【摘要】 为分析水电站不同蓄泄水期其流域滑坡灾害形变演化和滑坡灾害易发性规律，提出水电站流域蓄泄水期滑坡灾害识别的一般性时序InSAR监测方法流程，同时利用深度学习技术对光学遥感图像进行滑坡潜在区域识别。与InSAR结果对比验证，引入InSAR形变结果作为评价因子，同其它评价要素利用层次分析法进行滑坡灾害易发性评价。结果表明：三阶段蓄泄水期间水电站流域形变区域数量逐渐减少，形变量级先增大后减小，研究区地表形变速率值为-46.6～16.6 mm/a。基于U-net神经网络识别研究区滑坡潜在区域，与InSAR对比发现多数InSAR形变区域均在其识别范围内。滑坡灾害易发性评价结果显示三阶段蓄泄水期水电站流域滑坡灾害易发性逐渐下降。研究成果可为水电站的全生命周期监测运营管理提供技术支持。更多还原

【Abstract】 In order to analyze the deformation evolution and the regularity of landslide hazard in the basin of hydropower station in different storage and discharge periods, the general time-series InSAR monitoring method process for landslide hazard identification in the drainage period of hydropower station basin is proposed. Meanwhile, the deep learning technology is used to identify the potential landslide areas in optical remote sensing images, which are compared and verified with InSAR results. InSAR deformation results are introduced as evaluation factors, and the susceptibility to landslide hazards is evaluated using the hierarchical analysis method together with other evaluation factors. The results show that the number of deformation areas in the basin of the hydropower station gradually decreases during the three-stage drainage period, the magnitude of deformation increases first and then decreases, and the surface deformation rate in the study area ranges from-46.6～16.6 mm/a. The U-net neural network was used to identify potential landslide areas in the study area, and compared with InSAR, it was found that most of the deformation regions of InSAR were within the recognition range. The results of landslide hazard susceptibility evaluation show that the landslide hazard susceptibility of hydropower station basin decreases gradually during the three-stage drainage period. The research results can provide technical support for the whole life cycle monitoring operation management of hydropower station.更多还原