基于HJ-1C SAR数据的辽东湾海冰分类

Approach to the classification of sea ice in Liaodong Bay using HJ-1C SAR data

针对海冰遥感分类问题,使用我国首颗民用合成孔径雷达卫星环境一号星(HJ-1C)图像的S波段VV极化SAR数据进行辽东湾海冰分类,提出了一种针对单极化SAR数据的海冰分类方法。使用基于SAR数据的3种海冰信息作为分类依据,即灰度信息、灰度共生矩阵纹理信息及基于平整冰面积百分比提取的平整冰密集度信息。研究结果表明,平整冰密集度信息是区分碎冰和风致纹理粗糙开阔水的有效信息。使用最大似然法与决策树融合的分类方法可以有效地识别封冻期辽东湾海域的碎冰、平整冰和开阔水3种类型,为海冰分类提供了一种新思路。

Sea ice classification is an important basis for the sea ice forecast in that many parameters such as the maximum of sea ice extent can be derived from that. An approach to sea ice classification using HJ-1C SAR data, which is the first civil spaceborne SAR system of China, is presented in this paper. The data were S-band and VV polarized and acquired over Liaodong Bay. Three types of information were extracted from the SAR data, including the uncalibrated backscattering coefficients and the gray level co -occurrence matrices. Besides, the level ice concentration was introduced as the classification basis. It proves to be effective in sea ice classification, especially in the separation between brash ice and wind-roughened open water. Based on all the information, the authors implemented the classifier fusion which combines the maximum likelihood and the decision tree. With the optical data from HJ-1B for validation, a good result is obtained, in which the brash ice, level ice and open water are well distinguished.