Scholarlink

<p>Air pollution is a major environmental challenge worldwide, particularly in India due to rapid industrialization and urbanization, posing serious risks to public health and the environment. Accurate air quality forecasting is therefore essential for effective pollution control and mitigation. Traditional statistical techniques, while simple and computationally efficient, often show inconsistent accuracy due to the non-linear and dynamic nature of pollutants. In recent years, advanced machine learning methods have emerged, offering adaptive learning, high precision, improved temporal pattern modelling, and effective handling of high-dimensional data. Alongside these, numerical models have gained use for their detailed, physics-based simulations of pollutant dynamics, providing deeper insights into air quality behaviour. This review examines the evolution of forecasting methods in India from 2000 to December 2024, analysing 70 peer-reviewed articles classified by journal, year, region, input parameters, and performance. Results show an increasing number of publications, with PM<sub>2.5</sub> and AQI being the most frequently studied topics. Deep neural networks outperform other statistical methods, while hybrid models achieve lower errors. Key research gaps highlight the importance of integrating data-driven and physics-based approaches to advance forecasting accuracy and policy relevance.</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

Advances in air pollution forecasting in India: A review of modelling approaches

  • 期刊论文

  • 发表日期：

    2025年12月1日

查看全文
  • Nishant Behera,
  • Manoranjan Sahu

摘要

Air pollution is a major environmental challenge worldwide, particularly in India due to rapid industrialization and urbanization, posing serious risks to public health and the environment. Accurate air quality forecasting is therefore essential for effective pollution control and mitigation. Traditional statistical techniques, while simple and computationally efficient, often show inconsistent accuracy due to the non-linear and dynamic nature of pollutants. In recent years, advanced machine learning methods have emerged, offering adaptive learning, high precision, improved temporal pattern modelling, and effective handling of high-dimensional data. Alongside these, numerical models have gained use for their detailed, physics-based simulations of pollutant dynamics, providing deeper insights into air quality behaviour. This review examines the evolution of forecasting methods in India from 2000 to December 2024, analysing 70 peer-reviewed articles classified by journal, year, region, input parameters, and performance. Results show an increasing number of publications, with PM2.5 and AQI being the most frequently studied topics. Deep neural networks outperform other statistical methods, while hybrid models achieve lower errors. Key research gaps highlight the importance of integrating data-driven and physics-based approaches to advance forecasting accuracy and policy relevance.