Quantifying the Impact of Subsurface‐Land Surface Physical Processes on the Predictive Skill of Subseasonal Mesoscale Atmospheric Simulations

Abstract

AbstractIntegrated terrestrial system modeling platforms, which simulate the 3‐D flow of water both in the subsurface and the atmosphere, are expected to improve the realism of predictions through a more detailed physics‐based representation of hydrometeorological processes and feedbacks. We test this expectation by evaluating simulation results from different configurations of an atmospheric model with increasing complexity in the representation of land surface and subsurface physical processes. The evaluation is performed using observations during the (HD(CP)2) Observational Prototype Experiment field campaign in April–May 2013 over western Germany. The augmented model physics do not improve the prediction of daily cumulative precipitation and minimum temperature during this period. Moreover, a cold bias is introduced in the simulated daily maximum temperature, which decreases the performance of the atmospheric model with respect to its standard configuration. The decreased performance in the maximum temperature is traced in part to a higher simulated soil moisture, which shifts surface flux partitioning toward higher latent and lower sensible heat fluxes. The better reproduced air temperature profiles simulated by the standard atmospheric model comes, however, with an overestimated heat flux at the land surface caused by a warm bias in the simulated soil temperature. Simulated atmospheric states do not correlate significantly with differences in soil moisture and temperature; thus, different turbulent flux parameterizations dominate the propagation of the subsurface signal into the atmosphere. The strong influence of the lateral synoptic forcings on the results suggests, however, the need for further investigations encompassing different weather situations or regions with stronger land‐atmosphere coupling conditions.