The use of river runoff to test CSIRO9 land surface scheme in the Amazon and Mississippi River Basins

Point validation of land surface schemes in general circulation models (GCM s) can only provide Limited insight into the performance of the schemes whe n used over the large GCM grid cells. Streamflow, which integrates informat ion over large areas, is a potentially useful diagnostic for assessing the land surface schemes at large spatial scales. This paper discusses the use of streamflow for assessing the performance of the CSIRO9 land surface sche me over the Amazon and the Mississippi River Basins. The paper shows that a lthough streamflow can be a useful diagnostic tool for testing land surface schemes over large scales its utility is undermined by various problems: t he atmospheric control on evapotranspiration, the possibility of compensati ng errors within large basins, and most importantly the lack of reliable gr id-averaged atmospheric data at large spatial scales. Copyright (C) 2000 Ro yal Meteorological Society.