A process-based hydrological model for continuous multi-year simulations of large-scale watersheds

Abstract

Process-based hydrological models are useful tools to understand the mechanisms responsible for extreme floods and droughts and thus facilitate watershed management. This paper describes a watershed model, based on the coupled surface-subsurface distributed model Penn State Hydrologic Model (PIHM), to capture the hydrologic response of large watersheds over decades. Mathematical models for coupling surface-subsurface fluxes were developed to achieve mass conservation in multi-year simulations. A Voronoi technique was implemented to improve accuracy and computational efficiency. The model was used to simulate the Cedar River Watershed, IA from 2003 to 2018. Model predictions were compared against streamflow observations at four gauging stations. Results show good model performance with average Nash–Sutcliffe efficiency values greater than 0.70. The model results helped to identify the main hydrologic processes that contributed to the historic 2008 flood in the Cedar River Watershed. In addition to heavy storms, several other factors played an important role, including snowmelt and rainstorms in the spring, as well as low evapotranspiration values associated with traditional row crop agriculture in the first half of the year.