Rainfall trends in East Africa from an ensemble of IR-based satellite products

Abstract

The Africa Rainfall Climatology v2 (ARC2), Climate Hazards Group InfraRed Precipitation with Stations (CHIRPS) v2, and Tropical Applications of Meteorology using SATellite (TAMSAT) African Rainfall Climatology And Time Series v3 (TARCAT3) satellite rainfall products are exploited to study the spatial and temporal variability of East Africa (EA) rainfall between 1983 and 2017 through the time series of selected rainfall indices from the joint CCI/CLIVAR/JCOMM Expert Team on Climate Change Detection and Indices (ETCCDI). The indices total rainfall amount (PRCPTOT), Simple Daily Intensity (SDII), number of precipitating days (R1), maximum number of consecutive dry and wet days (CDD and CWD), and number of very heavy precipitating days (R20) were analyzed. The scope of the work is to draw the attention on the rainfall trend and variability identifying significant trend patterns regardless of the single satellite product, and also estimating the trend rate variability stemming from the multiplicity of the satellite products. The trend spatial patterns are recognized through the Mann-Kendall technique, considering the time series of the ensemble mean of the three satellite products and the corresponding time series of the standard deviations, which are interpreted as error bars associated with the ensemble mean time series. Indications on rainfall trends were extracted at annual and seasonal scales and the regions that more frequently exhibit statistically significant trends are located in eastern Kenya, Somalia at the border with eastern Ethiopia, northern Tanzania, and limited areas of South Sudan. At the seasonal scale increasing trends were identified for the October-November-December PRCPTOT, SDII, and R20 indices over eastern EA, with the exception of central Kenya, where negative trends with limited areas of significance stand out for R1 and CWD, distinguishable also at the yearly scale. In March-April-May rainfall decline is perceivable only through R1 and CWD in particular over the eastern EA region, whereas PRCPTOT, even though associated with negative trends, does not present any high confidence areas.