Evaluation of soil moisture in CMIP5 simulations over the contiguous United States using in situ and satellite observations

Abstract

This study provides a comprehensive evaluation of soil moisture simulations in the Coupled Model Intercomparison Project Phase 5 (CMIP5) extended historical experiment (2003 to 2012). Soil moisture from in situ and satellite sources is used to evaluate CMIP5 simulations in the contiguous United States (CONUS). Both near-surface (0-10 cm) and soil column (0-100 cm) simulations from more than 14 CMIP5 models are evaluated during the warm season (April-September). Multimodel ensemble means and the performance of individual models are assessed at a monthly timescale. Our results indicate that CMIP5 models can reproduce the seasonal variability in soil moisture over CONUS. However, the models tend to overestimate the amount of both near-surface and soil column soil moisture in the western US and underestimate it in the eastern US. There are large variations across models, especially for the near-surface soil moisture. There are significant regional variations in performance as well. Results of a regional analysis show that in the deeper soil layers, the CMIP5 soil moisture simulations tend to be most skillful in the southern US. Based on both the satellite-derived and in situ soil moisture, CESM1, CCSM4 and GFDL-ESM2M perform best in the 0-10 cm soil layer and CESM1, CCSM4, GFDL-ESM2M and HadGEM2-ES perform best in the 0-100 cm soil layer.