TECTONICS,VOL. 16,NO. 1,PAGES137-150,FEBRUARY1997
The B ilila-Mtakataka
fault in Malai:
An active, 100-kin long normal fault segment
in thick seismogenic crust
JamesJackson
BullardLaboratories,Universityof Caxnbridge,Cambridge,England
Tom Blenkinsop
Departmentof Geology,Universityof Zimbabwe,Haxaxe
Abstract.
Some parts of the east African rift sys-
at the southern ends of the two branches of the rift
temhavedeeperearthquakes
(30-40 kin), a largereffec- system,in Tanzania(east)and MalaCvi(west;seeFigtiveelasticthickness
(~35 kin), and widerhalf grabens ure la). In m0•t otherregionsof continentalextension,
(~50kin) thanaretypicalin otherregions
of continen- earthquake centroids are shallower than about 15 km
tal extension.One such region is the southernpart of [Jacksonand White,1989]. Parts of the eastAfrican
the western branch of the east African rift in Mala/vi.
lithosphereclearlyhavea thickerseismogenic
layerthan
In this regionwe describea normal fault scarpthat is
upto 15 m high and continuousover a distanceof more
than 100 kin. It representsthe latest incrementof slip,
is typical elsewhere.
2. Studies of the coherencebetween gravity and
topographyhave shownthat parts of the east African
lithospherehavean effectiveelastic thicknessas high as
perhaps
in a singleearthquake,on a fault (the BililaMtakatakafault) with a total offsetof about1000m. It
20-40 km, evenwithin rifted areas[e.g.,Bechtelet al.,
is the longestcontinuousnormal fault segmentthat we 1987;Ebingeret al., 1989].This contrastswith smaller
knowof on the continents,and it supportsthe sugges- effective elastic thickness values of 10-15 km that are
tionthat the thicknessof the seismogenic
layer is the typical in other rifted areas [e.g., Barton and Wood,
fundamental
controlon the scaleof geological
structures 1984;Fowlerand McKenzie,1989;Bechtel½tal., 1990].
These anomalous earthquake depths and elastic
that form within it. A consequenceof this association
is thepossibilityof very large(M• 8.0), thoughinfrequent,normal faulting earthquakesin east Africa.
Introduction
thicknesses
are often attributed to the old (ArcheanProterozoic),cold, and strongmaterial in whichparts
of the east African rift systemare forming, in contrastto
placeslike Greece,Nevada, and Tibet, which tmveun-
dergonemuchyoungerorogenies[e.g.,Shudofsky
et al.,
Althoughthe east African rift systemis perhaps 1987; Nyblade and Langston,1995; Jacksonand Blenkthe most famousexampleof continentalextensionon insop,1993].
Earth,it differsin severalwaysfromregions
of extension
This paper is concernedwith a possibleconsequence
elsewhere
on the continents,
suchas Greece,Turkey, of the east African lithospherebeing strongerthan in
Nevada,and Tibet.
other rifts; that is the associatedgeologicalstructures
1. Earthquakesin parts of east Africa are deeper may be biggerthan usual. The norma.1fault systems
thanin mostotherriftedregions.This is clearfromthe that bound continentMrifts are commonlysegmented,
earthquake
centroiddepths,whichcanbedetermined
to stepping en echelonor changingthe polarity of half
withinabout:k3km usingteleseismic
P and SH wave- grabenalongstrike [e.g.,Rosendahlet al., 1986; Crone
forms,
andwhichseveralauthorshavereportedin the and ttaller, 1991;Robertsand Jackson,1991]. Jackson
range25-40kmfor earthquakes
in eastAfrica(see,e.g., and White[1989]and Wallace[1989]pointedout the
Jackson.
andBlcnkinsop
[1993],Senoand$aito[1994], similarity betweenthe maximum fault segmentlength
andNyblade
andLangston
[1995]for summaries
ofthis of-020-25 km commonin maw rifts and the downdip
work).Thesedeeperearthquakes
occureitheroutside width of thesefaults in the seismogenicupper crust.
themaintopographic
expression
of the rift systemor (Thesefaultstypicallydip at ~45 ø andextendto depths
of-o15 km.) They suggested
that the thickness
of the
Copyright
1997
bytheAmerican
Geophysical
Union.
seismogenic
layer in someway influencesthe lateral continuity of the fa,ults tha.t form witlfin it,. Jackson and
Paper
number
96TC02494.
0278-7407/97/96TC.024
94512.00
White [1989]also suggested
that the ett•ctiveelastic
thicknessof the lithospheremight be expectedto con137
138
JACKSON AND BLENKINSOP: BILILA-MTAKATAKA FAULT
12
2o
20
30
40
50
a 60
16
33
36
Figure 1. (a) Epicentersof earthquakesin east Africa whosecentroiddepthsare constrainedto be
deeperthan 20 km from analysisof body waveforms
[fromJacksonandBlenkinsop,
1993;Nybladeand
Langston1995;A. Foster,personalcommunication,
1996]. The main faultsof the easternand western
rift systemsare shown. The March 10, 1989, earthquakein Mala/vi is marked M. Note how these deeper
earthquakesapparently occur at the southernends of the easternand westernrifts, or off the main axes
of the rifts altogether. The area of Figure lb is markedby a box. (b) Seismicityof southernMa!a-&i,
from the U.S. GeologicalSurvey Preliminary Determinationof Epicenterscatalogue1963-1989. Except
for the earthquakes
of March 1989(markedM), theseare all smalland poorlyrecorded:their locations
may be in error by at least 25 kin. LM is Lake Mala•i and LC is Lake Chilwa.
trol the maximum width of coherentlytilted blocksor
half graben if the shear stressesacting on the faults
are not to exceedthe typical range of 1-10 MPa stress
drops seenin earthquakes. There is no reasonwhy the
seismogenic
thicknessand the elastic thicknessshould
have the same values, but since bo•h are probably related to thermal structure[e.g., Wiensand Stein,1983;
BurovandDiament,1995],we might expectthemboth
to increaseor decreasetogether. The hypothesisis simple: the thicknessof the strong outer layer of the lithospheremight control both the lateral continuityof the
faults and the across-strike width of the tilted blocks or
half grabenthat form within it [seealsoHaywardand
Ebinger,1996].
in most regions of continental extensionthe thickness of the seismogeniclayer varies little from 10 to
15 km, so the above hypothesisis unrestable. However,
parts of the east African rift system appear to be colder
and strongerto greater depths than elsewhere.Are the
structurescorrespondinglybigger?
Someof the east African half graben have widths
of up to 60 kin, which is far wider than the typical
maximum
width
of ~25
km that is observed elsewhere
sonandBlenkinsop
[1993]andFosterandNimrao[1996]
showthat such wide structuresdo not require abnormally high shearstresseson the boundingfaults provided they are supportedby an elastic layer ~30 km
thick rather than the more usual 10-15 km.
This as-
pect.will not be discussed
further.
Faultsegment
lengths
aremore
pr•)blematic
because
they are harder to assessobjectively,particularly frorn
publishedmaps. One problemis scale;faults whichare
drawnascontinuous
at onescalemay appearsegmented
whenexaminedin moredetail [e.g., Walshand Walterson, 1991; Dawers and An&rs, 1995]. There is also
a problemof definition, since most fault surfaceshave
irregularitiesin the form of corrugationsand bendsat
all scales. A concernof this paper is fault segmenta•
_
lion that can potentiallylimit the extentof rupture
and hencethe sizeof earthquakes[e.g., Schwartzand
$ibson,1989]. Lateral or en echelonoffsetsas smallas
1-2 km acrossstrike are known to be able to halt rup-
ture [e.g.,Lindh andBoore,1981;Jacksonel al., !982;
Yielding,
1985],sofewseismic
surveys
(andnoneofthe
published
oneson the Africanlakes)are denseenough
overlongenoughdistancesto be of useto us here. It is
easierto demonstratecontinuityon land.
reachedobjectivelyfrom high-quality seismicreflection
In this paper we describethe Bilila-Mtakataka fault,
data,particularlyin eastAfricanlakes[e.g.,$pechtand southof Lake Mala•i, whichwe believeis a continuous
RosendaM,
1989;Flanneryand Rosendahl,
1990].Jack- fault, with no significantinterruptionsover a distance
[Jacksonand White, 1989]. This conclusion
can be
JACKSON
AND
BLENKINSOP'
BILILA-MTAKATAKA
FAULT
139
of at least100km. It is the longestcontinuous
normal pattern, these two aspectsare discussedfirst before de-
faultsegment
weknowofonthecontinents.
It islocated scribingthe fault itself in detail.
about50 km from the epicentersof earthquakes
in 1989
that had centroiddepthsof ~32 km, wherehalf graben Geomorphology
widthsare 40-50 km, and the effectiveelasticthickness
In the Bilila-Mtakataka region the west side of the
is around30 km [Jacksonand Blenkinsop,
1993]. The
rift valley is bounded by two subparallelfault escarpfaultis clearlyin a.regionwherethe Africanlithosphere
ments that separatea fiat plateau with someremnant
is strong.
hillsalongthe Mozambiqueborder(elevation~2000 m)
fromLakeMalafvi(~470 m) andthe Bwanje-Liwawadze
TectonicSettingof the Mala•i Rift
valley(Figures2 and3). The westernescarpment
isnot
The Mala/•i rift is located at the southern end of
well exposedor easilyaccessibleon the ground,as most
of it is covered in forest. It is not continuous but con•he westernbranch of the east African rift systemand
consists
of a seriesof basinstypically 100 km long and
sistsof threemain segments
(Figure2) (1) an unnamed
50 km wide which havethe morphologyof half grabens scarpup to 1000 m high that runs from north of Tsekthat stepor changepolarityalongstrike(seefbinger wereto the latitudeof Golomoti,(2) the Chirobwefault
½tal.,[1984,1987,1993]andFigure2, inset).Mostof [Walshaw,1965],forminga scarp300-1000m highand
the fault systemsboundingthesegrabensare offshore separatedfrom the scarp to the north by a step to the
oralongtheshoreof thelake.However,
at thesouthern right westof Golomoti,and (3) the Ncheufault [Walshaw,1965],forminga scarpup to 300rn high,whichis
endof the lake the basin-boundingfaults on the west
offsetin the north by a step to the right from the Chisideoccurentirely on land, and thesefaults are the subjectofthisstudy.In thisregiontherearenorockswith robwefault and decreasesin height to the south, dying
anageintermediate
betweena metamorphic
basement
complex
of presumedProterozoicageandlate Tertiary
sediments.
out at the Rivi Rivi river.
The eastern escarpment, bordering the BwanjeLiwawadze valley, is the Bilila-Mtakataka Fault. Between the eastern and western escarpmentslies the
The southernpart of the Mala•vi rift hashad fewsigshelf(namedafter the mainriversdrainnificantearthquakes
in moderntimes(Figurelb). The Nsipe-Livelezi
mostimportantfor this study are the two earthquakes ing it to the southand north). This shelfis an almost
of March9 (Mw 5.7), and March10 (Mw 6.1), 1989, fiat erosionsurface(Figure4) that mergeswith the Liwawadze Valley in the south near Makenzie and with
areasin Figures2 and 3. Thesehad centtolddepths the coastal plain of Lake Mala•i north of Mtakataka.
of 32•: 5 km [Jacksonand Blenkinsop,
1993;Nyblade The shelfis deeply dissectedby rivers, particularly near
andLangston,1995] and producedno faultingat the the easternescarpment,and is clearly the same erosion
surface.The larger earthquakehad a well-determined surfaceas that forming the Bwanje-LiwawadzeValley.
whichoccurrednear 13.7øS,34.4øE, just north of the
normalfaultingmechanismwith a strikeof 154•: 25ø
Accordingto Lister [1967]it may alsobe the samesur-
•nd an extension direction of 060 :k 30 ø. The total ex-
face as that of the high plateau west of the Chirobwe
tensionacrossthis part of the rift is unlikelyto exceed
10kin, andmay be muchless.All kinematicmodelsof
platemotionssuggestmaximumextensionrates across
the wholerift systemof <3-4 mm/yr [DeMetset al.,
1990;Jestinet al., 1994],probablydecreasing
to the
south.Suchrates,evenif operativeover 10 Myr, would
heightsof up to about 300 m abovethe generallevel
of the Nsipe-Livelezishelf, particularly near its eastern
edge(Figure4). The shelfitself may be gentlytilted to
the west;Lister[1967]suggests
by about30 m. Though
Lister[1967]describes
the erosionsurfaceas "African",
produceextensionssimilar to the 16 km estimatedby
it is of uncertain age.
Ebinger[1989]in Kivu-Rusizirift at the northernend
of the western branch of the east African rift system.
Bilila-Mtakataka
Fault
Figures2 and 3 show the western side of the rift
at the southernend of Lake Malafvi. The regionalgeologyof this area is describedin reports and 1:100,000
and Ncheu faults.
Several hills of resistant rocks rise to
The heightof the Bilila-Mtakatakafault escarpment
is much lessthan the Chirobwe-Ncheuescarpm.ents
to
the west,reachinga maximumof about 500 m near the
Nsipe-Livelezi
divideat 14ø45•S(Figures5-8). Except
at one locality discussedbelow, only sedimentsof the
Bwanje-Liwawadzevalley are exposedin the immedi-
ate hangingwall (i.e., the easternside)of the BililaMtakataka fault. These sediments are young, uncon-
scale
mapsby Walshaw[1965],DawsonandKirkpatrick solidated sandsand silts, some of which may originally
[1968],
•na rAafch•r[1968].Sincethe loc•]geomor- havebeenlakedeposits[DawsonandKirkpatrick,1968],
phology
providesthe best constrainton the total offset mixed with alluvium and reworked by surfacestreams.
on the Bilila.-Mtakataka
fault and since the basement
structure has a clear influence on the fault's outcrop
The sediments are known to be at least 50-100 m thick
'in places[Wal•haw,
1965;DawsonandKirkpatrick,
140
JACKSON
AND BLENKINSOP-
LAKE
BILILA-MTAKATAKA
MALAWI
FAULT
N
WtAK•IAKA
I
I
12-
13-
Figure 2. Summarytopographic
andlocationmapof thefaultsystemat theSW endof LakeMala-&i.
Topography
iscontoured
at 100-m
intervals
andshaded
above
900monlyontheNsipe-Livelezi
shelf(or
erosionsurface),whichoccupies
the regionbetweenthe Bilila-Mtakataka
fault and the Chirobwe-Ncheu
faultsystem.Spotheightsarein meters.The railwayis markedby a linewithticks.The linesof the
sections
S1toS3in Figure
7 andA-B-Cin Figure
8 arealso
shown.
Theareaoftheairphoto
in Figure
13
ismarked
bya box.Thelocations
ofthephotos
in Figures
4-6 and9-12areshown
bysolidnumbered
boxes.
JACKSON AND BLENKINSOP- BILILA-MTAKATAKA
FAULT
LAKE MALAWI
N
1
i
w
t
......
I
2
3
4
I
!
?
1
,
,
i
ß
I
y'- ..,{
I
/.
45'
3S"E
Figure 3. Structuralmap of the fault systemsSW of Lake Mala•vi,basedon work of Walshaw[1965],
Thatcher[1968]and Dawsonand Kirkpatrick[1968],aswellasourownobservations.
Thicklinesarethe
major faults. Medium width lines mark a lessimportant fracture pattern generallytrending WNW. Thin
lines showthe generaltrend of the foliation in the gneissicbasementrocks: Arrowsshowrepresentative
dip values and dip directions of this foliation. The shaded areas A and B show the only two outcrops
of gneissin the immediate hanging wall of the Bilila-Mtakataka fault. Numbersin circlesshow the locationsand directionsof views in Figures 4-6 and 9-12. The inset showsa lower hemisphereequaI-are•
projection of fracture planes found within the fault zone of the Bilila-Mtak•taka scarp. A total of 11
orientations
at sevendifferentlocationsareshown.Alsoshown(opencircle)is a slipvectorwith azimuth
060ø (the sameslip vectorazimuthas in the March 10, 1989,earthquake)and a plungeof 60ø to the
ENE.
141
142
JACKSON AND BLENKINSOP' BILILA-MTAKATAKA FAULT
Figure 4. View southacrossthe Nsipe-Livelezi
shelf(or erosionsurface)fromthe stepat the northern
endof the Chirobwefault segment(seeFigure2 for location).The Chirobwefault escarpment
is on the
right, and on the left are isolated small hills in the immediate footwall of the Bilila-Mtakataka fault.
1968]andto thin eastward,wheretheyoverlapexposed alongits strikeis small: -,-20m/kin. This gentlegradient is one of the most strikingfeaturesof the fault,
basementgneisseseast of the Bwanje and Liwawadze
rivers. Thus the total offset on the fault is unlikely to
exceed1000 m, and the average displacementgradient
which seems to have an ahnost constant footwall eleva-
tion overlargedistances(Figures5, 6, and 9).
Figure 5. ViewSW of a straightpart of the Bilila-Mtalaxtaka
faultbetween
Golomoti
andMua (see
Figure2 for location). The Chirobwefault escarpment
is in the background,
and the Bilila-Mtakataka
faultis theescarpment
separating
thewooded
Nsipe-Liv•lezi
shelffromthecultivated
Bwanje
valleyin
the foreground.
JACKSON AND BLENKINSOP' BILILA-MTAKATAKA FAULT
143
.•.
Figure 6. View of the Bilila-Mtakataka
fault about1 km southof Mtakataka,lookingSW (seeFigure2
for location).The Nsipe-Livelezishelfis narrowerherethan in Figure4, and this part of the fault is less
straight.In the backgroundis the escarpment
of the fault segmentnorth of the Chirobwefault.
Basement Geology
cially simple. A very prominent gneissicfoliation has a
fairly uniform strike betweenNW and NNW and gen-
The entire regionwest of the Bilila-Mtakataka.escarpmentconsistsof hornblende-biotite
gneisses
and erally dips NE at between40ø and 80ø (Figure 3). In
charnockitic
granulitesinterbandedwith somequartzo- detail the structure is more complex, with local deviafeldspathic
gneisses
and calc-silicategranuliteswhich tions in dip and strike and tight-to-isoclinal folds of the
formsomeof the hills and ridgeson the Nsipe-Livelezi sametrend that lead to repetition of somebanding. The
shelf.Theserocksare of uncertainage but are thought trend of the major rift-bounding faults clearly follows
to belongto the Mozambiquebelt and are probably the generalfoliation in the gneisses(Figure 3) [Walshaw, 1965; Dawso•tand Kirkpatrick, 1968] but with
Proterozoic.
somedeviationsfrom it, which are discussedbelow. The
foliation
has given a stronggrain to the gcomorphology
faultthe structureof the basementgneisses
is superfion the Nsipe-Livelezishelf, and most of the larger river
Over the entire area west of the Bilila-Mtakataka
courses follow it.
Another important structural fabric within the basement gneisses
are fracturesor joints with a WNW strike
that are particularly prominent in the northern half of
the area (Figure3). This fabric is very clearon satellite imagery and influencesthe coursesof rivers and
I[I
Ill
Ill
IllIlllllllllllllllllll!1111llllllllllllllllllllllllllllllll
llllllllllll
llllllllllllllll
,!
0
Figure 7.
10
20
30
4•
•
•0
70
•0
•
10•m
WSW-ENE topographicsectionsacrossthe
Nsipe-Livelezi
shelf(seeFigure2 for location).The positionof the Bilila-Mta'k•taka(BM) fault is markedin the neath is a parallel longitudin• section•ong the Bw•nj•
east.
Liwawadzi vMley.
144
JACKSON AND BLENKINSOP' BILILA-MTAKATAKA FAULT
Figure 9. View on the ground looking SW at the Bilila-Mtakataka fault scarp between Mua and Golo-
moti, closeto the area in Figure 5 (seeFigure2 for location).
minor streams, several of which make dramatic riglit
anglebendsor zigzagsas their coursesswitch between
from Makenzie
to north
of Tsekwere.
We will now de-
scribeits characterfrom southto north (Figures2 and
the directions of the foliation and the fracture fabric.
3). In what follows,when we refer to "the fault" we
mean the Bilila-Mtakataka fault as a whole (i.e., the
Bilila-Mtakataka
Fault
fault responsible
for the offsetof the erosionsurface),
The Bilila-Mtakataka
fault runs from Makenzie viland by "the scarp" we mean the clean topographicstep
lagein the south (Figure 2) to north of the Tsekwere 4-15 m high that marks the most recent movementon
Estate in the north, a straight-line distanceof about the fault.
The fault has a distinct southern end. Between the
110km (Figures2 and 3). Unlike the Chirobwe-Ncheu
faults to the west, the Bilila-Mtakataka fault is well ex- LinengweRiver, where the scarp height is about 4 rn,
posed,beingon the edgeof the cultivatedBwanjevalley. and Makenzie village the topographyassociatedwith
A roadanda railwayfollowthe bulkof its length,andit both the fault and the scarp die away completelyinto
is alsoaccessible
by the major riversthat cut through a gentle
•onoclineor warpthat merges
withtheLithe footwall in gorges. We visited many localitieson wawadzi valley. This is a well-known morphologyat
foot and flew along the entire fault at low level in a the end of a normal fault segment[e.g., Jacksonand
lightaircraft.We alsoexaminedcriticallocalitiesusing Leeder,1994]. BetweenMakenzieand about 5 km NW
air photographs.
of Bilila the fault more or lessfollowsthe gneissicfoAn astonishingfeatureof the Bilila-Mtakatakafault, liarion (Figure 3), and the scarpis continuousexcept
remarked
on in the earlierworkby Walshaw
[1965]and for two minor stepsat Bilila. These steps to the left are
Dawsonand t½irkpatrick[1968], is the existenceof a offsetabout 80 m and 40 m acrossstrike, and the length
steep,cleanfault scarp4-15 m highthat truncatesspurs of the short middle scarplinking the two long onesis
andis veryclearon the ground(Figures9-12), from lessthan 100 m. This disruptionappearsto coincide
theair (Figures
5 and6), andin air photographs
(Fig- with a local change in strike of the gneissicfohation
ure 13). Althoughno slickensides
or striatedsurfaces from about 140 ø in the south to 170 ø in the north. The
are visibleon this scarp(perhapsbecause
of the gran- scarp height does not changeobservablywhile crossular natureof the rocks),in many placeswe founda ing thesesteps, and the whole disruption is obviously
strongfracturefabric, sometimeswith gougematerial, of only minor significance:yet this is the biggestdisin the rocksformingthe scarp.Its originas a fault and continuity in the scarp we could detect over its entire
asthe mostrecentexpression
of movementon the Bilila- length.
Mtakatakafault •sclear.Thisscarpcanbe traced,with
nosignificant
breaks,alongtheentirelengthof thefault
Between about 5 km north of Bilila and the Liwadze
River the fault cutsacrossthe complicated
patternof
JACKSON
AND
BLENKINSOP-BILILA-MTAKATAKA
FAULT
145;
line, closelyfollowingthe basementfoliationonceagain.
Thesechangesin strikebetweenSharpevaleand north of
the LiwadzeRiver crossingare the mostdrmnaticseen
alongthe fault, yet the cleanscarpis continuousround
them all (Figure 13). In the cornerby the Liwadze
River crossing,and for the next 8 km north, are the
only exposuresof basementgneissin the hangingwall
of the fault (markedA and B in Figure 3). We are
uncertain how to interpret these. These outcropsmay
imply a smaller total offset on the fault at this point
(thoughtheyoungest
scarpitselfdoesnot changeheight
significantlyin this region). Alternatively,they may be
remnants of high topography in the erosionsurfaceof
the Nsipe-Livelezi shelf, which is studded with minor
hills, particularly in this region.
From Kasinje to Mua the fault is relatively straight,
followingthe gneissicfoliation most of the way (Figure 5). It makesa minor bend where it crossesthe
Livelezi River, but the scarpis identifiableon both sides
and hascauseda sing_le
terraceto be preserved
in the
uplifted
footwall
onthesouthsideat a•ei•ht ofabout
10 m.
Above the terrace is a more subdued
riser also
-.•10m high that becomesa fiat shelf in the gneissabout
100 m wide before rising again up to the higher erosion
surface. This morphologyis a common feature of the
fault' we observed it at several localities, particularly
at Bwanje and south of Mua, and it is remarked on
by Walshaw[1965]and Dawsonand I(irkpatrick[1968].
Their suggestion,which we think reasonable,is that
this higher bench in the gneissrepresentsa slip event
Figure 10. View westof a waterfallon the Bilila-Mtakataka
scarpwestof Kasinje (seeFigure 2 for location). The waterfall has retreated about 50 m upstream from the line of
the scarpand is formed in foliated gneiss.T. Blenkinsop,of
height1.70 m, providesthe scale.
foliationsin the basement rocks. Its outcrop is much
lessstraight and someof the curvesin its trace are severe,as it changesstrike from NNW to WNW in places.
prior to the one(or ones)that producedthe currentfree
scarp face.
At Mua mission itself the scarp is about 10-12 m
high and exposesunconsolidatedsedimentsoverlying
_
weatheredbasementgneissin the footwall [Dawso•,
and Kirkpatrick, 1968]. These depositsappearto have
formed in a channel and may represent a terrace in the
nearby Myendoparive River. From Mua northward the
fault makes a seriesof straight zigzagsas it alternates
Yet throughoutthis regionthe scarpitselfis continuous between following the WNW fracture fabric and the
(e.g.,Figure13). Westof Sharpevale
(Figure3), in a NNE gneissicfoliation. The scarp is continuousround
placewherethe scarpis exposedby burningof vegetation and is about 8 m high, a clear stream terraceis
preserved
by uplift in the footwall (Figure 11). Here
the scarpitself is a very steepfree facethat waspreservedfrom erosionby vegetationbut is nowdegrading.
It contains
a strongfracturefabricparallelto thescarp
in brecciated
gneissbut no slickensides.
BetweenSharpevaleand the LiwadzeRiver the fault
changes
strike to WNW as it crossesthe foliation in
calc-silicate
gneisses.
In this regionit may be following
theWNW trendingfracturepatternthat is prominent
inthebasement
rocks(Figure3). At theLiwadze
River
itselfthe fault changes
strikeabruptlyfromWNW to
NNW. It then continuesnorth as an almoststraight
all thesechangesin direction, maintaining its height at
about
8-10 m.
At 8 km north
of Mtakataka
it turns
abruptly WNW toward Tsekwere, and on this corner,
valley floor sedimentsare seenin the footwall abovethe
gneiss(Figure12).
We wereable to followthe scarp(frontthe air) an._
other 2 km NW of Tsekwere, where a farm is built on an
uplifted river terrace. Beyond that the fault is not clear:
the region is thickly forested, and the Nsipe-Livelezi
shel•is no iongeridentifiable
in the footwailbecause
the high western escarpmentthat marks the fault segment north of the Chirobwe fault in some way meets,
interferes, or merges with the NW trend of the fault
through Tsekwere. We could not confirm Walshaw's
146
JACKSON AND BLENKINSOP' BILILA-MTAKATAKA FAULT
Figure 11. View westof the steepscarpon the Bilila-Mtakatakafault westof Sharpevale(seeFigure 2
for location). The fiat surfaceis a river terraceoffsetabout8 m abovethe valleyfloor. The scarphas
recently been exposedby burning.
[1968]suggestion
that the Bilila-Mtakatakafault continues 30-40
km farther
north
of Tsekwere
to Salima.
Somefeelingfor the youth of the scarpitself may be
obtainedwhereit crosses
rivers. West of Kasinje(Figure 2) is a waterfall ~15 m high (Figure 10) that has
retreated upstream about 50 m from the place wherethe
scarp crossesthis moderate-sized stream. This rather
small distanceis typical of the amount of scarp retreat,
evenon biggerstreamscrossingthe fault south of Mua
(th mikokw)
(th god.i).
Figure 12. The Bilila-Mtakatakascarp8 km northof Mtakataka(seeFigure2 for location)whereit
turns abruptly WNW toward Tsekwere.This view is to the WNW just southof the turn. The scarpis
about8 rn high and herehas softvalleysediments
in the upliftedfootwall,whichhaserodedand gullied
more than is commonwhere only gneissis exposed.However,the sedimentcoverin the footwall is only
a few meters
thick.
On
JACKSON AND BLENKINSOP-BILILA-MTAKATAKA
0
1
....
FAULT
2
147
3
4
,
km
Mitongwe
N
,
ß
Yiga• 2 œo•lo½•[•o.). Th• s½•p •ts½1œ
i$ [h• ½1• white !•e.
muchbiggerLiveleziRiverwestof Golomotithe nick and at Mua) and uplifted river terracesare preserved
pointcausedby the scarpis now an area of rapids about
100m upstream. The dischargefrom theserivers in the
wet seasonis considerable,and the fact that the nick
pointsare still identifiable closeto the scarpsuggeststo
us that the motion on the scarp is not very old, but we
cannotquantify this.
Unresolved
Issues
There are various aspects of the Bilila-Mtakataka
fault that we do not understandand •vhichwe identify
below. However,first we wish to emphasizethe most
in its footwall,(2) nickpointsin moderate-sized
rivers
havenot migratedfar upstream,and (3) the scarpface
is still very steep suggestthat the movementis relatively young, but we cannot confidentlyqual•tify this
statement. If the scarp moved 10 m in one slip event
(seebelow)andif the long-termsliprate were1 mm/yr
(improbablyhigh, sincethere are other faults at this
latitude), suchan eventwould only be requiredevery
10,000 years.
What is the slip vector on the fault?
unable
to find
slickensides
or striations
ciated fault zone on the scarp.
We were
in the
brec-
Chorowicz and Sot-
important feature that we do understand and which is
lien [1992, Figure 2] indicate a NW-SE extensiondi-
the point of this paper: that the frontal scarp,which
marks the most recent episodeof movementon this
rection on this fault, suggestingalmost pure strike-slip
motion.
fault, is continuousfor at least 100 km from Makenzie
"on surfacesparallel to and within 100 m of the major
Their
measurements
of striations
were made
in thesouthto Tsekwerein the north. Thisis in spiteof fault" (page 1019). We cannot commenton whether
thevariousbendsand zigzagsin the fault trace,whos• their measurementsrecord the same age of deforma-
stepsat Bilila, which anyway are only 12 km from the
tion as the most recent slip on the fault. However, a
substantial strike-slip component, or NW-SE extension
direction, seeinsunlikely to us for three reasons.First,
the motion on the scarp is obviously mainly vertical
southern end of the fault.
or dip slip. We were unable to detect any convincing
outcropis clearly influenced by preexisting basement
foliationor fracturetrends. The only discontinuities
we
couldfind were the obviouslyminor and unimportant
The mostobviousuncertaintyis the ageof themove- strike-slip offsets,either on the ground or from the air
mentrepresentedby the frontal scarp. The observations (both aeroplaneand air photos). This was true even
that(1) softvalleyfloorsediments
(northof Mtakataka when the strike of the fault changesdramatically, as
145
JACKSON AND BLENKINSOP: BILILA-MTAKATAKA FAULT
it doesbetweenSharpevaleand Kasinje(Figure 13).
Second, because the fault strike varies so much, NW-
SE motion would inevitably produce local shortening
in someplacesthat would at the very least changethe
nature of the vertical displacementon the scarp. We
sawno evidenceof this anywhere. Third, the March 10,
1989, earthquake,only 40 km north of Tsekwere,has a
well-determinedslip vector in the direction 060 4-30 ø,
quite differentfrom the NW-SE directionsuggested
by
Chorowicz
and Sotlien[1992]. A slip vectorin thisdirectionwouldproducealmost pure normal faulting on
the Bilila-Mtakataka fault. A plot of the fracture planes
we foundin the faulted rocksof the scarpitself is shown
in Figure 3, inset. Most of these planes could have
a common slip vector with an azimuth close to that
of the March 10, 1989, earthquake(060ø, plungingat
60ø) withoutrequi•'ingmuchinternaldeformation
of the
footwallor hangingwall.
Does the continuousscarp along the fault represent
slip in a single large earthquake? From the observational point of view we saw no evidencefor severalslip
eventsproducing the steep frontal scarp. Where we
saw uplifted river terraces or waterfalls, we saw a single riser the height of the whole scarp: there were no
signsof multiple stepswithin the frontal scarp. However, theseobservationsmay not be conclusiveif, for example, severalearthquakeshappenedclosetogetherin
time. If the scarp were producedin a singleevent, with
an averageslip of ~10 m over 100 km length, it would
be the biggestnormal faulting earthquakeknown on
the continents.For comparison,two of the largestcontinental normal faulting earthquakesthis century are
(1) the 1959HebgenLake,Montana,earthquake
(total
faultlength~40 km;average
slip~6 m; M0~102øN m;
Mw 7.3 [Myersand Hamilton,1964])and (2) the 1915
PleasantValley,Nevada,earthquake(total fault length
~60 km; averageslip ~3-4 m; M0~0.7 x 10•ø N m;
Mw 7.2 [Wallace,1984]).In both of theseearthquakes
quakes[I(ikuchiand Kanamori,1995]. Theseevents
achieve
theirgreatsizeby rupturingoceaniclithosphere
to depths
of40kmormore[e.g.,Chapple
andForsyth,
1979;WiensandStein,1983].
Discussion
Our work in Mala/vi set out to discover whether
there was any evidencefor continuousfault segments
substantiallylonger than about 25 km in regions
where anomalouslydeep normal faulting earthquakes
occur. This questionhas been answeredby the Bilila,
Mtakataka scarp, which is certainly continuousfor at
least100km (as, indeed,waspointedout by Walsha•v
[1965]and DawsonandIfirkpatrick[1968]).Sincethis
scarp obviously representsthe most recent increment
of motion on the fault and is traceablealongits entire length, our conclusionregardingits continuityis
not subjective. This is lucky for us, as it is sometimes
difficult to be objectivewith argumentsbasedon mor-
phologyor structure alone. For our purposes,we can
bypassdifficult questionsabout the origin and evolution of the fault. For example, the dramatic bend in
the fault at the LiwadzeRiver westof Sharpevale,with
the exposedgneissin the hangingwall, is similarto the
geometryexpectedwhen two en echelonfaults merge
alongstrikeby breakingthe "relayramp" betweenthem
[e.g., Trudgilland Cartwright,1995]. Thus the fault
may have originatedas two segmentsthat joined to-
gether.Thisdoesnotmatterforusasthe (remarkable).
fact that the scarpis continuousthroughthis region
demonstrates
that now it behavesas a singlefault.
The Bilila-Mtakatakafault thus supportsthe notion
that the thickerseismogenic
crustin this regionis associatedwith longernormal faults as well as with wider
halfgraber/õ
and greaterelasticthickness[Jackson
and
Blenkinsop,
1993]andthat the thickness
of the seismogeniclayer may be the fundamental control on the scale
of geological
structures
in the crust[Jacksonand White,
!989; Wallace,1989].A consequence
of this association
is the probabilityof very large,but infrequent,normal
faulting earthquakesin parts of east Africa. However,
it is clear that one exampledoesnot prove the gencrustin
depth,wouldhavea momentof ~102•N m (Mw 8.0). eral point. Other areaswith thick seismogenic
the surfacefaulting occurredin discretesegmentsthat
probablymovedin discretesubeventsidentifiablein the
seismograms
[Doser 1985, 1986]. By contrast,a single
slip event on the Bilila-Mtakataka fault, with average
displacementof 10 m over 100 km extending to 40 km
Africa and elsewhere need to be examined.
One such
Such a large earthquake with a large slip increment
is a straightforwardconsequence
of the increasedfault regionis around Lake Baikal, which may also have a
[DiamentandKogan,1990]and
lengthandseismogenic
depth(whichincreases
thefault largeelasticthickness
earthquakes
as
deep
as
40 km [DiverchOre
et al., 1991].
area). Nonetheless,we know of no normal faulting
eventsof this size on the continents this century, and
the historical
record in Africa
itself is too sho•'t to be of
Acknowledgments. We thank NERC for supportof this work
use[Ambraseys
and Adams,I991]. A reasonable
ana- in Mala•i (grantGR9/997) andShellandAmeradaHessforgenloguefor suchan event may be the big normal faulting
earthquakesseaward of trenches in subduction zones,
which can even exceed this in size, such as the 1933
manuscriptand to Clive Oppenheimerfor help with satelliteim-
Sanriku(Mw 8.4) and 1977Sumbawa(M•o8.2) earth-
agery. Cambridge Earth Sciencescontribution 4740.
eral supportof our work in Africa. We are grateful to Cindy
Ebingez',Dan McKenzie, and Adrian Foster for commentson the
JACKSON
AND
BLENKINSOP:
BILILA-MTAKATAKA
FAULT
149
References
Ambraseys,
N.N., andR.D. Adams,Reap-
Ebinger, C.J., Tectonic developmentof the
praised
of major Africanearthquakes
south of 20øN, 1900-1930, Nat. Hazards,•l, 389-419, 1991.
Barton,P.J., and R. Wood,Tectonicevolution of the North Sea basin: Crustal
stretchingand subsidence,Geophys.J.
R. Astron. Soc., 7g, 987-1002, 1984.
Bechtel,T.D., D.W. Forsyth, and C.J.
Swain, Mechanismsof isostatic com-
pensationin the vicinity of the east
Africanrift, Kenya, Geophys.J. R. Astron. Soc., 90, 445-465, 1987.
Bechtel,T.D., D.W. Forsyth,V.L. Sharpton, and R.A.F. Grieve, Variations
in effective elastic thickness of the
North American lithosphere, Nature,
363, 636-638, 1990.
Burov,E.B., and M. Diament, The effective elasticthickness(Te) of the continental lithosphere: What does it really
mean?, J. Geophys. Res., 100, 39053927, 1995.
Chappie, W.M.,
and D.W. Forsyth,
Earthquakesand bending of plates at
trenches, J. Geophys. Res., 8,i, 67296749, 1979.
Chorowicz,J., and C. Sorlien, Oblique extensional tectonics in the Mala'&i rift,
Africa, Bull. geol. Soc. Am., 10,i, 10151023, 1992.
Crone, A.J., and K.M. Hailer, Segmentation and coseismic behaviour
and-range normal faults:
of basin-
Examples
from east-centred Idaho and southwest-
ern Montana, U.S.A., J. Struct. Geol.,
13, 151-164, 1991.
Dawers, N.H.,
and M.H.
Anders,
Displacement-length scaling and fault
linkage, J. Struct. Geol., 17, 607-614,
1995.
Dawson,A.L., and I.K. Kirkpatrick, The
geologyof the Cape Maclear peninsula
and Lower Bwanje valley, Bull. Geol.
Surv. Malalbi, 28, 71 pp, 1968.
DeMets, C., R.G. Gordon, D.F. Argus,
and S. Stein, Current plate motions,
Geophys.J. Int., 101,425-478, 1990.
D4verch•re, J., F. Houdry, M. Diament,
N.V. Solonenko, and A.V. So!onenko,
Evidencefor a seismogenicupper mantle and lower crust in the Balked rift,
Geophys. Res. Lett., 18, 1099-1102,
1991.
Diament, M., and M.G. Kogan, Long
wavelengthgravity anomaliesand the
deep thermal structure of the Baikal
rift, Geophys. Res. Lett., 17, 19771980, 1990.
Doser, D., Source parameters and faulting processesof the 1959 Hebgen Lake,
Montana, earthquakesequence,J.
phys.Res., 90, 4537-4555, 1985.
Doser, D., Earthquake processesin
the Rainbow Mountain-Fairview
Peak-
DixieValley,Nevada,region1954-1959,
J. Geophys.]•es., 91, 12,572-12,586,
1986.
western branch
of the east African
rift
system, Bull. geol. Soc. Am., 101,885903, 1989.
Ebinger, C.J., M.J. Crow, B.R. Rosendalai,
D.A. Livingstone, and J. LeFournier,
Structural evolution of Lake Malafvi,
Africa, Nature, 308, 627-629, 1984.
Ebinger, C.J., B.R. Rosendahl, and D.J.
Reynolds, Tectonic model of the Mala/vi
rift, Africa, Tectonophysics, 1,11, 215235, 1987.
Ebinger, C.J, T.D. Bechtel, D.W. Forsyth,
and C.O. Bowin, Effective elastic plate
thickness
beneath
the East African
and
Afar plateaus and dynamic compensation of the uplifts, J. Geophys.Res., 9•,
2883-2901,
1989.
Ebinger, C.J., A.L. Deino, A.L. Tesha, T.
Becker, and U. Ring, Tectonic controls
on rift basin morphology: Evolution of
the northern Mala•vi (Nyasa) rift, J.
Geophys. Res., 98, 17,821-17,836, 1993.
Flannery, J.W., and B.R. Rosendalai, The
seismic stratigraphy of Lake Mala•i,
Africa: Implications for interpreting geological processesin lacustrine rifts, J.
A.fr. Earth Sci., 10, 519-548, 1990.
Foster, A., and F. Ninuno, Comparisons
between the rift systems of East Afi'ica,
Earth and Beta Regio, Venus, Earth
Planet. Sci. Lett., 1.i3, 183-196, 1996.
Fowler, S., and D. McKenzie, Flexural
studies
of the
Exmouth
and
Rockall
plateaux using Seasat altimetry, Basin
Res.• 2, 27-34, 1989.
Hayward, N.J., and C.J. Ebinger, Variations in the along-axis segmentation of
the Afar rift system, Tectonics, 15, 244257, 1996.
Jackson, J.A., and T. Blenkinsop, The
Mala•i earthquake of March 10, 1989:
Deep faulting within the East African
rift system, Tectonics, 12, 1131-1139,
1993.
Jackson,J.A., and M.R. Leeder, Drainage
systemsand the developmentof normal
faults: An example from Pleasant Valley, Nevada, J. Struct. Geol., 16, 10411059, 1994.
Jackson, J.A., and N.J. White, Normal
faulting in the upper continental crust:
Observations from regions of active extension, J. $truct. Geol., i1, 15-36,
1989.
Shikotan earthquake of October 4, 1994:
Lithospheric earthquake, Geophys.Res.
Lett., •2, 1025-1028, 1995.
Lindh, G., and D.M. Boore, Control of
rupture by fault geometry during the
1966 Parkfield earthquake, Bull. Setstool. Soc. Am., 71, 95-116, 1981.
Lister, L.A., Erosion surfaces in Mala•vi,
Rec. Geol. Surv. Mala•i,
7, 15-28,
1967.
Myers, W.B., aaxd W. Hanfilton, Deformation accompanying the Hebgen Lake
earthquake of August 17, 1959, U.S.
Geol. Surv. Prof. Pap., ,435, 55-98,
1964.
Nyblade, A.A., and C.A. Langston, East
African earthquakes below 20 km depth
and their implications for crustal structure, Geophys.J. Int., 12i, 49-62, 1995.
Roberts, S., and J.A. Jackson, Active
normal faulting in central Greece: An
overview, in The Geometry of Normal Faults, edited by A.M. Roberts, G.
Yielding, and B. Freeman, Geol. Soc.
Spec. Publ. London, 56, 125-142, 1991.
Rosendahl,B.R., D.J. Reynolds, P.M. Lotbar, C.F. Burgess, J. McGill, D. Scott,
J.J. Lambiase, and S.J. Derksen, Structural expressions of rifting: Lessons
from Lake Tanganyika, Africa, in Sedimentation in the African Riffs, edited
by L.E. Frostick, R.W. Renaut, I. Reid,
and J.J. Tiercelin, Geol. Soc. Spec.
Publ. London, 25, 29-43, 1986.
Schwartz, D.P., and R.H. Sibson, Fault
segmentation and controls of ruptm'e
initiation and termination, U.S. Geol.
Surv. Open File Rap., 89-315, 447,
1989.
Seno, T., and A. Saito, Recent east Africaa•
earthquakes in the lower crust, Earth
Planet. Sci. Left., 121, 125-136, 1994.
Shudofsky, G.N., S. Cloetingh, S. Stein,
and R. Wortel, Unusually deep earthquakes in east Africa: Constraint. s on
the
thermo-mechanical
structure
of a
continental rift. system, Geophys. Res.
Left., 1.4, 741-744, 1987.
Specht, T.D., and B.R. Rosendahl, Architecture of the Lake Mala•i rift, east
Africa, J. Aft. Eart]• Sci., 8, 355-382,
1989.
Thatcher, E.C., The geologyof the Dedza
area, Bull. Geol. Surv. Malazbi, 29,
71 pp, 1968.
Jackson, J.A., J. Gagnepain, G. Houseman, G. King, P. Papadimitriou, C.
Trudgill, B., and J. Cartwright, Relayramp forms and normal-fault linkages,
Canyonlands National Paxk, Utah,
Soullefts, and J. Virieux, Seisnficity,
Bull. geol. Soc. Am., 106, 1143-1157,
normal faulting, mxd the geomorpholog-
ical development of the Gulf of Corinth
1995.
(Greece): The Corinth earthquakes Wallace, R.E., Fault scarps formed dm'of February and March 1981, Earth
Planet. Sci. Lett., 57, 377-397, 1982.
Jestin, F., P. Huchon, and J.M. Gaulier,
The Somalia plate and the east African
rift system: Present-day kinematics,
Geophys. J. [nt., 116, 637-654, 1994.
Kikuchi, M., and H. Kanamori, The
ing the earthquakes of October 2, 1915
in Pleasant Valley, Nevada, and some
tectonic implications, U.S. Geol. Surv.
Prof. Pap., i27.iA, 1-33, 1984.
Wallace, R.E., Fault plane segmentation
in brittle crust and anisotropy in load-
ing system, in Fault Segmentationand
150
JACKSON
AND BLENKINSOP:
Controls o] Rupture Initiation and rermi7•ation, edited by D.P. Schwartz and
R.H. Sibson, U.S. Geol. Sur.v. Open File
I{ep., 89-315, 400-408, 1989.
Walsh, J.J., and J. Watterson, Geometric and kinematic
cotterence
and scale
effects in normal fault systems, in The
Geometry of Normal Fe•ults, edited by
A.M. Roberts, G. Yieldi•g, and B. Freeman, Geol. Soc. Spec.Publ. London, 56,
193-203,
1991.
Walshaw,FI.D., The geologyof the Ncheu-
BILILA-MTAKATAKA
Balaka area, Bull. Geol. Surv. gfala•bi,
19, 96 pp, 1965.
Wiens, D.A., and S. Stein, Age dependence of oceanic intraplate seismicity
and implications for lithospheric evolution, J. Geophys. Res., 88, 6455-6468,
1983.
Yielding, G., Control of rupture by fault
geometry during the 1980 E1 Asna•n
FAULT
T. Blenkinsop, Department of Geology,University of Zimbabwe, P.O. Box MP
167, Mount Pleasant, Harare, Zimbabwe.
(e-mail: tblenki@geology.
uz.zw)
J. Jackson, Department of Earth Sci.
ences, University of Cambridge, Ballard
Laboratories, Madingley Road, Cam-
bridgeCB30EZ, England. (e-mail:[email protected])
(ReceivedDecember14, 1995;
(Algeria) earthquake, Geophys.J.
revised August 9, 1996;
Astron. $oc., 8I, 641-670, 1985.
acceptedAugust 9, 1996.)