PM_2.5 Quantitative Retrieval Based On The Remote Sensing And Ground-based Air Quality Measurement Data

Atmospheric aerosol is a general term for solid and gas particles suspended in air.As the key factor affecting the Earth's radiation budget and climate and environmental changes,aerosols have important impacts on regional and global atmospheric environment,climate and ecosystem.Among them,particulate matter with aerodynamic particle size less than 2.5 micron is called Fine Particulate Matter(PM_2.5),which is particularly important for environmental policy and public health research,and is a hot topic in atmospheric environmental research.With the rapid development of economy,severe and persistent haze weather has occurred frequently and extensively in China in recent years,which has become one of the regions with the most serious particulate pollution in the world.Using the extinction characteristics of PM_2.5 particles and retrieving aerosol optical depth?AOD?from satellite remote sensing data,the mass concentration of PM_2.5 near the ground can be obtained indirectly.It is an effective way to monitor the concentration distribution of PM_2.5 in the region.As a relatively new research field,remote sensing estimation of PM_2.5concentration can be obtained in a large range quickly.There is great advantage in the standardization of mass concentration data in space.However,due to the complexity of the atmospheric environment and the late start of remote sensing monitoring of particulate matter in China,and the long-term lack of large-scale and long-term PM_2.5ground monitoring data,the high-precision inversion of AOD in cities,the accuracy of remote sensing estimation model of particulate matter and its temporal and spatial applicability still need to be improved.Because of the large surface reflectance in urban area,it is difficult to determine the contribution of the ground surface,Therefore,the separation of the surface and the atmosphere is one of the key problems to be solved in high-precision aerosol AOD inversion.In this study,we use the V5.2 algorithm of dark target method to obtain the high-brightness surface reflectance in urban area,and define the red and blue band surface reflectance as a function of vegetation index and scattering angle,so as to solve the problem of separating the earth and atmosphere in urban bright surface.The partition of aerosol model is the second key problem to be solved in AOD inversion.Aiming at the problem that the current classification of aerosol types is extensive and can't meet the needs of large-scale regional inversion,a mathematical model for determining the volume percentage of aerosol components is proposed in this paper,and the volume ratio of aerosol components in the study area is defined.The results show that the model based on the self-defined component ratio is more comparative and less error value?<4%?than the traditional method.Aiming at the problems of redundancy of traditional look-up table structure,too tight step size setting and low efficiency,this study considers several coupling factors affecting atmospheric correction,such as sensor geometric imaging conditions,atmospheric conditions and surface elevation.In this paper,the influence of input variables on optical thickness is analyzed one by one with the help of 6S Model,so as to restrict the optical thickness reasonably.At the same time,select a reasonable look-up table difference algorithm,through research and analysis of the efficiency of different storage methods of look-up table,and research and analysis of multi-dimensional atmospheric parameter look-up table interpolation algorithm speed and accuracy.In order to maximize the precision of the look-up table,a multi-dimensional,efficient and high-precision look-up table for the parameters of AOD inversion is established.Compared with the traditional AOD inversion method,this paper improves the aerosol AOD inversion accuracy by determining the aerosol model accurately,restraining and adjusting the look-up table structure and step size reasonably.Due to the existence of a large number of soluble components in aerosols,the particle size,density,shape,complex refraction index and particle size distribution function will be changed under condensation and evaporation effects due to the influence of environmental humidity.At the same time,the extinction cross section of particles with the same concentration will increase,and the extinction characteristics of particles under different humidity conditions will be greatly different.Humidity correction is needed before fitting the relationship between the two quantitative models to reduce the uncertainty caused by the variation of aerosol extinction coefficient with humidity.In this study,meteorological data,AERONET data and visibility data were introduced to explore the hygroscopic growth law of aerosol extinction coefficient.The average mass extinction efficiency was used to describe the variation law of the overall extinction characteristics of urban composite aerosol particles with humidity.The relative humidity of the air was described based on the hygroscopic growth factor of the average mass extinction efficiency.On the basis of both data,the optimum hygroscopic growth model of PM_2.5 in the study area was fitted.AOD is the total integration of the extinction coefficients of aerosol particles in the vertical direction of light of a specific wavelength,while PM_2.5 measurements only represent the surface air quality.In order to reduce the uncertainty of the relationship model,it is necessary to explore the characteristics of the vertical distribution of aerosols,transform them into near-surface aerosol extinction coefficients,and match the PM_2.5 ground monitoring data in the vertical layer.In this study,the improved Peterson model is used to predict the seasonal elevation,and the change of atmospheric state is considered by adding time variables.For the impact of the model,the ASH is dynamically changed from fixed values to a more applicable AOD vertical correction model.After correcting the humidity and perpendicularity of AOD and PM2.5,this study used satellite remote sensing AOD to establish five fitting models,including linear function,logarithmic function,univariate quadratic function,power function and exponential function.The precision of each model was analyzed and verified.In order to verify the effect of meteorological factors on PM_2.5,the multi-linear and non-linear regression models were introduced to predict the concentration of PM_2.5 in the whole year and four seasons,respectively.Because of the strong non-linearity between the meteorological factors and PM_2.5,the prediction result of the machine learning algorithm is better than that of the multivariate regression model,which can better capture the non-linear influence between the mass concentration of PM_2.5 and the input factors.BP neural network is used to predict the mass concentration of PM_2.5,so as to achieve the purpose of quantifying the concentration of fine particulate matter based on AOD monitoring by remote sensing technology,and to make up for the shortcomings of traditional atmospheric monitoring by Point-Zone surface.Thus,a kind of real-time and dynamic monitoring new method of PM_2.5 concentration by satellite remote sensing is provided.