Extracting Subseasonal Scenarios: An Alternative Method to Analyze Seasonal Predictability of Regional-Scale Tropical Rainfall

Abstract

urrent seasonal prediction of rainfall typically focuses on 3-month rainfall totals at regional scale. This temporal summation reduces the noise related to smaller-scale weather variability but also implicitly emphasizes the peak of the climatological seasonal cycle of rainfall. This approach may hide potentially predictable signals when rainfall is lower: for example, near the onset or cessation of the rainy season. The authors illustrate such a case for the East African long rains (March?May) on a network of 36 stations in Kenya and north Tanzania from 1961 to 2001. Spatial coherence and potential predictability of seasonal rainfall anomalies associated with tropical sea surface temperature (SST) anomalies clearly peak during the early stage of the rainy season (in March), while the largest rainfall (in April and May) is far less spatially coherent; the latter is shown to contain a large noise component at the station scale that characterizes interannual variability of the March?May seasonal total amounts. Combining the empirical orthogonal function of both interannual and subseasonal variations with a fuzzy k-means clustering is shown to capture the most spatially coherent subseasonal ?scenarios? that tend to filter out the noisier variations of the rainfall field and emphasize the most consistent signals in both time and space. This approach is shown to provide insight into the seasonal predictability of long dry spells and heavy daily rainfall events at local scale and their subseasonal modulation.