Assessment of urban flood risk based on data-driven models: A case study in Fuzhou City, China. (November 2022)

• Record Type: Journal Article
• Authors: Ye, Chenlei; Xu, Zongxue; Lei, Xiaohui; Liao, Weihong; Ding, Xingchen; Liang, Yongshuai
• Is Part Of: International journal of disaster risk reduction. Volume 82(2022)
• Publication Date: 2022-11
• Subjects: Physically-based model -- Urban flood risk -- Cloud model -- XGBoost -- Fuzhou city; Emergency management -- Periodicals; Risk management -- Periodicals; Disaster relief -- Periodicals; Hazard mitigation -- Periodicals; 363.34
• DOI: 10.1016/j.ijdrr.2022.103318
• Languages: English
• ISSNs: 2212-4209

Abstract

Abstract: In response to the growing threat of urban flooding, this study developed an integrated method to evaluate the flood risk of the study area (Fuzhou City, China) to different types of urban flooding. First, a coupled hydrological–hydrodynamic model was developed to obtain the key dynamic flood factors forced by typical rainstorm events with river levee overtopping and pipeline transportation. Second, an integrated framework was developed to assess urban flood vulnerability under different flood hazard scenarios. Fifteen features were selected and their relative importance was analyzed by using the XGBoost model. Finally, a cloud model was used to realize qualitative to quantitative characterization of subregions of urban flood risk. The results highlighted the flooding nodes, and the inundation areas presented increasing trends with the increase of return periods. The inundation degree was divided into four categories based on the inundation degree, for which the flooding area was 324.30, 177.81, 43.42, and 20.93 ha, respectively. Six explanatory variables with greater importance were recognized in all schemes. The risk level of subregions changed dynamically, when the rainfall intensity increased, the risk level of subregions showed an increasing trend. Areas with intermediate and high risk were concentrated in the central and western of the study area. At the drainage subregion scale, two types of subareas are essential in urban flood management, respectively, the subregions with high risk in specific rainfall, and the subregions easily transforming into high-risk areas when rainfall increases. This study provides promising references in urban flood risk management. Highlights: The relative importance of indices differs to a certain extent under different rainfall. The flooding nodes and inundation area present an increasing trend when the return period increases. High-risk areas were recognized from qualitative expression to quantitative analysis using the cloud model. The intermediate and high-risk areas were concentrated in the central and western of the study area. Two types of risk zones, including the high-risk zones and dynamically changing risk zones, were recognized.