山崩
干涉合成孔径雷达
地质学
大地测量学
流离失所(心理学)
遥感
时间序列
运动学
加速度
地震学
合成孔径雷达
计算机科学
心理治疗师
物理
机器学习
经典力学
心理学
作者
Priyom Roy,Tapas R. Martha,Kirti Khanna,Namrata Jain,K. Vinod Kumar
标识
DOI:10.1016/j.rse.2022.112899
摘要
Landslides originating from remote steep slopes render people living downhill vulnerable, unaware of the impending danger. Identifications of slow-moving mountain slopes are possible now due to time series measurement from space using microwave satellite data and the InSAR technique, which potentially can detect displacement at millimetre level. Availability of open-source Sentinel-1 data has revolutionised the study involving landslide kinematics and predicting the time of failure. However, identification of accelerating trend, demarcation of release area and prediction of flow path after failure initiation are still challenging. In this paper, we present a novel method for time and path prediction of landslides using two large landslides (Kikruma and Kotropi) located in the Himalayas in India. Sentinel-1 data stack was processed using the Persistent Scatterer and Small Baseline Subset interferometric techniques to analyse the trend of ground deformation leading to slope failures. The displacement time series of the measurement points, analysed using inverse velocity and modified inverse velocity methods, show that the instability had commenced almost a year or more with the final onset of acceleration triggered by heavy rainfall, couple of weeks prior to the actual failure. The acceleration image created from displacement time series data was clustered using image segmentation techniques to demarcate the release area of landslides. The flow simulation was done using the Voellmy friction model with a high-resolution DEM to predict the flow path. The analysis done for Kikruma and Kotropi landslide case studies with the proposed method provided a safe prediction of the time of landslide with ~90% accuracy of the flow path prediction. Results show that the method demonstrated in this study may evolve as an effective tool for landslide early warning in hilly areas.
科研通智能强力驱动
Strongly Powered by AbleSci AI