Optimal sampling of soil depth variability for the prediction of hydrological response

It has been shown that the spatial pattern of soil depth is important for determining the hydrological response of a hillslope or catchment to storm events. This importance is due to the combined effect of differing water storage and the generation of the hydraulic routing surface. However, in general there is limited information available on soil depths compared to surface topography because information on soil depths is difficult and expensive to obtain. This problem therefore raises the question: given limited resources, how should soil depths be sampled and how can the acquired information be utilised to best effect within a hydrological modelling framework? The approach taken in this research is to sample a limited number of soil depths from a known soil depth map and utilise this information in stochastic soil depth map generation algorithms. Six different soil depth pattern generating algorithms have been investigated based on, sampling from different statistical distributions (uniform and exponential), smoothing of the resultant depth map and the use of fractal based approaches with differing amounts of roughness. The realisations of the different algorithms have been used in the hydrological model, CRUM3, which predicts the water table surface at each time step and hence responds dynamically to changes in the soil depth map. These predictions can then be compared with two hydrographs, one being the measured discharge and the second being the modelled hydrographwith the known soil depth map. These different hydrographs can then be statistically assessed to show how well each method is able to reproduce the measured hydrograph with limited data. The CRUM3 model has been applied to the Panola study hillslope (Georgia, USA). Results will be presented that explore the best of the six methods to generate soil depth maps from limited data, where on the hillslope is the best place to sample soil depth and which characteristics of the soil depth distribution are most important for determining the hydrological response of hillslopes to storm events.