A data-efficient deep transfer learning framework for methane super-emitter detection in oil and gas fields using the Sentinel-2 satellite
<p>Efficiently detecting large methane point sources (super-emitters) in oil and gas fields is crucial for informing stakeholder decisions about mitigation actions. Satellite measurements by multispectral instruments, such as Sentinel-2, offer global and frequent coverage. However, methane sig...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2025-04-01
|
| Series: | Atmospheric Chemistry and Physics |
| Online Access: | https://acp.copernicus.org/articles/25/4035/2025/acp-25-4035-2025.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849731305458106368 |
|---|---|
| author | S. Zhao S. Zhao Y. Zhang Y. Zhang S. Zhao S. Zhao X. Wang X. Wang D. J. Varon |
| author_facet | S. Zhao S. Zhao Y. Zhang Y. Zhang S. Zhao S. Zhao X. Wang X. Wang D. J. Varon |
| author_sort | S. Zhao |
| collection | DOAJ |
| description | <p>Efficiently detecting large methane point sources (super-emitters) in oil and gas fields is crucial for informing stakeholder decisions about mitigation actions. Satellite measurements by multispectral instruments, such as Sentinel-2, offer global and frequent coverage. However, methane signals retrieved from satellite multispectral images are prone to surface and atmospheric artifacts that vary spatially and temporally, making it challenging to build a detection algorithm that applies everywhere. Hence, laborious manual inspection is often necessary, hindering widespread deployment of the technology. Here, we propose a novel deep-transfer-learning-based methane plume detection framework. It consists of two components: an adaptive artifact removal algorithm (low-reflectance artifact detection, LRAD) to reduce artifacts in methane retrievals and a deep subdomain adaptation network (DSAN) to detect methane plumes. To train the algorithm, we compile a dataset comprising 1627 Sentinel-2 images from six known methane super-emitters reported in the literature. We evaluate the ability of the algorithm to discover new methane sources with a suite of transfer tasks, in which training and evaluation data come from different regions. Results show that DSAN (average macro <span class="inline-formula"><i>F</i><sub>1</sub></span> score 0.86) outperforms four convolutional neural networks (CNNs), MethaNet (average macro <span class="inline-formula"><i>F</i><sub>1</sub></span> score 0.70), ResNet-50 (average macro <span class="inline-formula"><i>F</i><sub>1</sub></span> score 0.77), VGG16 (average macro <span class="inline-formula"><i>F</i><sub>1</sub></span> score 0.73), and EfficientNet-V2L (average macro <span class="inline-formula"><i>F</i><sub>1</sub></span> score 0.78), in transfer tasks. The transfer learning algorithm overcomes the issue of conventional CNNs, which is their performance degrades substantially in regions outside regions with training data. We apply the algorithm trained with known sources to an unannotated region in the Algerian Hassi Messaoud oil field and reveal 34 anomalous emission events during a 1-year period, which are attributed to three methane super-emitters associated with production and transmission infrastructure. These results demonstrate the potential of our deep-transfer-learning-based method in contributing towards efficient methane super-emitter discovery using Sentinel-2 across different oil and gas fields worldwide.</p> |
| format | Article |
| id | doaj-art-59f0f8debccc429a94af501e84ca26e6 |
| institution | DOAJ |
| issn | 1680-7316 1680-7324 |
| language | English |
| publishDate | 2025-04-01 |
| publisher | Copernicus Publications |
| record_format | Article |
| series | Atmospheric Chemistry and Physics |
| spelling | doaj-art-59f0f8debccc429a94af501e84ca26e62025-08-20T03:08:36ZengCopernicus PublicationsAtmospheric Chemistry and Physics1680-73161680-73242025-04-01254035405210.5194/acp-25-4035-2025A data-efficient deep transfer learning framework for methane super-emitter detection in oil and gas fields using the Sentinel-2 satelliteS. Zhao0S. Zhao1Y. Zhang2Y. Zhang3S. Zhao4S. Zhao5X. Wang6X. Wang7D. J. Varon8College of Environmental and Resource Sciences, Zhejiang University, Hangzhou, Zhejiang Province, 310058, ChinaKey Laboratory of Coastal Environment and Resources of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, Zhejiang Province, 310024, ChinaKey Laboratory of Coastal Environment and Resources of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, Zhejiang Province, 310024, ChinaInstitute of Advanced Technology, Westlake Institute for Advanced Study, Hangzhou, Zhejiang Province, 310024, ChinaKey Laboratory of Coastal Environment and Resources of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, Zhejiang Province, 310024, ChinaInstitute of Advanced Technology, Westlake Institute for Advanced Study, Hangzhou, Zhejiang Province, 310024, ChinaCollege of Environmental and Resource Sciences, Zhejiang University, Hangzhou, Zhejiang Province, 310058, ChinaKey Laboratory of Coastal Environment and Resources of Zhejiang Province, School of Engineering, Westlake University, Hangzhou, Zhejiang Province, 310024, ChinaSchool of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138, United States<p>Efficiently detecting large methane point sources (super-emitters) in oil and gas fields is crucial for informing stakeholder decisions about mitigation actions. Satellite measurements by multispectral instruments, such as Sentinel-2, offer global and frequent coverage. However, methane signals retrieved from satellite multispectral images are prone to surface and atmospheric artifacts that vary spatially and temporally, making it challenging to build a detection algorithm that applies everywhere. Hence, laborious manual inspection is often necessary, hindering widespread deployment of the technology. Here, we propose a novel deep-transfer-learning-based methane plume detection framework. It consists of two components: an adaptive artifact removal algorithm (low-reflectance artifact detection, LRAD) to reduce artifacts in methane retrievals and a deep subdomain adaptation network (DSAN) to detect methane plumes. To train the algorithm, we compile a dataset comprising 1627 Sentinel-2 images from six known methane super-emitters reported in the literature. We evaluate the ability of the algorithm to discover new methane sources with a suite of transfer tasks, in which training and evaluation data come from different regions. Results show that DSAN (average macro <span class="inline-formula"><i>F</i><sub>1</sub></span> score 0.86) outperforms four convolutional neural networks (CNNs), MethaNet (average macro <span class="inline-formula"><i>F</i><sub>1</sub></span> score 0.70), ResNet-50 (average macro <span class="inline-formula"><i>F</i><sub>1</sub></span> score 0.77), VGG16 (average macro <span class="inline-formula"><i>F</i><sub>1</sub></span> score 0.73), and EfficientNet-V2L (average macro <span class="inline-formula"><i>F</i><sub>1</sub></span> score 0.78), in transfer tasks. The transfer learning algorithm overcomes the issue of conventional CNNs, which is their performance degrades substantially in regions outside regions with training data. We apply the algorithm trained with known sources to an unannotated region in the Algerian Hassi Messaoud oil field and reveal 34 anomalous emission events during a 1-year period, which are attributed to three methane super-emitters associated with production and transmission infrastructure. These results demonstrate the potential of our deep-transfer-learning-based method in contributing towards efficient methane super-emitter discovery using Sentinel-2 across different oil and gas fields worldwide.</p>https://acp.copernicus.org/articles/25/4035/2025/acp-25-4035-2025.pdf |
| spellingShingle | S. Zhao S. Zhao Y. Zhang Y. Zhang S. Zhao S. Zhao X. Wang X. Wang D. J. Varon A data-efficient deep transfer learning framework for methane super-emitter detection in oil and gas fields using the Sentinel-2 satellite Atmospheric Chemistry and Physics |
| title | A data-efficient deep transfer learning framework for methane super-emitter detection in oil and gas fields using the Sentinel-2 satellite |
| title_full | A data-efficient deep transfer learning framework for methane super-emitter detection in oil and gas fields using the Sentinel-2 satellite |
| title_fullStr | A data-efficient deep transfer learning framework for methane super-emitter detection in oil and gas fields using the Sentinel-2 satellite |
| title_full_unstemmed | A data-efficient deep transfer learning framework for methane super-emitter detection in oil and gas fields using the Sentinel-2 satellite |
| title_short | A data-efficient deep transfer learning framework for methane super-emitter detection in oil and gas fields using the Sentinel-2 satellite |
| title_sort | data efficient deep transfer learning framework for methane super emitter detection in oil and gas fields using the sentinel 2 satellite |
| url | https://acp.copernicus.org/articles/25/4035/2025/acp-25-4035-2025.pdf |
| work_keys_str_mv | AT szhao adataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT szhao adataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT yzhang adataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT yzhang adataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT szhao adataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT szhao adataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT xwang adataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT xwang adataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT djvaron adataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT szhao dataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT szhao dataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT yzhang dataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT yzhang dataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT szhao dataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT szhao dataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT xwang dataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT xwang dataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite AT djvaron dataefficientdeeptransferlearningframeworkformethanesuperemitterdetectioninoilandgasfieldsusingthesentinel2satellite |