## 1. Introduction

[2] Light scattering by atmospheric aerosols is a major factor in determining climate forcing on both global and regional scales. Climate modeling calculations generally rely on Mie theory to describe the light scattering and absorption properties of aerosols in modeling radiative transfer through the atmosphere. Mie theory is also commonly used in remote sensing data retrieval algorithms for determining aerosol loading and composition from both satellite data and LIDAR measurements [*Wang et al.*, 2002; *Del Guasta and Marini*, 2000]. Mie theory for light scattering by uniform spherical particles of size comparable to the wavelength of the light is well understood and straightforward to apply, given the size distribution and refractive index of the scattering particles [*Bohren and Huffman*, 1983]. However, there is much current interest in the effect of mineral dust aerosol on global climate [*Prospero*, 1999; *Prospero et al.*, 1989], and atmospheric mineral dust particles are not generally uniform spheres. Particles are often highly irregular in shape and composed of an inhomogeneous mix of different minerals [*Dick et al.*, 1998; *Claquin et al.*, 1999; *Sokolik and Toon*, 1999]. Experimental and theoretical studies carried out over many years have shown that particle shape effects lead to significant errors in the scattering phase function calculated by Mie theory, particularly at large scattering angles [*Holland and Gagne*, 1970; *Perry et al.*, 1978; *Asano and Sato*, 1980; *Jaggard et al.*, 1981; *Hill et al.*, 1984; *Mishchenko et al.*, 1997; *West et al.*, 1997; *Volten et al.*, 2001; *Kalashnikova and Sokolik*, 2004; *Kahnert and Kylling*, 2004; *Kahnert et al.*, 2007].

[3] The use of more advanced modeling methods to treat scattering by nonspherical particles, such as T-matrix based calculations or discrete dipole approximation (DDA) methods [*Mishchenko et al.*, 2002; *Draine and Flatau*, 1994] could help significantly to improve modeling accuracy. Because these theoretical methods are more numerically complex than Mie theory they have not been broadly applied, but as computer-processing power increases such calculations may become more commonplace [*Dubovik et al.*, 2002, 2006; *Kalashnikova et al.*, 2005]. These methods offer more accurate solutions than Mie theory, but still involve significant approximations and uncertainties.

[4] T-matrix based methods have been more widely explored and tested. The most advanced calculations involve averaging over polydisperse distributions of randomly oriented spheroids [*Mishchenko et al.*, 1997; *Dubovik et al.*, 2006] or various polyhedral particles [*Kahnert et al.*, 2001]. However, these methods require assumptions as to what range of particle sizes and shape parameters, and average refractive index values best approximate “typical” atmospheric mineral dust. DDA methods have also been applied to model the light scattering properties of particles characteristic of atmospheric mineral dust [*Kalashnikova and Sokolik*, 2004; *Kalashnikova et al.*, 2005; *Kalashnikova and Kahn*, 2006]. The DDA method offers more flexibility in modeling particles that are inhomogeneous and irregularly shaped, but the method is more computationally demanding, and is therefore restricted to smaller particles. Results from DDA modeling calculations have suggested that the neglect of sharp edges inherent in the assumption that particles can be treated as spheroids could lead to significant errors in some cases [*Kalashnikova and Sokolik*, 2004]. Other work [*Kahnert and Kylling*, 2004] suggests that the errors in the scattering phase function resulting from the spheroidal particle assumption may not be large. It is clear that more laboratory studies on well-characterized dust samples are needed to fully evaluate the reliability of these different theoretical approaches.

[5] Perhaps a more useful short-term approach to modeling the optical properties of “real” mineral dust aerosol is to use an empirical phase function based on laboratory data. Mishchenko and coworkers have recently developed an aerosol retrieval algorithm for AVHRR data [*Mishchenko et al.*, 2003] that relies on an experimentally determined aerosol phase function. The “synthetic” phase function was generated from laboratory scattering data of quartz aerosol particles [*Liu et al.*, 2003; *Volten et al.*, 2001]. However, a great deal of laboratory data is needed to determine a synthetic phase function that can best approximate results for authentic atmospheric dust. This again points to the need for additional laboratory studies of light scattering from well-characterized mineral dust particles.

[6] Significant experimental and theoretical work has been carried out to evaluate the scattering phase function for a range of mineral dust aerosols. Quantitative analysis is often limited by uncertainties in particle composition, shape, and size distribution. We have recently developed a new laboratory apparatus for measurement of the scattering phase function and linear polarization of mineral dust aerosol samples [*Curtis et al.*, 2007]. The apparatus includes a broadly tunable pulsed laser as the light source (a Nd:YAG pumped OPO), an elliptical mirror and CCD camera array for light detection (allowing us to measure the full angular scattering pattern on a shot-by-shot basis), and an aerodynamic particle sizer (for a real time determination of the particle size distribution in the aerosol flow). This apparatus allows for direct quantitative comparison of the observed light scattering from well-characterized dust samples with theoretical calculations based on the measured particle size distribution. In a recent paper we reported on testing and calibration of the system, and presented results for a synthetic phase function at 550 nm for quartz aerosol particles [*Curtis et al.*, 2007]. Here we extend these measurements of aerosol light scattering properties at 550 nm to a broader range of mineral dust aerosol particles including the silicate clays, kaolinite, illite, and montmorillonite, and the non-clay minerals, quartz, calcite, gypsum, and hematite, as well as to a sample of Arizona road dust. In each case the aerosol size distribution was simultaneously monitored with an aerodynamic particle sizer. The particle samples studied generally fall in the accumulation mode size range characteristic of mineral dust aerosols that are transported over long distances [*Prospero*, 1999; *Prospero et al.*, 1989]. Our results show significant discrepancies between the experimental and Mie theory-based phase functions. We also find intriguing differences in the scattering between the silicate clay components and the non-clay components of mineral dust aerosol in this particle size range. For the non-clay minerals the largest discrepancies are found at large scattering angles where Mie theory significantly overestimates the backscattering signal and, thus, underpredicts the asymmetry parameter, an important measure of aerosol light scattering. For the silicate clay minerals there is more variability in the comparison between experiment and theory.