The land-atmosphere coupling strength in the GEOS-5 atmospheric general circulation model (AGCM), i.e., the response of simulated precipitation in that model to variations in land state, was studied in two phases: in the context of the "single column model” (SCM) counterpart of the AGCM, and in the AGCM itself. The SCM allowed a particularly comprehensive analysis; the SCM includes all of the vertical physics of the atmospheric model (the radiation physics, the convection scheme, etc.) and can be can be used for fast, inexpensive sensitivity experiments. Sixteen parameters associated with the AGCM cloud moisture and turbulence parameterization schemes were identified as candidates for these experiments. Observations from the Atmospheric Radiation Measurement/Cloud and Radiation Testbeds (ARM/CART) measurement station in Oklahoma were used for the SCM's lateral boundary conditions. Idealized cycles of latent and sensible heat fluxes were prescribed at the land surface in each simulation; the cycles were given a two week period, with a tunable amplitude. The results of the SCM study suggest that two parameters have a particularly strong impact on land atmosphere coupling strength: a parameter related to the relative humidity threshold for cloud water formation, and a parameter associated with the assumed subgrid variability of surface temperature.

The impact of these two parameters on coupling strength was then tested in the full AGCM. The SCM results regarding the subgrid temperature parameter were roughly reproduced in the AGCM environment. The AGCM results, however, did not indicate that the relative humidity parameter was quite so important. Based on these results, and on the known ad hoc nature of the original subgrid temperature parameterization, we are constructing a version of the AGCM in which subgrid temperature variability is tied instead to spatially-varying MODIS-based estimates. Coupling strength results obtained with this modified parameterization will be presented.