{"title":"基于深度学习的东海浙闵沿海锋冬季时空变化","authors":"Qiang Lian, Xihong Jian, Tengfei Li, Shuwen Zhang, Jin Feng, Zhaoyun Chen","doi":"10.1016/j.jmarsys.2025.104104","DOIUrl":null,"url":null,"abstract":"<div><div>Oceanic fronts are crucial for marine ecology and the environment due to their vigorous mixing, high productivity, and abundant fishery resources. In this study, the Zhe-Min Coastal Front (ZMCF) in the East China Sea is identified from remote sensing Sea Surface Temperature (SST) images using a combination of the Belkin O'Reilly Algorithm and a deep learning network, addressing the shortcomings of traditional front gradient algorithms. The Ringed Residual U-Net network excels in detecting the ZMCF, particularly in discontinuous, fragmented, and multi-branch fronts. After detecting the ZMCF areas, the monthly variations in SST gradient, front probability, and offshore distance of the ZMCF are analyzed for winter. An empirical orthogonal function (EOF) method is employed to capture the dominant mode of spatiotemporal variations in the ZMCF probability, demonstrating that the cross-shore movement of the ZMCF is closely related to the magnitude of the northerly wind. Under the conditions of the low phase of the Pacific Decadal Oscillation, the intensified ZMCF is trapped near the coast, driven by strong northerly wind during La Niña years. El Niño events exert the opposite effect.</div></div>","PeriodicalId":50150,"journal":{"name":"Journal of Marine Systems","volume":"251 ","pages":"Article 104104"},"PeriodicalIF":2.5000,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Spatiotemporal variations in Zhe-Min Coastal Fronts in the East China Sea in winter by deep learning\",\"authors\":\"Qiang Lian, Xihong Jian, Tengfei Li, Shuwen Zhang, Jin Feng, Zhaoyun Chen\",\"doi\":\"10.1016/j.jmarsys.2025.104104\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><div>Oceanic fronts are crucial for marine ecology and the environment due to their vigorous mixing, high productivity, and abundant fishery resources. In this study, the Zhe-Min Coastal Front (ZMCF) in the East China Sea is identified from remote sensing Sea Surface Temperature (SST) images using a combination of the Belkin O'Reilly Algorithm and a deep learning network, addressing the shortcomings of traditional front gradient algorithms. The Ringed Residual U-Net network excels in detecting the ZMCF, particularly in discontinuous, fragmented, and multi-branch fronts. After detecting the ZMCF areas, the monthly variations in SST gradient, front probability, and offshore distance of the ZMCF are analyzed for winter. An empirical orthogonal function (EOF) method is employed to capture the dominant mode of spatiotemporal variations in the ZMCF probability, demonstrating that the cross-shore movement of the ZMCF is closely related to the magnitude of the northerly wind. Under the conditions of the low phase of the Pacific Decadal Oscillation, the intensified ZMCF is trapped near the coast, driven by strong northerly wind during La Niña years. El Niño events exert the opposite effect.</div></div>\",\"PeriodicalId\":50150,\"journal\":{\"name\":\"Journal of Marine Systems\",\"volume\":\"251 \",\"pages\":\"Article 104104\"},\"PeriodicalIF\":2.5000,\"publicationDate\":\"2025-07-10\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Marine Systems\",\"FirstCategoryId\":\"89\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S0924796325000673\",\"RegionNum\":3,\"RegionCategory\":\"地球科学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q2\",\"JCRName\":\"GEOSCIENCES, MULTIDISCIPLINARY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Marine Systems","FirstCategoryId":"89","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0924796325000673","RegionNum":3,"RegionCategory":"地球科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"GEOSCIENCES, MULTIDISCIPLINARY","Score":null,"Total":0}
Spatiotemporal variations in Zhe-Min Coastal Fronts in the East China Sea in winter by deep learning
Oceanic fronts are crucial for marine ecology and the environment due to their vigorous mixing, high productivity, and abundant fishery resources. In this study, the Zhe-Min Coastal Front (ZMCF) in the East China Sea is identified from remote sensing Sea Surface Temperature (SST) images using a combination of the Belkin O'Reilly Algorithm and a deep learning network, addressing the shortcomings of traditional front gradient algorithms. The Ringed Residual U-Net network excels in detecting the ZMCF, particularly in discontinuous, fragmented, and multi-branch fronts. After detecting the ZMCF areas, the monthly variations in SST gradient, front probability, and offshore distance of the ZMCF are analyzed for winter. An empirical orthogonal function (EOF) method is employed to capture the dominant mode of spatiotemporal variations in the ZMCF probability, demonstrating that the cross-shore movement of the ZMCF is closely related to the magnitude of the northerly wind. Under the conditions of the low phase of the Pacific Decadal Oscillation, the intensified ZMCF is trapped near the coast, driven by strong northerly wind during La Niña years. El Niño events exert the opposite effect.
期刊介绍:
The Journal of Marine Systems provides a medium for interdisciplinary exchange between physical, chemical and biological oceanographers and marine geologists. The journal welcomes original research papers and review articles. Preference will be given to interdisciplinary approaches to marine systems.