Wang et al. (2025) Extreme Climate Drivers and Their Interactions in Lightning-Ignited Fires: Insights from Machine Learning Models

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Forests
• Year: 2025
• Date: 2025-12-16
• Authors: Yu Wang, Yingda Wu, Huanjia Cui, Yilin Liu, Maolin Li, Xinyu Yang, Jikai Zhao, Qiang Yu
• DOI: 10.3390/f16121861

Research Groups

Not specified in the provided text.

Short Summary

This study integrates extreme climate indices with meteorological, vegetation, soil, and topographic data using machine learning to develop probabilistic models for lightning fire occurrence. The approach significantly improves prediction accuracy (up to 87.4%) over traditional methods, identifying extreme temperature and precipitation indices as key drivers and offering an interpretable framework for risk assessment.

Objective

• To integrate extreme climate indices with meteorological, vegetation, soil, and topographic data using machine learning methods to build probabilistic models for lightning fire occurrence, thereby improving short-term lightning fire prediction and providing quantitative support for risk warning and resource allocation in a warming climate.

Study Configuration

• Spatial Scale: Not specified in the provided text.
• Temporal Scale: Short-term prediction of lightning fire occurrence.

Methodology and Data

• Models used: Four machine learning methods, including XGBoost (highlighted as the best performer). Model interpretation used SHapley Additive exPlanations (SHAP).
• Data sources: Extreme climate indices, meteorological data, vegetation data, soil data, and topographic data.

Main Results

• Incorporating extreme climate indices significantly improved model performance for lightning fire occurrence prediction.
• XGBoost achieved the highest accuracy (87.4%) and AUC (0.903), substantially outperforming traditional fire weather indices (accuracy 60%–71%).
• Extreme temperature and precipitation indices contributed nearly 60% to fire occurrence.
• Key driving factors identified include growing season length (GSL), minimum of daily maximum temperature (TXn), diurnal temperature range (DTR), and warm spell duration index (WSDI).
• Heavy precipitation indices exerted a suppressing effect on fire occurrence.
• Compound hot and dry conditions amplified fuel aridity and markedly increased ignition probability.

Contributions

• Developed an interpretable framework for short-term lightning fire prediction by integrating extreme climate indices with machine learning.
• Demonstrated a significant improvement in lightning fire prediction accuracy compared to traditional fire weather indices.
• Provided quantitative insights into the driving mechanisms and interaction effects of extreme climate factors on lightning fire occurrence using SHAP.
• Offers an improved tool for risk warning and resource allocation in the context of increasing lightning fire frequency under climate change.

Funding

Not specified in the provided text.

Citation

@article{Wang2025Extreme,
  author = {Wang, Yu and Wu, Yingda and Cui, Huanjia and Liu, Yilin and Li, Maolin and Yang, Xinyu and Zhao, Jikai and Yu, Qiang},
  title = {Extreme Climate Drivers and Their Interactions in Lightning-Ignited Fires: Insights from Machine Learning Models},
  journal = {Forests},
  year = {2025},
  doi = {10.3390/f16121861},
  url = {https://doi.org/10.3390/f16121861}
}