Statistical Model of the Relationship Between Atmospheric Visibility and PM_2.5 in Shenzhen

With the development of economy, the level of air quality in main cities of China has experienced a continuously deteriorating process. Pearl River Delta Region are confronting with the disturbance of more and more haze weather especially. As one of the cities with the most serious haze problem in the region, Shenzhen experiences 231 haze days in 2007 according to the definition of haze that visual range is lower than ten kilometers and relative humidity is not higher than 80% at the same time. The degradation of atmospheric visibility is mainly caused by the extinction effect of aerosol particles, especially of fine particles, including scattering by inorganic components and absorption by black carbon. The average concentration of fine particle (particle whose aerodynamic diameter is lower than 2.5 μm) in Shenzhen is as high as 53.4±37.3 μg·m^-3 in 2007, which is a little lower than other main cities in China but two times higher than American national standard enacted by USEPA. Accordingly, the average visibility is as low as 13.4±9.3 km, and shows the same seasonal variation as fine particle concentration. However, few studies on the quantitative relationship between visibility and fine particles in Chinese cities are reported in the literature. Based on the analysis of the extinction mechanisms and relevant influential factors, statistic models are developed for describing the relationship between visibility and fine particles in the urban air of Shenzhen using multiple regression techniques. The data includes visual range acquired by visual observation, fine particles concentration generated by TEOM 1400a (an online instrument for monitoring the concentration of fine particles) and relative humidity (RH). All the data are monitored simultaneously in the year of 2007, and abnormal values are excluded before regression analysis. f_RH is used to eliminate the light extinction of humidity to particles, and four usual forms of it are discussed too. Multiple linear and nonlinear regression methods are used for regression analysis and the initial values of parameters come from literatures for nonlinear regression. Finally, a power function form of f_RH containing underlying physical mechanism of particles' extinction is selected to reflect humidity's effect due to good agreement and the compact form. The complete model expression is given at the same time. The correlative coefficient between the observed visibility values and the reconstructed visibility values using the best model is 0.40 for the 1-hour average data. The model for 24-hour average data is also established in the same form. The correlative coefficient reaches as high as 0.73, and the deviation of the reconstructed values is small, so the model can properly reflect a good relationship between visibility and fine particles concentration. In addition, the expression of extinction efficiency changing with relative humidity demonstrates the similar increasing patterns in existing study, and reasonably describes the relationship between extinction efficiency and relative humidity in Shenzhen.