Scholarlink

<p>Atmospheric humidity significantly influences air quality, thermal comfort, and public health outcomes, particularly in arid regions highly impacted by climate change. This study develops innovative environmental monitoring techniques through advanced prediction models for atmospheric humidity to support air quality management and health risk assessment in Kuwait City. Analysis of a 40-year dataset (1984–2023) comprising eleven atmospheric parameters revealed significant correlations between humidity variations and heat stress conditions. Multicollinearity assessment using Variance Inflation Factor, enabled exclusion of highly correlated variables (Tavg: VIF = 92.14; PS: VIF = 8.49), thereby enhancing model reliability. Four ensemble learning models– Random Forest, XGBoost, LightGBM, and GBR were applied to predict daily humidity levels. RF demonstrated superior performance (R<sup>2</sup> = 0.99329, RMSE = 1.31601) in capturing humidity dynamics. A novel environmental prediction framework comprising five HELMs was developed, with the optimal model (HELM-05) achieving enhanced accuracy (R<sup>2</sup> = 0.99364, RMSE = 1.28124). These technological improvements in humidity prediction accuracy enable better assessment of health risks associated with high humidity conditions, particularly during extreme heat events. The developed monitoring methodology provides an evidence-based tool for atmospheric monitoring and public health planning in arid regions, facilitating targeted interventions to reduce humidity-related health risks. This research contributes to enhanced understanding of environmental monitoring systems in arid environments and their implications for regional air quality management 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

Advanced hybrid ensemble learning models for atmospheric humidity prediction: an environmental monitoring solution in arid regions

  • 期刊论文

  • 发表日期：

    2026年1月10日

查看全文
  • A. A. Kafy,
  • Mst. T. Nowrin,
  • H. A. Altuwaijri,
  • M. M. T. Mukarram

摘要

Atmospheric humidity significantly influences air quality, thermal comfort, and public health outcomes, particularly in arid regions highly impacted by climate change. This study develops innovative environmental monitoring techniques through advanced prediction models for atmospheric humidity to support air quality management and health risk assessment in Kuwait City. Analysis of a 40-year dataset (1984–2023) comprising eleven atmospheric parameters revealed significant correlations between humidity variations and heat stress conditions. Multicollinearity assessment using Variance Inflation Factor, enabled exclusion of highly correlated variables (Tavg: VIF = 92.14; PS: VIF = 8.49), thereby enhancing model reliability. Four ensemble learning models– Random Forest, XGBoost, LightGBM, and GBR were applied to predict daily humidity levels. RF demonstrated superior performance (R2 = 0.99329, RMSE = 1.31601) in capturing humidity dynamics. A novel environmental prediction framework comprising five HELMs was developed, with the optimal model (HELM-05) achieving enhanced accuracy (R2 = 0.99364, RMSE = 1.28124). These technological improvements in humidity prediction accuracy enable better assessment of health risks associated with high humidity conditions, particularly during extreme heat events. The developed monitoring methodology provides an evidence-based tool for atmospheric monitoring and public health planning in arid regions, facilitating targeted interventions to reduce humidity-related health risks. This research contributes to enhanced understanding of environmental monitoring systems in arid environments and their implications for regional air quality management strategies.