Urban rainstorm floods have become one of the most significant natural disasters that restrict the healthy development of the economy and society. It is imperative to reduce the risk of urban rainstorm flood disasters, which is growing along with urbanization, during the process of urban planning and construction. However, because of the complexity of urban elements, it is quite difficult to identify sensitive indicators of urban rainstorm flood disasters to provide effective guidance on flood control. In this study, the composition of urban elements was systematically proposed based on Descriptive Framework for Cities and Communities, including natural, structural, and social elements, and then a specific indicator system was established. Furthermore, Random Forest Model was used to determine the relative importance of the indicators that are closely related to the direct economic loss that reflects the severity of urban rainstorm flood disasters. Considering four cities in China, Beijing, Tianjin, Chongqing, and Shanghai, with different topographical and hydrological conditions, as examples, the sensitivity indicators of urban rainstorm flood disasters were identified, and the results were verified based on the catastrophe evaluation method. The results indicate that urban rainstorm flood disasters are the most sensitive to surface water resources, and are the least sensitive to relief degree of land surface. Finally, specific measures to mitigate urban rainstorm flood disasters were proposed.@en

In recent years, dam failures have occurred frequently because of extreme weather, posing a significant threat to downstream residents. The establishment of emergency shelters is crucial for reducing casualties. The selection of suitable shelters depends on key information such as the number and distribution of affected people, and the effective capacity and accessibility of the shelters. However, previous studies on siting shelters did not fully consider population distribution differences at a finer scale. This limitation hinders the accuracy of estimating the number of affected people. In addition, most studies ignored the impact of extreme rainfall on the effective capacity and accessibility of shelters, leading to a low applicability of the shelter selection results. Therefore, in this study, land-use and land-cover change (LUCC) and nighttime lighting data were used to simulate population distribution and determine the number and distribution of affected people. Qualified candidate shelters were obtained based on screening criteria, and their effective capacity and accessibility information under different weather conditions were quantified. Considering factors such as population transfer efficiency, construction cost and shelter capacity constraints, a multi-objective siting model was established and solved using the non-dominated sorting genetic algorithm II (NSGA- II) to obtain the final siting scheme. The method was applied to the Dafangying Reservoir, and the results showed the following: (1) The overall mean relative error (MRE) of the population in the 35 downstream streets was 11.16 %, with good fitting accuracy. The simulation results truly reflect the population distribution. (2) Normal weather screening generated 352 qualified candidate shelters, whereas extreme rainfall weather screening generated 266 candidate shelters. (3) Based on the population distribution and weather factors, four scenarios were set up, with 63, 106, 73, and 131 shelters selected. These two factors have a significant impact on the selection of shelters and the allocation of evacuees, and should be considered in the event of a dam-failure floods.@en

Dam breaches often have catastrophic consequences in downstream areas. Hydrodynamic factors and the evacuation potential of the population at risk (PAR) have significant impacts on the loss of life (LOL) caused by dam breaches. However, the existing comprehensive evaluation models have not conducted in-depth research on the evacuation potential of populations. Thus, limited guidance is available for relevant departments to formulate emergency plans to reduce the potential LOL. Therefore, a new comprehensive evaluation model was proposed in this study. According to the relevant references and disaster theory, the main influencing factors and the process through which the LOL is caused by dam breaches were determined. The specific occurrence process was divided into six stages: a dam breach causes flood, the flood puts the PAR, the PAR complete the preparation work, the PAR evacuate, the un-evacuated population shelter themselves inside buildings, and flood causes the death of the exposed population. To calculate the LOL, the parameters relevant at each stage were defined. Furthermore, the Hydrologic Engineering Center's River Analysis System, Geographic Information System, and related materials were used to simulate the flood routing and evacuation potential of the PAR, quantifying the parameters in the model. The model was applied to 14 towns in the downstream areas of the Luhun Reservoir in Henan Province, China, and its accuracy was verified by comparing the results obtained from the two existing models. In addition, the specific suggestions for reducing the potential LOL were proposed based on the results of the simulation.@en