Exploration of spatial and temporal characteristics of PM2.5 concentration in Guangzhou, China using wavelet analysis and modified land use regression model

Abstract

This article attempts to detail time series characteristics of PM2.5 concentration in Guangzhou (China) from 1 June 2012 to 31 May 2013 based on wavelet analysis tools, and discuss its spatial distribution using geographic information system software and a modified land use regression model. In this modified model, an important variable (land use data) is substituted for impervious surface area, which can be obtained conveniently from remote sensing imagery through the linear spectral mixture analysis method. Impervious surface has higher precision than land use data because of its sub-pixel level. Seasonal concentration pattern and day-by-day change feature of PM2.5 in Guangzhou with a micro-perspective are discussed and understood. Results include: (1) the highest concentration of PM2.5 occurs in October and the lowest in July, respectively; (2) average concentration of PM2.5 in winter is higher than in other seasons; and (3) there are two high concentration zones in winter and one zone in spring.