Research On The Sea Ice Classification And Thickness Detection With High-resolution And Polarimetric SAR Data

The observation ability of synthetic aperture radar(SAR)has been greatly improved in recent years.The image resolution has been improved dramatically(from 100 m to 1 m),and the polarimetric information has become more abundant(from single polarization to full polarizations).The improvement of SAR observation ability brings new opportunities as well as challenges for sea ice classification and thickness detection.Firstly,for sea ice classification,the secondary classification method,based on the context of the sea-ice types,is commonly used.However,this method is proposed for medium-and low-resolution SAR,which cannot be applied to high-resolution SAR images.Therefore,new sea-ice classification method for high-resolution SAR images should be developed.Secondly,polarimetric SAR data enables the accurate detection of the sea-ice thickness.It has theoretical and practical value in taking advantage of sea-ice polarimetric information,to construct the detection methods of the sea-ice thickness,and increase the accuracy of the detection methods.Therefore,the thesis focuses on developing methods of sea ice classification and thickness detection for high-resolution and polarimetric SAR,which can provide new methods for sea ice monitoring.The work consists of four parts.(1)Research on sea ice classification and thickness detection is based on the sea-ice microwave scattering characteristics,which have never been studied in the Bohai Sea.The multi-band multi-polarization microwave scattering experiment was carried out for the first time in the Bohai Sea.With the experimental data,sea ice microwave scattering characteristics can be studied.The radar bands and incident angles applicable to sea ice classification and thickness detection were proposed,which can provide the criteria to select the scientific and reasonable data sources for sea ice classification and thickness detection in the Bohai Sea.(2)The secondary classification rules for high-resolution SAR data and the improved secondary classification method of sea ice were proposed.Firstly,the secondary classification rules for high-resolution SAR data were established based on the new context of the sea-ice types recognized from the high-resolution SAR images.The sea-ice types,which have been assigned correctly,may be changed to the wrong types with the classical secondary classification method.The classification confidence function was introduced to avoid the problem of the misjudgement.The secondary classification method for high resolution SAR images was proposed according to above works.It is proved that the proposed method can improve the classification accuracy of sea ice in the Bohai Sea to some extent.Compared to the classical method,the proposed method also shows a superior anti-noise(the speckle noise)capability in the simulation experiment.(3)The sea-ice thickness detection method based on the polarimetric alpha angle was proposed.Analysis of the response of the polarimetric characteristics to the first-year level sea-ice thickness showed that the polarimetric alpha angle is highly correlated to the sea-ice thickness.Therefore,the sea-ice thickness detection model based on the polarimetric alpha angle was developed.Verified by the sea-ice polarimetric SAR data and synchronous field sea-ice thickness data in the Arctic,the new model could provide more accurate results.(4)Considering the SAR data with both high resolution and polarization,the integration retrieval method of the sea-ice parameters combining the sea-ice classification and thickness detection results was proposed.The sea-ice classification results were acquired by the newly developed secondary classification method for high-resolution SAR images,and the sea-ice thickness results were acquired by the sea-ice thickness detection method based on the polarimetric alpha angle.The sea ice classification and thickness detection results corrected each other based on the constraint conditions.The integration retrieval method is proved to be superior to the independent retrieval method of the sea-ice parameters using the sea-ice high-resolution and polarimetric SAR data and the synchronous field sea-ice thickness data in the Arctic.