Scholarlink

<p>In the past two decades, California’s unprecedented wildfire activity has been driven by the combined effects of climate variability and human influence. This study investigates the sensitivity of number of fires (NF) and burned area (BA) to environmental and meteorological variability using machine learning and regression analysis. Monthly datasets (2000–2025) from the seven most severe wildfire years (2003, 2008, 2017, 2018, 2020, 2021, and 2025), encompassing 12 key climatic parameters, were analyzed to quantify environmental and climatic controls on NF and BA. Generalized additive model (GAM) and random forest (RF) analyses revealed that NF exhibited considerably stronger climatic predictability (R² = 0.73 and 0.74, respectively) than BA (R² = 0.27 and 0.36). Temperature and rainfall emerged as dominant predictors of NF and BA, respectively, with variable importance analyses ranking soil moisture as a secondary yet critical driver. RF modeling revealed a nonlinear escalation of fire activity above ~ 30&#xa0;°C and a rainfall threshold, with precipitation &gt; 1&#xa0;mm markedly reducing ignition and spread. Soil moisture and solar radiation modulated BA, while NF was predominantly driven by temperature and soil moisture. Spearman correlation analysis showed that wildfire activity is influenced by lagged climatic effects, with temperature positively (r = + 0.62) and precipitation negatively (<i>r</i> = -0.58) correlated, while vegetation indices and PDSI (r = + 0.41–0.47) indicate persistent drought and fuel effects. Validation showed higher accuracy for NF, emphasizing its stronger environmental control and greater predictability relative to BA. Overall, the findings highlight the nonlinear climate-wildfire dynamics and the dominant role of temperature and precipitation in driving extreme events in California.</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

Role of environmental variables and meteorology on California’s largest wildfires

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月2日

查看全文
  • Md. Tushar Ali,
  • Qauzi Hamidul Bari,
  • Jobaer Ahmed Saju

摘要

In the past two decades, California’s unprecedented wildfire activity has been driven by the combined effects of climate variability and human influence. This study investigates the sensitivity of number of fires (NF) and burned area (BA) to environmental and meteorological variability using machine learning and regression analysis. Monthly datasets (2000–2025) from the seven most severe wildfire years (2003, 2008, 2017, 2018, 2020, 2021, and 2025), encompassing 12 key climatic parameters, were analyzed to quantify environmental and climatic controls on NF and BA. Generalized additive model (GAM) and random forest (RF) analyses revealed that NF exhibited considerably stronger climatic predictability (R² = 0.73 and 0.74, respectively) than BA (R² = 0.27 and 0.36). Temperature and rainfall emerged as dominant predictors of NF and BA, respectively, with variable importance analyses ranking soil moisture as a secondary yet critical driver. RF modeling revealed a nonlinear escalation of fire activity above ~ 30 °C and a rainfall threshold, with precipitation > 1 mm markedly reducing ignition and spread. Soil moisture and solar radiation modulated BA, while NF was predominantly driven by temperature and soil moisture. Spearman correlation analysis showed that wildfire activity is influenced by lagged climatic effects, with temperature positively (r = + 0.62) and precipitation negatively (r = -0.58) correlated, while vegetation indices and PDSI (r = + 0.41–0.47) indicate persistent drought and fuel effects. Validation showed higher accuracy for NF, emphasizing its stronger environmental control and greater predictability relative to BA. Overall, the findings highlight the nonlinear climate-wildfire dynamics and the dominant role of temperature and precipitation in driving extreme events in California.