Modelling the Intra-Urban Variability of N02 for Estimating Human Exposure in Guangzhou, China

摘要

NO2 concentrations within cities are known to have high spatio-temporal variation and estimating intra-urban variability of NO2 accurately is important for human exposure assessments. Land-use regression (LUR) and dispersion models (DM) are widely used for estimating air pollution exposure. Few models have been developed in China on this scale due to scarcity of input data, especially from monitoring data. Integration of LUR with DM can help to overcome the lack of data. The aim of this study is to use both LUR and DM, including in combination, to simulate NO2 concentrations for the city of Guangzhou and to explore the differences between modelled results. N02 observations at 10 sites are obtained from http://beijingair.sinaapp.com for 2017. Hourly concentrations are averaged to annual mean values. DM with ADMS-Urban is applied in Guangzhou using input data including emissions from Multi-resolution Emission Inventory for China (MEIC), road geometry from OpenStreetMap, and hourly meteorological data from the National Oceanic and Atmospheric Administration (NOAA). The results are validated using NO2 observation, which are also used to develop a LUR model. Using a geographic information system, spatially explicit predictor variables in different buffer zones are regressed against monitoring data. The predictor variables include road networks, land-use classification, and population density. A stepwise multiple regression approach is used with a priori-defined predictor variables. These predictor variables are selected to maximize the adjusted percentage explained variance (R2). Model performance is evaluated by leave-one-out cross-validation. In the integrated model, DM generated concentrations of N02 at various receptors are used to develop LUR models. Modelled concentrations of the three approaches are compared with the aim to determine the best approach to derive urban pollution maps of N02 over the city to assess population exposure.