Assessing Large-Scale Flood Risks: A Multi-Source Data Approach

Flood hazards caused by intense short-term precipitation have led to significant social and economic losses and pose serious threats to human life and property. Accurate disaster risk assessment plays a critical role in verifying disaster statistics and supporting disaster recovery and reconstruction processes. In this study, a novel Large-Scale Flood Risk Assessment Model (LS-FRAM) is proposed, incorporating the dimensions of hazard, exposure, vulnerability, and coping capacity. Multi-source heterogeneous data are utilized for evaluating the flood risks. Soil erosion modeling is incorporated into the assessment framework to better understand the interactions between flood intensity and land surface degradation. An index system comprising 12 secondary indicators is constructed and screened using Pearson correlation analysis to minimize redundancy. Subsequently, the Analytic Hierarchy Process (AHP) is utilized to determine the weights of the primary-level indicators, while the entropy weight method, Fuzzy Analytic Hierarchy Process (FAHP), and an integrated weighting approach are combined to calculate the weights of the secondary-level indicators. This model addresses the complexity of large-scale flood risk assessment and management by incorporating multiple perspectives and leveraging diverse data sources. The experimental results demonstrate that the flood risk assessment model, utilizing multi-source data, achieves an overall accuracy of 88.49%. Specifically, the proportions of areas classified as high and very high flood risk are 54.11% in Henan, 31.74% in Shaanxi, and 18.2% in Shanxi. These results provide valuable scientific support for enhancing flood control, disaster relief capabilities, and risk management in the middle and lower reaches of the Yellow River. Furthermore, they can furnish the necessary data support for post-disaster reconstruction efforts in impacted areas.