Comparison Of Two Methods Of PM_2.5 Source Apportionment

Atmospheric particulate matter source apportionment is an important basis for effective pollution control.At present,a variety of particle source apportionment methods have been proposed at home and abroad,including the receptor model and source model.Among them,the Positive Matrix Factorization（PMF）and the Comprehensive Air quality Model with extensions（CAMx）based Particle Source Apportionment Technology（PSAT）are two representative methods that are relatively mature and widely used at present.A large number of application results showed that different source apportionment methods had their own advantages and disadvantages,and their analysis results often had large differences.Therefore,a deep understanding of the differences between the results of different source apportionment methods is of great significance for the reasonable selection of source apportionment methods and the correct understanding and application of analysis results.This study was aimed at Tianjin.By analyzing the atmospheric fine particles(PM_2.5)pollution characteristics from 2017 to 2018,26 pollution cases were selected,of which 6 were in autumn,12 were in winter,and 8 were in spring.The pollution events were classified according to the degree of pollution,season,dominant components,and meteorological conditions.Then,the source apportionment results of the two methods of PMF and CAMx-PSAT were compared and analyzed,and their differences in different seasons and different pollution sources were studied.Finally,preliminary ideas and suggestions were put forward for the comprehensive use of the two methods.The main findings of this study were as follows:（1）Characteristics of PM_2.5 pollution incidents in TianjinDuring the study period,most of PM_2.5 pollution in Tianjin occurred in the spring months were March,followed by the winter in January and December.According to the composition characteristics of PM_2.5,spring and autumn were mainly nitrate-dominated pollution,and pollution in winter was more complicated,with mixed pollution as the main type.PM_2.5average mass concentration was highest in winter.From the relative content of its components,NO₃^-was highest in spring,SO₄^2-and OC were highest in autumn,and NH₄⁺and EC were highest in winter.Most serious pollution cases were nitrate-dominated.Especially in the case of light wind and low humidity,it was easier to form heavy pollution.Sulfate-dominated pollution cases mostly occurred in the conditions of light wind and low humidity in autumn.Organic carbon-dominated and mixed pollution incidents mainly occurred in winters with low wind and low humidity.Meteorological conditions had a greater influence on the PM_2.5 concentration in Tianjin:wind speed and pollutant concentration were generally negatively correlated.When the wind speed was 1.5-2.5m/s,a pollution trough period was likely to occur;when the wind speed was greater than 2.5m/s,the diffusion period of pollution enterred;when the wind speed was low,the PM_2.5 mass concentration accumulated to the peak.Southeast wind played a major role in pollution dilution.High temperature and low humidity conditions would promote the reduction of PM_2.5 mass concentration.（2）Difference characteristics of PM_2.5 source apportionment results of two methodsFrom the analysis results of the average contribution of the two methods to different sources,the order of the contribution values of the two sources was basically the same,from the largest to the second,secondary conversion,coal combustion,motor vehicles and industrial emissions.The average contributions of secondary source were40%±14%and 46%±6%respectively.The average results of coal-fired were 22%±7%and 25%±8%,respectively.The average contribution to industrial sources were 6%±2%and 5%±2%,respectively.The biggest difference was in motor vehicles.The average results of PMF and PSAT were 20%±9%and 6%±4%,a difference of 14%.This might be because PMF had a certain degree of intersection when identifying the primary and secondary contributions of motor vehicles,and considered some secondary sources as primary vehicle sources.The low PSAT result might be due to errors in the source spectrum or the simulation process of pollutants.Regarding seasonal differences,the four sources analyzed by the two methods in the spring pollution process had the smallest differences.In terms of different pollution levels,the difference between coal-burning sources in light pollution was smaller.The difference between secondary sources in heavy pollution and serious pollution events was smaller.In terms of different pollution types,secondary sources and motor vehicle sources had the smallest differences in nitrate-dominated pollution events,and had the largest differences in organic carbon-dominated pollution events.But coal-fired sources had the smallest differences in organic carbon-dominated pollution events.In terms of different meteorological conditions,four types of source contributions had the smallest difference in low wind and high humidity weather.The secondary source contribution value varied greatly under low wind and low humidity conditions.Motor vehicles varied greatly under low humidity.（3）Comprehensive application of two PM_2.5 source apportionment methodsAccording to the advantages and disadvantages of the two methods and the different characteristics in different situations,this study proposed the following basic ideas for comprehensive utilization:To solve the problem that the PMF model could not resolve the source of the secondary components of PM_2.5,the contribution of various sources to the secondary component analyzed by the PSAT model could be used to decompose the PMF secondary source contribution value to various sources.In view of the problem that PMF could only give contributions from large categories of sources and it was difficult to distinguish refined sources,the relative proportion of the contribution of refined sources obtained by PSAT could be used to decompose the contribution of relatively reliable large-scale PMF sources,so as to obtain more refined source apportionment results.