Abstract:Modeling hydrologic processes for depression-dominated areas such as the North American Prairie Pothole Region is complex and reliant on a clear understanding of dynamic filling-spilling-merging-splitting processes of numerous depressions over the surface. Puddles are spatially distributed over a watershed and their sizes, storages, and interactions vary over time. However, most hydrologic models fail to account for these dynamic processes. Like other traditional methods, depressions are filled as a required preprocessing step in the Soil and Water Assessment Tool (SWAT). The objective of this study was to facilitate hydrologic modeling for depression-dominated areas by coupling SWAT with a Puddle Delineation (PD) algorithm. In the coupled PD-SWAT model, the PD algorithm was utilized to quantify topographic details, including the characteristics, distribution, and hierarchical relationships of depressions, which were incorporated into SWAT at the hydrologic response unit (HRU) scale. The new PD-SWAT model was tested for a large watershed in North Dakota under real precipitation events. In addition, hydrologic modeling of a small watershed was conducted under two extreme high and low synthetic precipitation conditions. In particular, the PD-SWAT was compared against the regular SWAT based on depressionless DEMs. The impact of depressions on the hydrologic modeling of the large and small watersheds was evaluated. The simulation results for the large watershed indicated that SWAT systematically overestimated the outlet discharge, which can be attributed to the failure to account for the hydrologic effects of depressions. It was found from the PD-SWAT modeling results that at the HRU scale surface runoff initiation was significantly delayed due to the threshold control of depressions. Under the high precipitation scenario, depressions increased the surface runoff peak. However, the low precipitation scenario could not fully fill depressions to reach the overflow thresholds in the selected sub-basins. These results suggest the importance of depressions as gatekeepers in watershed modeling.