Time series and predictability analysis of air pollutants in Delhi

Abstract

Air pollution is a challenging problem globally and the complete globe is facing the hazards caused by it. In the recent years, it draws an attention of all as it is directly related with the health concern of an individual. Air pollution is a big concern for Delhi. More than 15 million people are exposed to severely high pollutant concentrations. Air pollution have a high impact on the environment also. Government of Delhi performed a litmus test of vehicular emissions in the form of Odd-Even Scheme (15April 2016–30 April 2016). This paper deals with the estimation of regression coefficient, Hurst exponent, Fractal Dimension and Predictability Index of air pollutants NO, NO 2 , NO x , SO 2 , PM2.5 and the weather conditions relative humidity and temperature during (15April 2016–30 April 2016), pre (30 March 2016–14 April 2016) and post Odd-Even Scheme (1 May 2016–16 May 2016) in the Dwarka area adjoining Indira Gandhi International Airport, Delhi. The Hurst exponent is defined as the index of long-range dependence. It measures a relative tendency of a time series either regress strongly to the mean or to cluster in a direction. It is related to fractal dimension which gives measure for roughness of surface. The Predictability Index describes the behaviour of time series. The data for the above period is taken from Central Pollution Control Board, Government of India. It is observed that carbon mono oxide (CO) behavior is unpredictable with Relative Humidity and sulphur-di-oxide (SO 2 ) as they follows a Brownian motion for pre and post the Odd-Even Scheme simultaneously. During the period of Odd-Even Scheme, the behaviour of temperature with respect to PM2.5 is unpredictable. It is concluded that CO with PM2.5 follows a Brownian time series and hence the trend is unpredictable and SO 2 for pre and post Odd-Even Scheme follows a Brownian time series. Thus it is difficult to predict the behavior and trend of the pollutants.