Abstract
Land surface heterogeneities are important for accurate estimation of land-atmosphere interactions and their feedbacks on water and energy budgets. To physically introduce existing land surface heterogeneities into a mesoscale model, a land data assimilation system was coupled with a mesoscale model (LDAS-A) to assimilate low-frequency satellite microwave observations for soil moisture and the combined system was applied in the Tibetan Plateau. Though the assimilated soil moisture distribution showed high correlation with Advanced Microwave Scanning Radiometer on the Earth Observing System soil moisture retrievals, the assimilated land surface conditions suffered substantial errors and drifts owing to predicted model forcings (i.e., solar radiation and rainfall). To overcome this operational pitfall, the Coupled Land and Atmosphere Data Assimilation System (CALDAS) was developed by coupling the LDAS-A with a cloud microphysics data assimilation. CALDAS assimilated lower frequency microwave data to improve representation of land surface conditions, and merged them with higher frequency microwave data to improve the representation of atmospheric conditions over land surfaces. The simulation results showed that CALDAS effectively assimilated atmospheric information contained in higher frequency microwave data and significantly improved correlation of cloud distribution compared with satellite observation. CALDAS also improved biases in cloud conditions and associated rainfall events, which contaminated land surface conditions in LDAS-A. Improvements in predicted clouds resulted in better land surface model forcings (i.e., solar radiation and rainfall), which maintained assimilated surface conditions in accordance with observed conditions during the model forecast. Improvements in both atmospheric forcings and land surface conditions enhanced land-atmosphere interactions in the CALDAS model, as confirmed by radiosonde observations.