Understanding the Characteristics of Droughts over Eastern Africa in Past and Future Climates

Abstract Drought poses a threat to socio-economic activities across eastern Africa and its river basins. While there are indications that global warming may continue to enhance evaporation and intensify droughts at all scales, most drought projections over eastern Africa are based on rainfall alone and are limited to meteorological droughts. The present study combines rainfall and Potential Evapotranspiration (PET) to examine the characteristics of meteorological and hydrological droughts in present and future climates at the regional and river basin scales. To accomplish that we have applied five objectives; i) Study the temporal and spatial characteristics of eastern Africa droughts modes, ii) Investigate how some atmospheric teleconnections influence the characteristics of the Africa droughts modes, iii) Examine the influence of 1.5°C and 2°C global warming levels on drought modes in eastern Africa under two future climate scenarios, RCP 4.5 and RCP8.5 iv) Assess how increases in global warming will influence drought characteristics over eastern African river basins. v) Examine the potential impacts of climate change and land use change on water availability in the Rufiji River basin (RRB), Tanzania, with an emphasis of hydrological droughts in this basin. Different types of datasets, including gridded and station observation datasets, regional climate model simulations (CORDEX: Coordinated Regional Climate Downscaling Experiment) and hydrological simulations (SWAT: Soil and Water Assessment Tool), were analysed for the study. The meteorological drought were characterised using two indices (i.e. Standardized Precipitation Evapotranspiration Index, SPEI; Standardized Precipitation Index, SPI) at 3- and 12-month scales, while the hydrological droughts were characterised using four indices (i.e. soil water index, SWI; Surface Runoff Index, RFI; Water Yield Index, WYI; and Stream Flow index, SFI). The study combined principal component analysis (PCA) with wavelet analysis to identify the spatio-temporal structure of four dominant drought modes over the region. It also used wavelet coherence to quantify the influence of four atmospheric teleconnections (i.e. El Niño Southern Oscillation, ENSO; Indian Ocean Dipole, IOD; Tropical Atlantic Page | iv Dipole Index, TADI; and Quasi-Biennial Oscillation, QBO) on the drought modes. The study also projects the characteristics of future droughts over eastern Africa and its major river basins at different global warming levels (GWLs). Series of hydrological simulations were used to assess the sensitivity of future droughts to four land use change scenarios (i.e. increase in forestry, shrubs, cropland and agriculture) over the Rufiji River Basin (RRB), a prominent river basin in eastern Africa. Although eastern Africa have been documented with several drought studies, the application of a combination of PCA, Wavelet analysis, wavelet coherence and Self Organizing Maps provides more comprehensive representation of droughts in the region using SPEI/SPI derived from both models and observations The results of the study show that the four drought modes, which have their core areas over different parts of eastern Africa, account for more than 45% of drought variability in the region. All the drought modes are strongly coupled with either ENSO or IOD indices (or both); but, in addition, one of the modes is also strongly coupled with the TADI. CORDEX models give a realistic simulation of the relevant climate variables for calculating drought indices over eastern Africa and the river basins. However, the ensemble mean struggles to reproduce the spatial distribution and frequency of drought intensity in the region. The CORDEX simulations project no changes in the spatial structure of the drought modes but suggest an increase in SPEI drought intensity and frequency over the hotspots of the drought modes and elsewhere in the region. The magnitude of the increase, which varies over the drought mode hotspots, increases with increasing GWLs. The projections also show that the increase in intensity and frequency of drought can be attributed more to increased PET than to reduced precipitation. In contrast to the SPEI projection, the SPI projection shows a weak change in intensity and frequency of droughts, and the magnitude of the increase does not vary with the GWLs. Over the river basins, the SPEI projections are more robust than the SPI projections. Over the RRB, the future projections of some hydrological drought indices (i.e. RFI and SFI) follow the change in the SPEI projections, while others (i.e. SWI and WYI) follow that of SPI. Among the four land use scenarios considered, only forestry and shrubs show a substantial change in the hydrological drought indices. The results of the study thus give valuable insight Page | v into the characteristics of future droughts in eastern Africa and provide a useful guide to the effectiveness of using land cover to reduce the severity of hydrological droughts over river basins in the region. However, resolution of CORDEX dataset (50km, i.e. 0.44deg) could be among the potential limitation as it is too low to capture the influence of local-scale processes (e.g. sea breeze, mountain induced circulations) on drought over the region.