Study On Sea Ice Thickness Retrieval In Arctic Summer Using RadarSat-2 SAR Imagery

With the changing of global climate,the Arctic environment deteriorate much faster,resulting in a rapidly melt in summer arctic sea ice and the reduction rate is faster than the theoretical predictions of various models.This remind us that Arctic may be completely icefree in summer and this day perhaps come earlier than accepted.In order to explain the role of sea ice in Arctic environment better,it is necessary for us to have a deeper understanding of the properties of sea ice.There are various properties of the sea ice,thickness is one of the most importance parameter of sea ice,because it is sensitive to the change of the environment and it is the macroscopic expression of sea ice quality.But at the same time,ice thickness is the most difficult parameters to get.How to obtain large scale sea ice thickness by remote sensing is a widely discussed issue.This is also the main purpose of this paper.At present,there are four satellite remote sensing methods to get sea ice thickness: optical remote sensing,passive microwave remote sensing,SAR remote sensing(active microwave)and satellite altimeter.This paper compared and analyzed this four methods above.We introduced the most frequently used data source and its form,the basic way of data preprocessing and the common theory of sea ice thickness retrieval of each method in detail.It is concluded that SAR is the most suiTab.method for polar sea ice thickness retrieval after the comparison between the limitation and advantage of these method.This paper presents a novel method to estimate ice thickness depends on SAR texture feature.7 Arctic RadarSat-2 SAR imageries and data of ice thickness collected from the 6th Chinese National Arctic Research Expedition in 2014 were used in the study.Texture features were calculated by gray level co-occurrence matrix(GLCM)because of the actual situation of sea ice.Firstly,suiTab.values of texture parameters should be confirmed.The results show that we can get the optimum when calculate the average texture value of four angles with window size is 9×9,displacement is 1 and quantized gray tones is 64.Then calculate texture features using the parameters above.The result of single texture retrieval tells us that homogeneity?energy?entropy and correlation can estimate ice thickness better,the average relative error of correlation is 13.9%.Compared with the most commonly used method that only depends on backscattering coefficient,which average relative error is 20.5%,the present method gets smaller error and verify the reliability of using texture features for ice thickness retrieve.Considering single texture feature may can't explain the change of sea ice completely,ice thickness retrieval depends on double texture features was tested.Of which,using energy and correlation to estimate sea ice thickness obtain optimized result,which average relative error is 11.88%.Furthermore,there was no obvious improvement when using three texture features or texture features and backscattering coefficient to estimate ice thickness.