Space-Time Ground-Level PM2.5 Distribution at the Yangtze River Delta: A Comparison of Kriging, LUR, and Combined BME-LUR Techniques

Abstract

Ambient air PM2.5 is one of the major pollutants linked to respiratory and lung diseases in the Yangtze River Delta (YRD), which is China’s leading economic region and one of the top economic regions worldwide. The main objectives of this work is to compare the accuracy of some widely-used techniques to characterize and predict the space-time distribution of ground-level PM2.5 in the YRD, and to propose a synthesis of techniques that can yield better results than previous techniques. First, a land-use regression (LUR) model is implemented using the relevant data bases (such as air quality, aerosol optical depth, AOD, Modern -Era Retrospective analysis for Research and Applications, MERRA, meteorological monitoring, road networks information, longitude, latitude, elevation and land-use data). Then, the synthesis of the LUR and the Bayesian maximum entropy (BME) techniques is proposed and implemented, for the first time, in the study of PM2.5 concentrations over the YRD region. It was found that the combined (integrated) BME-LUR technique generated PM2.5 concentration estimates showing a 28.34% improvement in accuracy (R2 indicator) compared to the standard LUR technique, and a 12.53% improvement compared to the mainstream geostatistical Kriging technique.