Attention to Fires: Multi-Channel Deep Learning Models for Wildfire Severity Prediction
Wildfires are one of the natural hazards that the European Union is actively monitoring through the Copernicus EMS Earth observation program which continuously releases public information related to such catastrophic events. Such occurrences are the cause of both short- and long-term damages. Thus,...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-11-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-3417/11/22/11060 |
| _version_ | 1797511302239748096 |
|---|---|
| author | Simone Monaco Salvatore Greco Alessandro Farasin Luca Colomba Daniele Apiletti Paolo Garza Tania Cerquitelli Elena Baralis |
| author_facet | Simone Monaco Salvatore Greco Alessandro Farasin Luca Colomba Daniele Apiletti Paolo Garza Tania Cerquitelli Elena Baralis |
| author_sort | Simone Monaco |
| collection | DOAJ |
| description | Wildfires are one of the natural hazards that the European Union is actively monitoring through the Copernicus EMS Earth observation program which continuously releases public information related to such catastrophic events. Such occurrences are the cause of both short- and long-term damages. Thus, to limit their impact and plan the restoration process, a rapid intervention by authorities is needed, which can be enhanced by the use of satellite imagery and automatic burned area delineation methodologies, accelerating the response and the decision-making processes. In this context, we analyze the burned area severity estimation problem by exploiting a state-of-the-art deep learning framework. Experimental results compare different model architectures and loss functions on a very large real-world Sentinel2 satellite dataset. Furthermore, a novel multi-channel attention-based analysis is presented to uncover the prediction behaviour and provide model interpretability. A perturbation mechanism is applied to an attention-based DS-UNet to evaluate the contribution of different domain-driven groups of channels to the severity estimation problem. |
| first_indexed | 2024-03-10T05:43:27Z |
| format | Article |
| id | doaj.art-ab07aa6e9ce145faab961e7300fb4986 |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-03-10T05:43:27Z |
| publishDate | 2021-11-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-ab07aa6e9ce145faab961e7300fb49862023-11-22T22:23:03ZengMDPI AGApplied Sciences2076-34172021-11-0111221106010.3390/app112211060Attention to Fires: Multi-Channel Deep Learning Models for Wildfire Severity PredictionSimone Monaco0Salvatore Greco1Alessandro Farasin2Luca Colomba3Daniele Apiletti4Paolo Garza5Tania Cerquitelli6Elena Baralis7Department of Control and Computer Engineering (DAUIN), Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129 Torino, ItalyDepartment of Control and Computer Engineering (DAUIN), Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129 Torino, ItalyDepartment of Control and Computer Engineering (DAUIN), Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129 Torino, ItalyDepartment of Control and Computer Engineering (DAUIN), Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129 Torino, ItalyDepartment of Control and Computer Engineering (DAUIN), Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129 Torino, ItalyDepartment of Control and Computer Engineering (DAUIN), Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129 Torino, ItalyDepartment of Control and Computer Engineering (DAUIN), Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129 Torino, ItalyDepartment of Control and Computer Engineering (DAUIN), Politecnico di Torino, Corso Duca degli Abruzzi, 24, 10129 Torino, ItalyWildfires are one of the natural hazards that the European Union is actively monitoring through the Copernicus EMS Earth observation program which continuously releases public information related to such catastrophic events. Such occurrences are the cause of both short- and long-term damages. Thus, to limit their impact and plan the restoration process, a rapid intervention by authorities is needed, which can be enhanced by the use of satellite imagery and automatic burned area delineation methodologies, accelerating the response and the decision-making processes. In this context, we analyze the burned area severity estimation problem by exploiting a state-of-the-art deep learning framework. Experimental results compare different model architectures and loss functions on a very large real-world Sentinel2 satellite dataset. Furthermore, a novel multi-channel attention-based analysis is presented to uncover the prediction behaviour and provide model interpretability. A perturbation mechanism is applied to an attention-based DS-UNet to evaluate the contribution of different domain-driven groups of channels to the severity estimation problem.https://www.mdpi.com/2076-3417/11/22/11060wildfire severity predictiondeep neural networksmulti-channel attention-based analysis |
| spellingShingle | Simone Monaco Salvatore Greco Alessandro Farasin Luca Colomba Daniele Apiletti Paolo Garza Tania Cerquitelli Elena Baralis Attention to Fires: Multi-Channel Deep Learning Models for Wildfire Severity Prediction Applied Sciences wildfire severity prediction deep neural networks multi-channel attention-based analysis |
| title | Attention to Fires: Multi-Channel Deep Learning Models for Wildfire Severity Prediction |
| title_full | Attention to Fires: Multi-Channel Deep Learning Models for Wildfire Severity Prediction |
| title_fullStr | Attention to Fires: Multi-Channel Deep Learning Models for Wildfire Severity Prediction |
| title_full_unstemmed | Attention to Fires: Multi-Channel Deep Learning Models for Wildfire Severity Prediction |
| title_short | Attention to Fires: Multi-Channel Deep Learning Models for Wildfire Severity Prediction |
| title_sort | attention to fires multi channel deep learning models for wildfire severity prediction |
| topic | wildfire severity prediction deep neural networks multi-channel attention-based analysis |
| url | https://www.mdpi.com/2076-3417/11/22/11060 |
| work_keys_str_mv | AT simonemonaco attentiontofiresmultichanneldeeplearningmodelsforwildfireseverityprediction AT salvatoregreco attentiontofiresmultichanneldeeplearningmodelsforwildfireseverityprediction AT alessandrofarasin attentiontofiresmultichanneldeeplearningmodelsforwildfireseverityprediction AT lucacolomba attentiontofiresmultichanneldeeplearningmodelsforwildfireseverityprediction AT danieleapiletti attentiontofiresmultichanneldeeplearningmodelsforwildfireseverityprediction AT paologarza attentiontofiresmultichanneldeeplearningmodelsforwildfireseverityprediction AT taniacerquitelli attentiontofiresmultichanneldeeplearningmodelsforwildfireseverityprediction AT elenabaralis attentiontofiresmultichanneldeeplearningmodelsforwildfireseverityprediction |