Climate-driven susceptibility of natural wildfires using Random Forest under future climate scenarios in Mediterranean forests of Türkiye
Frontiers in Forests and Global Change · 20.03.2026 · 9
DOI: 10.3389/ffgc.2026.1771857 ↗Kurz gefasst
Fachartikel in der Zeitschrift Frontiers in Forests and Global Change, online erschienen am 20.03.2026. Verfasst von Nuri Bozali. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 77 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Introduction
Mediterranean forest ecosystems are highly susceptible to natural wildfires under climate change, driven by rising temperatures, reduced precipitation, and pro¬longed dry periods. This study aimed to develop a climate-based natural wildfire susceptibility model using the Random Forest (RF) machine learning algorithm for the Emet Forest Management Directorate in the Mediterranean climate zone of Türkiye. This study considers only natural wildfires and excludes human-induced fire events.
Methods
All fire occurrence data used in the modeling process consist solely of recorded natural wildfire ignitions. The model incorporated 19 bioclimatic variables with historical wildfire occurrence data, using 258 recorded natural fire locations from 2015 to 2025 as reference points. Model performance was evaluated using Receiver Operating Characteristic (ROC) analysis. Results and discussion An Area Under the Curve (AUC) value of 0.711, which indicates moderate but acceptable predictive performance and is consistent with values reported in similar climate-driven susceptibility studies. The model results showed that the most influential drivers of fire susceptibility were temperature seasonality (BIO4), dry season precipitation (BIO17), and the minimum temperature of the coldest month (BIO6). According to future projections based on the Representative Concentration Pathways (RCP) 2.6 scenario using the Beijing Climate Center-Climate System Model Version 2-Medium Resolution (BCC-CSM2-MR) climate model for 2050 and 2070, the proportion of areas with high and extremely high fire susceptibility is projected to increase from 56.4% in 2025 to 64.0% in 2070, while low-and moderate-suscep¬tibility zones decline. This study provides one of the first climate-only, machine learning-based evaluations of present and future natural wildfire susceptibility in Mediterranean forests of Türkiye. These results revealed the escalating threat of wildfires in Mediterranean forests under climate change.
Zitieren
APA
Bozali, N. (2026). Climate-driven susceptibility of natural wildfires using Random Forest under future climate scenarios in Mediterranean forests of Türkiye. Frontiers in Forests and Global Change, 9. https://doi.org/10.3389/ffgc.2026.1771857
Vancouver
Bozali N. Climate-driven susceptibility of natural wildfires using Random Forest under future climate scenarios in Mediterranean forests of Türkiye. Frontiers in Forests and Global Change. 2026;9. doi:10.3389/ffgc.2026.1771857
BibTeX
@article{Bozali2026,
author = {Bozali, Nuri},
title = {Climate-driven susceptibility of natural wildfires using Random Forest under future climate scenarios in Mediterranean forests of Türkiye},
journal = {Frontiers in Forests and Global Change},
year = {2026},
volume = {9},
doi = {10.3389/ffgc.2026.1771857}
}
EUVIMED