Sensitivity analysis of surface runoff generation in urban flood forecasting

Investigating relationships between landscape patterns and surface runoff from a spatial distribution and intensity perspective

Rapid urbanization changes landscape patterns and results in frequent urban waterlogging issues, which affect citizens' daily lives and cause economic loss. Understanding the spatial patterns and impact factors associated with urban waterlogging under different rainfall intensities has significant implications for mitigating this hazard. In this study, the runoff depth calculated according to the Storm Water Management Model (SWMM) simulation results was used to investigate the spatial characteristics of urban waterlogging. Multiple scenario-based designs, a correlation analysis, and a stepwise regression model were employed to detect the relationship between surface runoff depth and landscape patterns under different rainfall intensities. The results show that when the rainfall intensity reached 12.5 mm/12 h, the conversion rate of rainfall to runoff increased significantly, indicating an increased waterlogging risk. Areas with impervious surface proportions of 25-50% and 75-100% were shown to require more attention due to the strong sensitivity of the surface runoff depth to an increase in the impervious surface. It is most cost-effective to maintain the original high-density vegetation or increase the vegetation density from 0-25% to 25-50% for urban green space. Additionally, the landscape configuration also affects the surface runoff depth. The fragmented, scattered, or regular shape of impervious surface patches can reduce surface runoff effectively; larger and less fragmented green space was also shown to have a surface runoff controlling. The adjusted R² values were greater than 0.6 for all stepwise regression models, indicating that the landscape variables selected in the study can effectively predict the surface runoff depth. These models also showed that the landscape composition had a more profound contribution than the landscape configuration on runoff depth. These findings provide meaningful insights and perspectives for urban waterlogging hazard mitigation, quantitative landscape planning, and risk management. The method proposed by this study provides a referable framework for future studies on urban waterlogging and its response to the landscape in the context of global climate change.

A Sink Screening Approach for 1D Surface Network Simplification in Urban Flood Modelling

Sinks configure the surface networks for overland flow processes representations during 1D hydrodynamic modelling. The excessive number of sinks detected from high-resolution DEMs can boost 1D computational costs significantly. To pursue optimal sink numbers and their optimal spatial distribution, a Volume Ratio Sink Screening (VRSS) method was developed to screen for computationally important sinks, while compensating for volume losses from removed (unimportant) sinks, such that 1D hydrodynamic modelling yields faster computing times without significant loss of accuracy. In comparison with an existing geometry-based sink screening method, we validated this method by conducting sensitivity analyses for the proposed screening criteria in three Danish case areas of distinct topographies. Two iterative procedures were programmed to assess and compare their sink screening performances in terms of sink number reductions and volume loss reductions, and a volume loss solver was developed to quantify catchment-wide volume losses in the 1D surface network. Compared to a geometry-based sink screening method, the VRSS method performs more robustly and produces more efficient reductions in the number of sinks, as well as efficient reductions in volume losses.

Urban pluvial flood prediction: a case study evaluating radar rainfall nowcasts and numerical weather prediction models as model inputs

Flooding produced by high-intensive local rainfall and drainage system capacity exceedance can have severe impacts in cities. In order to prepare cities for these types of flood events - especially in the future climate - it is valuable to be able to simulate these events numerically, both historically and in real-time. There is a rather untested potential in real-time prediction of urban floods. In this paper, radar data observations with different spatial and temporal resolution, radar nowcasts of 0-2 h leadtime, and numerical weather models with leadtimes up to 24 h are used as inputs to an integrated flood and drainage systems model in order to investigate the relative difference between different inputs in predicting future floods. The system is tested on the small town of Lystrup in Denmark, which was flooded in 2012 and 2014. Results show it is possible to generate detailed flood maps in real-time with high resolution radar rainfall data, but rather limited forecast performance in predicting floods with leadtimes more than half an hour.