Using ARM Observations to Evaluate CAM5.1/CLM4 Simulations of Land-Atmosphere Coupling on the Southern Great Plains
Poster PDF
Authors
Thomas J. Phillips — Lawrence Livermore National Laboratory
Stephen Klein — Lawrence Livermore National Laboratory
Hsi-Yen Ma — Lawrence Livermore National Laboratory
Qi Tang — Lawrence Livermore National Laboratory
Shaocheng Xie — Lawrence Livermore National Laboratory
Williams Ian — University of Chicago
Joseph A. Santanello — NASA - Goddard Space Flight Center
David R. Cook — Argonne National Laboratory
Margaret S. Torn — Lawrence Berkeley National Laboratory
Category
Boundary layer structure, including land-atmosphere interactions and turbulence
Description
To estimate the range of land-atmosphere coupling (LAC) in the SGP region, we employed several independent ARM measurements of shallow-depth soil moisture SM and of the surface evaporative fraction EF and temperature T during 2003-2011 May-August warm seasons. We then used these observational estimates to evaluate the simulated LAC in version 5.1 of the global Community Atmosphere Model (CAM5.1), when coupled to the CLM4 Land Model. Two model experimental cases were considered:
1) a free-running Atmospheric Model Intercomparison Project (AMIP) simulation with historically observed SSTs specified, and
2) a constrained Hindcast (HC) run with the same SST specifications, but where the CAM5.1 atmospheric state was initialized each day from the ERA Interim reanalysis, while the CLM4 daily initial conditions were obtained from an offline run of the land model using observed surface forcings.
The constrained HC simulation deviated substantially less from the observed surface climate than its AMIP counterpart; but in both configurations, the SM-EF and SM-T couplings were much stronger than the SGP observational estimates. The source of the model’s overly strong LAC thus appears to be situated mainly in the detailed physics of its land-atmosphere coupling, rather than in biases displayed by individual variables.
This work was performed at LLNL under Contract DE-AC52-07NA27344. LLNL-ABS-744045.
In order to gain further insight into these coupling processes, we have constructed composite averages of AMIP and HC surface atmospheric variables at different days’ lag during dry-down periods following precipitation events. We then compared the simulated lagged composites with those determined from the ARM observations at the SGP Central Facility. While the dry-down characteristics of the HC run deviate less extremely than those of the AMIP run from observed behaviors, qualitative similarities are seen in both simulation configurations. This suggests a need to test the effects of alternative surface and boundary-layer parameterizations in the model, in order to produce a simulation of LAC that is closer to what the ARM observations imply.
