Inundation Hotspot: An insurance data and machine learning based fine-scale urban flood inundation map of China

Urban flood inundation usually causes unpredictable losses in cities, and thus establishing corresponding P&C insurance is of great importance to society. However, due to lack of related data, traditional hydrodynamic model is impossible to be applied to all regions for risk simulation and assessment. To solve this issue, based on the historical insurance claims caused by typhoon and rainstorm before 2021 in China, and combining with the weather data, digital elevation model (DEM), and building outline data, we generated an inundation risk map with adaptive clustering algorithm, and the accuracy of this map is validated by the official flood inundation spots. The urban flood inundation risk map has been embedded into the primary insurance workflow, which can benefit risk management, aid in quantifying risk exposures and improve risk prevention and relief strategies. During application, many assets with high flood risks are identified prior to underwriting, and therefore the losses due to floods are largely reduced. In summary, this study shows the possibility of building an efficient urban flood inundation risk map based on big data and machine learning method, which is more applicable than pure hydrology model in the scenario of insurance, indicating a new direction for flood risk assessment.