An observational study of the variability of East African rainfall with respect to sea surface temperature and soil moisture

Rainfall from the October-November-December (OND) short rains season over East Africa (EA, 5 degrees S-20 degrees N, 28-52 degrees E) were analysed during the 1983-2010 period using state-of-the-art observational datasets. Links among satellite-derived rainfall (Climate Hazards group InfraRed Precipitation with Station data, CHIRPS), sea surface temperature (Hadley Centre Sea ice and Sea Surface Temperature, HadISST1-SST), soil moisture (Climate Change Initiative, CCI-SM), and dynamical variables (European Centre for Medium-Range Weather Forecasts Re-Analysis, ERA-Interim) are investigated to disentangle their specific role in shaping the rainfall variability over the region. In general, interannual rainfall variability is highest in the area during OND. Empirical orthogonal function (EOF) analysis is applied to the rainfall dataset to extract the dominant spatial and temporal patterns of variability. Results show that the rainfall variability is directly influenced by the SST variability in the Indian Ocean (Indian Ocean Dipole, IOD) and in the Pacific Ocean (El Nino-Southern Oscillation, ENSO), and by the local SM. The strong positive correlation (0.78) between EA short rain index and the IOD index indicates that the positive-phase IOD (IOD+) plays a dominant role in driving OND rainfall. Moreover, IOD+ usually coincides with El Nino events and becomes stronger as the intensity of El Nino increases for years of joint events, and the Walker circulation is shifted accordingly. Furthermore, El Nino alone brings a reduction of rainfall over EA. The physical mechanism explaining the IOD+ link to EA rainfall consists of a Gill-type response to warm western Indian Ocean SST anomalies that induces anomalous low-level easterlies over the IO and leads to moisture convergence over EA. In general, the response of rainfall is opposite during the negative phase of the events (IOD- and La Nina).