Aiming at the problem that the deviation in the fitting of aerosol spectrum distribution function can seriously affect the calculation and prediction of visibility, a segmental fitting method combining Junge spectrum and Deirmendjian spectrum was proposed to improve the fitting accuracy of aerosol spectrum distribution function, and its experimental verification was conducted in the atmospheric environment simulation chamber. Firstly, a multi-modal spectral distribution of aerosols is generated in the atmospheric simulation chamber, and a parametric model of aerosol spectral distribution is established. Then, the relative errors and correlation coefficients of the spectral functions fitted by the proposed method and the single spectral function method are compared and analyzed. Finally, the accuracy of the propsed method in calculating visibility is evaluated based on relative deviation and correlation. The results show that the average relative deviation between the visibility calculated using the proposed method and the observed value is 4.87%, with a correlation of R² = 0.9847, which verifies that the fitting results have good accuracy, stability and consistency with the observed visibility, and can provide a support for improving the accuracy of visibility calculation and prediction.