Fig. 2: Predicted dust optical properties against laboratory data.

The index representing moderate hematite absorption is superior to other indices for predicting the complex refractive index (CRI) and single scattering albedo (SSA) of dust aerosols, as confirmed by optimal estimation techniques. a Comparison of the calculated imaginary part (y-axis) of the CRIs to laboratory data (x-axis) at the wavelength of 0.52 µm for the 19 sites (Fig. 1b). The black solid line is the 1:1 line. Black and gray dash lines represent a factor of 2 and 4 differences, respectively. The calculated dust imaginary CRIs were obtained using the volume averaging method, utilizing laboratory mineral fractions (Fig. 1a), distinguishing between hematite and goethite (CRI taken from a previous publication³), and considering various hematite CRIs (STRONG: violet, MODERATE: red, and WEAK: forest green). Also included are constant dust CRIs (CONSTANT: gray) with no spatial and temporal variation⁷⁹, and dust CRIs calculated based on the optimal hematite index (OPTIMAL: deep sky blue; derived using the optimal estimation techniques). For the calculated imaginary CRI of dust aerosols, the vertical standard-error bars denote uncertainty due to various error sources (see Methods). b Similar to (a) but for SSA (black dash lines represent a factor of 1.1 differences). The SSA was calculated based on the Mie Theory with the inputs of the calculated/constant dust CRI, as shown in (a), and the other inputs taken from the laboratory data⁴². The vertical standard-error bars represent uncertainty arising from error sources in the input to the Mie simulations, including those propagated from calculations for the imaginary CRI of dust aerosols (see Methods). The horizontal standard-error bars in both (a, b) indicate uncertainty in the laboratory data²². Metrics used to measure the distance between the calculations and measurements include the Pearson correlation coefficient (R; “*” denotes significance at the 95% confidence level) and root mean square error (RMSE). The RMSE obtained using a bootstrap procedure (see Methods) for imaginary dust CRI ranges between [0.0031, 0.010], [0.00034, 0.0014], [0.0018, 0.0021], and [0.00066, 0.00074] for the STRONG, MODERATE, WEAK, and OPTIMAL index cases, respectively. For dust SSA, it ranges between [0.069, 0.15], [0.022, 0.050], [0.064, 0.072], and [0.032, 0.034]. No RMSE range was estimated for the CONSTANT index case, since the resulting imaginary dust CRI and SSA show poor spatial variability.