基于多期SPOT-5影像的降雨型浅层滑坡遥感解译研究

Study on Interpretation of Rain-Induced Group Shallow Landslides Based on Multi-Period SPOT-5 Remote Sensing Images

2005年6月17—23日福建省建瓯市发生特大暴雨,诱发大量中小型浅层滑坡灾害。利用发灾前、后两期SPOT-5影像(包括全色和多光谱),采用几何精校正、影像配准、影像增强、数据融合等处理方法,结合野外调查,建立滑坡解译标志,利用两期影像对比分析,提取滑坡点信息,最终共解译出滑坡点2186个。选取300个解译的灾害点进行了野外验证,野外验证的准确率达92%。研究表明,SPOT-5融合影像在中小型滑坡信息提取中的应用是可行的。

During time from June 17-23 in 2005, an extraordinary storm event happened in Jian＇ou city, Fujian province; it induced lots of small and medium scale shallow landslides. This paper describes the treating process to get fusion SPOT-5 images using geometric correction, spatial calibration, image enhancement, data fusion and other processing methods based on panchromatic and multi-spectral SPOT-5 images before and after the event, taking the method of comparative analysis using the fusion images and field survey to gain the key interpretation to getting the total information of 2 186 landslides. The paper also gets the 92 percent of field affirming exactness using field validation method by choosing 300 interpreted landslides randomly from all of 2 186 ones at last. According to the research, using fusion SPOT-5 images in the application of abstraction of small and medium scale landslide information is available.