A comparison of satellite evapotranspiration estimation efforts.

Accurate estimation of surface heat fluxes is of considerable interest to meteorological, climatological and agricultural investigations as they identify the key physical processes that link the land surface with the atmosphere in this naturally coupled system. While characterising surface fluxes is critical in describing the partitioning of water and energy across Earth's terrestrial surfaces, accurately monitoring the spatial variation, particularly at daily and sub-daily temporal scales, is notoriously difficult. Spatial and temporal scaling issues, errors in forcing variables, heterogeneity in surface characteristics and simplifications in process understanding, all limit the capacity to accurately monitor flux development and variability. Various techniques to use satellite data to estimate evapotranspiration (ET) have been developed and are being evaluated globally as part of the GEWEX Landflux initiative. These comparisons have also included land-surface models as well as global coupled models and reanalysis. How the land-surface models use satellite data to inform the surface conditions, and hence the estimation of ET, varies substantially. Many models use satellite-based seasonal climatology of surface conditions, while some are attempting to assimilate satellite-based estimates of other variables such as soil moisture. In these efforts the land and atmosphere are uncoupled. How important the actual surface conditions are, as measured by satellite, in a coupled land-atmosphere system has been investigated over south-east Australia. Here satellite based estimates of albedo and vegetation fraction are used to inform the Noah land-surface scheme within the Weather Research and Forecasting (WRF) regional climate model. Output from WRF is also compared to common ‘combination’ type approaches such as the Priestley-Taylor formulation and surface energy balance retrievals to examine the variation and consistency of these different estimation approaches.