Sensitivity of a GCM simulation to subgrid infiltration and surface runoff

A subgrid parameterization of infiltration and surface runoff was evaluated using a land surface model coupled to an atmospheric general circulation model. Averaged over 5 year simulations, the subgrid parameterization resulted in significantly less infiltration of water into the soil compared to a simulation without subgrid hydrologic processes. As a result, the soils were drier, latent heat flux decreased, and surface air temperature increased. These results are consistent with other studies of subgrid hydrologic parameterizations, which also resulted in drier soils, decreased latent heat flux, and warmer surface temperatures. Several river basins were studied in detail. In the Amazon and Lena basins, the subgrid parameterization resulted in better annual runoff compared to observed annual river flow; surface air temperature was unchanged in the Amazon and better, compared to observations, in the Lena. In the Ob, Yenisey, and Amur basins, the subgrid parameterization resulted in too much annual runoff; July surface air temperature was unchanged or worse (Amur). Annual runoff for the Mississippi basin was better with the subgrid parameterization, but July surface air temperature was worse. These results sug gest the utility of subgrid hydrologic parameterizations vary among river basins depending on the relative importance of Horton and Dunne runoff and the geologic factors affecting runoff generation.