Variational surface analysis from screen level atmospheric parameters

A variational method to retrieve soil parameters from screen level parameters is investigated. A one-dimensional model including vertical diffusion and land-surface schemes is used. A sensitivity study evaluates the influence of soil moisture and temperature in 2 layers on the surface fluxes and the 2m parameters. The mean soil water content appears to be the most important soil parameter for modelling surface fluxes and 2m parameters. During stable conditions, the mean soil temperature is also important for a good forecast of screen level parameters. The screen-level observations are found to be informative on soil moistures during the day and on soil temperatures during the night. The characteristics of the variational assimilation method are presented in the framework of identical twin experiments. The general shape of the cost function with a 24-h assimilation period reveals the ambiguity in the repartition of water in the two soil reservoirs and the presence of secondary minima. The ability of this method to initialize soil parameters is tested with various soil water contents and vegetation fraction covers. Finally, some results show the influence of the frequency of the observations and the length of the assimilation period on the quality of the analysis.