A Study on Parsimonious Models in Catchments Generating Saturation Excess Runoff

Conceptual models in hydrology, e.g. TOPMODEL, represent an attractive alternative to distributed models, especially because of their simplicity, parsimony and computational efficiency. However, they often lack physical concepts and are unable to capture relevant hydrologic phenomena in certain areas. This study looks at surface runoff generation in the Northeastern US, namely Hubbard Brook, NH and Sleepers River, VT watersheds, where saturation excess is the dominant mechanism of runoff generation. We compared an established version of TOPMODEL, which uses watershed topography to predict subsurface and overland flow, against an alternate model where characteristics of soil and groundwater are also implemented. Results show that TOPMODEL tends to underpredict most of peak flows, whereas the alternate model overpredicts them, although the latter should be attributed to an imperfect calibration of the delay function for overland flow routing. We used water volume error to assess model performance in correspondence of peak flow events, and we found that errors vary across events, but are in general comparable between the two models. In conclusion, the study shows that conceptual models can reproduce saturation excess runoff with sufficient accuracy and relatively quickly. When applied to small watersheds, model results can be passed down to larger basins, overcoming the reliance of fully distributed models on coarse grids (> 50 m resolution) for hydrologic simulation.