Inversion And Analysis Of Aerosol Microphysical Characteristics Based On Principal Component Analysis Algorithm

The spatial distribution of aerosol microphysical properties is of great significance for the study of atmospheric physical and chemical processes.Multi-wavelength Raman lidar is an effective tool for detecting aerosols.The aerosol particle size distribution can be obtained by using the optical parameters detected with lidar,which can calculate the microphysical parameters of aerosol.However,this method has relatively high requirements for the inversion algorithm of aerosol particle size spectrum.The thesis mainly focuses on two parts.In the first part,the distribution of aerosol particle and microphysical parameters in vertical direction are studied and analyzed.Aerosol vertical distribution and microphysical properties are analyzed by using a set of 600 vertical profiles of aerosol number concentration with size distribution ranging from 0.1 to 3 ?m observed by airborne aerosol particle spectrometer.The multi-logarithm normal distribution model is applied to fit measured aerosol size distributions at different altitudes.Statistical analysis of fitting results is performed,which provide data support for the analysis of particle spectral inversion algorithm.In the second part,a method is introduced to derive integral properties of the aerosol size distribution and microphysical properties from aerosol extinction and backscatter data of multi-wavelength lidar,using an adapted form of the principal component analysis(PCA)technique.In addition,It is found that aerosol size distribution and microphysical properties can be usefully estimated in many kinds of particle spectrum simulation.The statistical analysis shows that the maximum inversion error can be controlled within 40%and the average error is within 20%.The anti-noise analysis of the algorithm is presented,and we find that the error of the backscatter coefficient of 355nm and the extinction coefficient 532nm has the greatest influence on the inversion result,the extinction coefficient of 355nm and the backscatter coefficient 532nm is the middle,and the 1064nm backscatter coefficient is the lowest.The principal component analysis algorithm(3?+2?)has lower requirements on optical parameters than the regularization algorithm.In this thesis,we use the extinction coefficient of 355nm,532nm and 1064nm to perform the inversion of microphysical parameters.It can be found that the relative error of effective radius can be controlled within 20%,and the relative error of volume concentration can be controlled within 40.The method can also be used to invert aerosol refraction index,with the real part ranging from 1.3 to 1.6 and the imaginary part ranging from 0.001 to 0.02,80%of the complex refractive index can achieve zero error inversion,the maximum relative inversion error of 20%complex refractive index is 30%.The actual airborne aerosol data is used to verify the algorithm.It is showed that the principal component analysis algorithm can realize the inversion of the microphysical parameters of the measured data.