Atmospheric Particulate Matter Concentration Monitoring By Combining GNSS/PWV And Meteorological Factors

With the economic and social development of human beings and the change of the natural environment,ashes in northern China are frequent,and Hebei Province is particularly serious.It has a significant impact on human life,production and health.Ash has become the most urgent environmental problem for Hebei's development.Particulate matter?PM?suspended in the atmosphere is the source of the occurrence of ash.Among them,inhalable particulate matter(PM₁₀)and fine particulate matter(PM_2.5)are the most important constituents of atmospheric particulate matter,which largely determines the quality of the Air Quality Index?AQI?.Research on PM_2.5 and PM₁₀ has become a hot topic.The most efficient method for monitoring PM_2.5 and PM₁₀ is a ground monitoring station that can monitor small-scale real-time monitoring.However,for economically underdeveloped areas,such as lack of coverage density,insufficient maintenance,and incomplete infrastructure,it is difficult to achieve reliable,stable high-time-resolution monitoring by means of this monitoring method alone.Other simple and effective monitoring methods with high time resolution are needed as a supplement.In the past two years,some scholars have found that there is a correlation between GNSS tropospheric delays and atmospheric particulate concentrations,and GNSS tropospheric delays can be solved in near real-time with high time fractions,which provides the possibility of using GNSS tropospheric delays to monitor atmospheric particulate concentrations.In this paper,based on the correlation between GNSS tropospheric delay and the concentration of atmospheric particulates,in taking the autumn and winter data of Baoding and Tangshan in Hebei province as an example,the sources of the correlations are analyzed,and the linear and Nonlinear regression model are established separately and compared.The main work and contributions are as follows:?1?Based on the two scales of hour and daily mean value separately,the regularity and correlation of atmospheric particulate matter concentration,atmospheric precipitable water volume,and surface water vapor content were systematically analyzed.The study found that the surface water vapor content and the atmospheric particulate matter concentration change at the time of 24-hours hourly average is overall similar,but with the atmospheric precipitable water in the daily mean value scale has the similar change rule.On the hour scale,there is a positive correlation between atmospheric precipitable water and atmospheric particulate matter concentrations.The correlation coefficient is mainly concentrated in the range of 0.3-0.5.There is also a positive correlation between surface water vapor content and atmospheric particulate matter concentration.The correlation coefficient is mainly between 0.5-0.7.It is confirmed that PWV and surface water vapor content can be used to monitor atmospheric particulate concentrations.At the same time,the correlation between atmospheric temperature and other meteorological elements was analyzed to provide a basis for subsequent modeling.?2?A multiple linear regression model based on atmospheric precipitable water,surface water vapor content and meteorological elements was proposed.After comparative analysis,the addition of atmospheric precipitable water and surface water vapor content can increase the model accuracy by about 20%.By analyzing the autocorrelation and partial autocorrelation of atmospheric particulate matter concentration,it is also proposed to add atmospheric particulate concentration in a multiple linear regression model.And the analysis demonstrates about 50%accuracy of the model can be improved.Considering that the error is accumulated in a recursive method,its actual effect need to verified through data.According to data analysis,within 6hours,the recursive model is reliable,the precision of PM_2.5 recursion is within 50-80?g/m³,and the precision of PM₁₀ recursive model can be within 80-110?g/m³.?3?Based on atmospheric precipitable water,surface water vapor content,atmospheric particle concentration basic value,meteorological elements,a generalized additive model for atmospheric particle concentration was proposed.Based on the daily periodicity of the atmospheric particle concentration itself and the nonmonotonic nonlinearity of the generalized additive model,it is proposed to add the hourly time value as an explanatory variable to the model.The results show that The addition of hourly time value can improve the accuracy of the model by about 5%;the precision of the PM_2.5 generalized additive model can be improved by 10%-30%compared with that of the linear model when no basic concentration value is added;the precision of the PM_2.5 general additive model can be improved by 15%compared to the linear model when the basic value is added.When interpolating data within 6 hours,the accuracy of the generalized additive model is basically the same as that of the linear model.After 6 hours,the generalized additive model with no concentration base value is more stable and higher in accuracy.To synthesize the actual utility of each model,when interpolating within6 hours of the data,a generalized additive model or linear model added to the basic value of concentration with a higher accuracy of the CV should be selected.For interpolating more than6 hours of data,the generalized additive model with no concentration base value is more reliable.This can basically ensure that the PM_2.5 concentration within 24 hours of missing data interpolation accuracy within 50-80?g/m³,PM₁₀ concentration within 24 hours of missing data interpolation accuracy within 80-110?g/m³.?4?The relationship between atmospheric particulate concentration and GNSS data processing was analyzed.The results show that there is no significant direct relationship between the concentration of atmospheric particulate matter and the precision of baseline of GNSS relative positioning and there is no correlation with accuracy of precision point positioning;the concentration of atmospheric particulates will not significantly affect the accuracy of GNSS retrieval of atmospheric precipitable water.After analysis,the amount of atmospheric precipitable water is due to the interaction with the surface water vapor content,which characterizes the vertical movement of the atmosphere and further shows the correlation with atmospheric particulate concentrations.