Method For Freezing-thawing Strength Extraction Of Ice Sheet In Polar Regions Based On SAR Images

The situation of freezing and thawing in polar ice sheet affects the exchange between atmosphere and water vapor,energy and heat on the surface,plays an important role in global balance.The information recorded in the polar ice sheet has the characteristics of large amount of information,strong authenticity and good time continuity.Therefore,monitoring of freezing and thawing in polar ice sheet is of great significance to the response research of global change.With the development of remote sensing,the study of freezing and thawing in polar ice sheet using SAR data has achieved initial results.There are unique advantages of SAR in the study of freezing and thawing in polar ice sheet due to SAR's characteristics of all-day,all-weather,high spatial and temporal resolution.At present,SAR has been used to monitor the freezing and thawing regions of the ice sheet.But the research on the freezing and thawing strength of the ice sheet is rare.In this paper,the freezing and thawing regions and strength of polar ice sheet are studied based on Sentinel-IA IW data and Radarsat-2 Full-polarization SAR data according to the methods of multi-polarization combination operation and polarization decomposition clustering.The results are as follows:(1)In this paper,the freezing and thawing strength of the polar ice cover is studied.The methods of threshold segmentation,operation of polarization and decomposition clustering are used to divide the freezing-thawing regions and strength.Based on the temperature data,the applicability of each method and the cause of the error are discussed.(2)In this paper,extraction of freezing-thawing regions and strength information of the ice sheet were studied by using the Sentinel-1A IW SAR data in Greenland.First,it is found that the backscatter coefficients of HH and HV are different at different points of freezing-thawing strength by analyzing the backscatter coefficients of HH and HV polarization data of Sentinel-1A SAR data in Greenland.Using histograms for statistic,the relationship between backscatter and freezing-thawing change in different polarization was investigated.The results demonstrate that the overall trend of backscatter coefficient of different polarization data is similar,but the corresponding values of peaks and valleys are different.HV polarization has a larger gradient in pixel value.It provides a theoretical basis for multi-polarization combinatorial operation.Second,the threshold of freezing and thawing region was determined according to the histogram.The backscatter coefficient of HV polarization was selected to extract the freezing and thawing regions based on threshold method.Verifying by MODIS temperature data,it is found that the dividing line of freezing-thawing regions is consistent with the isothermal determined by MODIS.Therefore,the determination of the backscatter coefficient of HV polarization can be used to divide the polar freezing-thawing region.Third,the algorithm of HH/HV was used to explore the law between the algorithm and temperature changes and study the research on the ectraction method of freezing-thawing strength.It can be found that that HIV HV is more sensitive to temperature than HH or HV unipolar datam,and the higher the temperature,the greater the HH/HV,the brighter the image.The reason is that the small increase of water content makes the surface scattering rise rapidly,the water content decreases continuously,the surface scattering decreases gradually,and the bulk scattering dominates.According to MODIS temperature data,it is found that the relationship between the image value and temperature after polarization operation is consistent with the overall trend.So HH/HV can represent the freezing-thawing intensity of the region to a certain extent.(3)In this paper,extraction of freezing-thawing regions and strength information of the ice sheet were studied by using the Radarsat-2 fully polarized SAR data in the Antarctic.First,getting different parameters by the data of different methods of polarization decomposition.Analying the parameters obtained by different decomposition methods and using the scattering angle(a),anisotropy(A)and elevation(E)are divided into freezing-thawing regions.Second,the polarization decomposition parameters obtained by different methods are clustered in different combinations,including Wishart unsupervised classification method,SVM supervised classification method,maximum likelihood method,minimum distance method and Mahalanobis distance method.Third,the accuracy of several supervised classification methods in different regions is analyzed,and the classification results are optimized according to the accuracy of different temperatures,so as to obtain a more accurate classification map of the study region.