Analysis of satellite monthly precipitation time series over East Africa

East Africa experienced in the 2001 -2011 time period some of the worst drought events to date, culminated with the high-impact drought in 2010-2011. The frequency and impacts of these extreme events require a continuous monitoring of precipitation, as a key variable for the inclusion of these phenomena in regional climatological studies and their timely forecast. Satellite precipitation products are particularly necessary in the region to enhance the observational capabilities limited sparse rain-gauge networks. Nevertheless, East Africa is characterized by a complex topography and highly varying climatic conditions ranging from the wetter mountainous regions to the arid lowlands with different precipitation seasonality, which can greatly affect the quality of satellite rainfall estimations. It is thus of utmost importance a satellite product validation and inter-comparison in order to assess their reliability and delimit the application domain. The monthly accumulated precipitation from six satellite products, TAMSAT, GSMaP, CMORPH, PERSIANN, RFE, and TRMM-3B42, are analysed for the time period 2001-2009, by dividing the studied region (5°S-20°N, 28°E-52°E) in eight sub-areas (clusters) characterized by a different annual cycle. Clusters are identified by applying a non-hierarchical k-mean cluster analysis to the GPCC Climatology Version 2011 product. Each satellite product correctly identifies the annual cycle characteristics of each cluster, whereas differences can be seen in the precipitations amount. The uncertainties in satellite-based precipitation estimations are evaluated by computing the variance from the ensemble of the six satellite products at the resolution of 0.25°. The variability among satellite products shows a dependence on season, precipitation intensity, and topography. The regions with higher variability among the satellite products are the mountainous area of West Ethiopia and the adjacent Rift Valley during summer (wet season) and for heavy precipitation, South Sudan and Congo during summer and fall, and the region surrounding Lake Victoria. Comparisons (correlation coefficient, mean error, root mean square error, and efficiency coefficient) are carried out with the GPCC Full Data Reanalysis at 0.5° resolution. From this analysis TRMM-3B42 stands out as the satellite product with the better performances, generally followed by CMORPH, RFE, and TAMSAT with performances depending on the considered cluster. Finally, monthly anomalies between the satellite products and the GPCC Climatology Version 2011 product at 0.25° are computed to evaluate the potential of satellite products for identifying the drought periods.