Multivariate Analysis for Air Contamination and Meteorological Parameters in Zonguldak, Turkey

Abstract

This study evaluates the concentration of PM10, PM2.5, NOx, NO2, CO and SO2 parameters and the four climatological parameters (temperature, wind speed, humidity and net radiation flux) during the four seasons. Various statistical techniques were utilized to study the behavior of the selected parameters during the seasons. Descriptive statistics exhibited that the studied parameters have high concentrations in winter, except for NO2 (which has a high concentration in autumn), while the concentrations of those parameters were the lowest in summer, except for NO2 and NOX (which have high concentrations in spring). Factor analysis (FA) showed that more than 80% of the total variation belongs to two factors, where 19.47% of the variation was due to wind speed and humidity, while other parameters were responsible for 62.90% of the total variation. Cluster analysis (CA) evaluated the similarity and dissimilarity between various elements through identifying four clusters representing the seasons; cluster 1: autumn, cluster 2: winter, cluster 3: spring and cluster 4: summer. This clustering indicates that the four seasons are entirely different. The highest dissimilarity was reported between summer and the other seasons. CA also classified all parameters into five statistically different clusters; cluster 1: PM10, PM 2.5 and CO; cluster 2: SO2, NOX and NO2; cluster 3: humidity; cluster 4: temperature and radiation and cluster 5: wind speed. This study illustrates the benefits of using multivariate techniques for the evaluation and interpretation of the total variation to get a better picture of the pollution sources/factors and understand the behaviors of the parameters in the air.