Monitoring the water balance of Lake Victoria, East Africa, from space

Using satellite gravimetric and altimetric data, we examine trends in water storage and lake levels of multiple lakes in the Great Rift Valley region of East Africa for the years 2003–2008. GRACE total water storage estimates reveal that water storage declined in much of East Africa, by as much as 60 mm year, while altimetric data show that lake levels in some large lakes dropped by as much as 1–2 m. The largest declines occurred in Lake Victoria, the Earth’s second largest freshwater body. Because the discharge from the outlet of Lake Victoria is used to generate hydroelectric power, the role of human management in the lake’s decline has been questioned. By comparing catchment water storage trends to lake level trends, we confirm that climatic forcing explains only about 50decline. This analysis provides an independent means of assessing the relative impacts of climate and human management on the water balance of Lake Victoria that does not depend on observations of dam discharge, which may not be publically available. In the second part of the study, the individual components of the lake water balance are estimated. Satellite estimates of changes in lake level, precipitation, and evaporation are used with observed lake discharge to develop a parameterization for estimating subsurface inflows due to changes in groundwater storage estimated from satellite gravimetry. At seasonal timescales, this approach provides closure to Lake Victoria’s water balance to within 17 mm month. The third part of this study uses the water balance of a downstream water body, Lake Kyoga, to estimate the outflow from Lake Victoria remotely. Because Lake Kyoga is roughly 20 times smaller in area than Lake Victoria, its water balance is strongly influenced by inflow from Lake Victoria. Lake Kyoga has been shown to act as a linear reservoir, where its outflow is proportional to the height of the lake. This model can be used with satellite altimetric lake levels to estimate a time series of Lake Victoria discharge with an rms error of about 134 m3 s . 2009 Elsevier B.V. All rights reserved.