitoids of the Pepiakese body at the northeast of the lakeshowed only minor fracturing of possible shock origin andshock pressure was estimated as P lt 8 GPa (W U Reimoldpersonal communication) In all likelihood the magneticproperties of the intrusives that are the source of the observedmagnetic anomalies did not change significantly except forthe remagnetization of the pre-impact formed pyrrhotite(Kontny et al 2007) which however occurs in volumes thatare not sufficiently large (mostly below 1 vol Kontny et al2007 W U Reimold personal communication) to have anoticeable effect on the measured total field either frommarine airborne or borehole platforms Consequently and ascan be observed on the amplitude of the analytic signal fromthe airborne total field on Fig 8 the magnetization patternsobserved on a macroscopic scale (airborne marine borehole)are similar inside and outside the crater

This rejects Plado et alrsquos (2000) hypothesis about thePepiakese granite (PG of Fig 9) as a probable indicator of theavailability of biotite-rich rocks that could have decomposedinto magnetite due to the impact However the Pepiakesegranitoid also contains a dioritic phase (W U Reimoldpersonal communication) which could have higher magneticproperties than what was measured by Plado et al (2000) onits granitic phase and that is not sufficiently high to generatethe observed magnetic anomalies to the northwest andsoutheast of the lake

CONCLUSIONS

Synthetic magnetic profiles of a body at equatoriallatitudes were helpful in showing that the magnetic anomaliesat low magnetic latitudes show distinct features which help inlocating the source body In plan view they show the centralnegative anomaly with asymmetric northern and southern

flank anomalies The amplitude of the analytic signal wasvaluable in locating the source of the observed magneticanomalies

Based on the high-resolution marine magnetic dataacquired at Lake Bosumtwi a new 3-D magnetic model wasconstructed for the Bosumtwi structure The model has thesame geometry as a 3-D gravity model of the structure and itis consistent with previous seismic models and thepetrophysical data from cores obtained from the ICDP deepholes LB-07A and LB-08A The model consists of a stack of3-D bodies with moderate magnetic properties across thelake but highly magnetic bodies located to the northeast ofthe lake are the most significant feature Integration of thenew 3-D model the newly published geological map of thestructure the ICDP borehole petrophysics and geology andthe airborne and marine magnetic data sets allowed tointerpret the observed magnetic anomalies based on themapped geology The magnetic anomalies on the lake arelikely caused by granodiorites of the Kumasi batholith orgranodiorites of the Pepiakese intrusion There is no need forhighly magnetic melt volumes which were not confirmedthrough the ICDP boreholes In order to verify this detailedmagnetic scanning of the Kumasi and Bansu intrusions isrequired

AcknowledgmentsndashDrilling at Bosumtwi was supported bythe International Continental Scientific Drilling Program(ICDP) the US NSF-Earth System History Program undergrant no ATM-0402010 Austrian FWF (project P17194-N10) the Austrian Academy of Sciences and by theCanadian Natural Sciences and Engineering ResearchCouncil (NSERC) Drilling operations were performed byDOSECC Local help by the Geological Survey Department(Accra) and the Kwame Nkrumah University of Science andTechnology (KNUST Kumasi) Ghana were invaluable Wethank the associate editor W U Reimold and A Kontny forhelpful comments and suggestions that really improved themanuscript

Editorial HandlingmdashDr Wolf Uwe Reimold

REFERENCES

Artemieva N Karp T and Milkereit B 2004 Investigating the LakeBosumtwi impact structure Insight from numerical modelingGeochemistry Geophysics Geosystems 51ndash20

Boamah D and Koeberl C 2003 Geology and geochemistry ofshallow drill cores from the Bosumtwi impact structure GhanaMeteoritics amp Planetary Science 381137ndash1159

Chao E C T 1968 Pressure and temperature histories of impactmetamorphosed rocksmdashBased on petrographic observations InShock metamorphism of natural materials edited by B MFrench and N M Short Baltimore Mono Book Corp pp135ndash158

Cisowski S M and Fuller M 1978 The effect of shock on themagnetism of terrestrial rocks Journal of Geophysical Research833441ndash3458

882 H Ugalde et al

Coney L Gibson R L Reimold W U and Koeberl C 2007Lithostratigraphic and petrographic analysis of ICDP drill coreLB-07A Bosumtwi impact structure Ghana Meteoritics ampPlanetary Science 42 This issue

Danuor S K 2004 Geophysical investigations of the Bosumtwiimpact crater and comparison with terrestrial meteorite craters ofsimilar age and size PhD thesis Kwame Nkrumah Universityof Science and Technology Kumasi Ghana

Deutsch A 2006 Lake Bosumtwi drilling project Shockmetamorphism in rocks from core BCDP-8A vs experimentaldata (abstract 1327) 37th Lunar and Planetary ScienceConference CD-ROM

Deutsch A Luetke S and Heinrich V 2007 The ICDP LakeBosumtwi impact crater scientific drilling project (Ghana) CoreLB-08A litho-log related ejecta and shock recoveryexperiments Meteoritics amp Planetary Science 42 This issue

Dressler B O Sharpton V L Morgan J Buffler R Moran DSmit J Stoeffler D and Urrutia-Fucugauchi J 2003Investigating a 65 Ma old smoking gun Deep drilling of theChicxulub impact structure Eos Transactions 84125ndash130

Elbra T Kontny A Pesonen L J Schleifer N and Schell C 2007Petrophysical and paleomagnetic data of drill cores from theBosumtwi impact structure Ghana Meteoritics amp PlanetaryScience 42 This issue

Ferriegravere L Koeberl C and Reimold W U 2007 Drill core LB-08ABosumtwi impact structure Ghana Petrographic and shockmetamorphic studies of material from the central upliftMeteoritics amp Planetary Science 42 This issue

Henkel H and Reimold W U 2002 Magnetic model of the centraluplift of the Vredefort impact structure South Africa Journal ofApplied Geophysics 5143ndash62

Halls H C 1979 The Slate Islands meteorite impact site A study ofshock remanent magnetization Geophysical Journal of the RoyalAstronomical Society 59553ndash591

Jones W B Bacon M and Hastings D A 1981 The LakeBosumtwi impact crater Ghana Geological Society of AmericaBulletin 92342ndash349

Junner N R 1937 The geology of the Bosumtwi caldera andsurrounding country Gold Coast Geological Survey Bulletin 81ndash38

Karp T Milkereit B Janle J Danuor S K Pohl J Berckhemer Hand Scholz C A 2002 Seismic investigation of the LakeBosumtwi impact crater Preliminary results Planetary SpaceScience 50735ndash743

Koeberl C Bottomley R Glass B P and Storzer D 1997Geochemistry and age of Ivory Coast tektites and microtektitesGeochimica et Cosmochimica Acta 611745ndash1772

Koeberl C and Reimold W U 2005 Bosumtwi impact crater Ghana(West Africa) An updated and revised geological map withexplanations Jahrbuch der Geologischen Bundesanstalt Wien(Yearbook of the Austrian Geological Survey) 14531ndash70(+1 map 150000)

Koeberl C Milkereit B Overpeck J T Scholz C A AmoakoP Y O Boamah D Danuor S Karp T Kueck J Hecky R EKing J W and Peck J A 2007 An international andmultidisciplinary drilling project into a young complex impactstructure The 2004 ICDP Bosumtwi Crater Drilling ProjectmdashAn overview Meteoritics amp Planetary Science 42 This issue

Kontny A and Just J 2006 Magnetic mineralogy and rock magneticproperties of impact breccias and crystalline basement rocksfrom the BCDP-drillings 7A and 8A (abstract 1343) 37th Lunarand Planetary Science Conference CD-ROM

Kontny A Elbra T Just J Pesonen L J Schleicher A M andZolk J 2007 Petrography and shock-related remagnetization ofpyrrhotite in drill cores from the Bosumtwi Impact Crater

Drilling Project Ghana Meteoritics amp Planetary Science 42This issue

Louzada K L Stewart S T and Weiss B P 2005 Shockdemagnetization of pyrrhotite (abstract 1134) 36th Lunar andPlanetary Science Conference CD-ROM

MacLeod I N Jones K and Dai T F 1993 3-D analytic signal inthe interpretation of total magnetic field data at low magneticlatitudes Exploration Geophysics 24679ndash688

Milkereit B Ugalde H Karp T Scholz C A Schmitt D DanuorS K Artemieva N Kuek J Qian W and LrsquoHeureux E 2006Exploring the Lake Bosumtwi cratermdashGeophysical surveyspredictions and drilling results (abstract 1687) 37th Lunar andPlanetary Science Conference CD-ROM

Moon P A and Mason D 1967 The geology of 14deg field sheets nos129 and 131 Bompata SW and NW Ghana Geological SurveyBulletin 311ndash51

Morris W A Ugalde H and Clark C 2007 Physical propertymeasurements ICDP boreholes LB-07A and LB-08A LakeBosumtwi impact structure Ghana Meteoritics amp PlanetaryScience 42 This issue

Nabighian M 1972 The analytic signal of two-dimensionalmagnetic bodies with polygonal cross-section Its properties anduse for automated anomaly interpretation Geophysics 37507ndash517

Pesonen L J Koeberl C Reimold W U and Brandt D 1997 Newstudies of the Bosumtwi impact structure Ghana The 1997 fieldseason (abstract) Meteoritics amp Planetary Science 32A72ndash73

Pesonen L J Koeberl C and Hautaniemi H 2003 Airbornegeophysical survey of the Lake Bosumtwi meteorite impactstructure (southern Ghana)mdashGeophysical maps withdescriptions Jahrbuch der Geologischen Bundesanstalt Vienna(Yearbook of the Austrian Geological Survey) 143581ndash604

Pilkington M and Grieve R A F 1992 The geophysical signatureof terrestrial impact craters Reviews of Geophysics 30161ndash181

Plado J Pesonen L J Koeberl C and Elo S 2000 The Bosumtwimeteorite impact structure Ghana A magnetic modelMeteoritics amp Planetary Science 35723ndash732

Reimold W U Brandt D and Koeberl C 1998 Detailed structuralanalysis of the rim of a large complex impact crater Bosumtwicrater Ghana Geology 26543ndash546

Roest W Verhoef J and Pilkington M 1992 Magneticinterpretation using the 3-D analytic signal Geophysics 57116ndash125

Schleifer N H Elster D and Schell C M 2006 Petrophysicalcharacterization of the core samples of the Bosumtwi meteoriteimpact crater (Ghana) (abstract 1273) 37th Lunar and PlanetaryScience Conference CD-ROM

Scholz C A Karp T Brooks K M Milkereit B Amoako P Y Oand Arko J A 2002 Pronounced central uplift identified in theBosumtwi impact structure Ghana using multichannel seismicreflection data Geology 30939ndash942

Spector A and Grant F S 1970 Statistical models for interpretingaeromagnetic data Geophysics 35293ndash302

Ugalde H A Artemieva N and Milkereit B 2005 Magnetizationon impact structuresmdashConstraints from numerical modeling andpetrophysics In Large meteorite impacts III edited byKenkmann T Houmlrz F and Deutsch A GSA Special Paper 384Boulder Colorado Geological Society of America pp 25ndash42

Ugalde H 2006 Geophysical signature of midsize to small terrestrialimpact craters PhD thesis University of Toronto TorontoOntario Canada

Ugalde H Danuor S K and Milkereit B 2007 Integrated 3-Dmodel from gravity and petrophysical data at Lake BosumtwiGhana Meteoritics amp Planetary Science 42 This issue