Remotely-sensed evapotranspiration estimates for an improved hydrological modeling of the early Holocene mega-lake Suguta, northern Kenya Rift

Abstract

The actual evapotranspiration is an important, but difficult to determine, element in the water balance of lakes and their catchment areas. Reliable data on evapotranspiration are not available for most lake basins for which paleoclimate reconstructions and modeling have been performed, particularly those in remote parts of Africa. We have used thermal infrared multispectral data for 14 ASTER scenes from the TERRA satellite to estimate the actual evapotranspiration in the 12,800km2 catchment of the Suguta Valley, northern Kenya Rift. Evidence from sediments and paleo-shorelines indicates that, during the African Humid Period (AHP, 14.8 to 5.5kyrs BP), this valley contained a large lake, 280m deep and covering ~2200km2, which has now virtually disappeared. Evapotranspiration estimates for the Suguta Basin were generated using the Surface Energy Balance Algorithm for Land (SEBAL). Climate data required for the model were extracted from a high-resolution gridded dataset obtained from the Climatic Research Unit (East Anglia, UK). Results suggest significant spatial variations in evapotranspiration within the catchment area (ranging from 450mm/yr in the basin to the north to 2000mm/yr in more elevated areas) and precipitation that was ~20% higher during the AHP than in recent times. These results are in agreement with other estimates of paleo-precipitation in East Africa. The extreme response of the lake system (~280m greater water depth than today, and a lake surface area of 2200km2) to only moderately higher precipitation illustrates the possible sensitivity of this area to future climate change.