The integration of vision transformers and SAM for automated methane super-emitter detection using TROPOMI data
Bibliographic record
Abstract
Methane (CH 4 ) significantly contributes to global warming, with a global warming potential approximately 84 times greater than carbon dioxide (CO 2 ) over 20 years. Numerous studies have shown that a small number of high-emitting point sources, known as super-emitters, account for a disproportionately large share of total anthropogenic CH 4 emissions, underscoring the urgency of targeted detection strategies. Recently, a growing focus has been on using remote sensing technology for CH 4 monitoring across various emission sources. As such, this study introduces an automated solution for identifying CH 4 super-emitters using Sentinel-5P (S5P) satellite data. Specifically, a deep learning (DL) framework that integrates a Vision Transformer (ViT) and the Segment Anything Model (SAM) for CH 4 plume detection is proposed. The ViT model is trained using CH 4 plume locations reported by the Netherlands Institute for Space Research (SRON) to classify the presence or absence of CH 4 plumes within image patches, achieving an overall accuracy (OA) of 0.92. Subsequently, SAM extracts plume boundaries from patches identified as plumes by the ViT. Integrated Mass Enhancement (IME) is then used to quantify emission rates based on the SAM-generated masks. This approach is applied to various known emission regions, including Turkmenistan, Spain, Algeria, Argentina, China, Iran, the United States, India, and Morocco, identifying significant CH 4 plumes with emission rates up to 92 t/h. The reported rates align closely with SRON values for the same dates and locations. While the ViT model requires supervised training, the SAM component operates without mask-specific training data, enabling automated and generalizable mask extraction. This study demonstrates the potential of combining advanced AI models with satellite data for effective CH 4 monitoring and environmental assessment.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.001 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".