Quantifying global evapotranspiration from remote sensing: Estimates, trends and uncertainties for terrestrial hydrology

Quantifying evapotranspiration (ET), a vital link between the terrestrial and atmospheric water cycles, is a key to understand regional to global scale hydrometeorology. Satellite remote sensing serves as the only practical means of observations at continuous spatial- and temporal scales. The current dissertation serves as one of the first comprehensive studies to quantify ET at daily timescales using multi-sensor and multi-platform inputs and multi-model analysis. Three process models were assessed: a modified Penman-Monteith (PM-Mu), modified Priestley-Taylor (PT-Fi), and the Surface Energy Balance System (SEBS). The three models either adjust the surface resistances using aerodynamic principles or provide ecophysiological constraints to account for changing environmental factors; thus scaling ET from its potential value to the actual estimate.;Polar orbiting satellites serve as a great potential for remote sensing retrievals of the physical states of the land surface and the near surface atmosphere. Remote sensing retrievals from Atmospheric InfraRed Sounder (AIRS), Clouds and Earth's Radiant Energy System (CERES), Moderate resolution Imaging Spectroradiometer (MODIS), and Advanced Very High Resolution Radiometer (AVHRR) were used as input forcings to generate moderate resolution (∼5 km) estimates of ET for the time period 2003-2006. The three model estimates were evaluated at different scales, which formed the basis of the first study. The current state of the predictive capability of the climate models was considered as part of the second study. The analysis consisted of intercomparison of energy and water fluxes from remote sensing to the output from seven operational global model analyses and a land surface model. In order to get a perspective on the inter-annual and inter-decadal variability in ET, long term estimates of ET were generated using input forcings from the International Satellite Cloud Climatology Project (ISCCP). To understand the effects of uncertainties in net radiation (Rnet) on the ET estimates, another estimate of Rnet from the Surface Radiation Budget (SRB) project was considered. The evaluation of the estimates using the ISCCP and SRB forcings and the uncertainties that limited the further evaluation of trends in ET formed the basis of the third study. Finally, the study is given a closure by evaluating the uncertainties that exist from the different model parameterizations and the sensitivity of the each of the models to the various input forcings. Overall, the results from the four studies provide the first ever quantification of evapotranspiration by considering a multitude of input forcings and three different model parameterizations. The intercomparison study highlights the bias that exists between the observational driven estimates from the current study and those predicted by the current state-of-the-art climate models.