Soil moisture is not a commonly observed geophysical variable. Soil moisture observations are scanty to non-existent except for a few sparse monitoring networks and for limited time periods in limited areas during major field experiments. For studying land/atmosphere interaction in data-poor environments, soil moisture patterns may be derived using the technique of variational assimilation. The Parameterization for Land-Atmosphere-Cloud Exchange (PLACE) model is a detailed land-surface process model of the partly cloudy atmospheric boundary layer and underlying heterogeneous land surfaces (Wetzel and Boone 1995). This study evaluates PLACE's performance in generating realistic patterns of soil moisture and latent and sensible heat fluxes by variational assimilation. The data-rich environment of SGP97 provides both forcing and verification data for study experiments. Because a number of atmospheric models are being modified to include PLACE, understanding PLACE's sensitivities will put into proper perspective future results from coupled model systems involving PLACE.
Study experiments initialized PLACE with climatological soil moisture and temperature values. Simulations were conducted both at a single point and over a mesoscale domain (1-km resolution) encompassing the entire watershed. Off-line simulations required atmospheric data (surface temperature, wind speed, station pressure, and humidity) and NEXRAD accumulated rainfall data. The convergence of surface soil moisture values determined the length of simulations. Model results were then compared to in-situ (Meso/Micronet) and remotely sensed soil moisture (ESTAR) and flux estimates (Twin Otter and Long-EZ). From this comparison, model sensitivities to parameters such as leaf area index and atmospheric variables such as wind speed were determined