Scholarlink

<p>Wildfires are a critical ecological disturbance that carries significant implications for biodiversity, the carbon cycle, and climate regulation. The boreal forests of Daxing’anling, Northeast China, have received particular attention in terms of wildfire prevention, yet remain susceptible to fire regime shifts under a changing climate. This study focuses on the influence of anthropogenic climate change on fire risk in the region by projecting the Canadian Fire Weather Index (FWI) under two climate scenarios (SSP245 and SSP585) using 18 CMIP6 Earth System Models (ESMs). Two multi-model ensemble (MME) approaches were compared: an arithmetic mean-based ensemble (MME-AM) and a Random Forest-based ensemble (MME-RF). Prior to the generation of ensembles, models were ranked based on their historical simulation performances, then the eight best ranked models were selected and bias-corrected. Results show that MME-RF outperforms MME-AM in reducing mean FWI bias, and aligns more closely with reference data. Under the SSP585 scenario, the area-averaged FWI is projected to increase by 57.1% by 2071–2100, with a signal particularly strong during summer. A scenario-induced post mid-century shift in annual FWI trajectories is found, with a continuous increase under SSP585, and a plateau under SSP245 until the end of the century. These findings underscore the importance of adaptive fire management to address ongoing seasonal trends in fire danger, and of the promotion of a lower emission scenario through climate mitigation strategies.</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

Projection of fire weather index in Daxing’anling, Northeast China using a machine learning-based CMIP6 multi-model ensemble

  • 期刊论文

  • 发表日期：

    2026年2月23日

查看全文
  • Stephane G. P. Debulois,
  • Min Sun,
  • Mingshi Li

摘要

Wildfires are a critical ecological disturbance that carries significant implications for biodiversity, the carbon cycle, and climate regulation. The boreal forests of Daxing’anling, Northeast China, have received particular attention in terms of wildfire prevention, yet remain susceptible to fire regime shifts under a changing climate. This study focuses on the influence of anthropogenic climate change on fire risk in the region by projecting the Canadian Fire Weather Index (FWI) under two climate scenarios (SSP245 and SSP585) using 18 CMIP6 Earth System Models (ESMs). Two multi-model ensemble (MME) approaches were compared: an arithmetic mean-based ensemble (MME-AM) and a Random Forest-based ensemble (MME-RF). Prior to the generation of ensembles, models were ranked based on their historical simulation performances, then the eight best ranked models were selected and bias-corrected. Results show that MME-RF outperforms MME-AM in reducing mean FWI bias, and aligns more closely with reference data. Under the SSP585 scenario, the area-averaged FWI is projected to increase by 57.1% by 2071–2100, with a signal particularly strong during summer. A scenario-induced post mid-century shift in annual FWI trajectories is found, with a continuous increase under SSP585, and a plateau under SSP245 until the end of the century. These findings underscore the importance of adaptive fire management to address ongoing seasonal trends in fire danger, and of the promotion of a lower emission scenario through climate mitigation strategies.