Using Land-Atmosphere Coupling for Process Oriented Model Evaluation

We use differences in model representation of land-atmosphere coupling and soil moisture to develop process level understanding of model skill for surface climate elements (temperature and precipitation) in a suite of regional climate simulations for the continental United States. Here we evaluate simulations over the North American domain of the Coordinated Regional climate Downscaling Experiment (CORDEX) using two regional climate models (WRF and RegCM4) driven by ERA-Interim reanalysis boundary conditions for 1989-2009. These evaluations were performed for simulations using both 25 and 50 km grid spacing. Multiple diagnostic measures of land-atmosphere coupling were used to interpret regional and seasonal variations in model skill as well as inter-model differences. Land-atmosphere coupling "hot spots" were found to differ in both location and magnitude depending on model physics and spatial resolution. Performance for temperature and precipitation over North America varies significantly between models, as well as between simulations using the same model but different physical parameterizations. Results show that land-atmosphere interactions play an important role in these model differences over the Great Plains region of North America where previous studies have shown land-surface moisture feedbacks are large.

This research was supported by the U.S. Department of Energy and the USDA National Institute of Food and Agriculture.