Forecasting Air Pollution: Case Study Investigating the True Impact of Covid-19 Lockdowns on Air Quality in South Asia

Air pollution is a modern anthropological catastrophe faced by all humanity. Most of the world’s pollution hotspots are located in the South Asian region where most countries are recognized as developing countries. Management of air quality in the region is directly linked to the insufficient infrastructure in the region to monitor and analyse air pollution. Regional air pollution studies are limited and excessively focused on analysing the current pollution landscape rather than developing forecasting tools suitable for the region. The study has focused on the analysis and development of an air pollution forecasting model to predict air quality variations based on the Auto Regressive Integrated Moving Average (ARIMA) method. The model was developed and validated using available historical PM2.5 pollution data from the region. Additionally, the model was used to analyse the true impact of COVID-19 lockdowns on regional air quality. The analysis includes investigations of PM2.5 pollutant levels in Colombo, Chennai, Dhaka, Delhi, Kathmandu, Islamabad, and Hyderabad. The developed model performed well with RMSE values of less than 12% in all instances relative to the pollutant concentration ranges for each city. Forecasted and historical data comparison identified that the impact of COVID-19 lockdowns is underestimated in cities such as Kolkata, Colombo, and Chennai by more than 10%. Vice versa was observed for Delhi, Islamabad, and Hyderabad where the pollution trend is decreasing. Traditional time series analysis will certainly underestimate or overestimate the impact of COVID-19 lockdowns as expected.