IMPACT OF SOME SIMPLIFYING ASSUMPTIONS IN THE NEW ECMWF SURFACE SCHEME

Two simplifying assumptions adopted in the current ECMWF surface schem e are explored: a uniform skin temperature for all grid-box fractions with variable latent heat release and a fixed value of an effective he at conductivity defining the soil heat flux density. This paper propos es relatively simple modifications of the ECMWF scheme with a better p hysical basis, without large input or computer infrastructure requirem ents. A uniform skin temperature overestimates evaporation from relati vely wet surface fractions when the other surface components are dry a nd warm. This is shown to be the case for an evaporating soil after ra in and vegetation evaporation in a sparse Mediterranean vineyard canop y. Allowing different temperatures for each surface fraction significa ntly reduces the overestimations and introduces only little additional computation. The default effective conductivity value (7 W m(-2) K-1) employed by the current ECMWF scheme is shown to be too low for the s parse vineyard canopy. By raising the conductivity to 17 W m(-2) K-1 f or the bare-soil part of the surface, the daytime simulated soil heat flux was improved considerably.