Research On Polar Sea Ice Detection Method Using Multi-source Satellite Scatteromete

Polar sea ice has been of interest to climate researchers as an important input source to global climate models and a sensitive indicator of climate change.The satellite scatterometer is one of the main sensors for remote sensing of polar sea ice,and existing studies are mainly based on a single satellite-based microwave（C or Ku-band）scatterometer data,while research on remote sensing of sea ice parameters based on multisource scatterometer measurements remains to be explored.In this thesis,the polar sea ice detection and Arctic sea ice type identification studies have been carried out by jointly using the data from domestic and foreign microwave scatterometers operating in orbit at the same period,including the European Met Op series satellite C-band ASCAT scatterometer,China’s HY-2B satellite Ku-band scatterometer（HSCAT）and Chinese–French Oceanography Satellite（CFOSAT）Ku-band scatterometer（CSCAT）,to explore the potential of joint use of multi-source satellite-borne microwave scatterometers in The feasibility of joint use of multi-source microwave scatterometer in polar sea ice detection studies was explored.In this paper,seven characteristic parameters are selected for sea ice extent detection based on multisource satellite scatterometer data,including the horizontal and vertical polarization backscatter coefficients (σ_HH⁰ and σ_VV⁰),the horizontal and vertical polarization backscatter standard deviations(σ_HH⁰ and σ_VV⁰),the polarization ratio(σ_HH⁰ / σ_VV⁰),the dependence of the vertical polarization backscatter coefficient on the incident angle(K_VV),and the ratio of the backscatter coefficient between the vertical polarizations in the C and Ku bands(named as the band ratio(BR=σ_VV-Ku⁰/σ_VV-C⁰)).The support vector machine（SVM）method is used to classify sea ice and seawater based on characteristic parameters,and the image erosion/expansion technique and area growth method are used to correct the SVM method classification results to reduce misclassification information.The results are verified by comparing with the sea ice concentration data from the National Snow and Ice Data Center（NSIDC）and Sentinel-1 Synthetic Aperture Radar（SAR）images,and the results show that the sea ice area obtained by the method in this thesis is closer to the area of sea ice calculated using15%as the threshold of NSIDC sea ice concentration in the Arctic and Antarctic,and the detected boundaries are in better agreement with the high-resolution SAR sea ice boundaries.In addition,the analysis in this paper shows that the coverage of multisource microwave scatterometer data with half day time resolution can reach over 97%in the Arctic and Antarctic,and the sea ice area obtained based on half day and daily multisource scatterometer data using the same SVM classification method is very close.In terms of Arctic sea ice type identification,the characteristic parameters used include vertically polarized backscattering coefficients σ_VV⁰,band ratio BR and the brightness temperature Tb_37h provided by the microwave radiometer AMSR-2 The random forest algorithm was used for sea ice type classification,and the correlation coefficients between the classification results and the sea ice type data provided by NSIDC and other data sources were all above 0.78,and the multi-year ice boundary is more consistent with the multi-year ice boundary displayed by high-resolution sea ice type image products based on SAR data inversion.In this paper,a method for detecting the extent of polar sea ice and identifying sea ice types has been studied using simultaneous multisource satellite scatterometer data.A daily polar sea ice cover map for the period 2019 to 2021 and a daily sea ice type map for the Arctic winter period from November 2019 to April 2022 have been produced.The detection results can accurately display the changing characteristics of the Arctic and Antarctic sea ice areas and sea ice types during this period.The research in this paper enriches the existing methods of remote sensing sea ice parameters based on microwave scatterometer data,and can provide a certain reference for constructing long time series sea ice remote sensing products based on multisource satellite observation data.