Information Extraction And Dynamic Monitoring Of Glacier Debris Flow Based On RS Technology

Sichuan-Tibet Highway is the backbone of the transportation network in the southeast region of Tibet and is of great significance for implementing the "One Belt and One Road" strategy and the development of the western region.Affected by warm and humid air currents and monsoon climate along the highway,glaciers and ice lakes spread widely.Geological conditions along the line are complex,the new tectonic activity is strong,the rock layers are broken,the topography is undulating,the joints are developed,the geological disasters such as collapse,stream sand slope,and landslides are developed,and debris flow frequently occurs.Among them,the glacial mud-rock flow,as a special disaster type in the alpine region,is characterized by abrupt and destructive power,which seriously threatens the safety of life and property of the residents along the road and along the route,and restricts the local socio-economic development.However,due to local terrain and climatic conditions,traditional large-scale field trips and large-scale regional surveys have made it difficult to achieve.Due to the abundant data,easy access and low prices,remote sensing is an effective means for the extraction of large-scale information on glaciers and mudslides and the dynamic monitoring of the region.This article takes the Sichuan-Tibet Highway Tongmai-Songzong Section as the research area.On the basis of data collection and field surveys,a glaciers identification and extraction model for the study area was established based on the characteristics of the glaciers in the study area,and their variation characteristics were analyzed.Based on the characteristics analysis,the technology for identifying and extracting glacial debris flows was established,and the dynamic monitoring and research of glacial debris flows was conducted in depth.The main results of this paper are as follows:1)According to the distribution characteristics and image characteristics of glacier in the study area,OOBR method based on feature partitioning is established.Introduce features such as slope,elevation,and texture to extract glacier covered and shadow covered glacier.The glacier cover,glacier covered glacier,and shaded area glacier extraction accuracy were 93%,100%,and 88%,respectively,and the glacier information in the study area was initially complete and effectively extracted.2)The overall glaciers in the study area continue to fade.The average annual retreat rate is greater between 1990 and 2015.The retreat rates of the glaciers reached the maximum at both phases in May and November of 2008-2015,reaching 1.96%and 1.48%respectively.3)Using correlation analysis and random forest algorithm to optimize the key features of the glacial debris flow image,making the accuracy of single-phase image extraction reach more than 80%.Therefore,features are preferred to effectively reduce feature redundancy and improve image classification efficiency and accuracy.4)Fully tapped the favorable information on the phase and texture of the glacial debris flows.Considering both scope and type,the identification and extraction technology of glacial debris flows was proposed.The technology initially solves the problem of range misinterpretation due to the seasonal variation of shadows and ice and snow area,and the misjudgment caused by other types of debris flow information interference.To a certain extent,this technology achieves the completeness and accuracy of information extraction of glacial debris flows.According to the field survey data,the accuracy of extraction of glacial debris flows is 92.59%.5)From 1991 to 2015,the area of the glacial debris flow and its provenance,ice and sediment,and landslides showed an increasing trend,indicating that the hazard of glacial debris flows has gradually increased.The area of glacial debris flows and landslides has been increasing year by year in three consecutive years since 2014-2016.The glacial mudslides and landslides mainly consisted of newly added disaster areas in 2014-2015,and the major disaster areas were mainly expanded from 2015 to 2016.