Evaluation of a fully coupled large-eddy simulation-land surface model and its diagnosis of land-atmosphere feedbacks

Daytime land-atmosphere interactions are the result of two-way coupled processes where the land states strongly affect the overlying atmospheric properties and vice versa. This study presents a numerical framework integrating a radiative parameterization, a large-eddy simulation of the atmospheric boundary layer (ABL), and a force-restore land surface model to investigate these coupled processes and the impact of local-scale atmospheric feedbacks on surface fluxes and states. Using field measurements collected during the Soil Moisture-Atmosphere Coupling Experiment 2002, a 12 h daytime diurnal cycle simulation is performed to evaluate the coupled model performance. The estimates of surface fluxes, near-surface micrometeorological air properties, surface temperature, and vertical profiles of boundary layer characteristics are compared to the measurements collected by meteorological towers, satellites, and radiosondes. Good agreement is obtained in comparisons between model outputs and observations at both footprint-and domain-averaged scales indicating that such a coupled model can reasonably represent the coupled land-ABL system. Using the verified model baseline, simulated results for cases with different degrees of coupling show that the estimates of surface fluxes and near-surface states are mainly dominated by the feedback of air temperature and wind speed, respectively. The feedback of each air property on surface fluxes and states for different vegetation cover can be either positive or negative, with the feedbacks via air temperature and wind speed playing opposite roles in the estimation of sensible heat flux. Thus, the total feedback of multiple air properties increases if all individual impacts are of the same sign; otherwise, the impacts partially cancel each other. Results indicate that the impact of ignoring atmospheric feedbacks (i.e., the local-scale spatial variability of near-surface air properties due to coupled interactions) results in a clear error in the estimation of sensible heat flux (up to 18%) and near-surface soil moisture (up to 10%). The direct impacts of feedbacks on latent heat flux and surface temperature are relatively unimportant for the case examined.