Machine learning insights into PM_2.5 changes during COVID-19 lockdown: LSTM and RF analysis in Mashhad

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Urban Air Quality

Background:

• COVID-19 lockdowns significantly altered human mobility and industrial activity globally.
• Understanding the impact of these interventions on urban air quality is crucial for public health and environmental policy.
• Particulate matter (PM2.5) is a key indicator of air pollution with significant health implications.

Purpose of the Study:

• To investigate the effects of COVID-19 lockdown measures on air quality in Mashhad.
• To compare air quality before, during, and after lockdown periods using statistical and machine learning methods.
• To quantify the impact of reduced human mobility on PM2.5 concentrations.

Main Methods:

• Statistical analysis, including paired sample t-tests, to compare hourly PM2.5 data.
• Machine learning models, specifically Random Forest (RF) and Long Short-Term Memory (LSTM) networks, for quantitative assessment.
• Evaluation of model performance using R-squared values (LSTM: 0.82, RF: 0.78).

Main Results:

• A modest 2.40% reduction in PM2.5 was observed immediately before the lockdown period.
• A significant decrease in air quality, with distinct seasonal patterns, was noted post-quarantine, unlike previous years.
• Machine learning models accurately predicted PM2.5 levels, demonstrating the link between mobility changes and air pollution.

Conclusions:

• Lockdown measures had a limited direct impact on PM2.5 levels in Mashhad, suggesting other factors influence air quality.
• Post-lockdown air quality degradation highlights the significant correlation between human mobility patterns and urban air pollution.
• Air pollution modeling is essential for understanding human intervention impacts and developing strategies for long-term air quality improvement.