29 / 2021-10-29 08:23:14

Non-contact monitoring technology of landslide with multi-source data fusion and landslide dynamic evolution

landslide; non-contact monitoring; multi-source data fusion; dynamic evolution

The mountainous region of southwest China is in the transition zone from Qinghai-Tibet Plateau to Sichuan Basin, with mostly high mountain valleys, abundant water system in the watershed and fragile geological conditions. Under the action of external factors such as rainfall, earthquake and glacial snow melt, it is very easy to trigger rockfall, landslide and other slope disasters, which pose threats to the implementation of national major strategic projects and regional economic development. Due to the complex natural environment and terrain conditions, the acquisition of post-slide terrain data as well as the monitoring of landslide deformation face the challenges of time constraint, difficulty and high risk. In the absence of revealing the landslide dynamic evolution and disaster-causing mechanism, the disaster prevention and control programs may lack theoretical basis, and their effectiveness and reliability cannot be guaranteed. To address these problems, this study proposes a non-contact monitoring technique of landslide based on terrestrial laser scanning (TLS), ground-based synthetic aperture radar (GB-SAR) and unmanned aerial vehicle (UAV). This method firstly uses non-contact control points and ICP algorithm to fuse the point cloud data of both TLS and UAV, which can obtain a large range of high-precision terrain data even in a limited active space. Then, this technique combines the high-precision 2D differential interference results of GB-SAR with the fused high-resolution 3D model of TLS and UAV to realize 3D visualization of slope micro-deformation for real-time monitoring and early warning. Besides, we propose a mean displacement vector of domain (MDVD) algorithm to resolve the spatial posture of point cloud data before and after deformation based on vector operations, and then realize the analysis of translational and rotational displacement of landslide in 3D space, which provides an important basis for revealing the mechanism of landslide and evaluating its potential danger. Compared with the traditional landslide deformation monitoring methods, the proposed method, based on multi-source data fusion, has made a breakthrough in terms of monitoring data acquisition methods, quality control and data analysis, which not only overcomes the limitations of terrain and traffic conditions on site monitoring, but also provides a more accurate response to the three-dimensional evolution process of landslides. We have successfully applied this method for the landslide deformation monitoring and emergency disposal of the Houziyan landslide, the Baige landslide and the Aniangzhai landslide. As the results, we have achieved remarkable results in monitoring the dynamic evolution of landslides, revealing the mechanism of landslide, and ensuring the safety of emergency treatment.