Mesoscale modeling of evapotranspiration using remote sensing data

For the purpose of spatial modeling of actual evapotranspiration (aET) at different scales from the field- to the landscape-scale the PROMET model-family has been developed. It is based on a gridded approach which allows the easy integration of parameters derived from remote sensing data. The model consists of a kernel for the process description ( a SVAT-model based on Penman-Monteith and a plant-physiological model for the influence of environmental factors on the canopy resistance) and a spatial parameter modeller, which provides and organizes the spatial input data at different scales from the field- to the mesoscale. It is shown, that the results of the kernel-model at held scale compares well with measured aET for different land-uses. After verification on the field scale themodel is run on a 100x150 km mesoscale test-site at a resolution of 1 km. Fractional land-cover was determined using time-series of NOAA-AVHRR data and an unmixing procedure for forest, grassland, agriculture, urban areas and water. It is shown, that this unmixed land-use information corresponds well with a LANDSAT land-use classification conducted in the test region. A digital elevation model and a soil map, interpolated meteorlogical data from the German Weather Service and data on the development of LAI and plant height was added to the data set. The model was run over one growing season on an hourly basis. To verify the spatial pattern of the modelled aET the model results of one date were compared with surface temperature distributions measured from NOAA/AVHRR at the same time. Both are in good agreement.