Evaluating land-atmosphere coupling in model simulations
Klein, Stephen - Lawrence Livermore National Laboratory
Area of research:
Journal Reference:Phillips T, S Klein, H Ma, Q Tang, S Xie, I Williams, J Santanello, D Cook, and M Torn. 2017. "Using ARM Observations to Evaluate Climate Model Simulations of Land-Atmosphere Coupling on the U.S. Southern Great Plains." Journal of Geophysical Research: Atmospheres, 122(21), doi:10.1002/2017JD027141.
The strength of the coupling of soil moisture with the lower atmosphere can influence regional surface temperature, evaporation, and precipitation. Researchers at LLNL, LBNL, and Argonne National Laboratories, and at NASA Goddard Space Flight Center, have investigated land-atmosphere coupling (LAC) strength over the U.S. Southern Great Plains (SGP) region, as estimated from multiple ARM field observations of regional soil moisture and surface atmospheric variables during the warm seasons of 2003-2011. These field estimates then were used to evaluate analogous simulations of LAC by version 5.1 of the Community Atmosphere Model (CAM5.1), when coupled to version 4 of the Community Land Model (CLM4). The model was run in two configurations over the study period: 1) a “free-running” atmospheric simulation with prescribed observations of ocean surface temperatures; and 2) a “controlled” atmospheric simulation having the same ocean boundary condition, but in which model soil moisture and atmospheric state variables were corrected daily from observationally based approximations of these fields. Even though the controlled simulation yielded a more realistic land-surface state than its free-running counterpart, its LAC was nearly as strong as in the free-running simulation, which greatly exceeded the ARM observational estimates.
This research confirms that CAM5.1/CLM4 soil moisture exerts an outsized influence on the model’s weather and climate predictions at SGP, possibly contributing to a large regional warm/dry bias that is typical of current models. Further, the unrealistically strong LAC in both the free-running and controlled climate simulation points to the model’s land-atmosphere coupling mechanisms--rather than its simulation of soil moisture or surface atmospheric variables--as the chief source of the problem. Future work will focus on analyzing these model coupling mechanisms in more detail.
Several independent measurements of warm-season soil moisture and surface atmospheric variables recorded at the ARM Southern Great Plains (SGP) research facility were used to estimate land-atmosphere coupling (LAC) strength. The observations revealed substantial variation in coupling strength, as estimated from three soil moisture measurements at a single site, as well as across six other sites having varied soil and land cover types. The observational estimates then served as references for evaluating SGP terrestrial coupling strength in the Community Atmospheric Model version 5.1 coupled to the Community Land Model version 4, in which the coupled system was operated in both a free-running mode and in a controlled configuration in which the atmospheric and land states were reinitialized daily. The controlled simulation deviated less from the observed surface climate than its free-running counterpart, but the LAC in both configurations was much stronger, and displayed less spatial variability, than the SGP observational estimates. These results imply that model physical parameterizations involved in the coupling of CAM5.1/CLM4 land and atmospheric components are likely to be the main sources of the problematical LAC behaviors.