Decadal rainfall variability modes in observed rainfall records over East Africa and their relations to historical sea surface temperature changes.

Abstract

Detailed knowledge about the long-term interface of climate and rainfall variability is essential for managing agricultural activities in Eastern African countries. To this end, the space-time patterns of decadal rainfall variability modes over East Africa and their predictability potentials using Sea Surface Temperature (SST) are investigated. The analysis includes observed rainfall data from 1920-2004 and global SSTs for the period 1950-2004. Simple correlation, trend and cyclical analyses, Principal Component Analysis (PCA) with VARIMAX rotation and Canonical Correlation Analysis (CCA) are employed. The results show decadal signals in filtered observed rainfall record with 10 years period during March - May (MAM) and October – December (OND) seasons. During June - August (JJA), however, cycles with 20 years period are common. Too much / little rainfall received in one or two years determines the general trend of the decadal mean rainfall. CCA results for MAM showed significant positive correlations between the VARIMAX-PCA of SST and the canonical component time series over the central equatorial Indian Ocean. Positive loadings were spread over the coastal and Lake Victoria regions while negative loading over the rest of the region with significant canonical correlation skills. For the JJA seasons, Atlantic SSTs had negative loadings centred on the tropical western Atlantic Ocean associated with the wet / dry regimes over western / eastern sectors. The highest canonical correlation skill between OND rainfall and the Pacific SSTs showed that El Niño-Southern Oscillation (ENSO)/La Niña phases are associated with wet/dry decades over the region.