Forecasting of air quality in Delhi using principal component regression technique

Abstract

Over the past decade, an increasing interest has evolved by the public in the day–to–day air quality conditions to which they are exposed. Driven by the increasing awareness of the health aspects of air pollution exposure, especially by most sensitive sub–populations such as children and the elderly, short–term air pollution forecasts are being provided more and more by local authorities. The Air Quality Index (AQI) is a number used by governmental agencies to characterize the quality of the air at a given location. AQI is used for local and regional air quality management in many metropolitan cities of the world. The main objective of the present study is to forecast short–term daily AQI through previous day’s AQI and meteorological variables using principal component regression (PCR) technique. This study has been made for four different seasons namely summer, monsoon, post monsoon and winter. AQI was estimated for the period of seven years from 2000–2006 at ITO (a busiest traffic intersection) for criteria pollutants such as respirable suspended particulate matter (RSPM), sulfur dioxide (SO2), nitrogen dioxide (NO2) and suspended particulate matter (SPM) using a method of US Environmental Protection Agency (USEPA), in which sub–index and breakpoint pollutant concentration depends on Indian National Ambient Air Quality Standard (NAAQS). The Principal components have been computed using covariance of input data matrix. Only those components, having eigenvalues ≥1, were used to predict the AQI using principal component regression technique. The performance of PCR model, used for forecasting of AQI, was better in winter than the other seasons as studied through statistical error analysis. The values of normalized mean square error (NMSE) were found as 0.0058, 0.0082, 0.0241 and 0.0418 for winter, summer, post monsoon and monsoon respectively. The other statistical parameters are also supporting the same result.