Remote sensing offers a very interesting means to estimate the soil moisture state of a hydrological system. However, practical use for small scale agricultural applications is still limited. Ground truth data remain necessary to validate the inversion from the measured quantities to soil moisture content, to understand small scale processes in the horizontal plane, and to assess the distribution of water over a soil profile. Additionally, land surface models offer basic knowledge of the physical and physiological processes affecting the soil moisture state. A combination of both sources of information yields an optimal estimate of the system state and offers the best knowledge available to decision makers.
In this study, ground measurements of soil moisture in the Optimizing Production Inputs for Economic and Environmental Enhancement (OPE<sup>3</sup>) field (near Washington D.C.) of the United States Department of Agriculture (USDA) were assimilated into the Community Land Model (CLM2.0). Some practical problems that prevent optimal state estimation are discussed, such as the presence of bias in the model or observations, and the limited knowledge of the correlation structure of e.g. model error. Some case studies revealed that the influence of assimilation of upper layer soil moisture, as provided by remote sensing, improves the model results, but is not as persistent for profile estimation as assimilation of soil moisture in deeper layers.