Capitalizing the predictive potential of machine learning to detect various fire types using NASA's MODIS satellite data for the mediterranean basin

Lassem, Nima KamaliGaafar, Obai Mohamed Hisham AbdelmohsenAli, Seyid Amjad2024-03-072024-03-072024-01-229798400708985https://hdl.handle.net/11693/114393Conference Name: ICAAI '23: Proceedings of the 2023 7th International Conference on Advances in Artificial IntelligenceDate of Conference: 13 - 15 October 2023This study investigates the realm of machine learning for the classification of different fire types using NASA's FIRMS MODIS satellite data for the Mediterranean basin. Concentrating on the Mediterranean basin and utilizing data spanning from 2019 to 2021 for model training, XGBoost and Random Forest models were subsequently validated for the 2022 data. The findings distinctly illustrate XGBoost's superior predictive precision as compared to Random Forest by showcasing an impressive overall F1 score surpassing 95% and 84% macro F1 score across various fire types. This study emphasizes the prospect of machine learning to improve worldwide wildfire monitoring and response by providing exact, real-time fire type forecasts.enWildfire predictionMODISMediterranean basinXGBoostRandom ForestCapitalizing the predictive potential of machine learning to detect various fire types using NASA's MODIS satellite data for the mediterranean basinConference Paper10.1145/3633598.3633603