Lake level change and total water discharge in East Africa Rift Valley from satellite-based observations

摘要

The measurement of total basin water discharge is important for understanding the hydrological and climatologic issues related to the water and energy cycles. Climatic extreme events are normal climatic occurrences in Africa. For example, extensive droughts are regular features in the last few decades in parts of East Africa, which suffers from a lack of in situ observations as well as a lack of regional hydrological models. In this study, multi-disciplinary different types of space-borne observations and global hydrological models are used to study total water discharge in the Great Rift Valley of East Africa (i.e. Lakes Victoria, Tanganyika, and Malawi) from January 2003 to December 2012. The data include the following: (1) total water storage (TWS) variations from Gravity Recovery and Climate Experiment (GRACE), (2) the lake level variations from Satellite Alimetric data, (3) rainfall from Tropical Rainfall Measurement Mission (TRMM) products, (4) soil moisture from WaterGAP Global Hydrology Model (WGHM), and (5) water fluxes from Global Land Data Assimilation System (GLDAS). Results show that a significant decline in the average lake level is found for all of the three lakes between 2003 and 2006. GRACE TWS variations of the whole basin area show the same pattern of variation as the average lake level variations estimated from Altimetric data. The TWS in the basin area of Lakes Victoria and Malawi is governed by the surface water stored in each lake itself, while for Lake Tanganyika, it is governed by both surface water and the soil moisture content in the basin area. Furthermore, the effect of rainfall on TWS is also studied. A phase lag of ～2 months is found between TRMM rainfall and GRACE TWS (generally, rainfall precedes the GRACE TWS) for the three lakes. In addition, the regional evapotranspiration ET is estimated from the water balance equation using GRACE land-water solutions, rainfall data from TRMM and runoff values obtained as a fraction of rainfall. It is found that the computed ET represents approximately 90% of the rainfall over the study region.