| Minjiang River upper reaches is located Longmen mountain area,where the topography is very complex and within the most significant gradient belt,landscape regional active faults are well developed,the earthquake distribution and tectonic activity is closely related to earthquake induced landslides and other geological disasters.Among them,the Maoxian Wenchuan section(Mao Wen section)is characterized by the significant development of large and giant landslides,especially famous for its giant landslide with its wide distribution,large scale disaster and serious damage and so on the world has seriously affected China Southwest of human life and property and major construction projects.On the basis of data analysis on the surface of geological hazards,taking the MaoxianWenchuan section of the Minjiang River upper reaches as the research area,landslide remote sensing interpretation and field geological hazard investigation were carried out.On the ENVI5.1 and e Cognition software platform,using Landsat 8 remote sensing satellite data,based on the object oriented classification method,the landslide information is analyzed by remote sensing analysis,the landslide remote sensing database is established,and the vulnerability of landslides is evaluated and analyzed in the scope of the study area,obtained the following conclusions:(1)The key features and target areas of landslides can be effectively obtained by the object oriented classification method,and then combined with visual interpretation,the detailed information of landslides can be obtained,and the success rate of the landslide interpretation can be improved.97 sites of landslide and collapse are interpreted in the Maowen section of the Minjiang River,and the results of the interpretation and the existing geological hazards are analyzed.The result shows that the correct rate of extraction area extraction is about 73.2%,and the number of the number of landslides in the extraction range is about 64.9%,and the result shows that the effect of the interpretation is better.(2)In the image segmentation,the multiscale segmentation algorithm is adopted,and the optimal segmentation parameters are obtained through the comparison of the different segmentation scales and weights.Finally,the segmentation scale is 50,the tightenness factor is 0.5,and the shape factor is the base of the 0.4 segmentation.(3)The recognition of the ancient landslides and the new landslides is discussed with the gray level co-occurrence matrix(GLCM)and the normalized vegetation index(NDVI).The GVI model is proposed based on the gray level co-occurrence matrix and the vegetation index,and the quality function IGVI of the model is constructed.The results of the statistical samples show that the IGVI value of the ancient landslides is obviously lower than the new landslides,indicating that the IGVI value of the ancient landslides is obviously lower than that of the new landslides.The GVI model proposed in this paper can provide a basis for identifying ancient landslides and new landslides.(4)On the basis of the results of remote sensing interpretation and the data of existing geological disaster data,combined with visual interpretation,the landslide database in the study area has been established.There are 64 landslides,of which there are 20 huge landslides,40 large landslides and 4 medium-sized landslides,and the height,ground shape slope,terrain slope and distance from the river are selected in the study area.Stratigraphic lithology,fracture,vegetation and other influencing factors make landslide susceptibility assessment.According to the evaluation results,the susceptibility of landslides is divided into 4 levels,namely,low prone area,middle prone area,high prone area and high prone area,which account for 31.56%,24.97%,24.71% and 18.76% of the area of the study area,which are extremely high prone areas.About 73% landslides have been developed,indicating that the results are in good agreement with the distribution of the known landslides,and the easily prone areas have high accuracy. |
PDF Full Text Request