Spatial Regression Models for Explaining AQI Values in Cities of Turkey

Abstract

The aim of this study is to determine the natural and anthropogenic factors affecting the air quality index (AQI) and to create a model that shows the effects of these factors on AQI values in cities of Turkey. Natural and anthropogenic factors, which are thought to have an effect on AQI, were determined and interpreted with kriging maps. The effects of these factors on AQI were examined by explanatory spatial data analysis (ESDA). Global Moran’s I and local Moran’s I (LISA) indices were examined for the presence of spatial relation. Spatial lag model (SLM) was proposed for parameter estimation instead of ordinary least squares method (OLS) and the average AQI values for 2014 and 2015 were compared. It was also concluded that the average AQI values of 2014 and 2015 were in a strong correlation relationship (Pearson correlation coefficient of 0.914). On the Southern Anatolia, desert dust transport decreases the air quality of the region, however on the Black Sea coast, meteorological factors have a strong effect on air quality. Both SLM and OLS models showed that higher wind speed increases air quality in the cities while increase in GDP increases AQI.