Development Of A Platform For Lanslide Monitoring Data Processsing And Its Application

Landslide disasters have occurred frequently in our country, which huge economiclosses every year. The rapid development of remote sensing technology and ground sensor technology provide a good technical support for landslide monitoring and prediction. The monitoring mode of typical landslide disaster-prone areas develops from one single sensor mode to intelligent multi-sensor network which is integration of remote sensing earth observation system with multiple ground sensors. The monitoring data have diverse resources and have different temporal and spatial resolution. It is very important for landslide monitoring and forecasting to transform the different time and spatial scales and build a unified temp-spatial reference. The paper analysis the current temp-spatial transformation methods of remote sensing data and the key landslide factor data comparatively, build a platform of processing the monitoring data and carry out the typical application, which is significant for upgrading monitoring and forecasting of landslide hazards and improving the ability of disaster prevention and mitigation. Research methods used and conclusions formed in this paper are mainly as follows:(1)By building systems design principles and overall design on the basis of analysis of the existing temp-spatial conversion method and landslide forecasting model, the platform of landslide monitoring data processing and forecasting is realized in the Microsoft Visual Studio development environment and using Arcgis Engine, ENVI/IDL and C# language. The platform has four main functions: the key landslide factor data processing, optical remote sensing data processing, temp-spatial reference transformation and the landslide forecasting.(2)Taking part of WenChuan disaster area as the study, three kinds of DEM: SRTM3 DEM, ASTER GDEM and GTOPO30 are used to carry out hazard zoning of information model for experimental areas. The result shows that SRTM3 DEM with a unified spatial scale is more suitable for modeling the information value in the region. By resampling the SRTM3 DEM, the data in the scale of 10 m, 20 m, 30 m, 40 m, 50 m, 60 m, 70 m, 80 m and 90 m are obtained. We find that the SRTM3 DEM resampled in the resolution of 60 m is best suited to predict the landslide hazard based on the information model by the evaluation parameters a proposed in the paper.(3)Based on the landslide forecasting module of landslide monitoring data processing and forecasting platform, MaoPing as the study place, is predicted the displacement for ten months by using the exponential smoothing model, GM(1,1)model and binomial regression model. There three main results are acquired. For the exponential smoothing model, the prediction accuracy is highest when the index is three and the smoothing coefficient is 0.2. For the binomial regression model, dimension gray recurrence method can improve the prediction accuracy of the model. For GM(1,1)model, according to the characteristic of the data, the proposed method of the combination of reduced sample and dimension gray recurrence can greatly improve the prediction accuracy of the model. Finally, the model was improved on the basis of the previous by using the constant variable weight combination and right combination method of combination forecasting model method. The results show that the prediction accuracy: Variable weight combination forecasting accuracy> constant weight combination forecasting precision> single model prediction accuracy. In seeking the right variety of methods, the most accurate method is inverse variance of constant weight combination, and the error is only 9.458.