Oceanic Eddy Identification Using Pyramid Split Attention U-Net With Remote Sensing Imagery

Zhao, Nan; Huang, Baoxiang; Yang, Jie; Radenkovic, Milena; Chen, Ge

doi:10.1109/lgrs.2023.3243902

Cited by 12 publications

(9 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A recent approach for our analysis involves the use of CNNs, which have already been used to detect eddies using two-dimensional SLA or sea surface height (SSH) (Lguensat et al, 2018 ; Santana et al, 2020 ), sea surface temperature (SST) (Moschos et al, 2020 , 2022b ), both (Zhao et al, 2023 ), or other grid map observations (Xia et al, 2022 ). In most cases, such as in the study of Lguensat et al ( 2018 ), a special CNN, known as U-Net, has been used, which is a state-of-the-art method for semantic segmentation of two-dimensional data has been used.…”

Section: Related Workmentioning

confidence: 99%

“…In radar altimetry, sea surface heights are measured that contain features (highs or lows) of an eddy within the satellite footprint. The current research on eddy detection infers results either from gridded sea level anomaly (SLA) maps that are processed from multiple radar altimetry missions (Lguensat et al, 2018 ), from sea surface temperature (SST) grid maps (Moschos et al, 2020 ), from both (Zhao et al, 2023 ), or other grid map observations (Xia et al, 2022 ). Along-track (AT) altimetry data capture the instant sea-level heights at the time of measurement.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Estimating daily semantic segmentation maps of classified ocean eddies using sea level anomaly data from along-track altimetry

Bolmer,

Abulaitijiang,

Kusche

et al. 2024

Front. Artif. Intell.

View full text Add to dashboard Cite

Mesoscale eddies, which are fast-moving rotating water bodies in the ocean with horizontal scales ranging from 10 km to 100 km and above, are considered to be the weather of the oceans. They are of interest to marine biologists, oceanographers, and geodesists for their impact on water mass, heat, and nutrient transport. Typically, gridded sea level anomaly maps processed from multiple radar altimetry missions are used to detect eddies. However, multi-mission sea level anomaly maps obtained by the operational processors have a lower effective spatiotemporal resolution than their grid spacing and temporal resolution, leading to inaccurate eddy detection. In this study, we investigate the use of higher-resolution along-track sea level anomaly data to infer daily two-dimensional segmentation maps of cyclonic, anticyclonic, or non-eddy areas with greater accuracy than using processed sea level anomaly grid map products. To tackle this challenge, we propose a deep neural network that uses spatiotemporal contextual information within the modality of along-track data. This network is capable of producing a two-dimensional segmentation map from data with varying sparsity. We have developed an architecture called Teddy, which uses a Transformer module to encode and process spatiotemporal information, and a sparsity invariant CNN to infer a two-dimensional segmentation map of classified eddies from the ground tracks of varying sparsity on the considered region. Our results show that Teddy creates two-dimensional maps of classified eddies from along-track data with higher accuracy and timeliness when compared to commonly used methods that work with less accurate preprocessed sea level anomaly grid maps. We train and test our method with a carefully curated and independent dataset, which can be made available upon request.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Estimating daily semantic segmentation maps of classified ocean eddies using sea level anomaly data from along-track altimetry

Bolmer,

Abulaitijiang,

Kusche

et al. 2024

Front. Artif. Intell.

View full text Add to dashboard Cite

show abstract

“…The U-Net model introduced in [28], initially designed for biomedical image segmentation, finds wide application across many different tasks due to its convolutional architecture, skip connections, and generative capabilities. Notably, the U-Net architecture has proven useful in many tasks in oceanography, including eddy detection [29][30][31], identification of algae blooms [32,33], and detecting or classifying sea ice [34,35]. The U-Net model is also a backbone of many recent diffusion models, such as in [36], where it is used for image super-resolution.…”

Section: Introductionmentioning

confidence: 99%

Ocean Satellite Data Fusion for High-Resolution Surface Current Maps

Kugusheva,

Bull,

Moschos

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

Real-time reconstruction of ocean surface currents is a challenge due to the complex, non-linear dynamics of the ocean, the small number of in situ measurements, and the spatio-temporal heterogeneity of satellite altimetry observations. To address this challenge, we introduce HIRES-CURRENTS-Net, an operational real-time convolutional neural network (CNN) model for daily ocean current reconstruction. This study focuses on the Mediterranean Sea, a region where operational models have great difficulty predicting surface currents. Notably, our model showcases higher accuracy compared to commonly used alternative methods. HIRES-CURRENTS-Net integrates high-resolution measurements from the infrared or visible spectrum—high resolution Sea Surface Temperature (SST) or chlorophyll (CHL) images—in addition to the low-resolution Sea Surface Height (SSH) maps derived from satellite altimeters. In the first stage, we apply a transfer learning method which uses a high-resolution numerical model to pre-train our CNN model on simulated SSH and SST data with synthetic clouds. The observation of System Simulation Experiments (OSSEs) offers us a sufficient training dataset with reference surface currents at very high resolution, and a model trained on this data can then be applied to real data. In the second stage, to enhance the real-time operational performance of our model over previous methods, we fine-tune the CNN model on real satellite data using a novel pseudo-labeling strategy. We validate HIRES-CURRENTS-Net on real data from drifters and demonstrate that our data-driven approach proves effective for real-time sea surface current reconstruction with potential operational applications such as ship routing.

show abstract

“…Conversely, convolutional neural networks (CNNs) have been shown to be highly capable of harvesting rich contextual information at different scales (Zheng et al, 2021). Some of the works, such as those by (Holail et al, 2023, Liang et al, 2022, focused on improving feature extraction through the use of spatio-temporal attention and differential feature fusion, to overcome problems of spurious differences caused by seasonal fluctuations, building shadows, and illumination differences (Liu et al, 2021a). However, these methods run into limitations, particularly when dealing with complex backgrounds such as desert sands, which can obscure building surfaces and make detecting changes difficult.…”

Section: Introductionmentioning

confidence: 99%

“…This paper delves into the investigation and evaluation of Freshly Built Locales (FBLs) using bi-temporal images through state-of-the-art computer vision networks that have been recently proposed. In addition, we have incorporated modifications to the AFDE-Net model (Holail et al, 2023) by introducing a novel residual pyramid attention fusion (RPAF) module. This enhancement enables more accurate identification of intricate details in complex change detection scenarios.…”

Section: Introductionmentioning

confidence: 99%

M-Afde-Net: Novel Deep Learning-Based Building Change Detection of Freshly Built Locales From Satellite Imagery in the Nile Valley, Egypt

Holail,

Saleh,

Xiao

et al. 2023

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. Urban land expansion is a defining characteristic of urbanization, necessitating the monitoring of this phenomenon and the detection of changes to promote sustainable land use and contribute to updating geospatial databases. Methods based on detecting changes in high-resolution satellite imagery have shown poor performance due to downsampling during image processing, resulting in the loss of boundary information. Furthermore, these methods struggle with complex backgrounds where the ground resembles building roofs. This paper delves into the investigation and evaluation of Freshly Built Locales (FBLs) using bi-temporal images through recently proposed computer vision networks. To address the limitations of existing approaches, we have introduced modifications to the AFDE-Net model, which include the novel residual pyramid attention fusion (RPAF) module. This enhancement enables more precise identification of intricate details in complex change detection scenarios. Our proposed model, M-AFDE-Net, has been evaluated on a newly captured dataset from the Nile Valley regions of Egypt, with a spatial resolution of 30 cm. Special attention has been given to New Mansoura and New Tiba as focal areas for analysis. The evaluation results reveal that the modified model, M-AFDE-Net, outperforms other state-of-the-art models in detecting FBLs. It achieves an impressive F1-score of approximately 89.2%, demonstrating its superiority and effectiveness.

show abstract

Oceanic Eddy Identification Using Pyramid Split Attention U-Net With Remote Sensing Imagery

Cited by 12 publications

References 22 publications

Estimating daily semantic segmentation maps of classified ocean eddies using sea level anomaly data from along-track altimetry

Estimating daily semantic segmentation maps of classified ocean eddies using sea level anomaly data from along-track altimetry

Ocean Satellite Data Fusion for High-Resolution Surface Current Maps

M-Afde-Net: Novel Deep Learning-Based Building Change Detection of Freshly Built Locales From Satellite Imagery in the Nile Valley, Egypt

Contact Info

Product

Resources

About