Chapter 10 - Applying LUR model to estimate spatial variation of PM2.5 in the Greater Bay Area, China

Abstract

PM2.5 pollution exposure assessment necessitates spatial distribution of PM2.5 concentration; however, sparse ground monitoring sites cannot meet the requirement. The Land Use Regression (LUR) model has been applied to simulate pollutants’ concentration successfully. This study took meteorological factors into account to develop a LUR model to predict winter mean PM2.5 concentrations in the Greater Bay Area. Five predictor variables were left in the final LUR model: BLH, V wind speed, percentage of forest area within the 4000-m buffer, density of all roads within the 1500-m buffer. These were negatively associated with PM2.5 concentration, and a percentage of buildings area within the 5000-m buffer were positively associated with PM2.5 concentration. The model adjusted R2 and the leave-one-out-cross-validation (LOOCV) R2 of PM2.5 LUR models were 0.756 and 0.717, respectively. The model would provide acceptable measurements for air pollution exposure of some acute disease for sensitive groups.