Through our participation in the evaluation and intercomparison of land-models under the Global Soil Wetness Project (GSWP), we identified a few problems with land-surface fluxes and hydrology simulations of Simple SiB (SSiB) in regions of significant snowmelt. Mocko and Sud (abstract, this volume) show that inclusion of a separate snowpack model for snow-cover with its own energy budget and flux exchanges with the atmosphere aloft and soil beneath helps to ameliorate the deficiencies of delayed snowmelt and excessive Spring season runoff. While the improvement in snowmelt timing happens everywhere, the snowmelt infiltration has became excessive in some regions, e.g., Lena River Basin. This leads to much reduced runoff as compared to observations in regions containing strong hill slopes. This is a consequence of neglect of the influence of subgrid-scale variations in orography for the production of surface runoff.
To solve the aforestated problem, we modified the SSiB model to allow for the influence of horizontal variations in topography on runoff. This is instituted by analyzing observations on river basins and latitude bands to determine how the lateral runoff varies as a function of subgrid-scale orography. We parameterize pre-infiltration runoff, Ro by a simple relationship: Ro = (Snowmelt + Rain) f (Orog. st. dv.); where functional f represents runoff fraction of the available surface infiltration flux = Snowmelt + Rain. Clearly f is a function of subgrid scale variance of orography which must be normalized. Such a surface runoff model is currently being evaluated for implementation into SSiB with the new snow model. Further evaluation of the scheme will be conducted with the ISLSCP-GSWP dataset for 1987-1988 as well as in the GEOS II GCM. The modified SSiB is expected to resolve the problem of too little runoff in regions of strong orography.