The Application Of GCV Regularization Algorithm In Multi Wavelength Lidar Inversion Of Aerosol Particle Distribution

Atmospheric aerosols play an important role in the earth-atmosphere radiation budget through direct and indirect effects,and have an important impact on human life and production.Aerosol particle spectrum is an important parameter that characterizes the microphysical properties of the atmosphere.Its source,shape,scale,composition and spatiotemporal variation mechanism are difficult to define,and it is one of the uncertain components in the atmosphere.The study of aerosol particle spectrum inversion is of great significance for revealing particle concentration,haze origin,particle size spectrum distribution,air pollution prevention and climatic origin change.However,Aerosol particle spectrum inversion is still a key problem and research focus at present.Therefore,combining with multi-wavelength lidar detection data,we implement a new GCV regularization algorithm for aerosol particle spectrum inversion.In this paper,GCV regularization algorithm is used to invert aerosol particle spectrum,and the solution of linear equations is solved by the solution of optical parameter integral equation,linear constraint inversion and generalized cross validation.The single peak logarithmic normal distribution is used to simulate aerosol particle spectrum,and the selection principle of inversion window is proposed.According to different inversion windows,the difference of specific inversion effect is analyzed,which theoretically proves the feasibility of GCV regularization inversion of aerosol particle spectrum.Is obtained by multiple wavelength lidar measurements,three wavelengths(355 nm,532 nm,1064 nm)of extinction coefficient and backward scattering coefficient,the 6 optical parameters as input mode,the actual inversion of the troposphere within 1.2 km of aerosol particles distribution,at the same time and the APS-3321 particle spectrometer validation comparison between the measured the aerosol particle spectrum proves that this method conformed to the aerosol distribution,can be applied to practical inversion of aerosol particle spectrum.A set of aerosol particle spectrum inversion program is designed based on GCV regularization algorithm.Through Mieplot,extinction coefficients and backscattering coefficients of different particle radius at three wavelengths of 355 nm,532nm and 1064 nm were calculated,and an optical data set of 1.26 million and 2520 sets suitable for single mode lognormal distribution and double mode lognormal distribution was established under SQL Server.Design the back-end calculation program,use Matlab to implement the GCV regularization inversion algorithm,and realize the call to the database.Matlab GUI was used to design the program interface,and the inversion of particle number concentration spectrum,surface area concentration spectrum and mass concentration spectrum was realized by calling different wavelength and complex refraction index.The characteristics of aerosol concentration and spectral distribution were studied.The spectral distribution characteristics of aerosol number concentration and mass concentration in four seasons were analyzed.The particle number concentration and mass concentration in fine weather,haze weather and sand dust weather were analyzed.