Mapping wildfire susceptibility in the tropical region of Brunei: a machine learning and explainable AI approach using google earth engine with remote sensing data

Download PDF
Open Webpage

Rufai Yusuf Zakari, Owais Ahmed Malik, Wee-Hong Ong

Published: 2025, Last Modified: 06 Nov 2025Earth Sci. Informatics 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: As wildfires escalate globally, the need for accurate and interpretable predictions of their occurrence has never been more pressing. While Brunei experiences wildfires less frequently, their potential to disrupt ecosystems, harm biodiversity, and impact socio-economic activities highlights the need for proactive management strategies. In this study, we utilized Google Earth Engine (GEE) and its freely available remote sensing image collection data spanning from 2019 – 2024 to map wildfire susceptibility in Brunei. We assessed the performance of four machine learning (ML) models including Random Forest (RF), Extreme Gradient Boosting (XGBoost), Adaptive Boosting (AdaBoost), and Support Vector Machines (SVMs) for mapping wildfire susceptibility. These models were evaluated using thirteen independent factors related to wildfires, grouped into four main categories: climate, environment, topography, and human impact conditions. A multicollinearity analysis was performed to examine the interrelationships among these factors. According to the Performance evaluation outcomes, the predictive capabilities of the ML algorithms were compared using different performance metrics. The results demonstrated that the RF model performed best, achieving an F1 score of 0.900, a precision of 0.845, an accuracy of 0.854, and an ROC of 0.959. Furthermore, SHapley Additive exPlanations (SHAP) analysis was applied to interpret the model’s learning process, identifying distance from road, population density, elevation and temperature as the most influential factors. Our findings revealed that in Brunei, peat swamp forests and non-forested areas, such as grasslands and urban regions near roads, are the most wildfire-susceptible zones, driven by vegetation density and proximity to human activity. This study provides insights for targeted wildfire management and resource allocation in Brunei.