Analysis Method of Landslide Early Warning and Prediction Supported by Ground-Based SAR Technology
摘要
Objectives Landslide early warning and prediction is the key and difficult problem in landslide dynamic disaster research. The timeliness and accuracy of early warning are two key indicators to evaluate the effect of landslide early warning and prediction. Methods Remote sensing monitoring is an important technology for landslide disaster prevention and control. Ground-based SAR(synthetic aperture radar)provides an advanced technology for remote sensing monitoring of landslides in short distance, and provides a wealth of data sources for accurate positioning of landslides, delineating the scale of landslides, analysing and judging the risk of landslides and predicting the time of landslides. Based on the classic three-stage theory of slope deformation evolution, we adopt the monitoring data of ground-based SAR, which has the characteristics of near real-time, coverage monitoring area and surface deformation information. Combined with the characteristics of the short period, high density and surface coverage of slope monitoring data, a dual index landslide early warning method based on deformation speed and deformation area is proposed. According to the negative correlation between deformation velocity time series curve and slope stability, we put forward the processing methods of velocity inversion, de-limit and unequal period smooth processing, and establish the landslide prediction model with fast convergence of stability. Results The landslide early warning method with double indexes of deformation velocity and deformation area can effectively improve the accuracy rate of landslide warning, and then the accuracy of landslide prediction can be effectively improved by taking the reciprocal of velocity, removing the limit value and unequal period smoothing treatment. Different periods of data processing methods have their own applicability, short processing cycle processing data curve, the prediction of landslide occurrence time is more accurate, the prediction o