The potential for using the ensemble square root filter data assimilation technique to estimate soil moisture profiles, surface heat fluxes, and the state of the planetary boundary layer （PBL） is explored. An observing system simulation experiment is designed to mimic the assimilation of near-surface soil moisture observations （θo ） and in-situ measurements of 2-m temperature （To ）, 2-m specific humidity （Qo ）, and 10-m horizontal winds [Vo =（Uo , Vo ）]. The background forecasts are generated by a one-dimensional coupled land surface-boundary layer model （CLS-BLM） with soil, surface-layer and PBL parameterization schemes similar to those used in the Weather Research and Forecasting （WRF） model. Soil moisture, surface heat fluxes, and the state of the PBL evolve on different characteristic timescales, so the minimum assimilation time intervals required for skillful estimates of each target component are different. Correct estimates of the soil moisture profile are obtained effectively when a 6-h update time interval is used, while skillful estimates of surface fluxes and the PBL state require more frequent updates. The CLS-BLM requires a shorter assimilation time interval to correctly estimate the soil moisture profile than previously indicated by experiments using an off-line land surface model （LSM）. Results from assimilating different subsets of observations show that θo makes a larger contribution to soil moisture estimates, while To , θo , and Vo are more important for estimates of surface heat fluxes and the PBL state. It is therefore necessary to combine these variables to accurately estimate the states of both the land surface and the PBL. Experimentation with different prescribed observational errors shows that the assimilation system is more sensitive to increases in observational errors than to reductions in observational errors.
