Transformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps

Mesoscale eddies are dynamic oceanic phenomena significantly influencing marine ecosystems’ energy transfer, nutrients, and biogeochemical cycles. These eddies’ precise identification and categorization can improve climate modeling, ocean circulation research, and environmental...

Full description

Saved in:
Bibliographic Details
Main Authors: Chen Ji, Wenyang Xu, Xiangtian Zheng, Yasmeen Ahmed, Saad Ahmed Jamal, Fakhar Imam, Mohammed Saleh Ali Muthanna, Maha Ibrahim, Sajid Ullah, Dmitry E. Kucher
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11015982/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849419926108897280
author Chen Ji
Wenyang Xu
Xiangtian Zheng
Yasmeen Ahmed
Saad Ahmed Jamal
Fakhar Imam
Mohammed Saleh Ali Muthanna
Maha Ibrahim
Sajid Ullah
Dmitry E. Kucher
author_facet Chen Ji
Wenyang Xu
Xiangtian Zheng
Yasmeen Ahmed
Saad Ahmed Jamal
Fakhar Imam
Mohammed Saleh Ali Muthanna
Maha Ibrahim
Sajid Ullah
Dmitry E. Kucher
author_sort Chen Ji
collection DOAJ
description Mesoscale eddies are dynamic oceanic phenomena significantly influencing marine ecosystems’ energy transfer, nutrients, and biogeochemical cycles. These eddies’ precise identification and categorization can improve climate modeling, ocean circulation research, and environmental surveillance. This study presents an innovative methodology for mesoscale eddy detection utilizing Transformer-based deep learning models, namely, Swin Transformer U-Net and SegFormer, to categorize ocean eddies from sea surface temperature (SST) maps sourced from the copernicus marine environment monitoring service. In contrast to traditional convolutional neural networks (CNNs) that have prevailed in the domain, Transformer-based models provide superior global attention mechanisms, facilitating greater feature extraction and segmentation precision. The models are trained on labeled SST datasets and assessed using intersection over union, Dice coefficient, precision, recall, and F1-score. Experimental findings demonstrate that Transformer-based designs surpass conventional CNN-based techniques, yielding enhanced generalization and superior accuracy in classifying cyclonic and anticyclonic eddies. This study illustrates the efficacy of attention-based segmentation algorithms for resilient oceanographic applications.
format Article
id doaj-art-8a0267a56b3d4010b491d1a218a71b82
institution Kabale University
issn 1939-1404
2151-1535
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
spelling doaj-art-8a0267a56b3d4010b491d1a218a71b822025-08-20T03:31:55ZengIEEEIEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing1939-14042151-15352025-01-0118152491526410.1109/JSTARS.2025.357400411015982Transformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature MapsChen Ji0https://orcid.org/0000-0001-5740-4872Wenyang Xu1Xiangtian Zheng2https://orcid.org/0000-0001-5819-1215Yasmeen Ahmed3Saad Ahmed Jamal4Fakhar Imam5Mohammed Saleh Ali Muthanna6https://orcid.org/0000-0002-1165-7812Maha Ibrahim7Sajid Ullah8https://orcid.org/0000-0002-5079-3761Dmitry E. Kucher9https://orcid.org/0000-0002-7919-3487School of Geography and Ocean Science, Nanjing University, Nanjing, ChinaNanjing Institute of Technology, Nanjing, ChinaNanjing Institute of Technology, Nanjing, ChinaDepartment of Building Construction Science College of Architecture Art and Design, Mississippi State University, Mississippi State, MS, USAInstitute for Research and Advanced Studies, Universidade de Evora, Évora, PortugalUniversity of the Punjab, Lahore, PakistanDepartment of International Business Management, Tashkent State University of Economics, Tashkent, UzbekistanDepartment of International Business Management, Tashkent State University of Economics, Tashkent, UzbekistanDepartment of Water Resources and Environmental Engineering, Nangarhar University, Nangarhar, AfghanistanDepartment of Environmental Management, Institute of Environmental Engineering, Peoples’ Friendship University of Russia, Moscow, RussiaMesoscale eddies are dynamic oceanic phenomena significantly influencing marine ecosystems’ energy transfer, nutrients, and biogeochemical cycles. These eddies’ precise identification and categorization can improve climate modeling, ocean circulation research, and environmental surveillance. This study presents an innovative methodology for mesoscale eddy detection utilizing Transformer-based deep learning models, namely, Swin Transformer U-Net and SegFormer, to categorize ocean eddies from sea surface temperature (SST) maps sourced from the copernicus marine environment monitoring service. In contrast to traditional convolutional neural networks (CNNs) that have prevailed in the domain, Transformer-based models provide superior global attention mechanisms, facilitating greater feature extraction and segmentation precision. The models are trained on labeled SST datasets and assessed using intersection over union, Dice coefficient, precision, recall, and F1-score. Experimental findings demonstrate that Transformer-based designs surpass conventional CNN-based techniques, yielding enhanced generalization and superior accuracy in classifying cyclonic and anticyclonic eddies. This study illustrates the efficacy of attention-based segmentation algorithms for resilient oceanographic applications.https://ieeexplore.ieee.org/document/11015982/Deep learningocean eddiesremote sensingSegFormersemantic segmentationsea surface temperature (SST) data
spellingShingle Chen Ji
Wenyang Xu
Xiangtian Zheng
Yasmeen Ahmed
Saad Ahmed Jamal
Fakhar Imam
Mohammed Saleh Ali Muthanna
Maha Ibrahim
Sajid Ullah
Dmitry E. Kucher
Transformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Deep learning
ocean eddies
remote sensing
SegFormer
semantic segmentation
sea surface temperature (SST) data
title Transformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps
title_full Transformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps
title_fullStr Transformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps
title_full_unstemmed Transformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps
title_short Transformer-Based Deep Learning for Mesoscale Eddy Detection in Sea Surface Temperature Maps
title_sort transformer based deep learning for mesoscale eddy detection in sea surface temperature maps
topic Deep learning
ocean eddies
remote sensing
SegFormer
semantic segmentation
sea surface temperature (SST) data
url https://ieeexplore.ieee.org/document/11015982/
work_keys_str_mv AT chenji transformerbaseddeeplearningformesoscaleeddydetectioninseasurfacetemperaturemaps
AT wenyangxu transformerbaseddeeplearningformesoscaleeddydetectioninseasurfacetemperaturemaps
AT xiangtianzheng transformerbaseddeeplearningformesoscaleeddydetectioninseasurfacetemperaturemaps
AT yasmeenahmed transformerbaseddeeplearningformesoscaleeddydetectioninseasurfacetemperaturemaps
AT saadahmedjamal transformerbaseddeeplearningformesoscaleeddydetectioninseasurfacetemperaturemaps
AT fakharimam transformerbaseddeeplearningformesoscaleeddydetectioninseasurfacetemperaturemaps
AT mohammedsalehalimuthanna transformerbaseddeeplearningformesoscaleeddydetectioninseasurfacetemperaturemaps
AT mahaibrahim transformerbaseddeeplearningformesoscaleeddydetectioninseasurfacetemperaturemaps
AT sajidullah transformerbaseddeeplearningformesoscaleeddydetectioninseasurfacetemperaturemaps
AT dmitryekucher transformerbaseddeeplearningformesoscaleeddydetectioninseasurfacetemperaturemaps