BEIJING — A group of researchers has proposed a new physics-based, cost-effective landslide displacement prediction framework, according to a research paper recently published in the journal Engineering Geology.
Prediction of landslide deformation is an important part of landslide early warning systems. Although displacement prediction based on geotechnical in situ monitoring works well, its high cost and spatial limitations prevent its frequent use within large areas.
Researchers from China University of Geosciences, Peking University, Hannover Leibniz University, and GFZ German Geoscience Research Center combine multitemporal interferometric synthetic aperture radar (MT-InSAR) and machine learning techniques to extract landslide displacement time series. did. Satellite imagery provides low-cost foundational data for early warning and prediction.
Applying this prediction method to the Three Gorges Reservoir area in China, we show that MT-InSAR can accurately monitor landslide deformation and that machine learning algorithms can accurately establish the nonlinear relationship between landslide deformation and its triggers. .
By integrating the advantages of MT-InSAR and machine learning techniques, the proposed prediction framework that considers the physical principles behind landslide deformation can cost-effectively predict landslide displacements within large areas, the research paper says. states.
