<![CDATA[Abstract. Thoha AS, Widyastuti DE, Fadhilla S, Hutahaean MI, Pasha SD. 2026. Analysis of forest and land fire vulnerability in the Toba Caldera UNESCO Global Geopark of North Sumatra, Indonesia. Biodiversitas 27 (8): d270815. https://doi.org/10.13057/biodiv/d270815. Forest and land fires in North Sumatra Province, Indonesia, are threatening a global tourist destination: the Toba Caldera UNESCO Global Geopark (Toba Caldera UGGp). The objective of this study is to analyze the spatial distribution of forest and land fire vulnerability in the Toba Caldera UGGp, North Sumatra Province, Indonesia. Spatial analysis was conducted using the Composite Mapping Analysis (CMA) method. Hotspots recorded during 2015-2024 were integrated with 2023 land-cover, population density, road, river, and village-center data. Factor-class scores were calculated from observed and expected hotspot frequencies, rescaled, and weighted using stepwise linear regression. Model agreement was assessed using 66 purposively selected field-survey points. Four variables were retained in the final model: distance from rivers, land cover, distance from roads, and distance from village centers, with weights of 36.7%, 24.5%, 19.5%, and 19.3%, respectively. High- and very-high-vulnerability zones covered 79,316.50 ha and 56,905.52 ha, respectively, representing 51.04% of the geopark. Samosir District contained the largest concentrations of very-high-vulnerability areas, particularly in Sianjur Mula Mula and Harian Sub-districts. The three-class map achieved 92.42% overall accuracy with a kappa coefficient of 0.751, whereas the five-class map achieved 93.94% overall accuracy with a kappa coefficient of 0.905. However, these values should be interpreted cautiously because validation points were purposively selected and unevenly distributed among classes. The resulting maps can support targeted patrols, community-based fire prevention, restoration, and geosite protection, particularly near rivers, settlements, shrublands, and dryland agriculture. Independent spatial or temporal validation is recommended before operational adoption.]]>
