How does imperviousness impact the urban rainfall-runoff process under various storm cases?

Dramatic changes in imperviousness exert significant influence on the rainfall-runoff process in urban catchments. In urban rainwater management, imperviousness is generally adopted as an effective indicator for assessing potential runoff risk. However, the effects of imperviousness on rainfall-runoff at the scale of small urbanized drainage areas have not been fully determined, particularly when various storm characteristics are considered. In this paper, a model-based analysis is conducted in a typical urban residential catchment in Beijing, China, in which 69 subareas are delineated within the catchment as the basic drainage units. Two metrics, total impervious area (TIA) and directly connected impervious area (DCIA), are employed to quantify the spatial characteristics of imperviousness of the subareas. Three runoff variables within the delineated subareas including total runoff depth (Q_t), peak runoff depth (Q_p), and lag time (T_lag) are simulated by using the Storm Water Management Model (SWMM) to represent the specific rainfall-runoff characteristics. Moreover, model input storms are designated to several typical flood-induced rainfall events with varying amounts, locations of rainfall peak, and durations for holistic assessment of imperviousness. Regression analyses are conducted to explore the contributions and relative significances of impervious metrics in predicting runoff variables under various storm cases. The results indicate that the performances of imperviousness with fine spatial scale (<1 ha) and heavy rainfall conditions (>34 mm) may vary markedly according to storm conditions. Specifically, TIA rather than DCIA acts as a dominate factor affecting total runoff, and its significance maintains relatively stable with various storm conditions. In addition, the combined use of both TIA and DCIA are more effective for predicting peak runoff than that using a single impervious metric; however, rainfall amount, peak location, and duration alter the contribution gaps between TIA and DCIA and the overall performance of the regression model. Moreover, DCIA is more likely to affect runoff lag time without the contribution of TIA; however, an increase in rainfall peak ratio or duration will significantly limit its performance. These results can provide insight into the hydrologic performance of imperviousness, which is essential for landscape design and runoff regulation in small urban catchments.