Research On The Extraction And Classification Of Arctic Sea Ice Based On Sentinel-1 Data

As arctic regions are affected by global warming,the melting rate of sea ice is accelerating gradually and sea ice cover is decreasing.It has an undeniable impact on the global climate,leading to a series of ecological problems.At the same time,the melting of arctic sea ice will become a prerequisite for its exploitation and utilization,which is conducive to the opening of the channel and the utilization of resources.Doing research on sea ice conditions in key areas of the arctic is important to master the information of relevant elements of Marine environment,accumulate reliable data,analyze arctic sea ice climate effect,maintain the safety of arctic waterway,exploit arctic resources reasonably,etc.Remote sensing technology is the most effective way to monitor sea ice.Compared with traditional methods,it has obvious advantages in large scale,real time and sustainable features.The polar sea ice concentration products are based on passive microwave radiometer data from different satellite sensors.The spatial resolution is mainly distributed in 4~25km.With the development of Synthetic Aperture Radar technology with higher spatial resolution,its application in sea ice monitoring is more and more extensive.Based on the SAR data of Sentinel-1 satellite,this paper studies the extraction of sea ice distribution and sea ice morphological feature parameters,analyzes their variation law,and studied the classification of sea ice based on grayscale and texture information.The main research results and innovation points are as follows:1.Using the long time serial images from Sentinel-1 satellite,the extraction and change analysis of sea ice in the Fram channel region is studied in this paper.After the image pretreatment,the sea ice was extracted by K-Means clustering algorithm.On this basis,the variations of sea ice distribution and sea ice morphology with time are analyzed.The distribution of sea ice is described in sea ice concentration.The morphological features of sea ice in floating ice area are characterized by area,perimeter and roundness.According to the classification of sea ice area,the variation of each parameter with time is analyzed.In addition,the relationship between the parameters and the physical process of ice breaking is studied.The results show that,based on high resolution SAR data,fine sea ice distribution can be obtained.Combining it with traditional sea ice concentration products,the multi-scale ice map with more abundant information can be constructed.It can provide support for polar sea ice monitoring.2.The texture and classification of sea ice are studied.And the research area is located in the McClure channel.The appropriate parameters are selected by experiments to calculate the Gray-level Co-occurrence Matrix.Based on the Gray-level Co-occurrence Matrix,the texture features of the sea ice target are obtained and the separability of them is evaluated.Combining with the spectral characteristics and texture features of sea ice targets,the classification and identification of sea ice in the study area is carried out by using the supervised classification algorithm based on Support Vector Machine.The results show that the classification effect of gray scale and texture feature is better than that of gray scale.For many texture feature parameters,the feature dimension can be reduced based on certain criteria.