TOPMODEL simulations of streamflow and depth to water table in Fishing Brook Watershed, New York, 2007-09

TOPMODEL, a physically based, variable-source area rainfall-runoff model, was used to simulate streamflow and depth to water table for the period January 2007-September 2009 in the 65.6 square kilometers of Fishing Brook Watershed in northern New York. The Fishing Brook Watershed is located in the headwaters of the Hudson River and is predominantly forested with a humid, cool continental climate. The motivation for applying this model at Fishing Brook was to provide a simulation that would be effective later at this site in modeling the interaction of hydrologic processes with mercury dynamics.
TOPMODEL uses a topographic wetness index computed from surface-elevation data to simulate streamflow and subsurface-saturation state, represented by the saturation deficit. Depth to water table was computed from simulated saturation-deficit values using computed soil properties. In the Fishing Brook Watershed, TOPMODEL was calibrated to the natural logarithm of streamflow at the study area outlet and depth to water table at Sixmile Wetland using a combined multiple-objective function. Runoff and depth to water table responded differently to some of the model parameters, and the combined multiple-objective function balanced the goodness-of-fit of the model realizations with respect to these parameters. Results show that TOPMODEL reasonably simulated runoff and depth to water table during the study period. The simulated runoff had a Nash-Sutcliffe efficiency of 0.738, but the model underpredicted total runoff by 14 percent. Depth to water table computed from simulated saturation-deficit values matched observed water-table depth moderately well; the root mean squared error of absolute depth to water table was 91 millimeters (mm), compared to the mean observed depth to water table of 205 mm. The correlation coefficient for temporal depth-to-water-table fluctuations was 0.624. The variability of the TOPMODEL simulations was assessed using prediction intervals grouped using the combined multiple-objective function. The calibrated TOPMODEL results for the entire study area were applied to several subwatersheds within the study area using computed hydrogeomorphic properties of the subwatersheds.