Although the important role of soil moisture in deep-convection development has been recognized, it remains the most difficult variable to obtain, because there are no routine high-resolution soil moisture observations at the continental scale. The soil moisture fields from the coupled land/atmosphere models suffer substantial errors and drift owing to model precipitation and radiation biases. An alternative strategy is to utilize observed rainfall, satellite-derived surface solar insolation, and meteorological analysis to drive an off-line simulation of a LSM, so that the evolution of soil moisture does not suffer from the model biases in surface forcing fields.

We describe here our effort to develop the high-resolution land data assimilation system (HRLDAS) to provide reasonable initial soil state for coupled WRF/LSM modeling system. RLDAS uses the same Noah LSM as in the coupled WRF/Noah model system. HRLDAS is able to characterize soil moisture and vegetation variability at small scales (~4km) over large areas in order to provide improved initial land state for the WRF/Noah coupled model. HRLDAS uses the following atmospheric forcing and surface conditions: 1) hourly 4-km NCEP Stage-IV rainfall analysis; 2) 0.5 degree solar downward radiation derived from GOES satellites, 3) atmospheric forcing conditions from model-based analysis; 4) 1-km USGS landuse map and 1-km STASGO soil texture map, and 5) satellite derived vegetation characteristics (e.g., green vegetation fraction).

We will present the HRLDAS configuration, its verification against observed soil moisture and temperature, surface heat fluxes, and 'natural' stremflow data. We will discuss the impact of using HRLDAS data on warm-season deep convection forecast.