Current Issue Cover


摘 要
摘 要:目的:中尺度海洋锋及涡旋均是重要的中尺度海洋环境特征。中尺度海洋锋及涡旋的提取及其时空分布、变化的研究对海洋生态系统的研究、渔业资源评估、渔情预报及军事等都有重要意义。遥感技术能在同一时间获取大面积海洋要素观测数据,遥感数据具有优良的连续性、同步性,因此遥感数据被广泛应用于中尺度海洋锋及涡旋提取的研究中。方法:本文对基于遥感数据进行中尺度海洋锋提取的梯度法、Canny算法、小波分析法和基于引力模型的方法,以及涡旋的提取的OW法、WA法、基于海面高度的无阈值等值线法和HD法进行总结和分析,并提出对中尺度海洋锋面及涡旋提取方法的见解及新思路。结果:利用2014年2月南海北部海表温度(SST)数据,分别采用梯度法中的Gradient法、Sobel算法以及Canny算法对南海北部温度锋进行提取并得到该区域温度锋分布图。结果表明在多种锋面提取方法中,Canny算法具有较高的效率且其提取结果的连续性和精度更好。中尺度涡的提取方法中,WA法的提取结果具有更好的准确性。早期的中尺度涡提取方法忽略了多中心结构涡旋存在的情况,而后来的HD法能较好的识别多中心结构涡旋。结论:阈值选取是中尺度海洋锋及涡旋提取的难点和提取结果好坏的关键。然而海洋要素图像弱边缘的特点使得传统边缘检测方法不一定适用于中尺度锋提取。文章通过对不同锋面及涡旋提取方法的总结与分析,为海洋锋面及涡旋提取的研究提供了参考依据。
Research progress of mesoscale ocean fronts and eddies extraction method based on remote sensing data

lianzhou,zhouweifeng,fanxiumei(Key Laboratory of Fisheries Resources Remote Sensing and Information Technology, East China Sea Fisheries Research Institute, Chinese Academy of Fishery Sciences)

Abstract: Objective: The mesoscale ocean front and eddy are both important mesoscale marine environment characteristics. Ocean front is the interface of water masses with different properties, in the area where there is front, corresponding hydrological factors (such as temperature, chlorophyll concentration, salinity, etc.) present a high horizontal gradient. The seawater converge and vertical motion are enhanced in the fields where fronts appear, then lead to enrichment of nutrient and provide a rich diet to plankton, fish, and so on. Therefore, sea area where the front appear can form a good fishing ground (such as Zhoushan fishing ground and Minnan fishing ground in China). The mesoscale eddy plays an important role in ocean circulation and is an important undertaker of energy transport and ocean material transfer in ocean. At the same time eddy can influence the distribution of hydrological factors such as temperature and salinity, thus it is one of the important factors of marine hydrological variation. In addition, the occurrence of eddies associated with local lifting flow, for instance, the upwelling associated with cold eddy carry nutrient to euphotic zone from bottom of ocean then greatly improve the primary productivity of the ocean and influence the distribution of fishing ground, thus affecting the development of the marine economy. The remote sensing data is excellent in continuity and synchronization, and it can reflect the spatial distribution characteristics of the marine hydrological elements and the sea surface height well. Therefore remote sensing data has been widely used in the extraction of mesoscale ocean fronts and eddies, such as MODIS-SST, AVHRR-SST, SSH( Sea Surface Height), SLA ( Sea Level Anomaly), etc. Therefore,the research on the extraction of the mesoscale ocean fronts and eddies based on remote sensing data is significant to research about marine ecosystem, fisheries stock assessment and fishing condition forecast. We aim at providing reference and ideas for the extraction of mesoscale ocean front and eddy by means of summarizing and analyzing the mesoscale ocean front and eddy extraction methods. Method: In this paper, the methods of fronts extraction such as Gradient method, Canny algorithm, wavelet analysis method and algorithm based on the law of universal gravity, the methods of ;eddies extraction such as OW method, WA method, SSH-based method and HD method were summarized and analysed, and then put forward insights and new ideas. In order to show the difference of various front extraction methods intuitively, extracting front from same area by means of Gradient algorithm, Sobel algorithm and Canny algorithm respectively and drawing the extraction results figure. When use Gradient algorithm and Sobel algorithm to extract front, the thresholds which are used to distinguish the background pixels and front pixels are got by iterative method. Afterwards, using Zhang-Suen method to implement binary image thinning and then get the center line of the front. When use Canny algorithm to extract front, choose 0.25 as low threshold and choose 0.9 as high threshold. Result: Using sea surface temperature (SST) data of northern South China Sea (SCS) in February 2014, adopt Gradient algorithm, Sobel algorithm and Canny algorithm to extract the temperature front in northern South China Sea and get the figure about temperature front distribution of this area. The results show that in a variety of front extraction methods, the gradient method is simple but is influenced greatly by noise; The Canny algorithm has great advantage in the front positioning accuracy, continuity and computational efficiency; Wavelet analysis can be used in multi-scale analysis, but the computation is complex; the algorithm based on the law of universal gravity takes into account the influence of value and position of the center pixel and neighborhood pixel thus it has better anti noise ability and accuracy than the gradient method. In a variety of eddy extraction methods, OW method can identify the eddy core region well, but the extraction result is greatly influenced by the selection of W value threshold; WA method is excellent in accuracy but it need large calculation to get streamline; SSH-based method is simple but it can only extract the eddy boundary but not the core area of eddy. Early eddy extraction methods tend to ignore the condition that the multi-core eddy structures eddies may appear in ocean, and this methods are unable to identify he multi-core eddy structures eddies. The HD method combines the advantages of OW method and SSH-based method, it can extract the boundary and core area of eddy simultaneously, furthermore, it’s able to identify the multi-core eddy structures eddies. Conclusion: Acording to the summary and analysis about various front and eddy extraction methods, considering that threshold choice is the difficulty on fronts and eddies extraction, and it is important to the quality of extraction result. Many researchers have done a lot of meaningful work on the threshold selection method. In addition, edge detection methods such as Gradient method and Canny algorithm that were used widely in front extraction currently are designed for extracting sharp edge (such as solid edge). However, sea water is fluid that with the character of weak edge, in other words, the edge of water masses with different properties is not obvious and is more difficulty to be identified than solid edge. Therefore, this traditional edge detection methodes is not suitable for the mesoscale front extraction due to the weak edge character of marine environment element image. Since ocean front is the interface of different water masses, we can adopt Region Growing Algorithm which is used widely in image segmentation to segment the marine environmental element image of study area into several independent parts that represents different water mass, then search their boundaries, this boundaries are the front we want to extract.