Spatial Scale Effect of Surface Routing and Its Parameter Upscaling for Urban Flood Simulation Using a Grid-Based Model

Urban catchments are characterized by a wide variety of complex juxtapositions and surface compositions that are linked to multiple overland flow paths. Their extremely high spatial heterogeneity leads to great sensitivity of hydrologic simulation to the scale variation of calculation units. Although extensive efforts have been made for investigating the scale effects and indicate its significance, less is understood of how routing features vary with spatial scales and further how the variation of routing features influences the hydrological response. In this paper, a grid-based distributed urban hydrological model is applied to study spatial scale effects ranging from 10 to 250 m. Two parameters are proposed to quantitatively depict the routing features of overland flow specified for impervious and pervious areas. The results show that routing features are quite sensitive to spatial resolution. Large differences among simulations exist in the infiltration amounts attributed to the combined effects of the two routing parameters, which leads to opposite effects for both total flow volume and peak flow for various rainfall events. The relationship of the key model parameters at different spatial resolutions can be explicitly expressed by corresponding routing features. With this relationship, parameters transfer among different spatial scales can be realized to obtain consistent simulation results. This study further revealed the quantitative relationship between spatial scales, routing features, and the hydrologic processes and enabled accurate and efficient simulations required by real-time flooding forecasting and land-atmosphere coupling, while fully taking the advantages of detailed surface information. Plain Language Summary Given the inherent complex underlying surface compositions and overland flow paths in urban areas, underlying high spatial resolution surface data eventually become necessary. Unfortunately, high-resolution modeling in urban catchment is still challenging in terms of computational restricts, proper setting up of parameters, and so forth due to the high spatial heterogeneity. Practical simulation requirements often limit the use of high-resolution models, as in the case of real-time prediction of urban flooding, the coupling of land-atmosphere processes. Therefore, it is necessary to investigate the scale effects and its mechanism and then to explore an accommodation approach to enable precise flooding prediction with a coarse model. For grid-based and distributed hydrologic models, the mosaic method can basically eliminate the scale effects on the runoff generation process. However, the scale effects on overland flow routing remain insufficiently understood, and to help understand the scale effects, simulations were performed under five different resolutions, ranging from 10 to 250 m, for various rainfall events. Two physical parameters are introduced to quantify the scale effects on routing features. Three variables are concurrently calculated to assess the effects on modeling outputs. The results indicate that routing features are sensitive to changes in spatial resolution, which results in opposite effects on simulation results under different rainfall conditions. In conclusion, an accommodation approach is proposed based on the affecting mechanism.