Development of land use regression models to characterise spatial patterns of particulate matter and ozone in urban areas of Lanzhou
Tian Zhou, Shuya Fang, Limei Jin, Xingran Li, Xiaokai Song, Yufei Wang, Xiaowen Zhou, Yana Bai, Xuying Ma
Abstract
There are still many challenges in Land use regression (LUR) application in cities in China due to insufficient air pollutants data. In this study, the LUR models of TSP, PM10, PM4, PM2.5, PM1, and O3 are developed by basing on the mobile monitoring in 2019 in Lanzhou, China. Our results show that the adjusted-R2 of six best models are rang of 0.45⁓0.87. Referring to adjusted-R2, the differences in cross-validation-R2 (CV-R2) using the training data are less than 9% excluding PM10, and the differences in CV-R2 using the test data are within 19% in the models of TSP, PM4, and O3. Overall, the models of TSP, PM4, and O3 are more robust than that of PM10, PM2.5, and PM1. The O3 model has a good fit. The spatial patterns of PMs exhibit high concentration in the west, center and east area, and the concentration being higher in the south than in the north. The predicted O3 concentrations decrease from west to east. All predicted concentrations indicate that there are the highest level and the largest area of air pollutants in Xigu Distinct. These results can provide scientific data for urban planning, land use regulation, prevention and control of air pollution.