Scholarlink

<p>Fine particulate matter (PM<sub>2.5</sub> and PM<sub>10</sub>) continues to present serious health hazards in fast‑growing urban centers such as Hyderabad. To address the pressing need for dependable forecasting systems that can inform both policy and public health responses, this study proposes a comprehensive evaluation framework tailored to wavelet–Gaussian process hybrid models. Six distinct formulations (IWG, GPWG, FAWG, IWA, FWA, and GPWA) were assessed using daily air quality records from Hyderabad. The framework integrates multiple dimensions, predictive accuracy, computational efficiency, statistical stability, and compliance with regulatory indicators, ensuring that methodological advances translate into practical utility. Among the tested approaches, the GPWA model consistently delivered superior outcomes, with a mean error of 0.737, a standard deviation of 0.843, and strong correlation with observed pollutant concentrations. It also achieved the most reliable episode detection, outperforming peers across False Alarm Ratio (FAR), Critical Success Index (CSI), and Probability of Detection (POD). Beyond conventional accuracy measures, the framework highlights the balance between precision and resource demands, while demonstrating the added benefits of ensemble learning. These findings confirm the promise of advanced hybrid wavelet models in strengthening urban air quality forecasting and providing actionable guidance for pollution mitigation and public health protection.</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

Novel quantitative framework for evaluating hybrid models in urban air quality prediction: a case study of Hyderabad

  • 期刊论文

  • 发表日期：

    2026年6月23日

查看全文
  • Guhan Velusamy,
  • Dharma Raju Akasapu,
  • Nagaratna Kopparthi

摘要

Fine particulate matter (PM2.5 and PM10) continues to present serious health hazards in fast‑growing urban centers such as Hyderabad. To address the pressing need for dependable forecasting systems that can inform both policy and public health responses, this study proposes a comprehensive evaluation framework tailored to wavelet–Gaussian process hybrid models. Six distinct formulations (IWG, GPWG, FAWG, IWA, FWA, and GPWA) were assessed using daily air quality records from Hyderabad. The framework integrates multiple dimensions, predictive accuracy, computational efficiency, statistical stability, and compliance with regulatory indicators, ensuring that methodological advances translate into practical utility. Among the tested approaches, the GPWA model consistently delivered superior outcomes, with a mean error of 0.737, a standard deviation of 0.843, and strong correlation with observed pollutant concentrations. It also achieved the most reliable episode detection, outperforming peers across False Alarm Ratio (FAR), Critical Success Index (CSI), and Probability of Detection (POD). Beyond conventional accuracy measures, the framework highlights the balance between precision and resource demands, while demonstrating the added benefits of ensemble learning. These findings confirm the promise of advanced hybrid wavelet models in strengthening urban air quality forecasting and providing actionable guidance for pollution mitigation and public health protection.