The Impact of Spatial Discretization Scale on Urban Hydrological Modeling Performance and Prediction

Urban catchments are typically high spatial variability and fast runoff processes resulting in short response times. Thus nowadays distributed model were widely used in urban hydrology due to the capability to better capture the spatial heterogeneity. The development of remote sensing technology also improved the use of high resolution catchment and precipitation information to properly represent catchment response. Selecting the proper spatial resolution (i.e., degree of aggregation) is not a trivial issue because this will affect the model output. However, a general consensus about the effect does not exist. This study investigated the impact of model input resolution on the outputs of detailed hydrodynamic models of an urban catchments in Sendai City of Japan. The aim was to identify critical resolutions for urban catchments to properly characterize catchment response and to quantify the impact of spatial resolution on model results and calibrated model parameters. Firstly a high resolution model was build up and calibrated. Based on this, a series of model was built via an upscale approach. The performances these models were compared and afterward each model were calibrated independently. The parameterization of EIA and prediction capacities were discussed during and after the calibration. Finally, the runoff of an adjacent small catchment were modeled to analyze the model fidelity of performance and prediction at local scale. There were obvious scale effect across models due to the non-linearity nature of model structure. The peak flow tend to decrease with the upscale process while the total flow generally keep constant. For the result of independent calibration, all the models showed good performances. The performance of the calibrated models were equal to or better than the high resolution model which meant that calibration could completely compensate models’ scale effect. However, the deterioration during the validation suggested a result of missing certain spatial information. The consideration of EIA (effective impervious area) as a calibration parameter increased performance of calibration period but decreased performance in validation period which indicated the importance of detailed EIA identification.