Effect of storm network simplification on flooding prediction with varying rainfall conditions

In the context of climate change and urban expansion, cities are increasingly vulnerable to floods for increased frequency of rainfall extremes. Timely and accurate flooding prediction is crucial to reduce the losses of life and property. Despite its crucial role in urban hydrologic modelling, storm network, as a key component of urban drainage system, has to be simplified because of both the data availability and computation power. Current literatures have noted the effects of storm network simplification (SNS), while the understanding is limited to certain models and conditions and still far from sufficient. In this study, a grid-based urban hydrologic model was employed to further investigate the effects of SNS on flooding prediction under varying rainfall conditions. The results show that SNS significantly affects both peak flow and total flow volume, while simplification to different degrees may lead to opposite effects. Larger degree of simplification leads to underestimation of flooding magnitude, while smaller degree of simplification results in overestimation of flooding magnitude. More importantly, the simulation bias caused by SNS is further amplified with increased rainfall intensity and peak ratio, especially for higher degree simplifications where the underestimation bias of river discharge can be as large as three times. Through qualitative and quantitative analysis, this study helps to further understand the effects of SNS, and provides some bias estimation for flooding prediction and management in urban areas.