Assimilation of land surface temperature data from ATSR in an NWP environment-case studies with ATSR data in Spain and The Netherlands

Land surface parameterizations in numerical weather prediction describe the exchange of energy and water at the land-atmosphere interface. They are generally based on a coupled energy-water budget equation, and the results depend critically on (prescribed) land surface characteristics, notably albedo, soil water storage capacity and aerodynamic roughness. In a recent case-study bi-angular (nadir and forward) observations provided by the ATSR instrument were used to estimate the component soil and foliage temperature. For two images overlying the Netherlands on 14 May 1998 and South-East Spain on 9 September 1998 vegetation and bare ground temperatures were derived from ATSR data, and aggregated to a grid of 25×25 km, compatible with a NWP model. This NWP model was equipped with a land surface parameterization scheme in which vegetation and bare ground temperature were calculated separately. In a variational assimilation scheme, the NWP model was forced to match the observed component temperatures by changing both the soil moisture content and the aerodynamic roughness for the heat exchange of the bare ground component. It appeared that the optimal solution differed significantly for the two regions. For The Netherlands, changing the aerodynamic roughness did not affect the correspondence to observations, and soil moisture had to be changed to increase the overall model performance.