To correctly simulate and retrieve dust distributions and estimate dust
impacts, global aerosol models and remote sensing retrieval algorithms
need accurate single-scattering properties of dust aerosols. However,
inconsistent and inaccurate quantifications of dust shape and shape
distributions are used in models and retrieval algorithms, generating
biases that propagate into the estimated dust distributions and dust
impacts. To improve models and retrieval algorithms, here we for the
first time account for the realistic dust shape distributions in
obtaining single-scattering properties of dust aerosols. We find that
approximating dust as spheres and neglecting dust asphericity, as most
global aerosol models do, result in substantial underestimations in the
extinction efficiency, the asymmetry factor, and the single-scattering
albedo for all dust sizes in both the shortwave and longwave spectra. In
addition, we find that the inaccurate quantification of dust shape in
retrieval algorithms causes them to generate an incorrect magnitude and
wavelength dependence of the linear depolarization ratio relative to
observations. Our new ellipsoidal dust optics accounting for realistic
shape distributions produce excellent agreement with the measured linear
depolarization ratio. Although these new dust optics show potential to
improve models and retrieval algorithms, they underestimate the
magnitude of the back-scattering intensity relative to laboratory and
field observations. This finding indicates that a realistic
quantification of dust body shape is not sufficient and that an accurate
quantification of dust surface texture is also critical to accurately
reproduce dust optical properties at back-scattering angles.