Effect of aerosol vertical distribution on aerosol-radiation interaction: A theoretical prospect

This study presents a theoretical investigation of the effect of the aerosol vertical distribution on the aerosol radiative effect (ARE). Four aerosol composition models (dust, polluted dust, pollution and pure scattering aerosols) with varying aerosol vertical profiles are incorporated into a radiative transfer model. The simulations show interesting spectral dependence of the ARE on the aerosol layer height. ARE increases with the aerosol layer height in the ultraviolet (UV: 0.25-0.42 μm) and thermal-infrared (TH-IR: 4.0-20.0 μm) regions, whereas it decreases in the visible-near infrared (VIS-NIR: 0.42-4.0 μm) region. Changes in the ARE with aerosol layer height are associated with different dominant processes for each spectral region. The combination of molecular (Rayleigh) scattering and aerosol absorption is the key process in the UV region, whereas aerosol (Mie) scattering and atmospheric gaseous absorption are key players in the VIS-NIR region. The longwave emission fluxes are controlled by the environmental temperature at the aerosol layer level. ARE shows maximum sensitivity to the aerosol layer height in the TH-IR region, followed by the UV and VIS-NIR regions. These changes are significant even in relatively low aerosol loading cases (aerosol optical depth ∼0.2-0.3). Dust aerosols are the most sensitive to altitude followed by polluted dust and pollution in all three different wavelength regions. Differences in the sensitivity of the aerosol type are explained by the relative strength of their spectral absorption/scattering properties. The role of surface reflectivity on the overall altitude dependency is shown to be important in the VIS-NIR and UV regions, whereas it is insensitive in the TH-IR region. Our results indicate that the vertical distribution of water vapor with respect to the aerosol layer is an important factor in the ARE estimations. Therefore, improved estimations of the water vapor profiles are needed for the further reduction in uncertainties associated with the ARE estimation.