Multi-sensor near-realtime burnt area monitoring using a superpixel-based graph convolutional network approach
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
Recent disastrous wildfire seasons highlight the urgent need for timely and accurate wildfire data to support relief efforts, to monitor the environmental impacts and to inform the public. While satellite-based thermal anomaly data is available in near real-time (NRT), deriving actual fire-affected areas from NRT imagery remains challenging. The proposed methodology combines a superpixel segmentation algorithm with rule-based and deep learning classification techniques to accurately derive burnt areas (BA) in NRT. This approach supports a range of mid- to high-resolution optical sensors and fuses data from diverse sources to continuously refine the burnt area during the monitoring of active fires. The NRT (DLRBAv2NRT) and the refined non-time critical (DLRBAv2NTC) BA product based on mid-resolution Sentinel-3 imagery were produced and tested against established global BA products for wildfire seasons in Greece 2023, British Columbia (Canada) 2023, and Central Chile 2023/2024. DLRBAv2NTC classified BA with the highest accuracies over all study regions (avg. IoU: 0.71; avg. F1-Score: 0.83). Despite its NRT processing capability, the DLRBAv2NRT achieved comparable accuracies (avg. IoU: 0.69; avg. F1-Score: 0.81) and could outperform the well-established and widely used global NASA burnt area product MCD64A1v061 by +2% (IoU) and +1% (F1-Score). Furthermore, the multi-sensor and fusion capability of the methodology was successfully demonstrated for the 2024 Valparaiso fire in Chile. The proposed mapping procedure demonstrates a fully-automated and flexible approach to derive burnt area delineations from satellite data in NRT with high accuracy. This allows for high-frequency monitoring of NRT burnt areas on a global scale.
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.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.002 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
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 it