Research on the PMF source apportionment input data quantity based on high-throughput monitoring data
To explore the impact of data input amount variation on source analysis results,the high throughput atmospheric heavy metal monitoring data collected by the automatic continuous monitoring ICP-MS equipped with a gas displacement device in an industrial zone in Shanghai were input into the Positive Matrix Factorization(PMF)model according to different data volumes.The influence of data volume on the source analysis results was analyzed by using the Similarity Level(Ls)of true Q value(Qtrue)and theoretical Q value(Qtheo),and the consistency of source classification and the contribution rates with the distribution characteristics of pollution sources in the study area.The results of Ls show that,as the amount of data gradually decreases,the reasonable number of factors reduces simultaneously.The optimal factor number decreased from 5 to 3 when the input data volume reduced from 120 to 35.In general,heavy metal pollution in the study area is dominated by industrial production(61.63％),followed by dust(20.58％)and transportation(17.79％).Of these 64.44％industrial sources,cable manufacturing sources(10.89％),and battery manufacturing sources(20.93％)can be identified separately by increasing the amount of data input to the PMF model.Through the investigation of PMF model input data quantity variation,it is found that optimal pollution source analysis result can be obtained when the data input amount is 60-120,when the data input falls within the range of 35-40,only three factors can be reasonably identified and distinguishing pollution sources became challenging.Considering reducing the monitor cost and the period of data acquisition,reasonable source analysis results can be obtained when the input volume is 60-80.The short-term and high-throughput minute-level data set is beneficial to the PMF model to output high-precision and time-efficient source analysis results and is the best means to solve emergency pollution monitoring.
environmental studiesatmospheric heavy metal pollutantshigh-throughput monitoring dataPositive Matrix Factorization(PMF)modeldata input quantity
万方数据
维普
