In this paper we present a variable bucket capacity (VBC) representation of TOPMODEL (Beven and Kirkby, 1979), generalizing the Manabe 'simple bucket' scheme widely used in climate models. In doing this, we have been motivated by the variable infiltration capacity (VIC) model presented by Wood et al. (1992). For catchments, or for parts of the land surface, whose hydrology fits the TOPMODEL assumptions, this lumped version offers an alternative representation that can be used as the building block for land surface hydrology models at large spatial scales. The model is first applied, with satisfactory results, to a small catchment near Canberra in Australia. Next, using the same model, the effects of random spatial variability of rainfall are studied, but in the context of land surface hydrological parameterizations for global climate models. Sub-grid rainfall variability, especially the partial coverage that characterizes rainfall events at such large scales, is found to cause significant biases in the estimation of land surface fluxes. Other manifestations of the effects of the spatial variability of rainfall are also investigated, also by means of model simulations, but with a view to gaining insights for developing simple parameterizations of land surface fluxes at large scales. This includes the investigation of the dynamics, i.e. spatial and temporal variability, of the soil moisture state variable during and after a rainfall event. In addition, empirical relationships linking the variable contributing area, subsurface runoff and evaporation, to a soil moisture state variable (e.g. soil moisture storage), are derived based on model simulations, and the effects of the sub-grid rainfall variability on these relationships are examined. Because of the strong non-linearity inherent in the process descriptions within the model, the above 'constitutive' relationships are strongly affected by rainfall heterogeneity. (C) 1997 by John Wiley & Sons, Ltd.
|Publication status||Published - 1997|