Understanding of regional air pollution over China using CMAQ, part I performance evaluation and seasonal variation
Xiao-Huan Liu,Yang Zhang,Shu-Hui Cheng,Jia Xing,Qiang Zhang,David G. Streets,Carey Jang,Wen-Xing Wang,Ji-Ming Hao
DOI: https://doi.org/10.1016/j.atmosenv.2010.03.035
IF: 5
2010-01-01
Atmospheric Environment
Abstract:The U.S. EPA Models-3 Community Multiscale Air Quality (CMAQ) modeling system with the process analysis tool is applied to China to study the seasonal variations and formation mechanisms of major air pollutants. Simulations show distinct seasonal variations, with higher surface concentrations of sulfur dioxide (SO2), nitrogen dioxide (NO2), and particulate matter with aerodynamic diameter less than or equal to 10 μm (PM10), column mass of carbon monoxide (CO) and NO2, and aerosol optical depth (AOD) in winter and fall than other seasons, and higher 1-h O3 and troposphere ozone residual (TOR) in spring and summer than other seasons. Higher concentrations of most species occur over the eastern China, where the air pollutant emissions are the highest in China. Compared with surface observations, the simulated SO2, NO2, and PM10 concentrations are underpredicted throughout the year with NMBs of up to −51.8%, −32.0%, and −54.2%, respectively. Such large discrepancies can be attributed to the uncertainties in emissions, simulated meteorology, and deviation of observations based on air pollution index. Max. 1-h O3 concentrations in Jan. and Jul. at 36-km are overpredicted with NMBs of 12.0% and 19.3% and agree well in Apr. and Oct. Simulated column variables can capture the high concentrations over the eastern China and low values in the central and western China. Underpredictions occur over the northeastern China for column CO in Apr., TOR in Jul., and AODs in both Apr. and Jul.; and overpredictions occur over the eastern China for column CO in Oct., NO2 in Jan. and Oct., and AODs in Jan. and Oct. The simulations at 12-km show a finer structure in simulated concentrations than that at 36-km over higher polluted areas, but do not always give better performance than 36-km. Surface concentrations are more sensitive to grid resolution than column variables except for column NO2, with higher sensitivity over mountain and coastal areas than other regions.