Improving the USGS Thornthwaite water balance model in semi-arid settings by including baseflow

The USGS Thornthwaite water balance model is used extensively in studies of climate change, flooding, and water resource management. Using monthly temperature and precipitation records and a set of six parameters, the simple, mechanistic model simulates actual and potential evapotranspiration, snow accumulation and melt, soil moisture, and runoff. In most temperate settings, careful parameter estimation can yield model results that match observations reasonably well. In the semi-arid watersheds of the intermountain west, however, the Thornthwaite model often simulates baseflow recession over the summer months poorly. By adding a simple baseflow component to represent the contribution of groundwater to runoff, the performance of the model in semi-arid watersheds can be improved significantly. The revised algorithm is presented, and model outputs are optimized based on a number of different objective functions. Optimized outputs from the original and the revised Thornthwaite models are compared for several semi-arid watersheds, and significant improvements are demonstrated in a number of different objectives. Given the scarcity of water resources in the American west, accurate simulation of the summer baseflow recession is particularly important - this revised model significantly improves the utility of the Thornthwaite water balance for water resource management in semi-arid settings.