Impact of COVID-19 lockdown on air quality analyzed through machine learning techniques

Area of Science:

• Environmental Science
• Atmospheric Chemistry
• Data Science

Background:

• The COVID-19 pandemic necessitated global lockdowns, leading to reduced industrial and vehicular emissions.
• This study investigates the impact of these restrictions on air quality in Lahore, Pakistan.

Purpose of the Study:

• To analyze the influence of the COVID-19 lockdown on air quality in Lahore.
• To compare the effectiveness of various machine learning models in forecasting air pollutant concentrations.

Main Methods:

• Collected historical air quality data for PM2.5, PM10, NO2, and O3 from four sites in Lahore.
• Employed multiple machine learning models including Decision Tree, SVR, Random Forest, ARIMA, CNN, N-BEATS, and LSTM.
• Evaluated model performance using RMSE, MAE, and R-SQUARE metrics, with a focus on LSTM's accuracy.

Main Results:

• Air quality in Lahore improved by approximately 20% during the lockdown.
• Significant reductions observed: PM2.5 (42%), PM10 (72%), NO2 (29%).
• Ozone (O3) concentration increased by 20%; LSTM demonstrated superior accuracy in pollutant estimation.

Conclusions:

• The COVID-19 lockdown positively impacted Lahore's air quality by reducing harmful pollutants.
• LSTM proved to be the most accurate model for predicting air pollutant levels during the lockdown period.
• Findings highlight the potential of emission reduction strategies for improving urban air quality.