Automated Classification of Thermal Hot Spot Types from the VIIRS Satellite Data Using Machine Learning Techniques

León M.-C.; Torres-Gil D.; Ccompi W.; Ticona W. · Lecture Notes in Networks and Systems (2026) · DOI: 10.1007/978-3-032-20749-4_16

Resumen

Thermal hotspots such as wildfires, agricultural burns, and industrial sources have increased significantly in Peru, creating major environmental and health concerns. This study presents an automated classification model using machine learning techniques applied to VIIRS satellite data. The methodology included five phases: data acquisition (over 409,000 records from 2019–2023), preprocessing (data cleaning, transformation, outlier treatment, normalization, and class balancing with SMOTE), feature selection (using Boruta, SHAP, and LightGBM), application of machine learning models (Random Forest, XGBoost, CatBoost, LightGBM, MLP, Naive Bayes, and Gradient Boosting), and evalua-tion with metrics such as accuracy, precision, recall, F1-score, ROC curve, and confusion matrix. Among all models, Random Forest achieved the best results with over 99.8% in all key metrics and an AUC of 1.00. The study demonstrates that machine learning models trained on satellite data can effectively classify thermal hotspot types in Peru, providing a powerful tool for early detection and response to fire-related emergencies. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Instituciones: Universidad ESAN; Universidad Tecnológica del Perú