Data-Driven Wildfire Spread Modelling Of European Wildfires
Bibliographic record
Abstract
Human-induced climate change is causing wildfires to intensify and become more frequent across the globe, as evidenced by recent extreme events in Greece (2023), Canada (2023), and Chile (2024). To effectively manage these risks, wildfire spread models play a crucial role in planning timely suppression efforts. Historically, wildfire spread models have been developed using semi-empirical approximations based on experimental burnings. Although used in an operational context, such models suffer from inaccuracies and transferability issues outside of their development region. Recent advances in the availability of remote sensing data, artificial intelligence, and computational resources allow for a new data-driven perspective on wildfire spread modelling that offers the opportunity to overcome the limitations of established semi-empirical models. We developed a novel data-driven wildfire spread modelling approach using a Spatio-Temporal Graph Neural Network (STGNN) trained on the historic burned area time series of European wildfires retrieved from Copernicus Sentinel-3 imagery. A training dataset was built by populating individual burned area perimeters with dynamic (e.g. meteorological data, Fire Weather Index (FWI), hotspots) and static (e.g. fuel map, land cover, topography) auxiliary datasets in a discrete, hexagonal grid system (H3), which allows to include neighbourhood relationships into the dataset. Each wildfire time series was then transformed into a spatio-temporal graph representation which formed the input for the model. The STGNN can simultaneously process and learn the spatial and temporal dependencies in the data by combining a Graph Convolutional Network (GCN) with a Gated Recurrent Unit (GRU). The model was iteratively trained on each time step of individual wildfire time series and can predict the next day’s burned area. Testing was done by feeding the first day of an unseen wildfire time series and predicting the wildfire’s burned area on the four following days. Validation was achieved by calculating the weighted macro-mean Jaccard Index (IoU) between the predicted daily burned area and the Sentinel-3 reference burned area. A first model was developed on Portuguese wildfires to test the ability of the STGNN to capture the spatial and temporal evolution of wildfires. To assess the generalization ability of the wildfire spread model, the STGNN was then trained and tested with Mediterranean wildfire time series from different countries with varying environmental conditions.
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 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 | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.006 |
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; both teacher heads agree on what is shown here.
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".