A Comparative Study of a Deep Reinforcement Learning Solution and Alternative Deep Learning Models for Wildfire Prediction
Wildfires pose an escalating threat to ecosystems and human settlements, making accurate forecasting essential for early mitigation. This study compared three deep learning models for wildfire prediction: Deep Reinforcement Learning (DRL) with Actor–Critic architecture, Convolutional Neural Network...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2025-04-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/15/7/3990 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849738579311329280 |
|---|---|
| author | Cristian Vidal-Silva Roberto Pizarro Miguel Castillo-Soto Ben Ingram Claudia de la Fuente Vannessa Duarte Claudia Sangüesa Alfredo Ibañez |
| author_facet | Cristian Vidal-Silva Roberto Pizarro Miguel Castillo-Soto Ben Ingram Claudia de la Fuente Vannessa Duarte Claudia Sangüesa Alfredo Ibañez |
| author_sort | Cristian Vidal-Silva |
| collection | DOAJ |
| description | Wildfires pose an escalating threat to ecosystems and human settlements, making accurate forecasting essential for early mitigation. This study compared three deep learning models for wildfire prediction: Deep Reinforcement Learning (DRL) with Actor–Critic architecture, Convolutional Neural Network (CNN), and Transformer-based models. The models were trained and evaluated using historical data from Chile (2000–2023), including wildfire occurrences, meteorological variables, topography, and vegetation indices. After preprocessing and class balancing, each model was tested over 100 experimental runs. All models achieved outstanding performance, with F1-Scores exceeding 0.999 and perfect AUC-ROC scores. The Transformer model showed a slight advantage over the CNN (99.94%) and Actor–Critic DRL (99.93%) in accuracy. Feature importance analysis identified wind speed, temperature, and vegetation indices as the most influential variables. While DRL offers theoretical benefits for adaptive decision-making, Transformer architectures more effectively capture spatiotemporal dependencies in wildfire dynamics. The findings can support the integration of deep learning models into early warning systems, contributing to proactive wildfire risk management. Future work will include validation with diverse regional datasets, real-time deployment, and collaboration with emergency response agencies. |
| format | Article |
| id | doaj-art-9aa11b17318d45ffa615f6fe253b4fbd |
| institution | DOAJ |
| issn | 2076-3417 |
| language | English |
| publishDate | 2025-04-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj-art-9aa11b17318d45ffa615f6fe253b4fbd2025-08-20T03:06:31ZengMDPI AGApplied Sciences2076-34172025-04-01157399010.3390/app15073990A Comparative Study of a Deep Reinforcement Learning Solution and Alternative Deep Learning Models for Wildfire PredictionCristian Vidal-Silva0Roberto Pizarro1Miguel Castillo-Soto2Ben Ingram3Claudia de la Fuente4Vannessa Duarte5Claudia Sangüesa6Alfredo Ibañez7Departamento de Visualización Interactiva y Realidad Virtual, Facultad de Ingeniería, Universidad de Talca, Talca 3460000, ChileCátedra Unesco en Hidrología de Superficie, Universidad de Talca, Talca 3467769, ChileForest Fire Laboratory, University of Chile, Santiago 8820808, ChileDepartamento de Visualización Interactiva y Realidad Virtual, Facultad de Ingeniería, Universidad de Talca, Talca 3460000, ChileDepartamento de Visualización Interactiva y Realidad Virtual, Facultad de Ingeniería, Universidad de Talca, Talca 3460000, ChileEscuela de Ciencias Empresariales, Universidad Católica del Norte, Larrondo 1280, Coquimbo 1781421, ChileCátedra Unesco en Hidrología de Superficie, Universidad de Talca, Talca 3467769, ChileCátedra Unesco en Hidrología de Superficie, Universidad de Talca, Talca 3467769, ChileWildfires pose an escalating threat to ecosystems and human settlements, making accurate forecasting essential for early mitigation. This study compared three deep learning models for wildfire prediction: Deep Reinforcement Learning (DRL) with Actor–Critic architecture, Convolutional Neural Network (CNN), and Transformer-based models. The models were trained and evaluated using historical data from Chile (2000–2023), including wildfire occurrences, meteorological variables, topography, and vegetation indices. After preprocessing and class balancing, each model was tested over 100 experimental runs. All models achieved outstanding performance, with F1-Scores exceeding 0.999 and perfect AUC-ROC scores. The Transformer model showed a slight advantage over the CNN (99.94%) and Actor–Critic DRL (99.93%) in accuracy. Feature importance analysis identified wind speed, temperature, and vegetation indices as the most influential variables. While DRL offers theoretical benefits for adaptive decision-making, Transformer architectures more effectively capture spatiotemporal dependencies in wildfire dynamics. The findings can support the integration of deep learning models into early warning systems, contributing to proactive wildfire risk management. Future work will include validation with diverse regional datasets, real-time deployment, and collaboration with emergency response agencies.https://www.mdpi.com/2076-3417/15/7/3990wildfire predictiondeep reinforcement learningactor–criticmachine learningconvolutional neural networkstransformer models |
| spellingShingle | Cristian Vidal-Silva Roberto Pizarro Miguel Castillo-Soto Ben Ingram Claudia de la Fuente Vannessa Duarte Claudia Sangüesa Alfredo Ibañez A Comparative Study of a Deep Reinforcement Learning Solution and Alternative Deep Learning Models for Wildfire Prediction Applied Sciences wildfire prediction deep reinforcement learning actor–critic machine learning convolutional neural networks transformer models |
| title | A Comparative Study of a Deep Reinforcement Learning Solution and Alternative Deep Learning Models for Wildfire Prediction |
| title_full | A Comparative Study of a Deep Reinforcement Learning Solution and Alternative Deep Learning Models for Wildfire Prediction |
| title_fullStr | A Comparative Study of a Deep Reinforcement Learning Solution and Alternative Deep Learning Models for Wildfire Prediction |
| title_full_unstemmed | A Comparative Study of a Deep Reinforcement Learning Solution and Alternative Deep Learning Models for Wildfire Prediction |
| title_short | A Comparative Study of a Deep Reinforcement Learning Solution and Alternative Deep Learning Models for Wildfire Prediction |
| title_sort | comparative study of a deep reinforcement learning solution and alternative deep learning models for wildfire prediction |
| topic | wildfire prediction deep reinforcement learning actor–critic machine learning convolutional neural networks transformer models |
| url | https://www.mdpi.com/2076-3417/15/7/3990 |
| work_keys_str_mv | AT cristianvidalsilva acomparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT robertopizarro acomparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT miguelcastillosoto acomparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT beningram acomparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT claudiadelafuente acomparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT vannessaduarte acomparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT claudiasanguesa acomparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT alfredoibanez acomparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT cristianvidalsilva comparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT robertopizarro comparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT miguelcastillosoto comparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT beningram comparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT claudiadelafuente comparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT vannessaduarte comparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT claudiasanguesa comparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction AT alfredoibanez comparativestudyofadeepreinforcementlearningsolutionandalternativedeeplearningmodelsforwildfireprediction |