VALIDATION OF SOIL-MOISTURE SIMULATION IN LANDSURFACE PARAMETERIZATION SCHEMES WITH HAPEX DATA

In the Soil Moisture Simulation Workshop, a number of numerical experi ments was conducted with fourteen representative schemes and the resul ts compared with HAPEX-MOBILHY data. This paper documents the general outcomes of the workshop, and provides the background information for the other papers presented in this special issue, which deal with more specific aspects of soil moisture simulation and the performance of i ndividual schemes. The results show that soil moisture simulation in c urrent landsurface schemes is profoundly different. After adjustment o f landsurface parameters, the disagreement in soil moisture for a 1.6 m soil layer remains around 100 mm. Correspondingly, the differences i n predicted annual cumulative evaporation as well as total runoff plus drainage are around 250 mm (annual precipitation being 856 mm for HAP EX-MOBILHY). The partitioning of surface available energy into sensibl e and latent heat fluxes is closely coupled to that of precipitation i nto evaporation and runoff plus drainage. ?These disagreements are rel ated to different causes but attempts to establish the causality betwe en the outcome and the responsible mechanism has had only limited succ ess to date because of the non-linear interactions embedded in the sch emes. This study implies that different schemes achieve different equi librium states when forced with prescribed atmospheric conditions and that the time period to reach these states differs among schemes; and even when soil moisture is fairly well simulated, the processes (parti cularly evaporation and runoff plus drainage) controlling the simulati on differ among schemes and at different times of the year. These resu lts suggest that prescription of landsurface scheme physics and bioche mistry may have to be a function of the type of predictions (short-ter m weather forecasting, mesoscale modelling or climate ensembles) requi red as well as the underlying scheme formulation and that scheme simul ations should be validated for all components of the prediction.