基于SNPP/VIIRS卫星数据的海温锋检测算法研究  

Research on Sea Surface Temperature Front Detection Algorithm Based on SNPP/VIIRS Satellite Data

作　　者：于杰[1,2] 管磊[1,2] Yu Jie;Guan Lei(Sanya Oceanographic Institution,Ocean University of China,Sanya 572024,China;College of Marine Technology,Faculty of Information Science and Engineering,Ocean University of China,Qingdao 266100,China)

机构地区：[1]中国海洋大学三亚海洋研究院,海南三亚572024 [2]中国海洋大学信息科学与工程学部海洋技术学院,山东青岛266100

出　　处：《中国海洋大学学报（自然科学版）》2024年第9期21-29,共9页Periodical of Ocean University of China

基　　金：三亚崖州湾科技城管理局2022年度科技计划项目(SKJC-2022-01-001);海南省自然科学基金项目(122CXTD519)资助。

摘　　要：本文采用SNPP/VIIRS(Suomi national polar-orbiting partnership/Visible infrared imaging radiometer suite)卫星海表温度数据对锋面检测算法进行对比分析,包括Sobel算法、Canny算法、引力模型法、直方图分析法和熵算法。通过分析上述算法中间结果、单日海温锋检测结果以及月度锋面概率分布的差别与特征,再结合各算法特点发现,在锋面识别中由于各算法的底层逻辑不同,造成了对海温锋定义难以统一的问题,因此各算法检测出的海温锋各有特点,适合不同的研究和应用需求,实际应用时要根据研究需要和实际情况选择合适的算法。各算法的特点为:Canny算法、引力模型法、Sobel算法、熵算法对强锋面的检测能力依次递减,对弱锋面的检测能力依次递增,抗噪性递减,直方图分析法检测出的锋面表现出较强的连续性,锋面光滑、清晰且连贯,抗噪性较好;各算法适用场合为:以梯度计算为基础的Sobel算法或者其他梯度类算法适用于精确计算温度变化程度的应用场景,Canny算法适用于强锋面的研究,引力模型法适用于应用噪声较大的数据检测锋面或者强锋面的检测,熵算法适用于弱锋面的检测,直方图分析法比较适合分析连续的长锋面。In this thesis,the sea surface temperature(SST)data of SNPP/VIIRS(Suomi National Polar-Partnership/Visible Infrared Imaging Radiometer Suite)are used to compare and analyze the front detection algorithms,including Sobel algorithm,Canny algorithm,the gravity model algorithm,the histogram analysis algorithm and the entropy algorithm.Through the difference and characteristics of the intermediate results,the one-day SST front detection results and the monthly frontal probability distribution of the above algorithms,combined with the characteristics of each algorithm,it is found that,because of the different underlying logic of each algorithm,it is difficult to unify the definition of SST front in front recognition.Therefore,the SST fronts detected by the algorithms have different characteristics and are suitable for different research and application needs.In practical application,the appropriate algorithm should be selected according to the research needs and the actual situation.The characteristics of each algorithm are as follows:Canny algorithm,the gravity model algorithm,Sobel algorithm and the entropy algorithm have decreasing ability to detect strong fronts and increasing ability to detect weak fronts,and the anti-noise ability is decreasing.The fronts detected by the histogram analysis algorithm are smooth,clear and coherent,and the anti-noise ability is good.The occasions suitable for each algorithm are:Sobel algorithm or other algorithms based on gradient calculation are suitable for the application scenario of accurately calculating the degree of temperature change,Canny algorithm is suitable for the study of the strong fronts,the gravity model algorithm is suitable for detecting fronts with noisy data or strong fronts,and the entropy algorithm is suitable for the detection of the weak fronts.The histogram analysis is more suitable for the analysis of continuous long fronts.

关 键 词：卫星 海表温度 海温锋 锋面检测算法 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]