Scholarlink

The COVID-19 pandemic led to unprecedented restrictions on mobility and industrial activity, creating a unique opportunity to study their impact on air pollution in urban India. This paper investigates the changes in air quality across four major Indian metropolitan cities—Delhi, Mumbai, Kolkata, and Bengaluru—during the lockdown period in 2020. Using publicly available air quality data from January to July 2020, we perform comparative statistical analyses and apply machine learning techniques to classify air pollution levels before and during the lockdown. Independent two-sample t-test confirm statistically significant reductions in major pollutants such as PM \(_{2.5}\) , PM \(_{10}\) , NO \(_2\) , and CO in most cities. Logistic Regression and Support Vector Machine (SVM) models are employed to classify pre- and during-lockdown periods based on pollutant profiles, achieving strong separation. The study highlights the responsiveness of specific pollutants to reduced anthropogenic activity and offers insights to inform environmental policy and urban planning.

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

Unmasking the Skies: Statistical and Machine Learning Insights Into the Air Quality Shift During COVID-19 Lockdowns in Indian Metropolises

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Shubhranshu Gorai,
  • Suchandra Banerjee,
  • Kinshuk Banerjee,
  • Saibal Majumder,
  • Chandan Bandyopadhyay,
  • Diganta Das

摘要

The COVID-19 pandemic led to unprecedented restrictions on mobility and industrial activity, creating a unique opportunity to study their impact on air pollution in urban India. This paper investigates the changes in air quality across four major Indian metropolitan cities—Delhi, Mumbai, Kolkata, and Bengaluru—during the lockdown period in 2020. Using publicly available air quality data from January to July 2020, we perform comparative statistical analyses and apply machine learning techniques to classify air pollution levels before and during the lockdown. Independent two-sample t-test confirm statistically significant reductions in major pollutants such as PM \(_{2.5}\) , PM \(_{10}\) , NO \(_2\) , and CO in most cities. Logistic Regression and Support Vector Machine (SVM) models are employed to classify pre- and during-lockdown periods based on pollutant profiles, achieving strong separation. The study highlights the responsiveness of specific pollutants to reduced anthropogenic activity and offers insights to inform environmental policy and urban planning.