Exploring hydro-meteorological drought patterns over the Greater Horn of Africa (1979-2014) using remote sensing and reanalysis products

Spatio-temporal patterns of hydrological droughts over the Greater Horn of Africa (GHA) are explored based on total water storage (TWS) changes derived from time-variable gravity field solutions of Gravity Recovery And Climate Experiment (GRACE, 2002-2014), together with those simulated by Modern Retrospective Analysis for Research Application (MERRA, 1980-2014). These hydrological extremes are then related to meteorological drought events estimated from observed monthly precipitation products of Global Precipitation Climatology Center (GPCC, 1979-2010) and Tropical Rainfall Measuring Mission (TRMM, 1998-2014). The major focus of this contribution lies on the application of spatial Independent Component Analysis (sICA) to extract distinguished regions with similar rainfall and TWS with similar overall trend and seasonality. Rainfall and TWS are used to estimate Standard Precipitation Indices (SPIs) and Total Storage Deficit Indices (TSDIs), respectively that are employed to characterize frequency and intensity of hydro-meteorological droughts over GHA. Significant positive (negative) changes in monthly rainfall over Ethiopia (Sudan) between 2002 and 2010 leading to a significant increase in TWS over the central GHA region were noted in both MERRA and GRACE TWS (2002-2014). However, these trends were completely reversed in the long-term (1980-2010) records of rainfall (GPCC) and TWS (MERRA). The four independent hydrological sub-regions extracted based on the sICA (i.e., Lake Victoria Basin, Ethiopia-Sudanese border, South Sudan, and Tanzania) indicated fairly distinct temporal patterns that matched reasonably well between precipitation and TWS changes. While meteorological droughts were found to be consistent with most previous studies in all sub-regions, their impacts are clearly observed in the TWS changes resulting in multiple years of extreme hydrological droughts. Correlations between SPI and TSDI were found to be significant over Lake Victoria Basin, South Sudan, and Tanzania. The low correlations between SPI and TSDI over Ethiopia are likely related to inconsistency between TWS and precipitation signals. Further, we found that hydrological droughts in these regions were significantly associated with Indian Ocean Dipole (IOD) events while El Niño Southern Oscillation (ENSO) plays a secondary role.