Research Article
Wildfire Risk Mapping Based on Multi-source Data and Machine Learning
@INPROCEEDINGS{10.1007/978-3-031-57523-5_9, author={Ghinevra Comiti and Paul-Antoine Bisgambiglia and Paul Bisgambiglia}, title={Wildfire Risk Mapping Based on Multi-source Data and Machine Learning}, proceedings={Simulation Tools and Techniques. 15th EAI International Conference, SIMUtools 2023, Seville, Spain, December 14-15, 2023, Proceedings}, proceedings_a={SIMUTOOLS}, year={2024}, month={4}, keywords={Machine Learning Wildfire Map}, doi={10.1007/978-3-031-57523-5_9} }- Ghinevra Comiti
Paul-Antoine Bisgambiglia
Paul Bisgambiglia
Year: 2024
Wildfire Risk Mapping Based on Multi-source Data and Machine Learning
SIMUTOOLS
Springer
DOI: 10.1007/978-3-031-57523-5_9
Abstract
The management and prevention of forest fires are crucial in fire-prone regions such as Corsica, a French island in the Mediterranean. In this study, an approach to mapping wildfire vulnerability is presented using different data sources, including meteorological, temporal, geographical and economic datasets. These heterogeneous datasets are seamlessly integrated to produce a comprehensive forest fire vulnerability map for Corsica. The methodology involves the collection and pre-processing of a variety of data, such as historical forest fire events, meteorological variables, land cover data, socio-economic indicators and temporal factors. Machine learning models are used to visualise the complex relationships between these variables and predict wildfire susceptibility. Finally, we were able to create a daily fire susceptibility map for the island of Corsica.
