Surface Hydrology in Global Climate Models: How Good is It?

Land surface schemes are used in general circulation models (GCMs) to simulate the surface processes and the surface-atmosphere interactions. This paper addresses several aspects of land surface modelling from a hydrological perspective. It investigates the energy and water fluxes estimated by a typical land surface scheme (CSIRO9) for the Amazon and Mississippi River Basins, using river runoff as a diagnostic variable. The results indicate that the mean monthly values of the key variables (runoff, evapotranspiration, sensible heat and soil moisture) estimated by the CSIRO9 land surface scheme are realistic and consistent with expectations. The study also shows that the incorporation of spatial variability of rainfall intensity and soil moisture provides a more realistic simulation of surface runoff. However, the model estimates are very sensitive to the atmospheric data used to drive the scheme, and until the quality of the atmospheric and land surface data at large spatial scales is comparable to the ever-increasing complexity of the land surface schemes, it is not possible to adequately assess the performance of the schemes.