INTRODUCTIONThe geologic structure and macrodeformation

associated with some small (<2 km), well-preserved, simple bowl shaped impact structures(Meteor Crater [Arizona], the Odessa crater[Texas], and the Pretoria Saltpan crater [SouthAfrica]) have been studied before (Brandt andReimold, 1995, and references therein). In con-trast, most larger, complex impact structures aretoo deeply eroded to allow detailed structuralanalysis of their crater rims. The young (1.07 Ma,Koeberl et al., 1997) Bosumtwi impact crater,located at 06°32′N, 01°25′W, about 30 km south-east of Kumasi in the Ashanti Province of Ghanais still well preserved (Fig. 3A). Although thestructure of the central part of this impact crateris completely obscured by Lake Bosumtwi, itmust be assumed, on the basis of the existingterrestrial impact crater record (cf. Grieve andPilkington, 1996) that Bosumtwi represents acomplex impact crater. At 10.5 km in diameter, itshould feature a collapsed central uplift structure(Melosh, 1989).

The origin of Bosumtwi has been controver-sial for several decades, but since the 1960s, sev-eral lines of evidence have supported an impactorigin (cf. Jones et al., 1981; Koeberl et al.,1997). The Bosumtwi impact crater has also beenassociated with the Ivory Coast tektite strewn-field, on the basis of similarities in chemical andisotopic composition, and identical ages fortektites and crater glasses (Koeberl et al., 1997,and references therein).

No structural geologic data have been avail-able so far, because of the extensive cover of thewhole region by rainforest and lateritic soils.

Bedrock and impact formations outside of thecrater are only accessible along road cuts and instream beds. In January 1997 we studied a seriesof fresh road cuts, produced in 1993 in the courseof complete reconstruction of the access road tothe crater lake through the northwestern crater-rim section. These exposures allowed detailedstructural analysis along a 7-km-long traversethrough the crater rim (Figs. 1 and 2). Two other,less complete cross sections were studied alonggravel roads in the northeastern and southernsectors of the rim and generally confirmed thefindings along the northwest traverse. Theseresults provide, for the first time, a detailedassessment of the deformation associated withthe crater rim of a complex impact structure.

GEOLOGY OF THE BOSUMTWI AREAA schematic geologic map of the area (Fig. 1)

shows that the crater is nearly completely filled byLake Bosumtwi and talus sediments in itsenvirons. The rim-to-rim crater diameter is about10.5 km. The main bedrock strata are metasedi-mentary rocks in greenschist facies belonging tothe Birimian Supergroup, dated at 2.1–2.2 Ga(Wright et al., 1985; Leube et al., 1990; Hirdesetal., 1996). To the southeast of the crater, someTarkwaian metasedimentary rocks occur as well.The regional geology is characterized by a strong,northeast-trending fabric with steep dips to eitherthe northwest or the southeast. Variations in thistrend, due to folding, were observed on a localscale. Lithology at and around Lake Bosumtwi isdominated by metagraywackes and metasand-stones, but some shale and mica schist were alsoobserved, especially in the northeastern and

southern rim sectors. A variety of granitoid intru-sions (mainly biotite or amphibole granites) wasmapped by Junner (1937) and Moon and Mason(1967). In addition, we observed numerous, butgenerally <1-m-wide, dikes of biotite granitoid atmany basement exposures in the crater rim. Manyof these dikes have vertical to subvertical orienta-tions. The overall granitoid component in theregion is estimated at about 2%.

Numerous breccia exposures have beenmapped around the crater in the past (e.g., Junner,1937; Moon and Mason, 1967; compare Fig. 1).Our investigations confirm the presence of suevitedeposits to the north and southwest of the crater. Itis, however, not certain at this stage whether any ofthe mapped occurrences of lithic (clastic) brecciarepresent impact breccia. Extensive sedimento-logical work is needed to establish how much ofthe reported breccias represents impact ejecta or isthe result of lateritization and secondary mass-wasting processes in this tropical and topographi-cally varied environment. Along the northwesttraverse through the crater rim, Birimian meta-graywacke and shale are the dominant rock types,besides minor occurences of shale and some dike-and pod-like granitoid intrusions.

STRUCTURAL ANALYSISLineament analysis on both Thematic Map-

per5 (TM5) Landsat and SPOTimagery con-firms the field observations that the crater istransected by numerous, generally steeply dip-ping radial fractures and/or faults. Pairs oflineaments commonly occur in a scissors-likearrangement. Concentric lineaments are promi-nent, but not very abundant. Field observations

Geology;June 1998; v. 26; no. 6; p. 543–546; 3 figures. 543

Detailed structural analysis of the rim of a large, complex impact crater: Bosumtwi Crater, Ghana

Wolf U. ReimoldDion Brandt

Department of Geology, University of the Witwatersrand, Private Bag 3, P.O. Wits 2050, Johannesburg, South AfricaChristian Koeberl

Institute of Geochemistry, University of Vienna, Althanstr. 14, A-1090 Vienna, Austria

ABSTRACTThe 1 Ma Bosumtwi Crater in Ghana is an 11-km-diameter, presumably complex, well-

preserved impact structure that is associated with the Ivory Coast tektite strewnfield. Detailedstructural geologic studies along a complete traverse through the northwestern rim section indi-cated fourzones characterized by distinct deformation styles from just outside of the craterrimto nearthe crater floor. Zone 1 is dominated by thick deposits of lithic impact breccia, intercalatedin places with products of local mass wasting. Zone 2 contains inward-dipping thrust planes,conjugate radial fractures, isoclinal folding, and overturned stratigraphic sequences. Zone 3represents a megabreccia zone, in which block size decreases upward and outward toward therim cr est. The innermost zone 4 is dominated by intense thrust faulting of multiple orientations,resulting in complex duplex- and lens-shaped bodies. These deformation styles generally corre-spond to those previously reported from the rims of simple bowl-shaped meteorite-impact cratersand appearto be characteristic of impact structures in general.

strongly suggest that inward-directed slumpingoccurred, possibly along concentric and listricfaults. The rim shape is complex (Fig. 2); the rimhas steep inner and outer slopes, and significantdepressions occur below the main rim crest—

presumably the result of faulting in the rim. Atthe inner end of the traverse, a relatively gentletalus slope is exposed around the lake (comparealso Fig.3A).

Earlier workers (e.g., Jones et al., 1981) pro-

posed that outside of the crater rim, about 1.5 kmfrom the rim crest, a shallow depression of about4 to 5 km width was surrounded by another slight(30–80 m) ring-shaped elevation. Our topo-graphic studies support the presence of a wide,shallow depression in the northern and north-eastern environs of the crater. This feature couldbe the result of formation of a ring depression dur-ing transient cavity collapse, or of preferentialerosion of impact ejecta along several streambeds, which follow the crater rim in a semicon-centric pattern. Jones et al. (1981) discussed thepossibility that this ring morphology could berelated to the multiple-ring features observed, forexample, around the Ries Crater (Germany).However, we favor a different scenario: a ringfracture, formed during the cratering phase, led tothe development of the specific drainage patternobserved around Bosumtwi Crater, which, in turn,caused the formation of a semiannular erosionchannel. The terrane farther away from the craterdoes not display any evidence of cratering-induced structural control on the drainage pattern.

The presence of shatter cones at Bosumtwi,which had been suggested by Rohleder (1934),could not be confirmed by us, but some curvi-

544 GEOLOGY, June 1998

Figure 1. Geologic map of area ar ound Lake Bosumtwi, with location of tra verse (A–B) along access r oad to the crater lake thr oughnor thwestern crater rim. Dashed line A–B: crater rim section based on obser vations along winding r oad, but pr ojected onto a straig htcross section (compare Fig. 2).

Figure 2. Zonation of nor thwestern crater rim along A–B tra verse (Fig. 1). Zone 1: dominantl yejecta breccia co ver and local mass wasting. Zone 2: inwar d-dipping thrust planes, conjugateradial fractures, isoc linal f olding, and over turned stratigraph y. Zone 3: megabreccia with upwar d-decreasing b loc k siz es. Zone 4: dominant thrust faulting of m ultiple orientations, resulting induple x- and lens-shaped bodies of pre-impact strata.

planar joints, partially displaying slickensides,were observed. Furthermore, melt-breccia occur-rences were also not found, whereas abundantlithic (clastic) breccias were observed. None ofthe >60 crater-rim samples collected by us dis-plays characteristic shock-metamorphic defor-mation. However, numerous metasedimentary orgranitic clasts in suevite samples exhibit shockeffects, such as single or multiple sets of planardeformation features (PDFs) or the presence ofdiaplectic quartz glass, that are characteristic forshock metamorphism at >10 to <35 GPa. Muchof the clastic component in melt fragments fromsuevite was derived from basement granites.Granitoid intrusions into the crater rim oftenexhibit, in microscopic patches, granophyric orspherulitic textures. However, lack of shock-

metamorphic and chemical evidence for their or-igin by impact melting leads us to prefer an originfor these dikes by pre-impact intrusion of a rela-tively hot felsic magma, which incorporatedBirimian country rocks and locally melted thesebiotite (volatile)–rich xenoliths.

At more than 60 stations, structural observa-tions were obtained along traverses through thecrater rim, in a series of distinct zones, which wewill discuss in detail for the northwesterlytraverse: Exposures in zone 1 (Fig. 2) are domi-nated by occurrences of breccias composedmainly of shale (at individual outcrops, >90vol%) and fine- to medium-grained metasand-stone and metagraywacke, with minor (<10vol%) schist. Most clasts are strongly weathered.Clasts may be as wide as 1.5 m, but they are gen-

erally smaller than 5 cm. Some larger clastsexhibit internal brecciation. Clast shapes are vari-able, ranging from angular to well rounded. Weinterpret this material as a mixture of lithic (frag-mental) impact breccia and locally accumulatedproducts of secondary mass-wasting processesalong the steeply dipping outer-rim slopes.

In zone 2, brittle deformation is ubiquitous.This regime consists mainly of highly fracturedrocks, which commonly exhibit conjugate radialfractures. Quartz veining occurs along manyfractures and joints of numerous orientationsand—together with some quartz-filled extensiongashes—presumably resulted from pre-impactdeformation and hydrothermal activity. Meta-graywacke and metasandstone are the dominantlithologies, but shale (up to several meters wide)

GEOLOGY, June 1998 545

A B

C D

E F

Figure 3. A: View of nor th-ern crater -rim section fr omlakeshore , demonstratingradial macr ofaulting andmegab loc k slumping to-ward crater interior . B: Rimsection zone 2; thrustplane dipping inwar d (toleft) and cutting Birimianmetagraywac ke. C: Zone2; isoc linal f olding in strat-ified Birimian metasedi-mentar y rocks. D: Zone3, near highest point ofrim; lithic impact brecciacomposed of centimeter -to decimeter -sized clasts.E: Zone 3; megabrecciawith meter -sized blocks ofBirimian metasedimentar yrocks and (lightest-colored)2 Ga granitoids; width offield of vie w = 10 m. F:Duple x- and lens-shapedshale and meta graywackebloc ks between v ariab lyoriented thrust faults;width of field of vie w = 8 m.

was also noted in some exposures. Some shalecontains significant amounts of carbonaceousmatter. Outward-thrusting faults (Fig. 3B) typi-cally dip inward at 40°–45°. Orientation of bed-ding is highly variable; it is generally inward dip-ping at angles between 20° and vertical, but atseveral locations it is outward dipping or evenoverturned. Symmetric gentle and open foldswere noted, but symmetric close and isoclinal(Fig. 3C) folds are more common. Faultingexploits less competent shale layers, resulting inbedding-parallel slip. Slickensides are preservedon some fault planes in metasedimentary rocksand generally confirm upward and outwardthrust movements.

On top of the rim, in zone 3 (Fig. 3D), ejectabreccia occurs. It consists of generally angularblocks of metasandstone and metagraywacke(with minor shale) of <60 cm clast sizes. Recentmass wasting, for example, in the form of rock-falls and debris slides, is related to the steepslope in this zone. Descending farther down theinner slope of the rim, rock fragments increase insize to tens of meters (Fig. 3E). Such mega-blocks are surrounded by relatively finer-grainedbreccia (clast sizes to tens of centimeters). It ispossible that some megablocks in the lower partsof this zone may still be in situ. Mostmegablocks are faulted and display internalbrecciation, on a centimeter-to-decimeter scale.In the lower parts of this zone, megablocks com-monly have bedding dips toward the inner craterand, often, are separated by dike-like zones offiner-grained clastic breccias. Main lithologiesrepresented in this zone are metagraywacke,minor shale, and an appreciable amount of pod-and dikelike granitoid intrusions.

Zone 4 is characterized by inward-dipping(~50°) shear zones and faults, which appear to berelated to thrusting. Vertically oriented blockswere also noted. Bedding-parallel faults, dippingat ~45° to the south-southeast are common in thiszone. Breccia lenses as much as 6 × 2 m in sizewere noted. The irregular orientations of target-rock blocks are attributed to complex thrust fault-ing as well as block rotation in this lower part ofthe crater rim. Duplex- and lensoid-shaped (oftenpinching and swelling) blocks (Fig. 3F) are abun-dant in this zone. Normal and antithetic faulting,related to slumping along the crater rim duringthe late modification stage of the cratering event(Melosh, 1989), was noted along this steep andunstable inner slope. Small-scale faults and folds,possibly related to an earlier (either pre-impactregional or early impact—i.e., shock compres-sional) phase of deformation, occur within largeupturned or rotated blocks. Some of these blocksare intensely fractured (shattered). Toward thelowermost inner part of the rim, blocks of allregional lithologies are highly fractured andsheared and locally internally brecciated. Thebedding attitudes observed in this strongly de-formed zone typically dip gently (<15°) inward.

COMPARISON WITH DEFORMATIONSTYLES AND ZONES IN SIMPLEBOWL-SHAPED IMP ACT CRATERS

Structural observations have been reportedfrom the crater rim of the 1.13-km-diameterPretoria Saltpan (Tswaing) meteorite crater inSouth Africa (Brandt and Reimold, 1995) andfrom the rims of the Odessa (Texas) and Meteor(Arizona) impact craters (Shoemaker, 1960;Shoemaker and Eggleton, 1961). In the cases ofboth bowl-shaped and complex impact struc-tures, ejecta breccia covers the top and outer zoneof the crater rim. In the lowermost inner zone,inward-dipping bedding orientations prevail,whereas upper parts of the inner slopes are char-acterized by highly variable bedding orientations,including overturning. On inward-dipping faultplanes, outward-directed thrusting is indicated. Incontrast, a deformation zone 2—as documentedat Bosumtwi—has not been observed in thesesmaller structures. Megablocks have not beenobserved in small craters, and anticlines are typi-cally observed in small craters in the middle partsof the inner rim.

Impact structures larger than Bosumtwi (e.g.,the 24-km-diameter Ries Crater, Germany) areknown to have well-developed terrace systemsalong the outer rim. This feature of large impactcraters is consistent with our observation ofblocks that apparently slumped inward alongsteeply oriented fault planes.

CONCLUSIONSWe have demonstrated that similar deformation

styles characterize crater rims of small (simple,bowl-shaped) and larger (complex) impact struc-tures. In particular, outward-directed thrust fault-ing and intense folding characterize the middle toupper parts of impact crater rims. In the light of thegenerally accepted sequence of impact-crateringstages (Melosh, 1989), the following generaliza-tion can be made: Most observed deformationstructures appear to be related to the cratering(compression and excavation) phase of the impactevent. For example, fold structures such as thosereported here from Bosumtwi Crater and earlier byBrandt and Reimold (1995), have been discussedby Gault etal. (1968) as the result of initialoutward-directed motion, during which rocks arefirst driven horizontally outward and thendeflected upward toward the surface. Only theantithetic normal faulting observed primarily onthe inner-rim slopes of the crater seems to berelated to postcratering modification of the highlyunstable rim.

ACKNOWLEDGMENTSThe support of C.E. Oduro, Director of the Geo-

logical Survey of Ghana, and G. O. Kesse, formerDirector of this institution, and the field assistance ofK. Atta-ntin were instrumental to the success of ourfield study. Landsat imagery was kindly provided by

Gold Fields of South Africa Limited. Supported by theAustrian Fonds zur Förderung der wissenschaftlichenForschung, Projects P09021-GEO and START Y-58(Koeberl), and by a Foundation for Research Develop-ment of South Africa grant (Reimold). Walter Alvarezprovided a much appreciated review.

REFERENCES CITEDBrandt, D., and Reimold, W. U., 1995, The geology of

the Pretoria Saltpan impact structure and the sur-rounding area: South African Journal of Geology,v. 98, p.287–303.

Gault, D.E., Quaide, W. L., and Oberbeck, V. R., 1968,Impact cratering mechanics and structures, inFrench, B.M., and Short, N.M., eds., Shockmetamorphism of natural materials: Baltimore,Maryland, Mono Book Corporation, p.87–99.

Grieve, R.A . F., and Pilkington, M., 1996, The sig-nature of terrestrial impacts: AGSO Journal ofGeology and Geophysics, v. 16, p.399–420.

Hirdes, W., Davis, D.W., Lüdtke, G., and Konan, G.,1996, Two generations of Birimian (Paleo-proterozoic) volcanic belts in northeastern Côted’Ivoire (West Africa): Consequences for the“Birimian controversy”: Precambrian Research,v. 80, p.173–191.

Jones, W.B., Bacon, M., and Hastings, D.A., 1981, TheLake Bosumtwi impact crater, Ghana: GeologicalSociety of America Bulletin, v. 92, p.342–349.

Junner, N. R., 1937, The geology of the Bosumtwicaldera and surrounding country: Gold CoastGeological Survey Bulletin, v. 8, p.1–38.

Koeberl, C., Bottomley, R.J., Glass, B.P., and Storzer,D., 1997, Geochemistry and age of the IvoryCoast tektites and microtektites: Geochimica etCosmochimica Acta, v. 61, p.1745–1772.

Leube, A., Hirdes, W., Mauer, R., and Kesse, G. O.,1990, The Early Proterozoic Birimian Super-group of Ghana and some aspects of its associ-ated gold mineralization: Precambrian Research,v. 46, p.136–165.

Melosh, H.J., 1989, Impact cratering—Ageologic proc-ess: New York, Oxford University Press, 245p.

Moon, P.A., and Mason, D., 1967, The geology of 1/4°field sheets nos.129 and 131, Bompata S.W. andN.W.: Ghana Geological Survey Bulletin, v. 31,p.1–51.

Rohleder, H. P. T., 1934, Über den Fund vonVergriesungs-Erscheinungen und Druckstruk-turen am Kesselrand des kryptovulkanischenBosumtwi-Sees, Ashanti: Zentralblatt für Min-eralogie, Geologie, und Paläontologie, Jahrgang1934, p.316–318.

Shoemaker, E.M., 1960, Penetration mechanics of highvelocity meteorites, illustrated by Meteor Crater,Arizona: International Geological Congress, 21st,Copenhagen, Det Berlingske Bogtrykkeri, v. 18,p.418–434.

Shoemaker, E.M., and Eggleton, R.E., 1961, Terrestrialfeatures of impact origin, in Milo, M. D., ed., Pro-ceedings of the geophysical laboratory/LawrenceRadiation Cratering Symposium, p.A1–A27.

Wright, J. B., Hastings, D.A., Jones, W. B., andWilliams, H.R., 1985, Geology and mineral re-sources of West Africa: London, George Allen &Unwin, p.38–45.

Manuscript received November 17, 1997Revised manuscript received March 19, 1998Manuscript accepted March 20, 1998

546 Printed in U.S.A. GEOLOGY, June 1998