Abstract
This study examines the impact of daily satellite-derived albedos on short-range forecasts in a limited-area numerical weather prediction (NWP) model over Europe. Contrary to previous studies in which satellite products were used to derive monthly “climatologies,” a daily surface (snow free) albedo is analyzed by a Kalman filter. The filter combines optimally a satellite product derived from the Meteosat Second Generation geostationary satellite [and produced by the Land Surface Analyses–Satellite Application Facility of the European Organisation for the Exploitation of Meteorological Satellites (EUMETSAT)], an albedo climatology, and a priori information given by “persistence.” The surface albedo analyzed for a given day is used as boundary conditions of the NWP model to run forecasts starting the following day. Results from short-range forecasts over a 1-yr period reveal the capacity of satellite information to reduce model biases and RMSE in screen-level temperature (during daytime and intermediate seasons). The impact on forecast scores is larger when considering the analyzed surface albedo rather than another climatologically based albedo product. From comparisons with measurements from three flux-tower stations over mostly homogeneous French forests, it is seen that the model biases in surface net radiation are significantly reduced. An impact on the whole planetary boundary layer, particularly in summer, results from the use of an observed surface albedo. An unexpected behavior produced in summer by the satellite-derived albedo on surface temperature is also explained. The forecast runs presented here, performed in dynamical adaptation mode, will be complemented later on by data assimilation experiments over typically monthly periods.
