Research On Glacier Boundary Extraction Method Based On Machine Learning And Remote Sensing Images

The sensitivity of regional glaciers in alpine Asia to climate change is more sensitive in the marginal mountain areas than in the middle region.Especially since the global warming in the 20th century,the glaciers in alpine Asia have been melting at a faster rate,which leads to the overall and accelerated retreat of the glacier tongue.Thus,the glacier surface area(mainly end)most likely to be table till sediment cover,however,because the table moraine and spectral similarity between adjacent rock,mountains and clouds are projected shadows and seasonal snow,mountain area table moraine cover,makes the mountain glacier mapping is still challenging in Asia.Focusing on glacier classification of remote sensing images and related theories such as remote sensing image analysis and processing,this study adopts related technologies combining Ensemble Learning and Deep Learning and utilizes data sources such as Landsat,Sentinel-1,Sentinel-2,and DEM to select typical glacier regions on the QinghaiTibet Plateau.A method based on Machine Learning was proposed to automatically extract the surface debris covering and surface debris-free glaciers in the glacial area,to realize the information extraction of clean-ice area and surface debris-covered area,and improve the classification accuracy of glaciers at the same time.The main research contents are as follows:(1)This study proposed a Random Forest algorithm for debris-covered glaciers mapping that automatically minify limitations of traditional monitoring such as mountain and cloud shadows,cloud cover,seasonal snow cover,and debris.The method consists of rule-based image segmentation and Random Forest classifier model.Rule-based technology extracts discrete objects of interest from Landsat 8 images.The predictive indicators are NDSI,NDWI,NDVI and LST.Then,according to the trained model to predict glaciers,debris-covered glaciers,and non-glaciers.The method was tested in the Eastern Pamir and demonstrated that the potential of the Random Forest method has high robustness in all glaciers of the eastern Pamir.(2)This study aimed to develop an automatic algorithm(using a Random Forest classifier model)implemented in the subregions of high mountain areas in the Tibetan Plateau to map debris-covered glaciers based on multi-source datasets such as Sentinel-1 Synthetic Aperture Radar data,Sentinel-2 Multispectral Instrument data,Landsat 8 Thermal Infrared Sensor and Digital Elevation Models.The main strength of this study is that our method overcomes most of the above-mentioned challenges and the great accuracy of the Random Forest classifier model represents a comprehensive success in identifying debris-covered glaciers,illustrating that if this method can be executed efficiently,it will bring benefits for glacier inventory management.(3)This research revolved around debris-covered glacier mapping,a fusion of related technologies combining Random Forest(RF)and Convolutional Neural Network(CNN)models,and using Landsat 8 Operational Land Imager(OLI)/Thermal Infrared Sensor(TIRS)data and Advanced Spaceborne Thermal Emission and Reflection Radiometer Global Digital Elevation Model(ASTER GDEMs),selecting Eastern Pamir and Nyainqentanglha as typical glacier areas on the Tibetan Plateau,to construct a glacier classification method system.Comparing the classification results of different classifiers,optimize different classifier construction strategies,and obtain multiple single classifier outputs with certain differences.Through a decision-level fusion of the output results,a fine classification of debris-covered glacier areas can be obtained.Through the representation relationship between the specific debris coverage strength and the Machine Learning Model parameters,it is expressed that the debris coverage directly determines the performance of the Machine Learning Model and overcame the challenge of detection on active and inactive debris-covered glacier.Integrate various classification models to explore the best model method suitable for the fine classification of glaciers,and improve the system of glacier classification methods.Figure[36]Table[13]Reference[134]...