Research Of Antarctic Ross Ice Shelf Thickness Estimation Based On The CryoSat-2 Radar Altimeter

The ice shelves and glaciers in Antarctica store about 70 percent of the earth's freshwater resources,so it is of great importance to monitor the Antarctic ice shelves.The satellite radar altimeter has become a reliable method to research the ice shelf thickness with the advantages of all-weather and all-day working ability and large observation scale.Based on the synthetic aperture interferometry,the radar altimeter on the CryoSat-2 satellite has better spatial resolution and measurement precision than the traditional altimeter.At present,the ice shelf thickness estimation based on the satellite radar altimeter mainly has the following problems.Firstly,it needs to use other data to classify the surface type,which may be different in the time of data acquisition and will introduce some unnecessary errors.Secondly,the common waveform retracking algorithms have low precision and poor applicability,so a new retracking algorithm is urgently needed to be proposed.Thirdly,the interference information on the CryoSat-2 satellite radar altimeter has important influence on the ice shelf thickness estimation.In view of the problems above,the research content of the thesis mainly includes:(1)In terms of waveform classification,waveform characteristic parameters having well separation on different surface(such as pulse peakiness,standard deviation,early tail to peak power ratio and amplitude)were combined with seven kinds of classifier(such as Threshold Classifier,K-Nearest Neighbor Classifier,Support Vector Machine Classifier,Random Forest Classifier,Naive Bayes Classifier,Ensemble Learning Classifier and Discriminant Analysis Classifier),and a total of 88 kinds combinations' classification effect were compared.A waveform classification algorithm based on the best combination of waveform characteristic parameters and classifier.The results proved that the combination of pulse peakiness,standard deviation,early tail to peak power ratio and Threshold Classifier had the best effect in distinguishing the lead,sea ice,open sea water and ice shelf,and the accuracy was up to 81.4%,which was about 8%higher than the combination of Support Vector Machine Classifier and the waveform characteristic parameters.(2)In terms of waveform retracking,an adaptive retracking algorithm b ased on the choice of dominant-wave(CDWAR)was proposed.It contained echo fitting,judging single-peak or multi-peak waveform and choosing the dominant-wave of multi-peak waveform,and doing adaptive retracking according to the type of waveform.And the results proved that the CDWAR method could accurately retrack the waveform,and the accuracy of ice shelf thickness estimation was better than the threshold method and OCOG method.(3)In terms of the interference phase information,an method of ice shelf freeboard correction was proposed,in which the waveform retracking point was set as the starting point for the one-dimensional phase unwrapping based on path integral,and the elevation difference between the geographical position corresponding to each sampling gate and the nadir point was obtained.The results showed that the elevation correction of sea surface was higher than the sea ice surface,and the freeboard correction of edge of the Ross Ice Shelf was higher than the flat surface of it.(4)The freeboard of Ross Ice Shelf was obtained based on above methods.Considering the influence of land on the ice shelf,a new ice thickness estimation model combining the hydrostatic balance principle with fracture mechanics principle was built,and the thickness of Antarctic Ross Ice Shelf was estimated based on it.Five ice shelf thickness estimation schemes were compared,and the results showed that the scheme containing the CDWAR method,the freeboard interference correction and the new estimation model was more accurate than other schemes.