A STOCHASTIC APPROACH TO STUDYING THE INFLUENCE OF THE SPATIAL VARIABILITY OF SOIL HYDRAULIC-PROPERTIES ON SURFACE FLUXES, TEMPERATURE AND HUMIDITY

The assessment of mass and energy fluxes at the soil-biosphere-atmosphere interface is a key point for the improvement and reliability of climate model predictions. At the local scale, numerous models have been developed to predict surface fluxes, These models are generally deterministic and the extension of their results to larger scales, such as the catchment scale or the scale of an atmospheric or climatic model mesh, is greatly complicated by the large variability of surface properties. To address this issue, stochastic approaches have been proposed to enable prediction of fluxes in terms of probability density functions. This paper is concerned with the influence of the spatial variability of soil hydraulic properties. This variability is expressed through one parameter, the scale factor for which a log-normal distribution is assumed. Values of the scale factor, drawn from this distribution, are introduced into a unidimensional model, called SiSPAT, which describes the soil-plant-atmosphere coupled heat and water transfers. Two land uses are considered: a bare soil and a rather densely vegetated one for which the same forcing is applied over 7 days. Comparison of the two cases shows that the vegetation tends to smooth the influence of the spatial variability of soil properties limiting the number of observations required to estimate a spatial mean with a prescribed degree of accuracy. Comparison of the deterministic and stochastic solutions also shows that the former approach leads to a bias for the bare soil case, the differences between the two being almost negligible in the presence of vegetation.