Retrieving historical ambient PM_2.5 concentrations using existing visibility measurements in Xi'an, Northwest China

Long term fine particulate matter (PM_2.5) data are needed to assess air quality and climate issues, but PM_2.5 data have only been monitored in the recent decade in Chinese cities. Considering strong correlations between PM_2.5 and visibility, regression models can be useful tools for retrieving historical PM_2.5 data from available visibility data. In this study, PM_2.5 and visibility data are both available during 2004-2011 in Xi'an, a megacity in northwest China. Data from 2004 to 2007 were used to develop a regression model and those from 2008 to 2011 were used to evaluate the model. An exponential regression model was then chosen to retrieve the historical PM_2.5 data from 1979 to 2003, which were then analyzed together with the measured data from 2004 to 2011 for long term trends. Seasonal PM_2.5 increased from 1979 to 2011 with the fastest increase in winter and the slowest in summer. Annual average PM_2.5 followed into three distinct periods with a slow decreasing trend from 1979 to 1996, a sharp increasing trend from 1997 to 2006, and a slow decreasing trend from 2007 to 2011. These increasing and decreasing trends are in agreement with the evolution of industrial development in Xi'an.