Development Of Ammonia Emission Inventories Based On Multiple-methods And Their Evaluation With Chemical Transport Modeling In Yangtze River Delta Region,China

Ammonia?NH₃?exerts important influence on atmospheric chemistry and biodiversity.Its reactions with sulfuric acid?H₂SO₄?and nitric acid?HNO₃?can generate sulfate,nitrate and ammonium,which are the critical components of fine particulate matter.As a non-restrained species in China,tropospheric ammonia columns show a rising trend in major agricultural regions.NH₃ emission inventory could reflect intensity and spatial and temporal distribution of emissions,and it is the crucial input to air quality modeling system.Therefore improving the accuracy and optimizing the methods of inventory development become an important research area.There are large differences in emissions and spatial and temporal distribution in existing inventories Exploring those differences and model performances could help deeply understanding the influence factors of ammonia volatilization,provide improvement directions of current inventory,and better serve atmospheric chemistry studies and pollution control management.In this study,two bottom-up ammonia emission inventories in Yangtze River Delta?YRD?region were developed using "emission factors" method?E1?and "agricultural process characterization" method?E2?based on statistical yearbooks,farming information and other geographic information.The total NH₃ emissions of E1 in Shanghai,Jiangsu,Zhejiang and Anhui were 44.5,791.9,270.1 and 658.2 Gg,and emissions of E2 were 33.2,496.5,147.2 and 389.9 Gg,respectively.Fertilizer application and livestock were the two largest contributors,together accounting for 74%-84%of the total emissions.These two inventories differed by a factor of 1.6 and 2 for all sources and agricultural source,respectively.Largest monthly emissions were found from June to August for both inventories,while emissions in March and September of E2 were also high due to incorparation of farming information.Two inventories were similar in spatial distribution with high emissions in Northern Anhui and Yancheng and Xuzhou in Jiangsu.Emissions in July and October of E2 were larger in high emission areas,and results were opposite for other cases.Through analysis on fertilizer application,we found that conisderation of agricultural process would lead to varied fertilizer rate and ammonia volatilization and thereby differences in spatial and temporal distribution.In order to evaluate both inventories and to analyze the influence of different estimation methods on simulation,we applied the Models-3/Community Multi-scale Air Quality?CMAQ?using the two inventories and compared the simulated NH₃ and aerosols concentrations with ground-based and Infrared Atmospheric Sounding Interferometer?IASI?satellite observations.Monthly average modeled NH₃ concentrations of E1 and E2 at Jiangsu Provincial Academy Of Environmental Science site?JS-PAES,urban site?in October were 7.75?g/m3 and 5.96?g/m3?7.62?g/m3 for groud-based observation?.The normalized mean bias?NMB?of NH₃ hourly concentration between ground observations and simulations were 1.73%and-21.75%,and the normalized mean errors?NME?were 6.94%and 53.68%,respectively.While the NME of daily average concentrations were 37.00%and 32.54%.Better simulation performance was achieved in E1 at JS-PAES site in October,while E2 probably was underestimated.Simulatio1 performances of ammonium?NH4+?,sulfate?SO42-?and nitrate?NO₃-?in 9km×9km cells were better than NH₃,and results from the two inventories were close.One possible reason is that secondary aerosols generated through atmospheric chemistry reaction could better indicate regional situation at the simulation resolution of 9km×9km.NH₃ columns from IASI were used to evaluate the spatial distribution of NH₃ emissions in two inventories.Improved performances of spatial distribution were indicatd by E2 in January and October with better results in NMB,NME and correlation coefficients,compared to E1.However,there were some uncertainties in the correction of emission factors by geographic information,which may result in little superiority shown in E2 in April and July.We further explore effects of soil pH on spatial distribution of NH₃ emissions and simulation.The results showed that the NME of NH₃ columns between satellite observations and simulations of E2 were smaller than E1 in neutral and alkaline soil.It proved that agricultural process could optimize the spatial distribution of NH₃ columns,and the correction effect for emission factors by pH was more reasonable in the alkaline soil in this study.Through the calculation of GR?gas ratio?,major part of YRD were identified as NH₃-rich throughout the entire year except southwestern Zhejiang.The overestimation in SO2 and NOx emissions led to higher simulation values for NH4+ and SO42-,and lower values for NO₃-.Deviation of secondary aerosol simulation could be smaller when SO2 and NOx emissions were reduced by 40%.Our finding also suggested that NO₃-would increase if no control measures were taken for NH₃ emissions,resulting in reduced benefits of air quality improvement through SO2 and NOx emission abatement.There were still some uncertainties in the analysis,which mainly came from emission inventories,model mechanism and the quality of observation data.Measurements on local emission factors and investigation of emission sources are recommended in the future to improve the accuracy of emission inventory.We also suggest improving long term observation of ammonia and modifying the mechanism of gas-to-particle partitioning of NH₃?g?to NH4+?g?to reduce the modeled bias.