Research On Earthquake-induced Landslides Extraction From Multi-source Remote Sensing Data In The Wenchuan

Wenchuan earthquake caused a lot of landslide which caused huge loss to people’s life and property. It had drawn great attention of the government and society’s widespread attention. In view of the serious situation of earthquake landslide and its losses, we need studying the extraction of earthquake landslide information and acquiring the disasters rapidly for the positive response to the earthquake emergency investigation and assessment. It is not only the urgent request of the affected areas, but also is the urgent demand for providing emergency response and disaster prevention and mitigation strategies. It has important scientific and realistic significance.In recent years, the multi-temporal, multi-resolution, multi-type of multi-source remote sensing data and remote sensing image processing technology has used in all aspects of natural disaster prevention and disaster information extraction. It plays a irreplaceable role for the disaster prevention and mitigation and post-disaster reconstruction.With the rapid development of remote sensing technology, high resolution remote sensing satellites have made great progress. High resolution remote sensing image which has rich geometric information can provide a lot of details and open a new era of remote sensing application. Domestic high resolution remote sensing satellite has pushed the vigorous development of domestic satellite application.This article selects the county of Yanmen in Wenchuan as the research area, using domestic high resolution remote sensing image such as ZY-1 02C、ZY-3、GF-1 and topography data to study the information extraction of earthquake landslide with objectoriented information extraction technology. In image segmentation,selecting estimation of scale parameter tool to determine the optimal segmentation scale and using multiscale segmentation technology, which avoided the blindness of selecting segmentation scale with experience.After analysising the characteristics of the landslide and highresolution images comprehensively, selecting image spectrum features, texture features, geometric features and terrain characteristics as the basis of classification. Finally,it’s important to determine the threshold of various features and select suitable membership function to establish the extraction rule set. Extraction results show that the seismic landslides distribute along the river and on both sides of the fault. Meanwhile, they distribute intensively around the source of the earthquake. Comparing the extraction results with the artificial visual interpretation results, the extraction accuracy is above 72%.This study shows that it is feasible to extract the information of landslides which induced in sudden earthquake(surface damage is serious, large scale) from multi-sourse Remote Sensing data by using the object-oriented classification method.On the other hand, it also can provide important technical support for surveying geological disasters and the macroscopic environment, disaster monitoring and evaluation, and post-disaster emergency investigation. What’s more, it has great application potential in information extraction of seismic landslides.