Restoration of Missing Patterns on Satellite Infrared Sea Surface Temperature Images Due to Cloud Coverage Using Deep Generative Inpainting Network

Kang, Song-Hee; Choi, Youngjin; Choi, Jae Young

doi:10.3390/jmse9030310

Cited by 17 publications

(8 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The mean value was calculated as the true value to compare with the MODIS SST in the same location. Indicators of bias, root mean square error (RMSE), mean absolute error (MAE), and standard deviation (STD) (Jang and Park, 2019;Liu et al, 2020;Kang et al, 2021) were used to evaluate the accuracy. The validation results showed that the MAE of MODIS SST in the SCS ranged from 0.69°C to 0.72°C, and the STD ranged from 0.78°C to 0.87°C (Table 3).…”

Section: Accuracy Assessment Of Modis Sst Datamentioning

confidence: 99%

Deriving fine-scale patterns of sea surface temperature in coral reef habitats using the Landsat 8 thermal infrared sensor

Zuo,

Yu,

Qin

et al. 2023

Front. Mar. Sci.

View full text Add to dashboard Cite

The available sea surface temperature (SST) products are too coarse to assess the fine-scale (<1 km) SST variations related to coral bleaching. In this study, we proposed an optimal SST inversion model using Landsat 8 thermal infrared sensor (TIRS) images to derive fine-scale SST patterns in the coral reef habitats of the Xisha Islands, South China Sea. Our study included two parts: 1) six SST inversion models were developed using the radiative transfer method and the split window (SW) algorithm in the hot season and cool season, from which the optimal SST inversion model was determined; and 2) the optimal model was applied to 47 Landsat 8 TIRS images to derive the SST spatial and temporal pattern among the geomorphic zones of six reefs in hot and cool season conditions. Compared with the measured sea temperature data and the verified MODIS SST products, the SST6 model using the SW algorithm was optimal, with an RMSE of approximately 0.64°C in the hot season. The average SST results usually had a pattern of reef flat > lagoon > reef slope/offshore sea. The reef flat was usually approximately 0.05°C–0.2°C hotter than the lagoon in the hot season. The SST in the lagoon also increased from south to north and the shallow lagoon was usually warmer than the deep lagoon in the hot season. Our results suggested that scleractinian corals in the reef flat and the lagoon were more susceptible to bleaching-level thermal stress than other geomorphic zones. During the cool season, the SST fluctuated markedly among coral reefs and geomorphic zones.

show abstract

Section: Accuracy Assessment Of Modis Sst Datamentioning

confidence: 99%

Deriving fine-scale patterns of sea surface temperature in coral reef habitats using the Landsat 8 thermal infrared sensor

Zuo,

Yu,

Qin

et al. 2023

Front. Mar. Sci.

View full text Add to dashboard Cite

show abstract

“…Zhang et al [ 77 ] proposed a unified spatial–temporal–spectral deep convolutional neural network (CNN) image inpainting architecture to recover information obscured by poor atmospheric conditions in satellite images. Kang et al [ 27 ] modified the architecture from [ 72 ] to restore the missing patterns of sea surface temperature from satellite images. Tasnim and Mondal [ 60 ] also applied the inpainting architecture from [ 72 ] to remove redundancies in satellite images and restore the imagery.…”

Section: Interpolation Of Spatiotemporal Data Using Deep Learningmentioning

confidence: 99%

Integrative urban AI to expand coverage, access, and equity of urban data

Howe

Brown

Han

et al. 2022

Eur. Phys. J. Spec. Top.

View full text Add to dashboard Cite

We consider the use of AI techniques to expand the coverage, access, and equity of urban data. We aim to enable holistic research on city dynamics, steering AI research attention away from profit-oriented, societally harmful applications (e.g., facial recognition) and toward foundational questions in mobility, participatory governance, and justice. By making available high-quality, multi-variate, cross-scale data for research, we aim to link the macrostudy of cities as complex systems with the reductionist view of cities as an assembly of independent prediction tasks. We identify four research areas in AI for cities as key enablers: interpolation and extrapolation of spatiotemporal data, using NLP techniques to model speech- and text-intensive governance activities, exploiting ontology modeling in learning tasks, and understanding the interaction of fairness and interpretability in sensitive contexts.

show abstract

“…In paper [12], authors proposed a novel deep generative inpainting network (GIN) trained under the framework of generative adversarial learning, which is optimized for the restoration of cloud-disturbed satellite sea surface temperature (SST) imagery. The proposed GIN architecture can achieve accurate and fast restoration results.…”

Section: Artificial Intelligence In Marine Science and Engineeringmentioning

confidence: 99%

Artificial Intelligence in Marine Science and Engineering

Márquez

Papaelias

Marini

2022

JMSE

View full text Add to dashboard Cite

show abstract

Restoration of Missing Patterns on Satellite Infrared Sea Surface Temperature Images Due to Cloud Coverage Using Deep Generative Inpainting Network

Cited by 17 publications

References 21 publications

Deriving fine-scale patterns of sea surface temperature in coral reef habitats using the Landsat 8 thermal infrared sensor

Deriving fine-scale patterns of sea surface temperature in coral reef habitats using the Landsat 8 thermal infrared sensor

Integrative urban AI to expand coverage, access, and equity of urban data

Artificial Intelligence in Marine Science and Engineering

Contact Info

Product

Resources

About