Scholarlink

<p>The intensification of global climate change has led to more frequent forest fires, making accurate prediction of fire occurrences critical for prevention and mitigation. This study utilizes forest fire occurrence data in Fujian Province from 2012 to 2020, combined with meteorological, vegetation, and topographic factors, to construct predictive models aimed at exploring the spatial and spatio-temporal characteristics of fire-driving factors. We compared the performance of the global Random Forest (RF), Geographically Weighted Random Forest (GWRF), and Spatio-Temporal Weighted Random Forest (STWRF) models. The results demonstrate that the STWRF model achieved the best performance, with superior fitting results and higher predictive accuracy. Elevation was identified as a key factor influencing forest fire occurrence in Fujian Province, where fires are most frequent from December to April in the central and northern regions, shifting to the central and southern regions during May to November. By incorporating spatio-temporal weights, the STWRF model enhances predictive performance and provides a solid theoretical foundation for policymakers to implement targeted forest fire prevention measures in response to spatio-temporal variations.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Spatio-temporal considerations enhance forest fire prediction in Fujian Province

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月23日

查看全文
  • Rongyu Ni,
  • Zhangwen Su,
  • Wei Guo,
  • Man Huang,
  • Jinwen Zhang,
  • Wenlong Wang,
  • Futao Guo

摘要

The intensification of global climate change has led to more frequent forest fires, making accurate prediction of fire occurrences critical for prevention and mitigation. This study utilizes forest fire occurrence data in Fujian Province from 2012 to 2020, combined with meteorological, vegetation, and topographic factors, to construct predictive models aimed at exploring the spatial and spatio-temporal characteristics of fire-driving factors. We compared the performance of the global Random Forest (RF), Geographically Weighted Random Forest (GWRF), and Spatio-Temporal Weighted Random Forest (STWRF) models. The results demonstrate that the STWRF model achieved the best performance, with superior fitting results and higher predictive accuracy. Elevation was identified as a key factor influencing forest fire occurrence in Fujian Province, where fires are most frequent from December to April in the central and northern regions, shifting to the central and southern regions during May to November. By incorporating spatio-temporal weights, the STWRF model enhances predictive performance and provides a solid theoretical foundation for policymakers to implement targeted forest fire prevention measures in response to spatio-temporal variations.