Combining Remote Sensing and Climate Reanalysis Data to Estimate Evapotranspiration in sub-Saharan Africa

Global warming is expected to lead to dramatic increases in evapotranspiration (ET) over many of the most climatically sensitive parts of sub-Saharan Africa. Evapotranspiration is a key input in climate/hydrologic models. Despite the importance of ET to regional climate, its application in statistical seasonal rainfall forecasting has been limited because ground data for calibration are limited. The new Fisher et al. (2008) model, however, utilizes the Priestly-Taylor formulation and performs well in the tropics using ground inputs. In this study, vegetation and climate inputs to the Fisher model are developed from the Moderate Resolution Imaging Spectroradiometer and Global Land Data Assimilation System (GLDAS). The model was compared against GLDAS ET(Penman-Monteith approach) and a thermal inertia (Simplified Surface Energy Balance-SSEB) model. The evaluation dataset came from 10 flux towers in sub-Saharan Africa. The initial results reveal that the model performs comparable to GLDAS and SSEB at a daily timestep (average R2=0.51, RMSE=30.46 Wm-2). The explained variance is mainly due to interannual cycles and to a lesser extent daily variability. Systematic biases in the results, particularly at high ET in the Fisher model, show that improvements in model parameterization are feasible. A residual and sensitivity analysis will be conducted on primary model inputs and parameters, in order to identify where the most significant improvements in the model can be made. The modified Fisher model will be reevaluated and spatio-temporal patterns in ET from 2000-2008 (length of MODIS series) will be summarized. The long-term goal of the study will be to synthesize the 2000-2008 series with a 1981-2000 series that utilizes the Advanced Very High Resolution Radiometer and historical GLDAS data. This extended series will then be used to study the spatio-temporal variability of ET and rainfall in sub-Saharan Africa. This innovative use of ground, climate reanalysis and remote sensing data is efficient and cost-effective, which is particularly important in the developing world.