The Great Sumatran Fault (GSF) is a 1650-km-long dextral strike-slip f
ault zone which accommodates part of the oblique convergence of the su
bduction between the Indo-Australian and Eurasian plates. To define th
e seismic hazard along this fault, we used paleoseismology and neotect
onics. To characterise the seismic history of the southern GSF we exca
vated four trenches, Within these trenches, the occurrence of only one
paleosol related to a seismic event indicates that in a wet, tropical
region, the degradation rate of organic matter could be faster than s
eismic recurrence. The trenching method permitted us to identify only
one recent earthquake, reactivating the southern GSF. As the trenching
method does not seem efficient to constrain knowledge of seismicity i
n this region, we have developed an active tectonic study to character
ise the seismic hazard along the GSF. We created a large-scale segment
ation map which allows 18 major fault segments with lengths ranging be
tween 45 and 200 km to be recognised, We complemented the segmentation
map reporting major earthquake ruptures on the basis of the historica
l seismicity which recorded 17 earthquakes since 1835. The segmentatio
n map indicates a northward increase of segment lengths which parallel
s the GSF slip-rate increase. This observation suggests a northward in
crease of seismic hazard along the GSF. Segmentation and historical se
ismicity provide evidence of a 300-km-long seismic Sap (between 3 degr
ees N and 5 degrees N) around a locked restraining bend which can be c
onsidered as having high potential for seismic hazard in Sumatra. The
magnitude of the maximum expected earthquake for each segment was esti
mated through two empirical methods. These estimates give higher maxim
um magnitude and shorter seismic recurrence intervals for segments in
northern Sumatra, confirming a northward increase of seismic hazard. (
C) 1997 Elsevier Science Ltd.