An intercomparison of AOD-converted PM2.5 concentrations using different approaches for estimating aerosol vertical distribution

Publication date: October 2017
Source:Atmospheric Environment, Volume 166
Author(s): Tianning Su, Jing Li, Chengcai Li, Alexis Kai-Hon Lau, Dongwei Yang, Chuanyang Shen
Due to the limited spatial coverage of surface PM2.5 monitoring sites, satellite AOD (aerosol optical depth) products have been widely used to estimate surface PM2.5 in different parts of the world. A major difficulty as well as source of uncertainty in converting AOD to PM2.5 is the determination of aerosol vertical distribution, usually represented by the boundary layer height (BLH). In this study, we evaluate the performance of different approaches of estimating aerosol vertical distributions in the AOD-PM2.5 conversion process, using long-term and multi-source data acquired at a super station, Yuen Long, Hong Kong. The monthly climatology of aerosol vertical distribution and BLH products derived from lidar, radiosonde, and MERRA reanalysis data are respectively applied for converting AOD to surface aerosol extinction coefficients. Seasonal empirical hygroscopic growth functions are constructed to convert aerosol extinction to dry PM2.5 mass concentration. Results indicate that different vertical distribution estimation approaches can have highly varying effect on the converted PM2.5 concentration. Using lidar-derived BLHs shows the best agreement, with a correlation coefficient of 0.73 and a relative bias of 30.6% between retrievals and observations. Since continuous lidar measurements are not available for most regions, the climatology pattern of aerosol structure and radiosonde-derived BLHs are found to be suitable alternatives with a correlation coefficient of ∼0.6, and considerably outperform the results using BLHs derived from reanalysis data. Elevated aerosol layers appear to be the major source of uncertainty and result in an overestimate of satellite results, especially during the spring and summer seasons.



http://ift.tt/2uo7YHb