By Joris Wever (The Netherlands)
Land use regression (LUR) is considered as a promising technique to predict air pollution concentrations within cities. The method entails that a statistical relationship is established between pollutant levels and land use factors using a limited number of sampling sites. Subsequently, pollution concentrations are predicted at sites where the value is unknown. A traditional land use regression model, however, does not take into account meteorological factors. Therefore, in this research LUR models were developed for different wind speed scenarios in order to improve the prediction of fine particulate matter (PM2.5) and ground level ozone (O3) concentrations.
Traditional LUR models explained 43% variation in fine particulate matter and 42% variation in ground level ozone concentrations. It was found that the effect of land use on the level of PM2.5 and O3 depended on wind speed. The influence of wind was further investigated by developing LUR models for different wind speed scenarios. LUR models explained 58%, 33% and 64% of the variation of PM2.5 in case of a high, moderate and low wind speed scenario, respectively. For the same wind scenarios, LUR models explained 50%, 29% and 23% of the variation of O3.
The results of the present study propose that land use factors explained a large amount of variation of both PM2.5 and O3. R2-values of the LUR models designed for the wind speed scenarios, however, showed in some cases an improvement of 10 to 20% over the traditional models. These results suggest that the incorporation of meteorological variables, such as wind speed, is a useful contribution to improve LUR models. In this way, more insight is obtained in the variability of air pollutants at the intra urban scale.
Keywords: air pollution; land use regression; fine particulate matter; ground level ozone; wind speed