| Debris flow disaster is a major natural disaster that threatens the safety and living environment of human life and property.For the mountain debris flow,the outbreak of the debris flow not only caused economic property loss,but the expansion of the debris flow area also had a serious negative impact on the local land use and ecological environment after the occurrence of the debris flow.The investigation and study of the debris flow region can not only understand the disaster situation,but also be of great significance for improving the local ecological environment.Remote sensing technology has great advantages in geological disaster detection and debris flow information extraction because of its safety,efficiency and speed.At the same time,GIS's powerful spatial analysis capabilities make environmental vulnerability spatial analysis technology possible.Dongchuan County of Yunnan Province is a typical frequently-occurring region of debris flow in China and even in the world.This paper was based on multi-temporal remote sensing image data,DEM data,and geographic element data,took the typical debris flow branch ditch of Dongchuan County-Dabaini ditch as the study area,applied multi-scale research ideas,then the debris flow ditch was analyzed at three different research levels: large scale,medium scale and small scale.At the large scale-the basin level,taking the small watershed as the research object,the remote sensing information interpretation of the Dabaini River Basin was carried out by using different classification methods combined with different classification features,and the land cover of the small watershed was monitored to select the information extraction method which is most suitable for the small watershed in the plateau mountainous area;At the medium scale-the debris flow area level,taking the debris flow area as the research object,the mountain environment vulnerability assessment was carried out according to the evaluation model and the evaluation factors,and the basic range of the severe vulnerable area,that was,the debris flow area was obtained.Then the land use classification of remote sensing interpretation was used as the limiting factor to divide the exact range of debris flow,finally,the debris flow area was divided into sections according to the geological characteristics and spectral characteristics of the debris flow sections;At the small scale-thesections level,taking the sections as the research object,the trend of each section of debris flow was discussed from the two aspects of time characteristic and spatial characteristic,and its influence factors were explored.Finally,the following research results were obtained:(1)The detection methods of remote sensing classification technology application to the spatial information extraction of land cover in a typical debris flow was studied:The Landsat image was used as the data source to monitor the land cover of the small watershed in the mountainous area.The support vector machine classification based on Band Ratios,the support vector machine classification based on PCA,the object-oriented classification based on Band Ratios,and the object-oriented classification based on PCA were used for classification and accuracy evaluation.Compared with the two methods of support vector machine and object-oriented,the object-oriented classification method had low classification accuracy,exited salt and pepper effect,and even “holes” appeared.which indicated that the object-oriented classification method was not applicable to Landsat images in small watersheds under complex terrain and broken ground situations.Under the support vector machine classification,the overall classification accuracy of support vector machine based on PCA was higher,reaching 75.36%,and the Kappa coefficient was 0.51,which indicated that the support vector machine classification based on PCA method had a better overall classification effect and the consistency of the results was better.Therefore,the most suitable method for extracting ground information from small watersheds in plateau mountainous areas was support vector machine classification based on PCA method.(2)A debris flow detection method based on basin environmental vulnerability assessment analysis and comprehensive land cover modification corrected by remote sensing interpretation results was carried out: According to the environmental vulnerability of regional natural disasters,5 factors including slope,gully density,NDVI,soil brightness index and population density were used,and the layer overlay model was used for evaluation and spatial difference analysis.The weights were determined according to the analytic hierarchy process(AHP),and the scope of severe vulnerability was determined by spatial overlay analysis.Then,the range assessment of severe vulnerability was corrected by the results of remote sensing interpretation,so as to determine the range of debris flow area.Finally,the sections was dividedaccording to the geological characteristics and spectral characteristics.(3)The analysis of the division of the debris flow area and the dynamic change characteristics of the sections in the typical valley-type watershed were carried out:From the time characteristics of the debris flow sections change,all three sections showed a growth pattern.The most obvious segment of expansion was the source section,which expansion area was 37.56 ha,and the smallest expansion occurred in the accumulation section,which was 14.09 ha.The expansion rate of varied depending on the influence factors;From the spatial characteristics of the debris flow sections change,the source section had the largest expansion in the east direction,and the centroid moved 19.61 meters to the southwest;the circulation section had the largest expansion in the north direction,and the centroid moved 19.71 meters to the northwest;the accumulation section had the largest expansion in the northwest direction,and the centroid moved 31.39 meters to the northwest.The change of time and spatial of debris flow area was affected by many factors,but the main factors were three: gully bed ratio,precipitation,and source.At the same time,the expansion of debris flow areas was also affected by soil,vegetation,engineering construction facilities,mining and so on.The application analysis research of typical debris flow area detection by combining remote sensing classification technology with GIS spatial analysis technology has practical guidance and scientific significance for deepening the dynamic changes and driving factors of mountain plate debris flow area.But how to use richer multi-source data,apply more diverse research methods,introduce more scientific factors,and build a more reasonable model method is the next step we need to deepen. |
PDF Full Text Request