Due to the rapid urbanization and increasing energy consumption, air pollution, especially some fine particulates like PM2.5 rise in the context of fast urbanization. PM2.5 pollution has been given considerable attention recent years. High PM2.5 concentration is the main reason for the atmospheric haze in Beijing-Tianjin-Hebei region. Air pollution has become the key issue restricting the sustainable development of Beijing-Tianjin-Hebei region and even the whole country. Long-term exposure to PM2.5 is likely to cause adverse effects on human health. The spatial-temporal variation of air pollution can be characterized by the land use regression model. It is significant to have a good knowledge of spatial characteristics of PM2.5 concentration, which could assist air pollution management and the epidemiological research. This manuscript used air quality data of 104 monitoring sites of Beijing-Tianjin-Hebei region from 1st January 2014 to 31st December, 2014, combined with VⅡRS (visible infrared imaging radiometer) AOD(aerosol optical depth), land use, meteorological factors, road network, population, and pollutant sources distribution to establish the land use regression model by least square method and geographically weighted method respectively. The four models established were least square land use regression model with VⅡRS AOD data, geographically weighted land use regression model with VⅡRS AOD data, least square land use regression model without VⅡRS AOD data and geographically weighted land use regression model without VⅡRS AOD data. And the adjusted R2 values for these four models were 82.13%, 84.87%, 80.45% and 81.99%, respectively. Research results demonstrated that the geographically weighted method performed better than the least square method and improved the land use regression model to a certain extent.
Keywords: PM2.5; air pollution; geographical weighted regression; land use regression; spatial variation.