Applications Of GRAPES-CUACE Adjoint Model In Sensitive Source Areas Tracking And Emission Inversion

GRAPES-CUACE aerosol adjoint model was used to track sensitive source areas and key emission periods of PM_2.5 pollution in Beijing of typical episode,different weather pollution types and strong and weak East Asian winter monsoon years.Based on the adjoint sensitivity results,the contributions of local and surrounding emissions to PM_2.5concentration in Beijing were quantitatively analyzed on the time scales from short-term to sub-seasonal?seasonal?.In addition,combined with the optimization algorithm and pollutants observations,the GRAPES-CUACE-3D-Var assimilation system was further developed on the basis of GRAPES-CUACE aerosol adjoint model,and used in the inversion of BC emissions in Beijing-Tianjin-Hebei region.The main conclusions are as follows:Adjoint sensitivity analysis of the heavy pollution process from November 27 to December 2,2015 shows that Beijing,Tianjin,Hebei,and Shanxi emissions contribute 31%,9%,56%and 4%,respectively,to PM_2.5 peak concentration in Beijing within 72 hours before the objective time point.The corresponding key emission periods are 1-13h,1-33h,1-57h and17-33h,respectively,before the objective time point.Based on PM_2.5 observations from 2012to 2015,the heavy pollution processes were screened,and the weather clustering classification method was used to classify PM_2.5 pollution processes in Beijing into five types.The simulation studies showed that there are significant differences in the distributions of sensitive source areas of the five pollution types.The sensitive source areas of low pressure and weak high pressure alternating and weak high pressure types are located in the southwest of Beijing,low pressure control and high rear types are distributed in the south of Beijng,and uniform pressure type is located in the southeast of Beijing.Within 72 hours ahead of the objective time point,the contributions of local source for the five types are 37.5%,25.0%,39.4%,31.2%and 42.4%,respectively,the contribution of Hebei source is 37-64%,and those of Tianjin and Shanxi sources are 6-16%and 3-7%,respectively.It's better to reduce local source within 20 hours before the objective time point,when the local source is reduced by50%within 72 hours priori to the objective time point,the decrease rate of PM_2.5concentration of the five types are 11.6%,9.4%,13.8%,9.9%,and 15.2%,respectively.And the reduction effect of surrounding source is better within 3-57 hours ahead of the objective time point.Statistical analysis showed that PM_2.5 concentration in Beijing is significantly anti-correlated with East Asian winter monsoon index,and December 2014 and 2016 were selected as strong and weak monsoon years,respectively.The adjoint simulation showed that in strong and weak East Asian winter monsoon years,there is no obvious difference in the distributions of sensitive source areas of PM_2.5 pollution events in Beijing,but the contributions and key emission periods of sensitive source areas are significantly different.In the weak monsoon year?December 2016?,the contributions of Beijing,Tianjin,Hebei and Shanxi sources in the first 72 hours before the objective time period are 2.2,5.9,1.6 and 1.4times higher,respectively,and the key emission period is about 2.6 times longer than those of the strong monsoon year?December 2014?.Compared to Hebei and Shanxi sources,the contributions of Beijing and Tianjin sources have increased greatly,emphasizing the importance of the source areas in the southeast direction of Beijing in weak monsoon year.It can be seen that before the advent of unfavorable meteorological conditions or extreme climate events,the adjoint model can be used to locate sensitive source areas and key emission periods,and then we can take control measures in conjunction with surrounding areas to improve emission reduction benefits.Besides,combined with pollutants observations,the newly built GRAPES-CUACE-3D-Var assimilation system was used to inverse BC emission in Beijing-Tianjin-Hebei area.The results of two inversion experiments showed that the simulation effect of posterior BC emissions is significantly better than that of prior sources,the correlation coefficient is increased from 0.2 to 0.7 and 0.64,respectively,and the variance is reduced from 0.38 to 0.22 and 0.24,respectively.