固定冰是南极海冰的重要类型和重要组成部分，对南极近岸区域的大气环流、海洋热交换、南极冰架稳定性、生态系统平衡以及人类在南极的科学考察活动都具有重要的影响。然而，由于南极恶劣的自然环境，固定冰的现场观测受到了很大的限制，遥感观测和数值模拟在固定冰的研究中起到了重要作用。本文结合这两种手段，对南极固定冰的时空变化特征及其与周围环境的相互影响进行了研究，主要研究内容如下：

（1）基于净梯度差算法，结合影像边缘检测和自动矢量化方法，发展了一套适应于全南极的固定冰快速识别和提取方法。该方法以约20天内获取的2-3景C波段SAR影像为输入数据，计算净梯度差影像，通过中介边缘检测的方法得到固定冰区的边缘，剔除噪声点，使用边缘矢量化方法得到固定冰的矢量边界。研究结果表明本文方法对SAR影像入射角不敏感，可以忽略入射角对固定冰边缘提取的影响。该方法提取的固定冰边界与基于光学影像目视解译结果非常吻合，两者提取的固定冰面积差异最大为3.6%，最小差异为0.5%。因此，本文提出的固定冰边缘提取方法可用于C波段SAR影像提取全南极、长时间序列的固定冰边缘。

（2）首次获取了全南极高分辨率、长时间序列的固定冰最大范围数据集，并对南极固定冰范围、分布、宽度等特征开展了详细的时空分析。基于本文发展的固定冰边缘提取方法，使用2006-2011年的ENVISAT ASAR和2016-2017年Sentinel-1A/B SAR数据，获取了南极固定冰最大范围分布情况。研究显示，固定冰平均面积约为4.9±0.32×10⁵ km²，约70%位于东南极沿岸，2006-2011南极固定冰年平均减少1.2×10⁴ km²，与南极海冰的变化完全不同。南极固定冰在空间分布上具有显著的区域性差异，印度洋和太平洋扇区固定冰面积分别为1.65±0.11×10⁵ km²和1.32±0.20 ×10⁵ km²，显著高于罗斯海的0.79±0.09×10⁵ km²和别林斯高晋海的0.71±0.10×10⁵ km²，威德尔海的固定冰范围最小，仅有0.48±0.10×10⁵ km²。固定冰区最大宽度为太平洋的56.4 km，其次为印度洋的49.2 km，威德尔海与别林斯高晋海的宽度均为25.6 km，比罗斯海的小3.5 km。固定冰的分布与海底地形具有很强的相关性，95%的固定冰位于1000 m以浅的大陆架区域。南极固定冰占总海冰面积的3.64±0.29%，尽管固定冰比例较低，但从海-气能量交换来看，固定冰应该要单独作为一个变量来研究。

（3）发现了冰山影响南极固定冰形成和发展的机制。由于接地小冰山群或大型冰山的存在，Thwaites冰架和Filchner 冰架前方、Mertz冰舌东侧和普里兹湾东部四个海域，固定冰宽度超过150 km，最大约为200 km。2008年东南极普里兹湾冰障湾海域固定冰与往年相比异常偏多，分析发现主要是由B15B与B15R两座冰山共同作用造成的。两座冰山接地后阻碍了海冰的运动，导致海冰的堆积，使得原本不存在固定冰的区域形成固定冰，而冰山漂离后固定冰区消失。这表明冰山的突然搁浅或者漂移会导致南极沿岸区域海冰在较短时间内发生剧烈变化。但冰山与海岸和其他冰山的距离多少会影响固定冰的发展呢？通过分析威德尔海费尔希纳-龙尼冰架前端的固定冰与A23A冰山的关系发现，A23A冰山对固定冰的锚定仅能在距离海岸约190公里的范围发生作用，在冰山和海岸的共同锚定作用下两者之间可以形成连续的固定冰区。但当超过A23A冰山与海岸距离超过194 km后，固定冰范围急剧减小至110 km。这表明冰山与海岸之间距离在一定范围内，共同锚定作用明显，但当距离超过一定范围后这种共同作用消失。本研究通过对固定冰与冰山的相互作用阐述了冰山对固定冰的影响机制，为进一步研究冰山与固定冰之间的相互作用提供了研究基础。

（4）首次使用热力学冰雪模型HIGHTSI对南极固定冰冰厚展开了区域性模拟。厚度是研究海冰物质平衡的关键参数，本文基于固定冰现场观测数据对海洋热通量开展敏感性试验。结果表明当海洋热通量为20 W/m²时，模拟的固定冰厚度与实测数据最为接近，平均偏差为-0.012 m。对2006-2011年和2016-2017年麦克罗伯逊地沿岸固定冰的模拟结果显示，该区域固定冰厚度通常于每年的10月份达到最大值，最大值约为1.96 m，结合本文提取的固定冰数据，得到研究期内固定冰平均最大体积约为59.3 km³。其厚度分布显示，距离海岸越近固定冰厚度越大，固定冰厚度与负积温的相关性可达98%。这为后续开展环南极固定冰冰厚模拟和物质平衡研究提供了基础。

综上所述，本研究提出了一种可以快速提取南极沿岸固定冰的方法，首次获取了环全南极的高分辨率、长时间序列沿岸固定冰数据集，详细分析了南极固定冰的分布及变化，研究了海冰、海底地形以及冰山等与固定冰之间的相互作用，为进一步的南极固定冰研究提供了基础。同时，本文首次利用海冰热力学模型HIGHTSI开展了南极固定冰区域尺度的模拟，为后续开展环南极固定冰冰厚和物质平衡研究提供了基础。

Fast ice is an important type and important component of Antarctic sea ice. It has an important impact on atmospheric circulation, ocean heat exchange, Antarctic ice shelf stability, ecosystem balance, and human scientific exploration activities in the Antarctic. However, the harsh natural environment of the Antarctic has greatly limited the on-site observation and research of fast ice. Therefore, the remote sensing and modelling play important roles in fast ice studies. This paper combines remote sensing and numerical model to study the temporal and spatial variation characteristics of Antarctic fast ice and its interaction with the surrounding environment. The main research contents are as follows:

(1) Based on the net gradient difference algorithm, combined with image edge detection and automatic vectorization method, the method of quickly identifying and extracting the maximum edge of Antarctic fast ice is developed. The method takes the 2-3 scene C-band SAR image acquired in about 20 days as the input data, calculates the net gradient difference image, and then obtains the edge of the fast ice by means of the intermediate edge detection method, and finally eliminates the noise point based on the prior knowledge. The edge vectorization method is used to obtain the vector boundary of the fast ice. The results show that the proposed method is insensitive to the incident angle of SAR images, and the influence of incident angle on the extraction of fast ice edges can be neglected. The fast ice boundary extracted by the method is very consistent with the visual interpretation based on visible image. The difference of fast ice area extracted by the two methods is 3.6%, and the minimum difference is only 0.5%. Therefore, the method proposed in this paper can be used to extract the fast ice of the entire Antarctic with long time series of C-band SAR images.

(2) For the first time, the high-resolution and long-term series of fast ice dataset of the entire Antarctic was obtained, and spatio-temporal analysis of fast ice distribution, extent, width, variations was carried out. Based on the fast ice extraction method of this paper, the maximum range distribution of Antarctic fast ice was obtained by using ENVISAT ASAR from 2006-2011 and Sentinel-1A/B SAR data from 2016-2017. During the study period, the average area of fast ice in the Antarctic was about 4.9±0.32×10⁵km², about 70% of which was located on the coast of the East Antarctic. The average annual Fast ice of the Antarctic was reduced by 1.2×10⁴ km² in 2006-2011, which is completely different from the change of Antarctic sea ice. The Antarctic fast ice has significant regional differences. The fast ice area of the Indian Ocean and Pacific sectors are 1.65±0.11×10⁵ km² and 1.32±0.20×10⁵ km², respectively, which is significantly higher than that of Rose Sea with 0.79±0.09×10⁵km² and of Bellingshausen sea wiht 0.75±0.10×10⁵ km². Fast ice area in Weddell Sea is the smallest, only with 0.48±0.10×10⁵ km². The maximum width of the fast ice is 56.4 km in the Pacific Ocean, followed by 49.2 km in the Indian Ocean. The width of the Weddell Sea and the Bellingshausen Sea is 25.6 km, which is 3.5 km smaller than that of the Ross Sea. The distribution of fast ice is strongly correlated with seabed topography, with 95% of fast ice located in a 1000 m shallow continental shelf area. Antarctic fast ice accounts for 3.64±0.29% of the total sea ice area. Although the fast ice ratio is low, from the perspective of sea-air energy exchange, fast ice should be studied as a variable alone.

(3) The mechanism of icebergs influencing the formation and development of Antarctic fast ice is revealed. In front of the Thwaites Ice Shelf and the Filchner Ice Shelf, the east side of the Mertz Ice Tongue and the eastern part of the Prydz Bay, the fast ice width is over 150 km and the maximum is about 200 km. The analysis found that there are grounded small icebergs or large icebergs in these four areas. In 2008, the fast ice in the Barrier bay of Prydz Bay in the East Antarctic was abnormally more than that in other years. The analysis found that it was mainly caused by the interaction of B15B and B15R. After the two icebergs were grounded, the movement of the sea ice was hindered, resulting in the accumulation of sea ice, which made the Fast ice not formed in the area where there was no fast ice, and the fast ice area disappeared after the iceberg drifted away. This indicates that the sudden grounded or drift of the iceberg will cause the sea ice in the Antarctic region to change drastically in a short period of time. But how much does the distance between the iceberg and the coast and other icebergs affect the development of fast ice? By analyzing the relationship between the fast ice at the front of the Filchner-Ronne Ice Shelf in Weddell sea and the iceberg A23A, it was found that the A23A anchored to fast ice only in the range of about 190 km from the coast. Under the common anchoring effect, a continuous fast ice zone can be formed between the A23A and coast. However, when the distance between the A23A iceberg and the coast exceeds 194 km, the fast ice range is drastically reduced to 110 km. This indicates that the distance between the iceberg and the coast is within a certain range, and the joint anchoring effect is obvious, but when the distance exceeds a certain range, the joint effect disappears. This study illustrates the influence mechanism of icebergs on fast ice, and provides a research basis for further study of the interaction between icebergs and Fast ice.

(4) For the first time, the high-resolution sea ice thermodynamic model HIGHTSI was used to simulate regional fast ice thickness in Antarctic. Thickness is a key parameter in studying the mass balance of sea ice. In this study, the model sensitivity test was carried out based on the fast ice field measurements. The results show that the simulation results of fast ice are the closest to the measurements when the ocean heat flux is 20w/m², with mean deviation -0.012 m. A multi-year (2006-2011 and 2016-2017) simulation of fast ice along the coast of Mac Robertson Land in the East Antarctic was carried out. The results show that the fast ice thickness in this area usually reaches the maximum value in October of each year (maximum 1.96 m). The average maximum volume of fast ice is about 59.3 km³. The closer to the coast, the greater the thickness of the fast ice. The correlation between the fast ice thickness and FDD can reach 98%. This provides a basis for the subsequent simulation of the fast ice thickness and mass balance research of the Antarctic.

In summary, this study proposes a method for quickly extracting circum-Antarctica fast ice using SAR data. For the first time, a high-resolution, long-tiem series of fast ice dataset of the entire Antarctic is obtained, and the spatial distribution of the Antarctic fast ice is comprehensively analyzed. The changes, the interaction between sea ice, seafloor topography and icebergs and fast ice were studied, which provided a basis for further research on Antarctic fast ice. For the first time, the sea ice thermodynamic model HIGHSSI was used to carry out the regional simulation of fast ice in Antarctic. This provides a basis for the subsequent simulation of the fast ice thickness and mass balance research of the Antarctic.