Search

Study region: Major urban areas in Henan Province of central China. Study focus: data fusion technology is also a key and difficult point in the field of flood research. Remote sensing and text data have different modalities and scales, making fusion difficult. This study proposed a remote sensing and text bimodal data fusion model based on UFCLI, and we validated the spatiotemporal distribution of floods and the calculation results of disaster losses. The results show that through the coupling analysis of remote sensing and text bimodal data, rainstorm and flood events can be fully reproduced in space and time. The proposed UFCLI effectively improves the accuracy of remote sensing single-data inversion for urban flood disaster losses. The flood loss in Henan is 121.98 billion yuan, and the accuracy improvement result is R² increased by 0.08 and MAPE decreased by 0.88. New hydrological insights for the region: In the case of sudden urban storm flooding with complex spatial and temporal evolution, the traditional hydrological-hydraulic model has many pending parameters, which makes it difficult to accurately calculate large-scale disaster losses. By establishing a theoretical model of bimodal data fusion, we effectively use the complementary spatiotemporal information using remote sensing and text to solve the differences in spatiotemporal scales existing between remote sensing and text data. The timeliness and accuracy of urban flood damage estimation have further improved. Data Availability Statement: Not applicable.