Rainfall variability, expressed through the LVRS, has a demonstrable and significant (5% level) relationship
with Lake Victoria levels. NCEP rainfall is found to be poor in representing the 1961 anomaly in East
Africa. Running correlations indicate poor and mainly negative associations between NCEP rainfall and
the observed LVRS.
2. ENSO is the predominant factor responsible for long-term variability of Lake Victoria rainfall, as revealed
through spectral analysis of the LVRS and analysis of OND and MAM EOF(∗) series.
3. Indian Ocean SSTAs are more influential in determining moisture fluxes and supplying latent heat,
particularly in the 1961 (and the subsequent MAM 1962) rainfall episode, than the Pacific Ocean (following
Reverdin et al. (1986)).
4. The anomalous OND 1961 rainfall episode that contributed to the greatest measured rise in Lake Victoria
level was not a consequence of an Indian Ocean cell in the circumglobal Walker circulation.
5. OND EOF5 (Figure 8) illustrates that a stronger meridional, rather than zonal, circulation develops over
the Indian Ocean. This mode of variability correlates significantly with LVRS at the 5% level and may be
associated with other anomalous high rainfall events (e.g. OND 1977) than OND 1961 alone. However,
given that EOF5 explains only 2.35% of the variance, further work is necessary to explore the importance
of this meridional interpretation.
6. The 1961 rainfall anomaly is found to be the consequence of a reversal in the prevailing wind regime in
the Indian Ocean, positive SSTAs in the western Indian Ocean, and a burst of cool air from the southern