Time Series Simulation and Forecasting of Air Quality Using In-situ and Satellite-Based Observations Over an Urban Region

Abstract

Air quality is directly associated with the health of society. So, it becomes essential to forecast air pollution, which assumes an imperative part in air pollution warnings and control. A time-series simulation approach was adapted for the forecasting of monthly mean ambient air pollutants (PM2.5, O3, NO2) concentration and Aerosol Optical Depth (AOD) at an urban traffic site (Mathura Road, CSIR-CRRI) in New Delhi, India. Satellite-based aerosol loading (AOD550) retrieved from the Terra MODIS (Collection 6) enhanced Deep Blue (DB) algorithm was used for further analysis. The analysis considered the average monthly mean concentration of air pollutants and AOD between 2012-2017 and, simulates the concentrations of PM2.5, O3, NO2, and AOD for the same period and then forecasts air quality for the years 2020-2023. The forecasted results were validated with 24 months of in-situ and satellite data from 2018-to and 2019. In the year 2020, observed and simulated results are in lower agreement due to the shutdown of anthropogenic activities to combat pandemic situations. Otherwise, modeled and forecasted results are in good harmony with the in-situ and satellite observations. The results also signify that the time series Autoregressive Integrated Moving Average (ARIMA) modeling approach can be an effective and simple tool for air pollution simulation and future forecast. The results are evocative concerning the forecast of near future aerosol loading information and will also be profound to address the problems.

Keywords

• arima, modis aod, pm2.5, o3, no2, simulation, forecast