| In recent decades,more and more scholars have begun to pay attention to the hot issue of global warming,which is bound to have an impact on the ecological environment.As an important part of the cryosphere,glaciers are important indicators of climate change.The rising temperatures and the intensifying human activities will cause a large number of glaciers to retreat.Mountain glacier densely develop in arid land of Northwestern China,and the change of the glaciers is bound to have an impact on the change of the reserves of glacial meltwater and fresh water resources in China,which will greatly affect the sustainable development of the region.With the development and application of remote sensing technology,more and more effective methods for monitoring glacier change information have emerged.We selected Tibet and xinjiang as research areas,using Landsat remote sensing images,Sentinel-2 remote sensing images,digital elevation model,the Second Glacier Inventory Dataset of China,and Google images,combining the technology of ENVI5.3,IDL,and Arc GIS10.1,finally we implemented the accurate recognition of glacier boundary under the influence of the shadow,proglacial lakes and debris cover.The main research results of this paper are as follows:(1)The Automated Glacier Extraction Index(AGEI)was constructed to improve the mapping accuracy of debris-free glaciers under the influence of shadow and proglacial lakes.The AGEI makes full use of the spectral advantages of Red,NIR and SWIR bands,automatically reduces or eliminates the interference of shadow and proglacial lakes on glacier recognition by ratio weighting,and determines the optimal threshold by visual evaluation glacier in shadow areas,finally realizing accurate identification of debris-free glaciers boundary.This paper using AGEI index,ML supervised classification,NDSI,Red/SWIR,and NIR/SWIR to carry out glacier mapping in the four study areas,the results show that the AGEI can quickly and accurately identify the debris-free glaciers,and has the highest accuracy.(2)A comprehensive accuracy assessment for the new AGEI and the existing four glacier classification methods were carried out using three different scale validation methods: Perpixel,sub-pixel,and plot-scale accuracy assessment.The per-pixel validation showed that,compared with the other four methods,AGEI has the highest OAs and kappa coefficients,and the lowest glacier total errors and non-glacier total errors.The sub-pixel validation showed that AGEI could identify the mixed pixel of glacier boundary most accurately,with an accuracy rate of 74.5%.The plot scale validation showed that AGEI had the highest correlation with the true value,and the correlation coefficient r2 was up to 0.878,which was the closest to the actual distribution of the glacier.In addition,robustness tests showed that the sentinel-2 glacier classification results were 97% similar to Landsat classification results,demonstrating AGEI's extensibility.(3)Based on AGEI Index,this paper further constructs the Automated Debris-covered Glacier Extraction Index(ADEI),which can be used to realize the automatic recognition of the debris-covered glaciers.The ADEI combined the advantages of optical image and thermal infrared data,by the ratio of brightness temperature to AGEI,amplify the differences of debris-covered glacier and bare land region.The density slicing method was adopted to determine the optimal threshold value range.Combined with the slope information generated by the digital elevation model,the separation of the supraglacial debris and periglacial debris was realized.The experimental results showed that the glacier area identified by this method is 2.7% different from the Second Glacier Inventory Dataset of China,and is 2.1% different from the glacier area digitalized based on Google image.To sum up,the ADEI method combined with slope information achieves the debris-covered glacier mapping with less input data,easier operation and higher efficiency. |
PDF Full Text Request