Research On Flood Disaster Risk Assessment Methods In Urbanized Areas Under Different Spatial Scales

As one of the most frequent and widespread natural disasters,flooding has brought many negative effects on humans and the environment.Urbanized areas are areas with dense population and developed economy.Meanwhile,their land uses are undergoing drastic changes,which makes them the crucial areas for studying flood disaster risk.As an important technical measure of non-engineering means,flood risk assessment can provide effective technical support for disaster prevention and mitigation.Exploring the applicable methods for urban areas with different spatial scales and investigating risk laws have important theoretical research value and practical guiding meaning.Taking Xiangzhou District of Zhuhai City and the Pearl River Delta as the research regions,this paper constructs different assessment models for flood resilience and flood risk,deeply explores the inherent laws of risk,and summarizes flood risk assessment methods at different spatial scales.The main research achievements of this paper are as follows:(1)The Xiangzhou District,Zhuhai city was selected as the urban-scale research region,of which the rainfall-runoff model was constructed based on Info Works ICM.Recorded data were used to verify the rationality of the model,demonstrating that the model had good accuracy and was consistent with the actual situation.(2)Based on the system performance curve and previous research,a new flood resilience assessment method was proposed and applied to the flood resilience assessment in Xiangzhou District,Zhuhai City.This method can identify areas with low resilience and pose great threats to human production and life.Then,four heavy rain scenarios with different return periods and two time-scenarios including work and rest hours of people were set to apply this novel resilience assessment metric,after which the evaluation and zoning of flood resilience under eight scenarios were conducted.Results show that under the same rain pressure,the research region exhibits lower resilience during rest hours.Raising building doorsills can significantly reduce areas with low and very low resilience,and this reduction effect is more pronounced during rest hours.In addition,raising building doorsills can significantly improve the resilience of residential lands.(3)The flood resilience was integrated into the "Hazard-Vulnerability" flood risk assessment system as a part of hazard,after which scenario simulation and multi-source data were used to assess the flood risk in Xiangzhou District,Zhuhai City.Based on the analysis of flood risk,a zoning map of the highest-risk return period is further proposed.Results show that the multi-source data used in this study can obtain accurate vulnerability index data such as population density and GDP density with the spatial resolution of 4 m and 130 m,respectively.As the rainstorm return period increased,the regional risk became higher,and areas with very high risk had the highest growth rate.Moreover,the highest-risk return period zoning map could be utilized to quickly identify highest-risk areas under different rainstorm scenarios,which could provide efficient guidance for disaster prevention and mitigation.(4)Taking the Pearl River Delta as the research region,12 indices were selected and six machine learning methods were applied to evaluate the flood risk,where the gradient boosting decision tree(GBDT),XGBoost,and convolutional neural network(CNN)were firstly applied in this field.In the dataset construction process,the flood risk results in multiple local areas including Xiangzhou District were utilized.Based on the best-performing model,the inherent laws of risk were investigated and the corresponding disaster prevention and mitigation strategies were proposed.Results show that GBDT performed best among all model,its prediction accuracy on the test set reaches 96.83%,and the corresponding AUC value reached0.9728.The disaster-inducing factor,disaster-breeding environment,and disaster-bearing body were not definitely becoming more serious as the flood risk increased.In the highest-risk areas,rural areas were featured by worse disaster-breeding environment than urban areas,and the disaster-inducing factors of coastal areas were more serious than those of inland areas.