Radiance and flux simulations for mineral dust aerosols: Assessing the error due to using spherical or spheroidal model particles

The use of simple model particles to represent mineral dust aerosols may introduce errors in radiative transfer calculation. A modeling study has been performed to assess these errors. At a wavelength of 550 nm it is found that spheres can cause errors in the diffuse spectral radiance between −18% and 26% at the bottom of the atmosphere (BOA) and between −16% and 115% at the top of the atmosphere (TOA). These errors correspond to an uncertainty in the extinction optical depth τ between 0.7τ and 1.5τ (BOA) and (disregarding extreme errors in the solar nadir direction) between 0.5τ and 2τ (TOA). When using spheroidal model particles, the spectral radiance errors are reduced by roughly a factor of 2 (BOA) and are almost zero at the TOA over a large angular range. For a latitude of 20° we find, as a conservative estimate, that spherical model particles overestimate the yearly averaged TOA spectral net flux by almost 10 W m−2 nm−1. This error is mainly due to the misrepresentation of the aerosol phase function by the spheres. A rough estimate using Mie calculations indicates that this error is 5 times higher than (50% of) the error due to the uncertainty in the real (imaginary) part of the refractive index. Spheroids, on the other hand, only differ by 1.3 W m−2 nm−1 from the reference case. This indicates that using spherical model particles may cause significant positive errors in radiative forcing simulations for dust aerosols and that using simple spheroids can substantially improve the results.