Late Quaternary tectonic activity and lake level change in the ......Late Quaternary tectonic...

PergamonJournal of African Earth Sciences, Vol. 26. No.3. pp. 397-421, 1998

c 1998 Elsevier Science LtdAll rights reserved. Printed in Great Britain

PII:S0899-5362(98)00023-2 0899-5362/98 $19.00 + 000

Late Quaternary tectonic activity and lake levelchange in the Rukwa Rift Basin

D. DELVAUX,1 F. KERVYN,l E. VITTORI,2 R. S. A. KAJARNand E. KILEMBE4

1Royal Museum for Central Africa, B-3080 Tervuren, Belgium2ANPA, Via Vitaliano Brancati, 48, 00144 Roma, Italy

3MADINI, Box 903 Dodoma, Tanzania4TPDC, Exploration-Production, Box 2774 Dar-es-Salaam, Tanzania

Abstract-Interpretation of remotely sensed images and air photographs, compilationof geological and topographical maps, morphostructural and fault kinematicobservations and 14C dating reveal that, besides obvious climatic influences, thelake water extent and sedimentation in the closed hydrological system of LakeRukwa is strongly influenced by tectonic processes. A series of sandy ridges,palaeo lacustrine terraces and palaeounderwater delta fans are related to an EarlyHolocene high lake level and subsequent progressive lowering. The maximum lakelevel was controlled by the altitude of the watershed between the Rukwa andTanganyika hydrological systems. Taking as reference the present elevation of thepalaeolacustrine terraces around Lake Rukwa, two orders of vertical tectonicmovement are evidenced:i) a general uplift centred on the Rungwe Volcanic Province between the Rukwa andMalawi Rift Basins; andii) a tectonic northeastward tilting of the entire Rukwa Rift Basin, including thedepression and rift shoulders.This is supported by the observed hydromorphological evolution. Local uplift is alsoinduced by the development of an active fault zone in the central part of thedepression, in a prolongation of the Mbeya Range-Galula Fault system. The Ufipaand Lupa Border Faults, bounding the Rukwa depression on the southwestern andnortheastern sides, respectively, exert passive sedimentation control only. Theyappear inactive or at least less active in the Late Quaternary than during the previousrifting stage. The main Late Quaternary tectonic activity is represented by dextralstrike-slip movement along the Mbeya Range-Galula Fault system, in the middle ofthe Rukwa Rift Basin, and by normal dip-slip movements along the Kanda Fault, inthe western rift shoulder. © 1998 Elsevier Science Limited.

Resume-L'interpretation d'imageries satellitaires et de photographies aeriennes. lacompilation de cartes geologiques et topographiques, des observationsmorphologiques et cinernatiques des failles et des datations au 14C montrent que,en dehors des influences climatiques evidentes. l'etendue lacustre et la sedimentationdans Ie svsterne hydrologique ferrne du lac Rukwa sont influences par la tectonique.Une serie de rides sablonneuses (sandy ridges), de terrasses et de deltas souslacutres fossiles sont mis en relation avec un ancien haut niveau du lac Rukwa auHolocene precoce, et son abaissement progress if. Le niveau lacustre maximum eta itcontrole par I' altitude du seuil de partage des eaux entre les bassins de Rukwa etTanganyika. En prenant comme reference l'elevation des terrasses paleolacustresen bordure du bassin de Rukwa, deux types de mouvements verticaux sont mis enevidence:i) un soulevement en forme de dome, centre sur la Province volcanique du Rungwe,entre les bassins de Rukwa et de Malawi; etii) un basculement vers Ie nord-ouest de I'ensemble du bassin de Rukwa, y compris

Journal of African Earth Sciences 397

D. DELVAUX et al.

avec les epaules de rift, mais sans mouvement differentiels entre les epaules et ladepression.Ceci est deduit de I'observation de l'evolution hydro-morphologique. Un soulevernentlocal est egalement associe aune zone de faille active dans Ie centre de la depression,en prolongement du svsterne faille du Mbeya Range-Galula. Les failles bordieres deUfipa et l.upa, qui limitent la depression de Rukwa, respectivement au sud-ouest etau nord-est, n'exercent qu'un controls passif de la sedimentation. Elles apparaissentinactives ou moins actives durant Ie Quaternaire tardif que dans Ie stade de rittoqeneseprecedent. l.'ectivite recente principale est representee par des mouvementsdecrochants dextres Ie long du svsterne faille de Mbeya Range-Galula, au centre dubassin de Rukwa, et par des mouvements normaux Ie long de la faille de Kanda,dans lepaule ouest du bassin de Rukwa. © 1998 Elsevier Science Limited.

(Received 3 March 1997: revised version received 4 October 1997)

INTRODUCTIONThe tectonic depression of Lake Rukwa belongsto the western branch of the East African RiftSystem (McConnell, 1972). The Rukwa RiftBasin is located in the relay zone between theTanganyika and Malawi (Nyasa) rift valleys (Fig.1). With the latter it forms the TanganyikaRukwa-Malawi (TRM) lineament trendingnortheast. Kazmin (1980) first proposed that theTRM lineament formed an intracontinentaltransform fault zone, along which the RukwaRift Basin is believed to have opened as a pullapart basin in an oblique, northwest-southeastextension. This model was adopted byChorowicz and Mukonki (1980) and supportedby many others on the basis of the analyses ofsatellite images, the observation of fault slipindicators in the Precambrian basement alongthe major border faults and the interpretation ofindustrial seismic profiles (e.g. Tiercelin et al.,1988; Kilembe and Rosendahl, 1992; Wheelerand Karson, 1994). In contrast, Morley et al.(1992) favoured an opening of the Rukwa RiftBasin in a northeast-soutwest direction, suborthogonal to its general trend.

The major problem with the proposedmechanisms is the lack of precise timing. Mbede(1993) insists on the multistage evolution of theRukwa Rift Basin, with various extensiondirections during its evolution since the LatePalaeozoic. In the Rungwe Volcanic Province andin north Malawi, Delvaux et al. (1992) and Ringet al. (1992), recognised several changes inkinematic regime during the Late Cenozoic.Palaeostress investigations in dated sedimentsand volcanics and the use of the inverse methodfor stress reconstructions (as in Angelier, 1989)showed that the Late Tertiary period (Late

Miocene-Pliocene) is characterised by a semiradial extensional stress field, with the openingof both the northwest trending basins (RukwaBasin, Songwe Valley, Livingstone Basin) andthe northeast trending Usangu depression(Delvaux et al., 1992). In the Middle Pleistocene,a stress inversion led to a new, strike-slip stressregime with a north-south horizontal principalcompression. The kinematics of faulting changedcompletely. Normal faulting along the major faultzones ceased, or was strongly reduced, and onlysome of them were reactivated by strike-slipfaulting (dextral movement along northwesttrending faults and sinistral movement alongnortheast trending faults). This kinematic changewas correlated with a similar change thatoccurred in the Kenya Rift at the same period(Strecker et al., 1990; Delvaux, 1993a).

In this study, the Late Quaternary activetectonics and lake level change of the RukwaRift Basin are investigated to help answer thefollowing questions;

i) what is the expression of Late Quaternarytectonic activity and how does it differ from theolder tectonic activity?

ii) where is the main fault activity at thesurface?

iii) were the major rift border faults still activein the Late Quaternary?

iv) what is the Late Quaternary kinematics offaulting?

v) what kind of vertical movement affects thearea? and

vi) what are the respective roles of activetectonics and climate in controlling the recentsedimentation and hydrology of Lake Rukwa?

Presently Lake Rukwa is a closed hydrologicalsystem, however during an earlier more humid

Figure 1. Simplified neotectonic map of the Rukwa Rift and surrounding area in the western branch of the East African RiftSystem.

398 Journal ofAfrican Earth Sciences

1;-

~c;;;:;-~.

~sr"~~

o Basement

_KamoFi',l Cenozoic rift sediments

VvVv\V Rungwe volcanicso Major volcanoes

~Deep lakes

~ Shallow lakes and swamps

31°00'

TANZANIAN

CRATON

33°00'

\i.......,

......

"./

7°PO'

o 50 100kmI ; I

8°00'

~&9 RUNGWE I-9°00'

VOLCANICPROVINCE

(It"111.

G.s'h011.~

~~

34°00'

rQl

CDI:)

sCDsQl..,

"<:

CD

'"gC').Ql

'"~:~Ql~Q.

Q),..<b

<il

~g.~<b

sS<b

~~

@~tos

D. DELVAUX et al.

climatic phase it overflowed into LakeTanganyika. A series of palaeolacustrine terracesand sandy ridges are well preserved along theflanks of the rift and in the valley floor. Thealtitude of the upper terrace and sandy ridgecan be correlated with the altitude of thewatershed. The age of the last high stand ofthe lake is estimated at the Early Holocene, alsocorroborated by preliminary 14C dating. Thealtitudes of the lacustrine terraces with referenceto the present lake level reveal differential verticalmovements since the last high lake stand. Thisis also supported by the long-term trends inhydromorphological evolution and the deformation of palaeoerosional surfaces. Observationof tectonic deformation in the lacustrine terracesand in the recent sediments of the lake floorhave allowed the identification of active faults.Morphostructural and microstructural investigations were made to determine the LateQuaternary fault kinematics.

GEOLOGICAL SETTINGThe northwest trending Rukwa Rift Basin is 360km long and 40-60 km wide. It is occupied in itseastern part by a wide but shallow lake, whosewater level presently lies at an average elevationof 802 m. The lake level and the surface areaoccupied by the water change rapidly due toclimatic fluctuations. The lake, which nowcovers about half of the depression, was almostdry during the preceding century, but used tooccupy almost the whole basin in the EarlyHolocene (Kennerley, 1962).

The Rukwa Rift developed within thenorthwest-southeast trending Palaeo proterozoicUbende Belt, from which it reactivates mainlyNeoproterozoic sinistral shear zones of the sametrend (Theunissen et al., 1992; Lenoir et et.,1994; Theunissen et el., 1996). The northwesttrending Rukwa depression is bordered on itsnortheastern side by the Lupa Fault and theTanzanian Craton, and on its southwestern sideby the Ufipa Fault and the Ufipa uplifted block.The stratigraphy and structure of the Rukwadepression is relatively well known (Wescott etaI., 1991; Morley et al., 1992; Kilembe andRosendahl, 1992). The depth of the Precambrianbasement ranges from 4000 m in the northern

half, to 11,000 m in the southern half (Peirceand Lipkov, 1988; Morley et al., 1992). Theoverall structure is mostly that of a symmetricgraben in the northern part of the depression,grading into two opposing half-grabens in thesouthern part, with diverging dips of sedimentarylayers towards the border faults. At thesoutheastern extremity of the depression, thetwo half-grabens are progressively isolated fromeach other by the Mbozi block. The eastern branch(Songwe Valley) is connected to the northernextremity of the Malawi Rift (Livingstone orKaronga Basin) and to the transverse Usangudepression, in the Rungwe Volcanic Province. Thewestern branch (Msangano Trough) dies outprogressively to the southwest.

The earliest sedimentary record within theRukwa Graben consists of deposits of the LateCarboniferous to Early Jurassic KarooSupergroup. The next sedimentary unit is thefluviatile Red Sandstone Group of controversialLate JurassiclEarly Cretaceous or Miocene age(Dypvik et al., 1990; Wescott et al., 1991;Kilembe and Rosendahl, 1992; Mbede, 1993;Damblon et al., 1998). A shallow lacustrine tofluviatile environment was then established, withthe deposition of Pliocene-Early Pleistocene OlderLake Beds and Late Pleistocene-HoloceneYounger Lake Beds (Quennell et al., 1956).

Volcanism in the Rungwe Province started inthe Late Miocene, about 8 Ma ago, and is stillactive in the Holocene (Ebinger et al., 1993).Williams et al. (1993) reveal the presence ofdiscrete ash horizons in Holocene sediments fromnorthern Lake Malawi, providing evidence for atleast six eruptive episodes in the Rungwevolcanic field between 9000 and 150 yr. BP.The youngest tuff deposits in the Rungwe areahave 14C ages of around 11,000 yr. BP (Crossley,1982; Ebinger et al., 1989) and the youngesteruption occurred in the Kiejo volcano, 200 yearsago (Harkin, 1960).

Recent tectonic activity in the Rungwe region,which acts as a structural link between the northMalawi and Rukwa Rift Basins, wasdemonstrated by Delvaux and Hanon (1993).Recent and active faults, fossil travertine, activehydrothermal springs and gas vents, LateQuaternary volcanic eruptions and seismicityattest to this neotectonic activity. A local

Figure 2. Hydromorphological map of the Rukwa Rift Basin showing the succession of sandy ridges and swamps of decreasingaltitude from the northern extremity of the Rukwa Depression to the present-day northern shore, and the surface occupied bywater in 1950, 1977 and 1984. Circled numbers indicate sites cited in the text. 1: Luika River falls; 2: Saza River falls; 3:Mkwajuni; 4: fault scarp between Luika and Saza Falls; 5: Kilamba River falls near Muse; 6: Kavuu River section near RungwaRidge; 7: Kanda Fault scarp near Sumbawanga; 8: Kanda Fault scarp near Kaengesa; 9: Njelenje (Mbya Range Fault); 10:canyon of the Song we River; 11: canyon of the Ipwizi River. S: Sumbawanga; T: Tunduma; M: Mbeya.

400 Journal of African Earth Sciences

7'00'r-tll

<b0~<bstll..,-cCii(')

C;:)?;.tll(')

~.

:So~tll;:)Q..Ql

""C1l

iC<:~8'00'g.tll;:)tQC1l

S·SC1l

~~tll

~;::-

~'"S·

"'</ ~UNG~E+ 9'00'

o 50kmI ! ! ! I I

33'00'

riller

32'00'

_......._................. -,

\ii"

.....'......_.- - - - -".

\\,

31'00'

__ Rift Faults

I 0 1 Sandy Ridge

I...... -t~ ....1 Swamps

I Rift Sediments

1;-

~oh'"~.~s.'"~.II"a

D. DELVAUX et al.

network of seismic stations operating since 1992in the Mbeya area provides evidence for highseismic activity (Camelbeeck, 1992; Camelbeeckand Iranga, 1996). Moderate to largeearthquakes periodically hit the region (e.g.Ambraseys and Adams, 1991).

The Neotectonic period is considered here tocorrespond to the last first-order geodynamicregime that affects the region underconsideration. For the Rukwa-Malawi region, thelast change in tectonic regime is believed to haveoccurred between 1 and 0.5 Ma (Delvaux et al.,1992; Ring et al., 1992), and therefore theNeotectonic period covers the MiddlePleistocene-Holocene.

PALAEOLACUSTRINE TERRACES AND LAKELEVEL FLUCTUATIONS

This section investigates the palaeolacustrineterraces and Late Quaternary lake levelfluctuations in the Rukwa depression,considering successively the sandy ridges in thebasin floor, the watershed and palaeolacustrineterraces along the basin flanks, and terminatingwith an age estimation of the last high stand ofthe lake. The elevations used here were eithertaken from the 1:50000 topographical map ofTanzania or measured at ± 10m accuracy withan altimeter calibrated to the lake level.

Sandy ridges and palaeo-outlet of Lake RukwaIn the basin floor, north of Lake Rukwa, welldefined, slightly curved low ridges are evidencedin air photos and Landsat MSS images. Someare mentioned on geological and topographicalmaps as "Sandy Ridges". They are almostperpendicular to the rift trend and wereinterpreted by Tiercelin et al. (1988) as curvednormal faults. Following this interpretation andassuming that the Lupa and Ufipa Faults have adextral strike-slip movement, Tiercelin et al.(1988) further suggested that the Rukwa Riftdepression developed as a typical pull-apartbasin. A different interpretation of these ridgeswill be given here.

Detailed examination of Landsat images andtopographical maps reveals a succession ofcurved ridges of decreasing elevation from thenorthern extremity of the Rukwa depression tothe present-day lake (Fig. 2). The floor of thedepression on which they lie slowly but regularlydescends towards the southeast. The mostprominent of these ridges are the lIyandi Ridge(980-998 m a.s.I.), the Major Maimba Ridge(920-927 m a.s.I.) and the Rungwa Ridge (840


m a.s.l.). In addition, less pronounced ridges arefound at 900 m (Lower Maimba Ridge), 880and 860 m a.s.1. For reference, the lake shorewas at ± 802 m a.s.1. in the years 1977-1978.These ridges are separated by 10 to 45 km,with a total distance from the lIyandi Ridge tothe northwestern shore of Lake Rukwa of 115km. The ridges modify the drainage pattern andsmall lakes or swamps often develop behindthem. Examples are the Katavi Lake and swampnorth of the lIyandi Ridge, the Chada Lake andNsakasa Mbuga behind the Major Maimba Ridgeand the Isimba swamp behind the unnamed ridgeat 860 m a.s.1. (Fig. 2). Direct field observationof the lIyandi and Rungwa Ridges has shownthat they are relatively symmetrical, up to 40 mhigh and composed mostly of sand.

Based on these characteristics, the "SandyRidges" are unlikely to be tectonic scarpsassociated with curved normal faults. They areinterpreted here as abandoned shore lines causedby the stepwise retreat of the lake. They resembleoffshore sand bars that form by the combinedaction of waves and wind when the water depthis shallowing to a critical value, causing the wavesto break. Typical swampy pools have formedbehind the sand bars and are isolated from theopen lake. Such modern equivalents are knownfrom both the northern end of Lake Malawi (Nyasa)in Matema and Lake Tanganyika (southern sideof the Mahali uplift). They are all orientated at ahigh angle to the long axis of the lakes andsubjected to strong winds.

The watershed between the Rukwa and theTanganyika drainage basins lies at an elevationof 970-980 m a.s.l. at the western extremity ofthe lIyandi sandy ridge. It separates the swampyarea belonging to the Rukwa depression at 953m a.s.1. from the swamps of the KinywangiriMbuga (762 m a.s.l.), which feeds the NkambaRiver. The latter flows westwards to LakeTanganyika (762 m a.s.I.) via Karema (Fig. 2).The lIyandi sandy ridge develops mostly above980 m a.s.I., culminating at 998 m a.s.l..therefore, there can be little doubt that the lIyandiRidge corresponds to the uppermost level of LakeRukwa and formed when it was overflowing intoLake Tanganyika through the watershed. Thetop of the lIyandi Ridge is a few metres higherthan the probable corresponding lake level, asis normally the case for such sand bars.

Palaeolacustrine terraces along the Lupa andUfipa Border FaultsAlong the escarpments of both the Lupa andUfipa Border Faults, palaeolacustrine terraces are

Late Quaternary tectonic activity and lake level change in the Rukwa Rift Basin

8"25

/

u Normal fault (inactive)

I

l]JJ111IT M"Jrphological scarp

• ~:I~~~;nOf underwater

-A- +-~ -'!'J. Swamp

~ Tilted paleo-morphologicalsurface

1091m...

J~ +~ ~ ~ ~~ ~

--"-¥-- -4- --*- ~~ -:i!c---

~- --'~ 4- -4- 4- ~

~~ 4- -4- 4 +--'.lIt- ~ -~ ~ ~ ~~

1012m...I

33°00

Figure 3. Lupa Fault scarp near Mkwajuni: morphostructural map interpreted from air photographs and field observations(indices 1-4 in Fig. 2). Location of sections for Figs 4 and 5.

observed locally, in connection with the maindrainage lines from the Lupa and Ufipa Plateaus.The main rivers developed waterfalls of a fewtens of metres, when crossing the faultescarpments (e.g. the Saza and Luika Riversalong the Lupa Border Fault and the Kilambwaand Mbede Rivers along the Ufipa Border Fault).

Along the Lupa Border Fault, Quaternarydeposits are found at the mouth of importantrivers emerging from the escarpment (Fig. 3).They lie against the fault and have thecharacteristic shape of a horse shoe with awaterfall in the centre. They are made oflacustrine and fluvial deposits. The surface ofthese deposits, departing from the escarpment,display a step-like profile gently sloping towards

the lake. They are interpreted as sedimentsdeposited under a higher lake level, formingunderwater delta fans at the mouth of the riversflowing through the fault scarp. The observeddeposits lie undisturbed against the faultescarpment, apparently sealing the Lupa BorderFault. Air photo interpretation and direct fieldobservations reveal no traces of neotectonicactivity in these deposits.

The palaeodelta fans have a typical steppedprofile in different places along the Lupa BorderFault, probably corresponding to a series oflacustrine terraces of decreasing altitude,developed during the stepwise progressivelowering of the lake level. At the mouth of theLuika River (index 1 in Figs 2 and 3), the top of


O. OELVAUX et al.

LUPA FAULT ESCARPMENT (Saza falls)

Figure 4. Schematic cross-section of Quaternary deposits at the Sasa falls location (index 2 in Figs 2 and 3.)

Figure 5. Schematic profile at the base of the Lupa Escarpment between Luika and Saza falls (index 4 in Figs 2 and 3).

the water fall is at 840 m a.s.1. At the Saza Falls(index 2 in Figs 2 and 3; Fig. 4), the top of faultscarp is at 850-855 m a.s.l., the upper surfaceof Quaternary deposit against the fault scarp is


at 840 m a.s.I., an intermediate terrace is at830-835 m a.s.1. and the lower terrace is at820 m a.s.1. At Mkwajuni Village (index 3 inFig. 3), a similar alluvial fan of sand and pebbles


UFIPA FAULT ESCARPMENT

Figure 6. Schematic cross-section of the Quaternary deposits at the base of the Ufipa Fault scarp, north of Muze (index 5 inFig. 2). Terraced profile of the underwater palaeodelta.

is observed along the Lukolongo River, with thesummit of the fault scarp at 900 m, and gentlysloping surfaces at 840-860 m (upper terrace)and 810-815 m a.s.1. The base of the fault scarpbetween the Luika and Saza Deltas (index 4 inFig. 3; Fig. 5) also have a stepped profile, withgently sloping surfaces to 840, 830-835 and815 m a.s.l. All these altitudes are remarkablycoherent. This suggests a past succession ofhigh lake levels at 840-860 m, 830-835 m and810-820 m a.s.1. along the central portion ofthe Lupa Border Fault.

Along the Ufipa Border Fault, at the mouth ofthe Kilambwa River falls near Muze Village (index5 in Fig. 2), a palaeounderwater delta fan alsoabuts the base of the fault scarp (Fig. 6). Noevidence of recent faults cutting the Quaternarydeposits was found. Any tectonic mobility alongthis escarpment should therefore be on smallerand much younger faults in front of the mainescarpment and terraces. There is thus asimilarity with the opposite side of the Rukwadepression, along the Lupa Fault, however theterraces and waterfalls are at a higher elevationwith respect to the ones on the Lupa side. NearMuze, the terraced surfaces gently slopingtoward the valley axis are at 1000 m (uppermostone, at contact with the fault scarp), 925-930m, 905 m, 880-870 m and 850 m (village) a.s.1.This relative difference in elevation will bediscussed later.

Water level fluctuations in the HoloceneThe palaeodelta fans and sandy ridges areevidence of a higher lake level. The maximumaltitude of their occurrence (990 m at the

northern extremity of the Rukwa depression)corresponds to the lowest altitude of thewatershed between the Rukwa and Tanganyikadrainage basins (between 969 and 980 rn): Thissuggests that at that the time of the high lakelevel, the water of the Rukwa Rift Basin wasflowing out to Lake Tanganyika, via theKinywangiri swamps and the Nkamba River.

The high lake stand in the closed East Africanlakes could be achieved by an increase inprecipitation and/or a decrease in evaporation,both resulting from climatic changes (Haberyanand Hecky, 1987). In much of East Africa, aridconditions prevailed before 20,000 yr. BP, whenmost lake levels were low (e.g. Scholz andRosendahl, 1988). The last major humid climaticphase in Central Africa is known to haveoccurred between 12,000 and 9000 yr. BP,maybe up until 7000 yr. BP in places. This isindicated by a high lake level in Lakes Victoria(Adamson et el., 1980) and Kivu (Haberyan andKecky, 1987), by low sedimentation rates inLakes Kivu and Tanganyika (Haberyan andKecky, 1987) and by trace element analysis inthe Nile Delta sediments (Hamroush and Stanley,1990). Also in Lake Magadi (south Kenya), anEarly Holocene wet phase corresponding tolaminated clays and zeolites is 14C dated to 9120yr. BP (Butzer et al., 1972). Detailed sedimentological and geochemical analysis of two coresfrom Lake Magadi document a wet and high lakelevel between 12,000 and 10,000 yr. BP(Damnati and Taieb, 1995). For Lake Malawi,major recession had already occurred in 10,740± 130 yr. BP (Owen et al., 1990; Finney andJohnson, 1991), although this lake is known to


D. DELVAUX et al.

have a slightly different evolution than the lakesnorth of it.

The high lake level in Lake Rukwa mightcorrespond to this Late Quaternary climaticoptimum, between 12,000 and 9000 yr. BP.This is supported by a new 14C age of 9615 ± 95yr. BP, obtained on a charcoal sample taken fromthe highest terrace near Muze (1000 m a.s.l.:index 5 in Fig. 2). This result fits remarkablywell with the 14C age of 9740 ± 140 yr. BP on amollusc shell from a mollusc rich layer on top ofthe lacustrine sequence, south of Galula (Clarket al., 1970). From the description of Clark etal. (1970) and the observations made in thisstudy, the elevation of the sampling site shouldbe at 900-920 m a.s.l., not far from the LakeRukwa high stand. This age may date theemergence of this area at that time, and thecorresponding palaeolake level.

A high lake level in Lake Rukwa during 12,700and 4400 yr. B.P. had already been inferred byHaberyan (1987) on the basis of fossil diatomassemblages in a 23.1 m sediment core fromLake Rukwa. Since then, Lake Rukwa dried outprogressively and became a closed hydrologicalsystem. It has completely dried out at least threetimes since 3300 yr. B.P. (Talbot andLivingstone, 1989). In recent times, from 1740to 1840 AD, Lake Rukwa was again completelydry (Owen et et., 1990). Since then, the surfaceoccupied by the lake is progressively increasing,as shown by a compilation of lake contours fromcartographic documents and satellite images(Fig. 2). The 1:120000 geological maps of theTanganyika territory, based on an aerophotographical survey made in 1950, show twoisolated pools separated by a zone of seasonalflooding. The 1:50 000 topographical maps ofTanzania, based on a new aerophotographicalsurvey in 1977, show a greater and unified lake.A Landsat MSS image taken in 1984 during thesame season shows that the lake has continuedto expand in the northern plain and near Ivuna.During the last field work in June-July 1994, itwas observed that the contour of the lake is alittle wider, mainly at both extremities wherethe floor of the depression is very flat.

ACTIVE FAULTINGThe tectonic activity of the Rukwa-Mbeya regionis demonstrated by its high seismicity, often

characterised by unusually deep hypocentres of15-35 km (Camelbeeck and Iranga, 1996) andby field observations made during the geologicalmapping of this area. In this section evidence isexamined for possible modern tectonicmovements along the major rift faults: the Lupa,Mbeya Range-Galula, the Ufipa and the KandaFaults (Figs 1 and 2).

The Early Holocene lacustrine terracesmentioned above will be used as a stratigraphicalreference to evaluate the recent tectonic activityalong the Lupa and Ufipa Faults. The MbeyaRange-Galula Fault will be investigated usingmorphostructural and microstructural observations, and the results of a high resolution seismicsurvey made by Cermicola et al. (1996) in thesouthern part of Lake Rukwa.

In several places, microstructural data werecollected on minor faults with slip indicators,conjugated pairs of joints and single joints. Theywere used to identify the kinematics of faultingand to differentiate between normal and strikeslip faulting. Additionally, they were inverted todetermine the reduced palaeostress tensorsfollowing the method of Angelier and Mechler(1977) and Angelier (1989), using the TENSORprogram (Delvaux, 1993b). Although palaeostress reconstruction is not the main topic ofthis paper, the palaeostress results representadditional constraints for the tectonicinterpretation. More details of this method areprovided in Delvaux et al. (in press).

Lupa FaultThe Lupa Fault separates the Rukwa Rift Basinfrom the Tanzanian Craton and the Lupa block(Fig. 1), topped by the African I lateriticmorphological surface of Late Miocene age (King,1963). The fault appears relatively rectilinear,but in detail it is composed of several curvedsegments (Fig. 3). It is associated with a wellexpressed morphological scarp up to 200 m high.To the southeast the scarp progressively fadesout and disappears near Mbeya. In between, theLupa Fault splays in several segments alsorunning inside the Lupa block, and some of themcontinue into the Usangu depression. The seismicactivity recorded by the Mbeya seismic stationis precisely concentrated along some of theselateral branches (Delvaux and Hanon, 1993).

The structure of the Lupa Fault itself is madeof a series of thin (1-5 m). steeply inclined (45-

Figure 7. Lupa Fault scarp near Mkwajuni: fault kinematics and stress tensor reconstruction from data of the Luika River falls(index 1 in Figs 2 and 3). Stereograms (Schmidt net, lower hemisphere) with traces of fault planes, observed slip lines and slipsenses; histogram of observed slip-theoretical shear deviations for each fault plane and stress map symbols. Rose diagramsof the dip and strike of fault planes and joints, and inclination and azimuth of slip lines.


rOJ

~

c~~sOJ

""'<:

~C")

~".OJC")

~:~

§c,

'",..<b

co~g.OJ

c2<b

S'S<b

~~

@~enS'

azhlUth

weight: Ystrike

aziflluth

weight: Ystrike

N

N

N

N

14 planes

18 planes

12 lines

9 lines

O·

a·

O·

O·

dip

dip

plunge

plunge

Older movement: n=14/29dd610.dat: Luika falls, Lupa fault scarp

Younger movement: n=18/47dd610.dat: Luika falls. Lupa fault scarp

WeighedSLOWERnt= 14

®

fWeighedS LOWERnt= 18N

r>.:,(fI ~,/~,_.~

'~

""~~~\ ~\lt-. \\\ ® ~\'

\\ \\

~

dd61a.dat Luika falls,

dd610.dat Luika falls. Lupa fault scarp

frequency x weight 12

Younger movement: n=18/47

~,

{// ~ (-;)\/\. rv " ~ I

~,--/'/

0: (0)

36

6~a frequency x weight 23

Older movement: n=14/29

@ 1: 69/152~ 2: 21/319[!] 3: 65/651R, gO: .67 •. 26Pure EX TENS IUE

0: (0)

36

66

~~ ·00 )\ t I\, J'~/

@ 1: 10/163~ 2: 21/277[!] 3: 66/a69R, gO: .41 6.05Pure COMPRESSIUE

66

~

~c;;:'"1)'

~

~sg>it'

~...Cl'J

~~

""!!.c;:'"Q"c

~S.

~"~

Normal fault (inactive)

Recent strike -slip reactivation

o 5kmI ; ; ! ; I

33°15'

Figure 8. Mbeya Fault near Njelenje (index 9 in Fig, 2), (a) Morphostructural map with the location of the palaeostress sites,

~

tJ~§

~;P.9C

tO>

Cbo~Cbs~

'<::

Cb'"§r:;'0>

'"~.~'

~Cl..1il,.,.C1l

Cil

~3-~C1l

S·so.C1l

~~

~2''"S·

llzimuth

weight:stri~eII

II

ill planes

31 lines

0'

0"

dip

plunge

WeighedSLOWERnt= ill

ddiili.dat IIjelenje uiliage Ilbeya range scarp

frequency x weight 2B

se

« (")

lie

C00•0<0® .-1: 77/197&. cr2: 05/300~ cr3: 12/039R, «": .31 9.72Pure EXTEIIS WE

b)

Figure 8. continued. (b) Fault kinematics and stress tensor reconstruction for site 00446: normal faulting only (location in a).

~

lC>

;;;~C>

~S-en"is'

~'"&5

~

t:::li!!§

~C1l.-t

tll

azillluth

weight: Ystrike

weight: Ystrike

azillluth

N

N

N

N

13 planes

21 planes

15 lines

13 lines

0°

6°

6°

eo

dip

dip

plunge

plunge

Younger movement: 13/44 faults & jointsdd601.dat: Mbeya range scarp. Njelenjele village

Older movement: 21/44 fault & jointsdd661.dat: Mbeya range scarp. Njelenjele village

N

,

rz"~"'" )

'v'

18

dd601.dat Mbeya range scarp. Njelenjele villageWeighedSLOWER'# nt> 13

:'\.~\\

dd601.dat Mbeya range scarp. Njelenjele villageWeighedS LOWER

N ~........ nt> 21

Older movement: 21/44 fault & joints

frequency x weight


0: CO)

30-1.-

0: (")

36

Younger movement: 13/44 faults & joints

60

/ ......--~"

(C~):#\)\ #("1

\£.-'

/1

"'---""--/® ..1: 23/219~ ..2: 65/015[!] ..3: 09/125R. «0: .16 7 .11Pure STRIHE-SLIP

.~

( ~!?)\{} ~/

""---,----/® ..1 62/101~ ..2 26/310[!] ..3 10/215R. «0 .21 8.51Radia EXTENSIUE

60

c)..c;

~3!l<.c;;:'"s~~

[;;1~

~"a


65°) tectonic slices of basement rock (massivegranite in the investigated area). The few sliplines observed on these slices provide dip-slipto oblique-slip directions. The inclination of thefault plane and the presence of curved faultsegments are typical for normal faulting in anextensional stress regime (e.g. Stewart andHancock, 1991). In addition, the tectonic slicescut older minor fault planes with well expressedslip lines, indicating a dominant strike-slipmovement (Fig. 7). These observations are inaccordance with the succession of (1) strikeslip and (2) normal faulting, described by Tiercelinet at. (1988) for the Lupa Fault.

Using the Late Quaternary lacustrine terracesas a reference, the major normal faulting alongthe Lupa Fault seems to have occurred beforethe Early Holocene. The observed terraces lieundisturbed against the major fault scarp. Limitedtectonic movement might have occurred at thefoot of the scarp since the deposition of theterraces, but more detailed analysis of airphotographs, field geology, topographicalmeasurements and trenching are necessary tosolve this issue. In any case, examination ofavailable air photographs reveals no evidencefor large recent offsets. In particular, horizontaldisplacements can be excluded for the LateQuaternary. No offset of drainage, no pull-apartsor push-ups, typical of strike-slip faults (e.g.Sylvester, 1988) are evident. This is especiallythe case near Saza Falls, where the fault scarppresents a series of curved sections withoutperturbations of the sedimentary layers restingon it. It is clear that no important tectonic activityoccurred along this portion of the Lupa Faultafter the deposition of the alluvial fans into theformer lake with higher water level. However, itcannot be ruled out that some recent movementsmay have occurred on some still undetected faultinside the graben. A shift of faulting with timetowards the centre of the depression iscommonly observed in normal faulting (e.g.Stewart and Hancock, 1991), or in the case ofkinematic inversion (e.g. Ring et al., 1992).

Ufipa EscarpmentThe Ufipa Escarpment is up to 1000 m highabove the lake level and bounds thesouthwestern side of the Rukwa Rift. The latteris comparable to the Lupa Fault, but lesscontinuous (deep valleys cut through the rangefront) and rectilinear. Based on seismic reflection

profiles (e.g. Morley et al., 1992; Kilembe andRosendahl, 1992), the total vertical offset ofthe Ufipa Fault (2-4 km) is less than that of theLupa Fault (4-11 krn), but the total vertical reliefdifference is much greater (1000-1200 m) thanthat of the Lupa scarp (100-200 rn). As for theLupa scarp, field observations andaerophotographical interpretation reveal theabsence of significant vertical or horizontalmovements displacing the Late Quaternarydeposits along the Ufipa Fault scarp.

Mbeya Range-Galula Fault ZoneThe Mbeya Range Fault (Figs 2 and 9a) runsalong the southwestern foot of the northwestsoutheast trending Mbeya Range. It is an upliftedbasement block protruding the main LupaEscarpment between the Rukwa Rift Basin andMbeya. The summit of the Mbeya Range (2818m a.s.l.) lies above the mid-Miocene referencesurface of the African I laterite peneplain, whichhas an elevation of 1300-1600 m a.s.l. in theadjacent Lupa block and 1600 m a.s.l. in theMbozi block. The uplift of the Mbeya Rangecaused the narrowing of the Rukwa depressionat its southeastern extremity, which is therebyrestricted to the 10 km-wide Songwe-RukwaValley.

The Mbeya Range appears to be an upliftedbasement block tilted to the northwest. On airphotographs, the northeastern flank looks like astrongly tilted remnant of the African I peneplain(Fig. Ba). The southwestern flank is bounded bya weakly inclined normal fault with dip-slipslickensides (Fig. 8b). This flank correspondstherefore to a major fault scarp in whichintermediate steps are recognised (Fig. 8a).

The foothill of the Mbeya Range scarp alongthe southwestern flank has steep triangular spurswhich, although partly following a pre-existingstructural grain, on the whole strongly suggesta recent rejuvenation of the range front (Fig.8a). A survey conducted in the Njelenje area atthe contact between the basement andQuaternary deposits did not reveal any evidenceof recent tectonic movement, however the areais covered by thick vegetation and agriculturalactivities may shroud some critical exposures.The steep slope along the faceted spurs, almostfree scarp faces at their base, and the generalmorphology with back-tilted faulted Quaternarysurfaces, suggest some recent movements inthe area. The presence of push-up slices of the

Figure 8. continued. (c) Fault kinematics and stress tensor reconstruction for site DD604: older normal faulting and youngerstrike-slip faulting (location in a).


~

tlPJ§

~~

~

aziMuth

weight :strike

weight:strike

aziMuth

"

"

"

"

Z3 planes

22 planes

Z3 lines

7 lines

e'

e'

O'

e'

dip

dip

plunge

plunge

dd6a&.dat: Galula road, deep trenches

i1d619.dat: Ipwlsl river, -~kw-aplalii;'loungerLake Beds

t=)

WeighedSLOWERnt= zz

dd6G6.dat Galula rDad. deep trenches

dd619.dat Ipwlsl river, Rukwa plain, Younger Lake BedsWeighedSLOWER

N -~ nt= Z3.«it::)

frequency x weight ZB

b)

o~e frequency x weight 16

30

U (0)

60

@ at 1'1/11M&,,,.Z 71/016~".3 91/27'5R. UO .'11 3.66Extensluc STRIKE-SLIP

0: (0)

60

30

a)

@".1 03/Zez£".2 B7/352~".3 02/112R, UO .Z3 6.6BCDMpresslve STRIKE-SLIP

..;;;

Ich'"~.~S

~~a

Figure 9. Mbeya Range-Galula Fault in the Rukwa Plain. ie) Fault kinematics and stress tensor reconstruction from data of the Ipwizi River Isite DD619, index 11 in Fig. 2). (b} Faultkinematics and stress tensor reconstruction from data of the Song we River (site DD606, index 10 in Fig. 2).


basement rocks covered by Quaternary depositsin front of the main scarp, secondary obliquetrending faults and displaced river valleys aresuggestive of a strike-slip component. A smallquarry excavated in a limonitised rock along thefault plane (index 9 in Fig. 2) shows the presenceof two generations of minor faults: a dip-slipand a strike-slip system with right-lateralmovement along the trend of the Mbeya RangeFault (Fig. 8c). Cross-cutting relationships allowthe strike-slip system to be defined as theyoungest.

In the valley floor, in the continuation of theMbeya Range Fault, minor faults with slipindicators and joints were observed in the LateQuaternary volcanic, lacustrine and fluvial gravelbeds in the canyons of the Ipwizi River (index11 in Fig. 2; Fig. 9a) and in one tributary of theSongwe River (index lOin Fig. 2; Fig. 9b). Theseobservations suggest a northwestern continuationof the Mbeya Range Fault Zone, at least up tothe Galula Village.

The Mbeya Range-Galula Fault Zone probablycontinues beneath the lake along the same trend.This is ascertained by a high resolution seismicsurvey in the southern half of Lake Rukwa,recording the upper 200 m of sediment(Cermicola et et., 1996). A series of steeplydipping faults have affected the uppermostreflectors and are typically syndepositional withthe youngest lake sediments (Cermicola et al.,1996).

In the multichannel seismics carried out byAMOCO in the same area, Morley et al. (1992)demonstrates the existence of an important faultline that affects most of the sediment infill in apositive flower structure. However, themultichannel seismics probably did not recordthe upper layer. The faults seen on the highresolution seismics might well be branched tothe deeper flower structures.

The palaeostress tensors obtained for thethree sites along the Mbeya Range-Galula FaultZone (Njelenje, Ipwizi River and Galula) displayboth principal compression (0,) and extension(0) horizontal stress axes, indicating aprevalent strike-slip style of deformation. TheSHmax (0) axes trend north-northeast - southsouthwest, ranging from N05°E to N39°E andthe SHmin (03) axes trend west-northwest - eastsoutheast, ranging from N096°E to N125°E.The relative succession of (i) normal faultingand (ii) dextral strike-slip faulting observed inNjelenje, at the foot of the Mbeya Range, isin full agreement with the previous results frompalaeostress investigations for the Late

Quaternary period in the Mbeya area (Delvauxet al., 1992).

Kanda FaultThe Kanda Fault (Figs 1 and 2) appears to bethe most active structure in the Rukwa area. Itruns almost continuously for about 180 krn. fromnear Tunduma to north of Sumbawanga. TheKanda Fault dissects the Ufipa Plateau along itstrend, between the Rukwa and Tanganyika RiftBasins. The height of the free scarp facedecreases progressively towards the northwest,from about 50 m near Mpui to a few metresnear Sumbawanga. In some segments thefaulting appears to be distributed on severalsubparallel scarps that look younger towards thevalley axis. Elsewhere, the faulting isconcentrated on a single plane. It strikesgenerally to the northwest-southeast and dipsto the northeast, displacing the laterite surfacerelated to the African I mid-Miocene peneplainof King (1963). On its hanging wall side a large"Mbuga" (swamp) corridor forms a percheddepression at the top of the Ufipa Plateau. Twoother similar faults are known further west: theMkunda and Mwimbi Faults (Fig. 2). They allaffect the drainage pattern profoundly.

The Kanda Fault is hosted either in thebasement gneisses or in the Karoo sediments.To the north it is linked with the fault structurethat affects the Namwele coal field. The KandaFault is described by van Loenen and Kennerley(1962) as reactivating an ancient structural line.Microstructural indicators found along the KandaFault (Fig. lOa) and the in Namwele coal fieldnear Sumbawanga, the structure of the coal fielditself (McConnell, 1947) and the presence ofisolated lenses of Karoo sediments along theKanda Fault near Sumbawanga all suggest thatan important right-lateral strike-slip movementoccurred along this line, after the deposition ofthe Karoo beds (Late Carboniferous-Permian).The presence of the laterite surface truncatingall these structures constrains the age of thestrike-slip activity as pre-Middle Miocene. It islikely that this movement took place before orat the beginning of the Late Jurassic-EarlyCretaceous phase of denudation revealed by vander Beek et al. (1998) by apatite fission trackthermochronology. Recent activity along theKanda Fault is dominantly dip-slip, displacing thelaterite surface.

An important section across the fault zoneis exposed along a trench of the road thatclimbs the fault scarp near Kaengesa (index 8in Fig. 2; Fig. lOb). The difference in elevation

Journal of A {dean Earth Sciences 413

!='

"~§

~~

l'!-

azi~th

azi~th

weight: Ystrike

..eight:striket1

t1

t1

N

26 planes

42 planes

25 lines

25 lines

O·

O·

O·

O·

dip

dip

plunge

plunge

Younger movement: 42/104 faultsdd396.dat: Kanda fault, SUMbawanga (1992-1994)

Older movement: 28/ 104 faultsdd396.dat: Kanda fault, Su~bawanga (1992-1994)

WeighedSLOWERnt= 42

WeighedSLOWERnt= 26

N

dd396.dat Kanda fault, Su~bawanga (1992-1994)

dd396.dat Kanda fault, Su~bawanga (1992-1994)



0: (.)

60

60

Older movement: 28/ 104 faults

0: (.)

~

( <>-1 \\i'»

~----,----

@ .-1: 07/204£ .-2: 07/295[!J .-3: 60/070R, 0:.: .19 10.26Strike-slip COMPRESSIVE

Younger movement: 42/104 faults

.~~(/' "~ ~.<:'.)\\r?"'- /

'---,-_.---/@ .-1: 61/245£ .-2: 06/146[!J 0-3: 29/053R, 0:.: .96 H.l.63Strike-slip EXTENSIVE

30

30

...:;:~

~!!!.c;;:

*""II;'s.g>1;;'

""a


Figure 11. Digital shaded topographical model of the southern extremity of the Rukwa Depression, computed from digitised1:50 000 topographical maps.

between the surface topping the footwall andthe pounding area in front of the scarp is about30 m. Several minor fault planes parallel themain escarpment with one displacing the topof the basement by 1.5 m of vertical offset.A weathered mylonitic gneiss, topped by athick soil, is found opposite fluvial andlacustrine deposits dragged and tilted near thefault. A sample of charcoal collected in thefluvial deposit, just on the footwall of thescarp, yielded a radiocarbon age of 8340 ± 700yr. BP (Vittori et el., 1997). This is a maximumage for the major event that caused thisdislocation. It should not be very far from themaximum age, since this sample comes fromthe first sediments deposited after theescarpment formation.

At the same location a few metres below, alayer of black organic clay covers the footwallbasement. It yielded a radiocarbon age of 13,600± 1240 yr. BP (Vittori et al., 1997). The fluvialdeposit above the fault scarp also contains debrisof a similar organic rich black clay, suggestingthat this layer was originally continuous anddisrupted by the 8340 ± 700 yr. BP event. TheseLate Pleistocene-Holocene deposits now hangmore than 15 m above the present valley floor,suggesting a Holocene slip rate of more than1.0 mm a'.

Due to its high rate of activity in the recenttimes, the Kanda Fault may have produced the1910 Rukwa M 7.4 earthquake, whoseinstrumental epicentre was located very near thefault (Ambraseys, 1991). The Kanda Fault is then

Figure 10. Kanda Fault. Fault kinematics and stress tensor reconstruction near Sumbawanga (site DD393, index 7 in Fig. 2):older transpressional faulting and younger transtensional faulting.


D. DELVAUX et al.

I'!iiPCI

54

540000m E.

5040

52

20

3010 20

10

46

Kilometres _",,10;'F3o;F33::F3:F3::EF3~0""''''''''''''''~=;::=:;=~''''''''''~~==~''''''''''~.Miles 10 0

E'3 ........ E3E3F?

Figure 12. fa) Same digital topographical model as in Fig. 11. Grey intensity is as a function of altitude. Lines 1-3 locate thetopographical sections of b.

comparable with the 100 km long BililaMtakataka normal fault in southern Malawi,along which occurred the march 9 (M = 5.7) andMarch 10 (M = 6.1) earthquakes (Jackson andBlenkinshop, 1997). More details on the recentactivity of the Kanda Fault are provided in Vittoriet al. (1997).

VERTICAL MOVEMENT AND REGIONALTILTING

The elevation of the highest lacustrine terracesand sandy ridges at different locations aroundthe Rukwa depression is not constant.Preliminary altitude measurements, using analtimeter calibrated to the lake level, reveal thatthe terraces along the central part of the LupaFault, at the northeastern side of the depression(840-860 m a.s.l.). are more depressed than theterraces against the Ufipa Fault, on the oppositeside (up to 1000 m). The upper terraces along

the Lupa Fault are more elevated towards itssoutheastern termination, in the direction of theRungwe Volcanic Province (900-950 m) and atthe northern termination of the Rukwadepression (llyandi sand ridge: 980-998 m). Thisindicates that the northeastern central part ofthe Rukwa depression is presently subsidingrelative to the rest of the depression. Thisinterpretation supposes that all the highestterraces at each site and the highest sandy ridgeswere formed at the altitude of the outflow sill ofLake Rukwa during the last humid period, in theEarly Holocene. However, some terraces mayhave not been deposited at this time and othersmay have been entirely removed by erosion,depending on local conditions. In the absenceof the precise stratigraphy and morphology ofthe terraces, stratigraphical markers andsystematic 14C dating, it is necessary to look forother clues to the recent differential verticalmovements.



b)

I100,000

NE

Lupa plateauwith residual hills

I80,000

NE

Lupa

fault\arp

Io

NE

Section 3

Section 2

Mbeya Section 1Range

20,000

I20,000 40,000 60,000

Distance from origin (m)

i60000 40000 20000

Distance from origin (m)

Mbozi plateau

1500j

1000~<

i,-I

I500 -t-~'--~----'----r--- I -,-----,----,--,---,-----.-

40,000 60,000 80,000Distance from origin (rn)

0SW

2000-,~ I

en j«i 1500

I

.s iQ)"0 1000 -1.a+='«

500!0

SW

~

en«i

.s 1500 -iQ) j"0.a 1000+='

i«500 I

SW

Figure 12. continued. (b) Topographical sections (location in a).

Evidence for the northeastward tilting of thesouthern part of the Rukwa Rift Basin is bestprovided by the morphology and asymmetryof the basin floor topography (Fig. 12). Thenortheastern side of the Rukwa depression ismostly occupied by water and swamps, withthe lake extending very close to the LupaEscarpment. The southwestern side of thedepression is a dry plain, with deeply incisedrivers in the Late Quaternary sediments. TheIpwizi, and Momba Rivers are incised into theRukwa lake beds for more than 20 m, and theSongwe River up to 80 m (Fig. 11).

The Late Quaternary regional tilting of theRukwa Depression is in accordance with longterm differential vertical movements betweenthe Ufipa Plateau and the Lupa Plateau, withreference to the African I (mid-Miocene) lateritepeneplain. In the Lupa Plateau this surface issloping progressively westwards, from 1570 mnear Chunya, 25-30 km west from the Lupa

Fault scarp, to 900-1000 m a.s.l. near the fault(Fig. 12b). Residual inselbergs from an olderpeneplain surface range from 1725 m to 1400m a.s.1. In the Ufipa Plateau, the laterite AfricanI surface lies at 1800-1900 m a.s.1. atSumbawanga near the Ufipa Escarpment, anddecreases westwards to 1500 m a.s.1. atNamanyere towards Lake Tanganyika.

At the southern end of the RukwaDepression, a broad zone of doming hasdeveloped in the vicinity of the Mbeya Rangeuplift. The dome coincides with the MbeyaRange-Galula zone of active strike-slipdislocation. This doming causes an uplift inthe axial part of the depression. This hascaused divergent tilting of the former lakefloor, with apparent deepening towards theLupa and Ufipa Border Faults. This isparticularly well illustrated by the northeastflowing Ipwizi River, which has deeply incisedinto the Quaternary sediments (Fig. 11).


D. DELVAUX et al.

Preliminary analysis of the high resolutionseismic profiles in Lake Rukwa (Cermicola etel., 1996) suggests that sedimentation in thesouthern part of the lake is controlled by acombination of subsidence towards the LupaBorder Fault and uplift in the central part of thedepression. The upper sedimentary layers in thecentral part of the depression are stronglyaffected by faulting and associated deformation,while they are less disturbed and progressivelythicker towards the northeastern margin. Recentsubsidence associated with normal faultingalong the Lupa Border Fault can be invoked toexplain the progressive thickening of thesedimentary layers towards the northeast.However, the lack of marked recent tectonicactivity along this border fault, and the presenceof the active Mbeya Range-Galula dislocationzone in the axial part of the depression, favourthe idea that the tectonic factor controlling thesedimentation is a combination of axial doming(uplift) linked to the Mbeya Range-Galuladislocation zone and subsidence in the area ofthe Lupa Border Fault in the region of GalulaMkwajuni. This means that the basin floor andrift shoulder subside together, restricting faultmovement between them.

This differential movement can be related to acombination of large scale tilting of the centralpart of the depression towards the northeastand general uplift of the southeastern extremityof the depression. The latter is in agreementwith the Quaternary uplift centred on the RungweVolcanic Province (Delvaux and Hanon, 1993),whereas the former is similar to the tilting ofthe Kyela plain at the northern extremity of theMalawi (Nyasa) Depression (Delvaux and Hanon,1993). It is also revealed by morphostructuraland high resolution seismic investigations (deVos, 1994).

This asymmetric subsidence may also reflecta possible local post-rift subsidence of thisregion. Post-rift subsidence may be driven byan isostatic response to sedimentary loading andby thermal relaxation, contraction of thelithosphere and compressional intraplate stresses(Cloetingh and Kooi, 1992; Ziegler, 1996). Infact, most of the subsided area corresponds tothe thickest Cenozoic sedimentary accumulation,up to 7 km according to Morley et al. (1992).Moreover, the region has been submitted tocompressive intraplate stresses since the MiddlePleistocene (Delvaux et et., 1993). However,more investigations and kinematic modelling arenecessary to understand the causes of thesevertical movements.

418 Journal ofAfrican Earth Sciences

SYNTHESIS AND CONCLUSIONSResults of morphostructural investigation revealthat the major border faults of the Rukwa RiftBasin underwent moderate tectonic activity inthe Late Pleistocene-Holocene. Observation ofthe palaeolacustrine deltas and sandy ridgesalong the margin of the Rukwa Depression allowsthe recognition of both climatic and tectoniceffects.

Long term variations of water level and arealextent of the lake are indicated by palaeolacustrine terraces 60 to 170 m above thepresent-day level of the lake and up to 115 kmnorth of its present northern shore. Short termfluctuations of the lake surface are recorded byhistorical and cartographic documents andsatellite images. The higher level of terraces andsandy ridges corresponds to the altitude of thewatershed, over which water from Lake Rukwawas outflowing into Lake Tanganyika during thelast humid period. The high lake level in LakeRukwa is preliminarily dated as Early Holocene.

The present altitude difference of the highestterraces in several locations around the RukwaDepression indicates that substantial differentialtectonic movement occurred in the Holocene.The terraces along the escarpment of the LupaBorder Fault, on the northeastern side, are moredepressed than the ones along the escarpmentof the Ufipa Border Fault on the southeasternside. This demonstrates an active northeastwardtilting of the entire depression, including boththe rift shoulder and the adjacent depressionfloor. Along trend, the terraces against the LupaBorder Fault are also more depressed in thecentre of the depression than in both extremities.This may reflect the doming associated with theLate Quaternary activity of the Rungwe volcaniccentre.

The kinematics of faulting along the majorfaults of the Rukwa Rift Basin are more complexthan previously thought. An earlier episode ofstrike-slip faulting can be seen from slickensidespresent in the footwall of the Lupa Border Fault,and along the Kanda Fault in the Ufipa Plateau.Strike-slip faulting along the Kanda Faultdisplaced the Karoo sediments of the Namwelecoal field, but did not affect the mid-Miocenelaterite crust of the Ufipa Plateau. Along theLupa Border Fault, normal faulting occurred laterthan strike-slip faulting and seems to be relatedto the deposition of the Late Cenozoic lake beds.The Mbeya Range appears to be a stronglyuplifted and tilted block formed by asymmetricnormal faulting. The normal fault system wasreactivated in a right-lateral way in recent times.


In the Late Quaternary, both the Lupa andUfipa Fault scarps appear to be weakly active.Geomorphology, air photographs and field workhave failed to prove the presence of any recentmovement in the Late Quaternary. In thesouthern extremity of the Rukwa Rift Basin, anaxial zone of active deformation can be observedin Landsat (TM) imagery, field work and highresolution seismic profiles in the lake itself. Itcorresponds to a right-lateral northwest trendingstrike-slip fault system, disrupting theQuaternary lake beds in prolongation of theMbeya Range Fault. On the Ufipa rift shoulder,the Kanda Fault, together with the Mkunda andMwimbi Faults, form an active normal faultsystem. Movement along the former may havecaused the M 7.4 Rukwa earthquake of 13December 1910.

In conclusion, the major border faults seeminactive or weakly active and the brittledeformation is transferred to the centre of thedepression (Mbeya Range-Galula Fault Zone) andto the southwestern rift flank (Kanda Fault). TheLate Quaternary tectonics of the Rukwa region ischaracterised by a complex combination of strikeslip and normal movements between the MbeyaRange-Galula and the Kanda Fault Zones. Activedeformation is also expressed by regional tiltingof the whole rift depression and the domingassociated with the Rungwe volcanic activity.

Besides a marked climatic control of the lakewater level fluctuation and sedimentation, thesediment transfer was and is strongly controlledby tectonic activity. Vertical differential tectonicmovements control the location of thedepocentres and uplifted parts of the basin floor.In consequence, the Late Quaternarysedimentation and lake level in the Rukwa RiftBasin are controlled, besides the obvious climaticfactors, by active tectonics.

ACKNOWLEDGEMENTSThis work is a contribution to the RUKWA

Project "Investigation of the sediments of LakeRukwa (Tanzania): A clue for reconstructing thesouth equatorial climate during the last 130,000years" funded by the EEC, DG XII EnvironmentProgramme, Area 1.1. Natural Climatic Change,co-ordinated by M. Taieb and initiated by J.Klerkx. This project received the support of thefollowing Tanzanian institutions: Ministry ofWater, Energy and Minerals (5. Bugaisal, MADINI(P. Kenyunko) and TPDC (Y. S. Mwalyego). Thispaper benefitted from the comments andsuggestions of H. Wopfner and H. Dypvik.

This is also a contribution to projects IGCP400"Geodynamics of Continental Rifting" and INTAS93-134 "Continental Rift Tectonics andSedimentary Basin Evolution".

REFERENCESAdamson, D. A., Gasse, F., Street, F. A. and Williams, M.

A. J. 1980. Late Quaternary history of the Nile. Nature288, 50-55.

Ambraseys, N. N. 1991. The Rukwa earthquake of 13December 1910 in East Africa. Terra Nova 3, 202-211 .

Ambraseys, N. N. and Adams, R. D. 1991. Reappraisal ofmajor African earthquakes, south of 20 oN, 1900-1930.Natural Hazards 4, 389-419.

Angelier, J. 1989. From orientation to magnitudes inpaleostress determinations using fault slip data. JournalStructural Geology 11, 37-50.

Angelier, J. et Mechler, P. 1977. Sur une methode graphiquede recherche des contraintes principales eqaleme ntutilisable en tectonique et en seismologie : la methodedes diedres droits. Societe geologique France Bulletin 7119,1309-1318.

Butzer, K. W., Isaac, G. L., Richardson, J. L. and WhashbournKamau, C. 1972. Radiocarbon dating of East-African lakelevels. Science 175, 1069-1076.

Camelbeeck, T. 1992. Seismic activity in Southern Rukwawith data from Panda Hill station (September-November1991). UNESCO Newsletter Geology for EconomicDevelopment 9, 183-194

Camelbeeck, T. and Iranga, M. D. 1996. Deep crustalearthquakes and active faults along the Rukwa trough,Eastern Africa. Geophysical Journal International 124,612-630.

Cermicola, S., Vanhauwaert, P., Kilembe, E. and de Batist,M. 1996. High-resolution reflection seismic investigationof Lake Rukwa. Natural Climatic Change, Rukwa projectProgress report (unpubl.l, 33p. EC, DGXII Environmentprogramme 1.1 .

Chorowicz, J. et Mukonki, IVI. B. 1980. Lineamentsanciens, zones transformantes r e c ent es etqeotectonique des tosses dans I'est Africain, dapresla teledetection et la microtectonique. Rapport Annuel1979, pp143-146. Departement de Geologie etMineralogie, Mus ee Royal de I' Afrique Centrale,Tervuren, Belgium.

Clark, J. D., Haynes, C. V., Mawby, J. E. and Gautier, A.1970. Interim report on paleo-antropological investigationsinthe Malawi rift. Quaternaria 13, 305-354.

Cloetingh, S. and Kooi, H. 1992. Intraplate stress anddynamics aspects of rift basins. In: Geodynamics ofRifting,vol. II. Thematic discussions (Edited by Ziegler, P. A.)Tectonophysics 215, 67-85.

Crossley, R. 1982. Late Cenozoic stratigraphy of theKaronga area in the Malawi rift. Palaeoecology Africa15, 139-144.

Damblon, F. D., Gerrienne, Ph., dOutrelepont. H., Delvaux,D., Beeckman, H. and Back, S. 1998. Identification of afossil wood speciemen in the Red Sandstone Group ofsouthwestern Tanzania: stratigraphical and tectonicimplications. In: Tectonics, sedimentation and volcanismin the East African Rift System (Edited by Delvaux, D.and Khan, M. A.) Journal African Earth Sciences 26,387-396.

Damnati, B. and Taieb, M. 1995. Solar and ENSO signaturesin laminated deposits from lake Magadi (Kenya) duringthe PleistocenelHolocene transition. Journal African EarthSciences 21, 373-382.

Journal ofAfrican Earth Sciences 419

D. DELVAUX et al.

Delvaux, D. 1993a. Quaternary stress evolution in East Africafrom data of the western branch of the East African rift.In: Geoscientific Research in Northern Africa (Edited byThorweihe, U. and Schandelmeier, H.) pp315-318.Balkema, Rotterdam, Brookfield.

Delvaux, D. 1993b. The TENSOR program for paleostressreconstruction: examples from the east African and theBaikal rift zones. EUGVII, 4-8 april 1993, Strasbourg. TerraNova Abstract Supplement N° 1 5, 216.

Delvaux, D. and Hanon, M. 1993. Neotectonics of the Mbeyaarea, SW Tanzania. Rapport Annuel 1991-1992, pp8797. Departernent de Geologie et Mineraloqie, Muses Royalde I' Afrique Centrale, Tervuren, Belgium.

Delvaux, D., Levi, K., Kajara, R. and Sarota, J. 1992.Cenozoic paleostress and kinematic evolution of the Rukwa- North Malawi rift valley (East African rift system). BulletinCentres Recherches Exploration Production Elf-Aquitaine16,383-406.

Delvaux, D., Moeys, R., Stapel, G., Petit, C., Levi, K.,Miroshnichenko, A., Ruzhitch, V. and Sankov, V. In press.Paleostress reconstructions and geodynamics of the Baikalregion, Central Asia. Part II: Cenozoic tectonic stress andfault kinematics. Tectonophysics.

De Vos, A. 1994. Pliocene en Kwartaire evolutie vanhet Livingstone bekken (Malawi rift, Tanzanie)afgeleigd uit hoge-resolutie reflectieseismischeprofielen. M. Sc. dissertation 91 p. University ofGent, Belgium.

Dypvik, H., Nestby, H., Rudent, F., Aagaard, P., Johansson,T., Msindai, J. and Massay, C. 1990. Upper Paleozoicand Mesozoic sedimentation in the Rukwa- TukuyuRegion, Tanzania. Journal African Earth Sciences 11,437-456.

Ebinger, C. J., Deino, A. L., Drake, R. E. and Tesha, A. L.1989. Chronology of volcanism and rift basin propagation:Rungwe volcanic province, East Africa. JournalGeophysical Research 94, 15,783-15,803.

Ebinger, C. J., Deino A. L., Tesha A. L., Becker, T. andRing, U. 1993. Regional tectonic controls on Rift BasinMorphology: Evolution of the Northern Malawi (Nyasa)Rift. Journal Geophysical Research 98, 17,821-17,836.

Finney, B. P. and Johnson, T. C. 1991. Sedimentation inlake Malawi (East Africa) during the past 10,000 years: acontinuous paleoclimatic record from the southern tropics.Palaeogeography, Palaeoclimatology, Palaeoecology 85,351-366.

Haberyan, K. A. 1987. Fossil diatoms and the paleolimnologyof Lake Rukwa, Tanzania. Freshwater Biology 17, 429436.

Haberyan, K. A. and Hecky, R. E. 1987. The Late Pleistoceneand Holocene stratigraphy and paleolimnology of LakesKivu and Tanganyika. Palaeogeography, Palaeoclimatology,Palaeoecology 61, 169-197.

Hamroush, H. A. and Stanley, J. D. 1990. Paleoclimaticoscillations in East Africa interpreted by analysis of traceelements in Nile delta sediments. Episodes 13, 264-288.

Harkin, D. A. 1960. The Rungwe volcanics at the northernend of Lake Nyasa. Geological Survey Tanganyika, Memoir11,172p.

Jackson, J. and Blenkinshop, T. 1997. The BililaMtakataka fault in Malawi: An active, 1OO-Iong, normalfault segment in thick seismogenic crust. Tectonics 16,137-150.

Kazmin, V. 1980. Transform faults in the East African riftsystem. In: Geodynamic Evolution of the Afro-Arabian RiftSystem. Academia Nazionale Lincei, Roma. Atti deiconvegni Lincei 47,65-73.

Kilembe, E. A. and Rosendahl, B. R. 1992. Structure andstratigraphy of the Rukwa Basin. Teet, nophysics 209,143-158.


King, L. C. 1963. South African Scenery 308p. Oliver andBoyd, Edinburgh.

Lenoir, J.-L., Lieqeois, J.-P., Theunissen, K. and Klerkx, J.1994. The Palaeoproterozoic Ubendian shear belt inTanzania: geochronology and structure. Journal AfricanEarth Sciences 19, 169-184.

Mbede, E. I. 1993. Tectonic development of the Rukwa Riftbasin in SW Tanzania. Berliner GeowissenschaftlicheAbhandlungen A 152, 92p.

McConnell, R. B. 1947. Geology of the Namwele-NkomoloCoal Field. Geological Survey Tanganyika Short Paper 27,54p.

McConnell, R. B. 1972. Geological development of the riftsystem of eastern Africa. Geological Society AmericaBulletin 83, 2549-2572.

Morley, C. K., Cunningham, S. M., Harper, R. M. andWescott, W. A. 1992. Geology and geophysics of theRukwa rift, East Africa. Tectonics 11, 68-81.

Owen, R. B., Crossley, R., Johnson, T. C., DavisonHirschmann, S., Tweddle, D. and Engstrom, D. E. 1990.Major low levels of Lake Malawi and implications forevolution rates in cichlid fishes. Royal Society LondonProceedings B240, 519-553.

Peirce, J. and Lipkov, L. 1988. Structural interpretation ofthe Rukwa rift, Tanzania. Geophysics 53, 824-836.

Quennell, A. M., McKinlay, A. C. M. and Aitken, W. G.1956. Summary of the geology of Tanganyika. Part I:Introduction and stratigraphy. Geological SurveyTanganyika Memoir 1, 264p.

Ring, U., Betzler, C. and Delvaux, D. 1992. Normal vs. strikeslip faulting during rift development in East Africa: theMalawi rift. Geology 20, 1015-1018.

Scholz, C. A. and Rosendahl, B. R. 1988. Low lake standsin Lakes Malawi and Tanganyika, East Africa, delineatedfrom multifold seismic data. Science 240, 1645-1648.

Stewart, I. S. and Hancock, P. L. 1991. Scales of structuralheterogeneity within neotectonic normal fault zones inthe Aegean region. Journal Structural Geology 13, 191204.

Strecker, M. R., Blisniuk, P. M. and Eisbacher, G. H. 1990.Rotation of extension direction in the central Kenya Rift.Geology 18, 299-302.

Sylvester, A. G. 1988. Strike-slip faults. Geological SocietyAmerica Bulletin 100, 1666-1703.

Talbot, M. R. and Livingstone, D. A. 1989. Hydrogen indexand carbon isotopes of lacustrine organic matter as lakelevel indicators. Palaeogeography, Palaeoclimatology,Palaeoecology 70, 121-137.

Theunissen, K., Lenoir, J.-L., l.ieqeois. J.-P., Delvaux, D. etMruma, A. 1992. Empreinte mozambiquienne majeuredans la chaine ubendienne de Tanzanie sud-occidentale:geochronologie U-Pb sur zircon et contexte structural.Compte Rendus Acedemie Sciences Paris, Serie II 314,1355-1362.

Theunissen, K., Klerkx, J., Melnikov, A. and Mruma, A. 1996.Mechanism of inheritance of rift faulting in the westernbranch of the East African Rift, Tanzania. Tectonics 15,776-790.

Tiercelin, J.-J., Chorowicz, J., Bellon, H., Richert, J.-P.,Mwambene, J. T. and Walgenwitz, F. 1988. East Africanrift system: offset, age and tectonic significance of theTanganyika-Rukwa-Malawi intracontinental fault zone.Tectonophysics 148, 241-252.

Van der Beek, P., Mbede, E., Andriessen, P. and Delvaux,D. 1998. Denudation history of the Malawi and RukwaRift flanks (East African Rift System) from apatite fissiontrack thermochronology. In: Tectonics, sedimentation andvolcanism in the East African Rift System (Edited byDelvaux, D. and Khan, M. A.) Journal African EarthSciences 26, 363-385.


Vittori, E., Delvaux, D. and Kervyn, F. 1997. Kanda Fault: amajor seismogenic element west of the Rukwa rift (East Africa,Tanzania). In: Paleoseismology: using Quaternary geology tounderstand past earthquakes (Edited by Hancock, P. L. andMichetti, A. L.) Journal Geodynamics 24, 139-153.

Wescott, W. A., Krebs, W. K., Engelhardt, D. W. andCunningham, S. M. 1991. New biostratigraphic age datesfrom the Lake Rukwa Rift Basin in Western Tanzania.American Association Petroleum Geologists Bulletin 75,1255-1263.

Williams, T. M., Henney, P. J. and Owen, R. B. 1993. Recenteruptive episodes of the Rungwe volcanic field (Tanzania)recorded in lacustrine sediments of the Northern Malawirift. Journal African Earth Sciences 17, 33-39.

Wheeler, W. H. and Karson, J. A. 1994. Extension andsubsidence adjacent to a "weak" continental transform:an example of the Rukwa Rift, East Africa. Geology 22,625-628.

Ziegler, P. A. 1996. Geodynamic processes governingdevelopment of rifted basins. In: Geodynamic evolutionof sedimentary basins (Edited by Roure, F., Ellouz, N.,Shein, V. S. and Skvortsov, I.) pp19-67. Technip, Paris.

Geological and Topographical mapsGeological map of ltaka, Sheet 243; 1: 125 000. 1962.

Compiled by Kennerley, J. B. Geological Survey ofTanganyika, Dodoma.

Geological map of Mpui, Sheet 225; 1: 125 000. 1962.Compiled by van Loenen, R. E. and Kennerley, J. B.Geological Survey of Tanganyika, Dodoma.

Topographical maps of Tanzania, Sheets 207, 208, 242,243,244; 1:50 000. 1983. Sheets 169, 188,225,226,227,228; 1:50 000. 1984. Sheet 189; 1:50 000. 1985.Compiled by the Surveys and Mapping Division, Ministryof Lands, Housing and Urban Development, Dar-es-Salaam,Tanzania.


Late Quaternary tectonic activity and lake level change in the ......Late Quaternary tectonic...

Documents

Transcript of Late Quaternary tectonic activity and lake level change in the ......Late Quaternary tectonic...