The representation of land surface fluxes and runoff generation in atmospheric circulation models has always been limited by computer run times. The variability that is important to the hydrologist has, until recently, generally been ignored by atmospheric models which have estimated fluxes on the basis of grid square average conditions. Some predictions of the atmospheric models, including rainfall and runoff forecasts, will, however, be sensitive to such sub-grid variability. Work at Lancaster is exploring these sensitivities using spatially distributed hydrological models and fine scale atmospheric modelling, together with the uncertainties associated with such predictions. One interesting question is, given the uncertainties associated with the predictions, how complex a land surface parameterisation can be justified by the data.

Current work is looking at sensitivities of rainfall forecasts to land surface conditions and is supported by NERC.

Some recent references