Relative influence of vertical and horizontal processes in large-scale water and energy balance modelling

The role of large-scale land-atmosphere transfer schemes is to partition the massive radiative energy from the sun into evaporation and sensible heat. In order to do so, the models need to trace where the radiation energy is being absorbed and how much water is available at the point of absorption. Therefore, great emphasis has been placed on the vertical structure of the land surface and on the vertical movement of water through the soil and the vegetation system. As part of this tracking of the vertical movement of water, there is a necessity to include some representation of the horizontal flow of water. If this is ignored, the land is assumed to be flat and homogeneous, and no surface runoff will be generated, resulting in modelled soil moisture, and hence evaporation, that is too high. This is partly why hydrological models are more commonly being embedded in meteorological land surface schemes. The other reason is that the runoff itself is now required for the purposes of predicting water resources, freshwater flows into the oceans and for model validation using river flow. There is a need to assess how complex these hydrological models need to be. On the one hand, they need to be complex enough to model the processes accurately. On the other hand, the accuracy of the model needs to be weighed against the overhead of obtaining parameters on a global scale. This paper attempts to address this issue by comparing the sensitivity of the desired output (long-term water and energy balance) to the parameters in the horizontal and vertical models. The land surface scheme of the Met Office (Met Office Surface Energy Scheme-MOSES) is used with a standard rainfall-runoff model, Probability Distribution Model (PDM) embedded in the top soil layer. The parameters of the vertical process model are varied from a coarse soil to a fine soil. The parameter for the horizontal process model is varied from a flat homogeneous landscape to one with high variability. It is shown that the long-term water balance in this model is equally sensitive to parameters that describe the horizontal flow of water and the vertical flow of water.