A Short-Term Autoregressive Model for the Prediction of Daily Average NO2 Concentration in Nagercoil, Tamil Nadu, India

Abstract

Nitrogen dioxide (NO2) is one of the pollutants that can cause potential damage to the ecosystem. NO2 emitted from vehicles forms the primary precursor for ground-level ozone. In this study, an analysis of the daily average of NO2 concentration with meteorology measured for two years 2021 and 2022 is being carried out. It is evident from the analysis that NO2 concentration followed an apparent diurnal pattern with a maximum value in the morning hours and a minimum during the afternoon hours. Summer months recorded the highest, and North East Monsoon (NEM) recorded the lowest values of NO2. A statistically significant positive correlation was found between NO2 and Temperature. An autoregressive model was formulated to forecast the daily average values of NO2 concentration. Unit root test was performed to check the stationarity of the data points, which is important in determining trends and seasonal changes. From the model procedure, the order that best fits the data was identified as AR (4), in which the process has the current value based on the previous three values. The Akaike Information Criterion (AIC) and Schwartz Criterion (SC), which are estimators of prediction error for AR (4), are low. The Jarque confirmed the normal distribution-Bera test, which again approves the satisfactoriness of the model.

Keywords

• akaike information criterion, autoregressive model, diurnal changes, jarque-bera test, no2