P wave velocity structure of Proterozoic upper mantle beneath central
and southern Africa was investigated by forward modeling of Pnl wavefo
rms from four moderate size earthquakes. The source-receiver path of o
ne event crosses central Africa aid lies outside the African superswel
l while the source-receiver paths fur the other events cross Proterozo
ic lithosphere within southern Africa, inside the African superswell.
Three observables (Pn waveshape, PL-Pn time, and Pn/PL amplitude ratio
) from the Pnl waveform were used to constrain upper mantle velocity m
odels in a grid search procedure. For central Africa, synthetic seismo
grams were computed for 5880 upper mantle models using the generalized
ray method and wavenumber integration; synthetic seismograms for 216
models were computed for southern Africa. Successful models were taken
as those whose synthetic seismograms had similar waveshapes to the ob
served waveforms, as well as PL-Pn times within 3 s of the observed ti
mes and Pn/PL amplitude ratios within 30% of the observed ratio. Succe
ssful models for central Africa yield a range of uppermost mantle velo
city between 7.9 and 8.3 km s(-1), velocities between 8.3 and 8.5 km s
(-1) at a depth of 200 km, and velocity gradients that are constant or
slightly positive. For southern Africa, successful models yield upper
most mantle velocities between 8.1 and 8.3 km s(-1), velocities betwee
n 7.9 and 8.4 km s(-1) at a depth of 130 km, and velocity gradients be
tween -0.001 and 0.001 s(-1) Because velocity gradients are controlled
strongly by structure at the: bottoming depths for Pn waves, it is no
t easy to compare the velocity gradients obtained for central and sout
hern Africa. For central Africa, Pn waves turn at depths of about 150-
200 km, whereas for southern Africa they bottom at similar to 100-150
km depth. With regard to the origin of the African superswell, our res
ults do not have sufficient resolution to test hypotheses that invoke
simple lithospheric reheating. However, our models are not consistent
with explanations for the African superswell invoking extensive amount
s of lithospheric thinning. If extensive lithospheric thinning had occ
urred beneath southern Africa, as suggested previously, then upper man
tle P wave velocities beneath southern Africa would likely be lower th
an those in our models.