The representation of spatial heterogeneity of soil moisture is essential for modeling processes that are nonlinearly related to soil moisture, such as the partitioning of sensible and latent heat fluxes. Remotely sensed soil moisture data is becoming increasingly available, however the variability within the remotely sensed footprint is spatially averaged. A number of studies have suggested that the spatial variability of soil moisture varies with wetness. This relationship varies with the characteristics of the location and scale, due to the different dominant controls on soil moisture. At different locations, scales, and wetting and drying conditions, soil moisture patterns have been linked to topography, soil characteristics such as porosity and wilting point, vegetation characteristics, and rainfall distribution.

A principal goal of the Department of Energys (DOE) Water Cycle Pilot Study (WCPS) is to balance the water budget in a small watershed in the Southern Great Plains using observations of as many water cycle components as possible. Another goal is to evaluate various model components, both atmospheric and hydrologic, that could be joined to form an analysis and forecast system of this watershed. An important consideration is how the spatial heterogeneity should be represented to give reasonable land surface hydrological fluxes. As part of the DOE WCPS and an NSF water cycle study, we have investigated the controls on soil moisture under different wetting and drying conditions in the Whitewater Watershed, Kansas, and used a land surface model to determine how subgrid spatial variability of soil moisture might be represented. In the absence of detailed spatially variable soil moisture measurements, TOPLATS, a land surface model that has been shown to reproduce soil moisture patterns in the Southern Great Plains, was used in fully distributed mode to generate small scale (30m) spatially variable data over a subdomain of the Whitewater catchment. The model was then run with distributions of 30m topography and vegetation characteristics over 1km grid cells, and at the 1km scale. Comparison of the states and fluxes from the 3 different modes can give insight into whether distributions of topography and vegetation will improve the representation of soil moisture variability and average fluxes, and whether this varies under different wetting and drying conditions due to different controls.