Comparing ERT- and scaling-based approaches to parameterize soil hydraulic properties for spatially distributed model applications

Optimal management of water resources depends on the prediction capability of process-based hydrological models. New generation Richards equation-based distributed models provide detailed descriptions of hydrological processes that, however, depend highly on the proper assessment of the spatial variability of soil hydraulic properties (SHPs). Field surveys and sampling campaigns commonly provide sparse local scale measurements of SHPs. Nevertheless, the challenge is to transfer this information from the observation scale to the model scale by increasing as much as possible the prediction efficiency with affordable practical burdens. The performances of two parameterization approaches are compared in this study in terms of outputs of water budget simulated by the HydroGeoSphere (HGS) model in a small catchment. The first approach relies on knowledge of soil hydraulic data from soil cores (centimeter scale) collected in many locations of the study site and the application of the Miller-Miller scaling technique enabling the spatial variability of the so-called aggregated SHPs to be described over the spatial domain of interest. The second approach, instead, estimates a set of equivalent SHPs through the numerical inversion of geophysical data measured with the electrical resistivity tomography (ERT) during an infiltration experiment (meter scale) performed in only one plot-transect of the study catchment. The SHPs obtained by these two approaches are then assigned to each numerical grid cell in HGS to simulate the catchment-scale soil water budget at daily time resolution. Within the framework of a functional evaluation, we compare the performance of these two soil hydraulic parameterization approaches in terms of streamflow (viewed as a lumped flux) and near-surface soil effective saturation patterns (viewed as distributed state variables). Better comparisons between observed and simulated streamflow values are obtained when the study site is parameterized using the equivalent SHPs. Conversely, the HGS water budget simulations using the aggregated SHPs are more effective when one should describe the spatial patterns of near-surface soil moisture values. This study highlights the pros and cons and potential flaws when choosing two different parameterization techniques in spatially distributed modeling applications.