When impervious cover doesn't predict urban runoff: Lessons from distributed overland flow modeling

We used the distributed hydrologic model GSSHA to simulate overland runoff from impervious and semipervious (low vegetation) land covers in a 0.3 km2 urban catchment. The model is hypothetical, which allowed us to test alternative land use configurations and better understand the effects of urban heterogeneity. To track the arrival of water to a stormwater basin inlet from distinct parts of the catchment, we applied a conservative solute tracer to portions of the gridded model area, a new application for GSSHA's 'contaminant transport' module. We rearranged the positions of pervious and impervious land uses over the catchment during four storm events and found that the portion of total runoff volume from impervious areas varied from 50 to 75% while the relative proportion of impervious cover remained constant at 54%. We also applied solute to equalsized zones with different slopes and land uses and found that percent imperviousness did not predict solute mobilization. Furthermore, variation in total runoff volume and peak discharge for different land use arrangements and storm sizes showed that percent imperviousness could not be used to estimate a runoff-rainfall ratio. Fluctuating solute proportions over time suggested that grab samples might not be adequate for capturing average overland runoff chemistry. We quantified factors contributing to non-linearity in rainfall-runoff responses from an urban watershed with stormwater infrastructure and determined that using models to find hotspots for stormwater infiltration and to explore resilience to future storm events can improve design and function of stormwater control measures.