Assessing forest fire risks using geo-spatial techniques in western Nepal

Abstract

Forest fires are a significant threat to Nepal’s ecological and socio-economic stability, particularly in vulnerable regions. Despite increasing fire frequency and intensity, research on spatially explicit risk assessment in Nepal remains limited; therefore, this study maps forest fire risks, identifies key contributing factors, and proposes actionable management strategies for Lumbini Province. Fire incidents and burned areas (2012–2024) were derived from MODIS and VIIRS datasets, while topographic, climatic, biophysical, and anthropogenic data were sourced from open-access repositories. Primary data from key informant interviews revealed causes, impacts, challenges, and preventive measures. Spatial layers were processed using GIS/RS techniques and integrated using a Weighted Linear Combination (WLC) model to create a forest fire risk map, categorising the region into five risk levels. The WLC model identified land cover (20%), road proximity (15%), settlement proximity (15%), NDVI (15%), and temperature (10%) as the dominant drivers, revealing that 57.42 per cent of forests are in very-high to high-risk zones. The findings revealed fires peak in the pre-monsoon season (March-May), with April accounting for 76.85 per cent of occurrences. High-risk areas feature broad-leaved closed forests, temperatures >40°C, gentle slopes <15%, south-facing aspects, and proximity to humans. The risk model correctly predicted 78.22 per cent of historical fire incidents within the identified high-to-very-high risk zones, supported by a confusion matrix (Kappa: 0.94) and ROC curve analysis (AUC: 0.803). Furthermore, perception analysis of forest managers emphasised the urgent need for proactive measures, including increased funding, specialised training, and the development of early warning systems to support sustainable forest management.