Review of aerosol mass scattering efficiencies from ground-based measurements since 1990

Abstract

[1] We performed a survey of ground-based estimates of aerosol mass scattering efficiencies (α_sp, m² g⁻¹) for various aerosol species and size modes from peer-reviewed literature published since 1990. Accurate estimates of α_sp are important in aerosol modules of global circulation and chemical transport models to compute radiative forcing effects of aerosols and in chemical extinction budgets used for visibility regulatory purposes. The variety of techniques used to compute α_sp can be categorized into four basic methods. We separate reported α_sp on the basis of the methods used to derive them and normalize estimates of α_sp to a common dry relative humidity (RH, RH ≈ 0%) and mass composition basis, removing some of the variability inherent in α_sp reported under different monitoring conditions. The values of α_sp reviewed here represent common aerosol species and correspond to data from a variety of time periods and global locations. For the 60 studies reviewed, the average (and one standard deviation) α_sp (at visible wavelengths) for fine mode dry ammonium sulfate and dry ammonium nitrate are 2.5 ± 0.6 m² g⁻¹ and 2.7 ± 0.5 m² g⁻¹, respectively. The fine mode particulate organic matter α_sp is 3.9 ± 1.5 m² g⁻¹. The fine mode dust α_sp is 3.3 ± 0.6 m² g⁻¹and the fine mode dry sea salt α_sp is 4.5 ± 0.9 m² g⁻¹. Coarse mode α_sp estimates are also reported. The range of α_sp may reflect differences in aerosol morphology, age, physicochemical properties and mixing state; however, this survey suggests that the type of method used to derive α_sp can contribute considerably to the variability observed.

1. Introduction

[2] The ability of aerosols to scatter and absorb solar radiation has led to considerable research of their role in climate forcing, and has motivated regulatory efforts to mitigate their contribution to visibility degradation. The magnitude of their radiative impact depends largely on their optical properties; uncertainties in these properties contribute to uncertainties in climate forcing and visibility estimates. To reduce these uncertainties, numerous studies have been performed and many monitoring networks exist for the purpose of accurately assessing aerosol optical properties.

[3] Assessing the role of aerosols in radiative forcing and visibility degradation often requires reducing their physicochemical properties to a set of parameters that describe their optical properties (as a function of wavelength). These parameters are aerosol optical depth (τ, the vertical integral of the aerosol extinction coefficient, b_ext, which is the sum of the aerosol scattering, b_sp, and aerosol absorption coefficients, b_ap), single scatter albedo (ω_o, the ratio of the aerosol light scattering to total extinction, ω_o = b_sp/b_ext), asymmetry parameter (g, a measure of the amount of radiation scattered in the forward versus backward direction), scattering phase function which describes the angular dependence of scattered light, scattering enhancement factor (f(RH), a factor describing the dependence of b_sp on relative humidity, RH), and the mass extinction (or scattering or absorption) efficiency (α_ext, a measure of the aerosol light extinction per mass of the aerosol).

[4] As will be discussed in more detail in section 3, the aerosol mass extinction efficiency, which is the focus of this review, is dependent on particle composition and size distribution. Particle composition is important because it determines refractive index (particles with a higher refractive index scatter more light), and its hygroscopic properties. When assessing mass extinction efficiency, particle density is also important; particles with a lower density scatter more light on a per mass basis. Fine mode particles (typically defined as particles with diameters, D_p < 1.0 or 2.5 μm) have higher α_ext compared to coarse mode particles (D_p > 1.0 or 2.5 μm) because smaller particles scatter light more efficiently at visible wavelengths. The variability in mass extinction efficiency is typically more dependent on mass size distribution than on density or refractive index. Several investigators have reported on the size dependence of mass scattering efficiencies [e.g., Malm and Pitchford, 1997; Li et al., 2000; Hand, 2001; Malm et al., 2003]. Other authors report the dependence of mass scattering efficiencies on mass concentrations [Malm et al., 2003; Malm and Hand, 2007] or on light scattering coefficients [Lowenthal and Kumar, 2004]; however, for the most part, these relationships arise because of the dependence of b_sp and mass on particle size. Particles that are hygroscopic and have absorbed water have higher values of α_ext (with respect to dry mass) because of the increase in the particle cross section that leads to higher b_ext. All of these quantities are important for accurately estimating α_ext and thereby reducing uncertainties in aerosol radiative effects.

[5] Chemical extinction budgets require knowledge of mass extinction efficiencies for major aerosol species in order to estimate their contribution to overall extinction. An example of a chemical extinction budget is:

assuming an externally mixed aerosol model consisting of fine ammonium sulfate [AS], ammonium nitrate [AN], particulate organic material [POM], sea salt [SS], coarse mass [CM] and light absorbing carbon [LAC]. The scattering enhancement factor (f(RH)) takes into account RH effects for each hygroscopic species. One example of the use of this type of budget is its application in evaluating compliance under the Regional Haze Rule, which was implemented in 1999 to track trends in visibility in protected visual environments [U.S. Environmental Protection Agency, 1999]. Clearly, reasonable estimates of α_ext are crucial for regulatory purposes in this type of application.

[6] Estimates of α_ext are also important in aerosol modules of global circulation and chemical transport models that compute the radiative forcing effects of aerosols. Typically, aerosol radiative properties are computed by converting an aerosol mass to optical depth using a mass extinction efficiency as a function of aerosol species and relative humidity. The comparison of optical depths between models or to observational (i.e., remote sensing) data is dependent (among many factors) on the values of α_ext applied in the model. Recently, efforts to assess the aerosol modules in global models has led to the AeroCom model intercomparison study to evaluate aerosol properties ranging from emissions to optical properties [Kinne et al., 2003, 2005; Schulz et al., 2006]. Part of the model assessment discussed by Kinne et al. [2005] is the intercomparison of aerosol optical properties from twenty different models. Kinne et al. [2005] provide the values of α_ext used by each of the twenty models, and the fairly wide range of α_ext as a function of species demonstrates the need for more accurate estimates of α_ext as a function of composition, size distribution and RH.

[7] Mass extinction efficiencies are not measured directly, they are derived from measurements of aerosol optical and physicochemical properties. There are several approaches used to derive them, depending on the data available. Estimates of α_ext can be computed by applying Mie theory to aerosol size distributions with assumptions of optical and chemical properties, or from bulk mass measurements and measured light scattering or extinction coefficients from nephelometry or transmissometry, or using multilinear regression techniques and speciated or size segregated mass data. Understanding the contributions to uncertainties in α_ext that can arise from these methods is the motivation behind a critical investigation and comparison of estimates from these methods.

[8] We performed a survey of approximately 60 studies that report aerosol mass extinction efficiencies for various aerosol species that were performed under a variety of monitoring conditions, from pristine environments to urban regions around the globe. Although there are several studies that report mass extinction efficiencies from aircraft measurements, we focus specifically on ground-based data. Other less extensive reviews previously have been performed, for example, the Intergovernmental Panel on Climate Change [2001] reported summaries of α_ext for rough geographical locations and aerosol types (e.g., continental anthropogenic, biomass burning, etc.). However, this summary was based on reported size distribution data and does not reflect the range of values that are obtained when using other, more commonly used methods. The review also assumes a constant refractive index for all accumulation mode particles and does not report α_ext for individual species nor is it clear whether variations in monitoring conditions (e.g., relative humidity) have been taken into account. An even earlier review by White [1991] reported α_sp for individual species but derived these using mass apportionment schemes using measured mass and b_sp to compute α_sp. Charlson et al. [1999] discussed methods for deriving mass scattering efficiencies and briefly reviewed estimates for sulfate species. They did not account for differences in monitoring conditions (e.g., RH), and the estimates they report vary by a factor of two. Our study differs from previous reviews in that we distinguish efficiencies on the basis of the particular method used to derive them. We also normalize, where possible, efficiencies to the same mass type and RH, removing some of the inherent uncertainty that exists from different monitoring conditions.

[9] We organize this review by first presenting a description of the methods used to compute mass extinction efficiencies (section 2), followed by a summary of reported efficiencies for common aerosol species for each method (section 3). Section 4 summarizes the review, and recommendations for future measurements and analyses can be found in section 5.

2. Methods for Deriving Mass Scattering Efficiencies

[10] In the following sections we discuss the most common methods used to derive mass extinction efficiencies. The motivation for presenting these various methods is to emphasize the differences in the physical meaning of the derived mass extinctions efficiencies from each method and discuss under which situations they are directly comparable. For the remainder of the paper, we will refer to mass scattering efficiencies instead of mass extinction efficiencies, as we focus only on aerosol scattering in the visible spectral range. For a review of aerosol absorbing properties the reader is encouraged to refer to the recent article by Bond and Bergstrom [2005].

[11] We apply the following definitions of mass scattering efficiency pertaining to the type of aerosol mixture assumed. Typically, aerosol species are assumed to be mixed externally, such as shown in equation (1) for a chemical extinction budget. The simplest example is a particle composed of a single chemical compound such as ammonium sulfate; for this case the efficiency is referred to as a “mass scattering efficiency.” However, realistically, particles in the atmosphere are composed of a variety of species. These types of particles are referred to as internal mixtures, and their mass scattering efficiencies are termed “specific mass scattering efficiencies”; for example, a mixed fine mode aerosol has a specific mass scattering efficiency that corresponds to the mixture mass. Internally mixed particles can also be externally mixed from other particle populations; the most obvious case would be internally mixed fine mode aerosols externally mixed from coarse mode aerosols. We will also apply the term “specific mass scattering efficiency” to aerosols that include water mass. For example, an inorganic salt particle with associated water mass has a “specific mass scattering efficiency” that corresponds to the combined aerosol and water mass.

2.1. Theoretical Method

[12] Mass extinction (or scattering) efficiencies can be calculated from size distributions of particle mass or number concentrations. For the following discussion, the aerosol size distribution refers to an externally mixed aerosol of a single chemical component, j. For an aerosol number distribution f_N,j(D_p) and complex refractive index n_j = m_j + ik_j, the light extinction coefficient (b_ext,_j) is computed with equation (2) and has units of inverse length:

[13] The Mie extinction efficiency is given by Q_ext(n_j, D_p, λ) and is a function of refractive index, particle diameter (D_p), and wavelength (λ), assuming spherical particles. Writing equation (2) in terms of the mass size distribution of species j and normalizing it by the total mass concentration of species j (M_j) results in a mass extinction efficiency for species j (α_Mj):

where f′_M,j(D_p) is the normalized mass size distribution and the single particle mass extinction efficiency (α_ext,j) has units of m² g⁻¹ and is a function of the Mie extinction efficiency, particle diameter and species density (ρ_j):

If f′_M,j(D_p) does not vary with total mass concentration, then the average mass extinction efficiency for species j can be considered a constant and the light extinction coefficient corresponding to species j can be written as

The total extinction is a linear combination of the species mass concentrations:

Equation (6) also holds for an internally mixed aerosol where the chemical species are mixed in fixed proportions to each other, the index of refraction is not a function of composition or size, and the specific mass extinction efficiency is prorated to its chemical constituents on the basis of their relative densities.

[14] The theoretical method requires assumptions of the chemical form, and the optical and mixing properties of the aerosols. If size distributions are measured using an impactor, even though the particles are most likely internally mixed, an externally mixed aerosol is assumed and the efficiencies derived are mass scattering efficiencies. However, if number size distributions are obtained by mobility measurements (differential mobility analyzer, DMA), optical measurements (optical particle counter, OPC) or aerodynamic measurements (aerodynamic particle sizer, APS), an internally mixed aerosol is usually assumed, but perhaps for separate size modes, (and specific mass scattering efficiencies are derived). Species mass scattering efficiencies can be computed from number size distributions of internally mixed aerosols by apportioning a fraction of the size distribution (and therefore b_ext) to a given species [e.g., Quinn et al., 2004].

[15] If finely size-resolved number size distributions are available (e.g., DMA data), the theoretical method provides the most accurate estimate of efficiencies out of those discussed in this section. This method has been shown to accurately reconstruct light scattering on fine time scales [e.g., Hand et al., 2002; Quinn et al., 2004; Malm et al., 2005; McMeeking et al., 2005b], and it has the added advantage of allowing for the investigation of the functional dependence of efficiencies on particle size, composition, and humidity. On the other hand, for impactor and some optical measurements, the size resolution of the size distributions can be fairly coarse, especially in the size range that is more efficient for scattering light, reducing the sensitivity of the method. A major disadvantage to the theoretical method is that size distribution measurements are more analytically intensive and not often available in routine monitoring programs, so mass scattering efficiencies computed by this method are not as common.

2.2. Partial Scattering Method

[16] Estimating the change in extinction due to the removal or addition of a single species is referred to as a “partial mass extinction efficiency” (α_ext,part).

Estimating mass extinction efficiencies as the change in b_ext resulting from the removal (or addition) of a single species (M_j) is different from apportioning a fraction of measured extinction to a chemical species or a mixture of species (i.e., equation (1)). As species are added or removed from an aerosol, the efficiency is dependent on the change in composition and mass as well as on change in aerosol size [White, 1986; Malm and Kreidenweis, 1997].

[18] Computing partial mass scattering efficiencies by this method requires a model to account for the effects of removal of mass on light scattering coefficients, such as the ELSIE model (Elastic Light Scattering Interactive Efficiencies) [e.g., Sloane, 1983; Sloane et al., 1991; Lowenthal et al., 1995; Omar et al., 1999]. The theoretical basis for the ELSIE model is similar to the theoretical model described previously in that size distributions are incorporated; however, the ELSIE model has two options for the change in aerosol mass. Mass can be removed (or added) by assuming that the particle size changes but not the particle number, or that the particle number changes but not the particle size [Lowenthal et al., 1995]. Unless the option is chosen to keep the size constant, the efficiencies derived from this method cannot be compared to efficiencies from other methods. Although this method may be useful in a regulatory context, in practice it is not possible to measure the change in light scattering due to the removal or addition of an aerosol mass concentration from the atmosphere.

2.3. Measurement Method

[19] The simplest method for computing efficiencies is with measured mass concentrations from filter samples and measured light scattering coefficients from nephelometry or light extinction coefficients from transmissometry. For an aerosol population, a specific mass scattering efficiency α_{sp_spec} can be defined as the ratio of a light scattering coefficient (b_sp) corresponding to the mass concentration (M) of that population.

[20] The average specific mass scattering efficiency can be estimated by dividing the average scattering coefficient by the average mass concentration for a given sampling period, or the slope from a linear regression of b_sp and M can be interpreted as the specific mass scattering efficiency. Specific mass scattering efficiencies derived by this method represent average conditions of an aerosol that could be changing because of variations in relative humidity, size distribution, and composition during the sampling period. Mass concentrations used in this method could be either gravimetric mass or the sum of chemically analyzed masses, or be computed from integrated volume distributions and an assumed density.

2.4. Multilinear Regression (MLR) Method

[21] A number of investigators have taken advantage of the form of equation (6) (b_ext = Σα_MjM_j) to construct a multilinear regression (MLR) model with b_ext as the independent variable, and the measured aerosol mass concentrations for each species j (M_j) as the dependent variables [e.g., White, 1986; Eatough et al., 1996; Malm and Day, 2000; Quinn et al., 2001; Vasconcelos et al., 2001]. The regression coefficients are then interpreted as mass extinction (or scattering or absorption) efficiencies, and this method is referred to as the “MLR” method. The assumptions required in this formulation are that all the aerosol species contributing to extinction are included, equation (6) is a reasonable approximation to the relationship between extinction and the various aerosol species, the number of samples is large enough to give stable results, and the concentrations of the species are uncorrelated.

[22] Regression-derived efficiencies are subject to a variety of systematic and random errors. Meteorologically driven fluctuations of the existing aerosol concentrations may cause different aerosol species to be collinear, which can make regression results very sensitive to the choice of species included in the analysis. Also, regression results can be biased by uncertainties in the data. The coefficients for species with lower measurement uncertainties (e.g., sulfates) can be artificially high, while coefficients for species with larger measurement uncertainties (e.g., organics) may be too low [White and Macias, 1987].

[23] Collinearity issues can affect fine mode mass scattering efficiencies associated with coarse mode species, depending on how the regression is performed. For example, in the chemical extinction budget described by equation (1), aerosol total extinction (b_ext) is regressed against a number of fine mode species (e.g., sulfate and POM) as well as coarse mode species measured in the fine mode (e.g., soil dust and sea salt). In addition to these species, coarse mass (CM) is included in the regression. Soil dust and sea salt are typically characterized by formulas based on their chemical composition from measurements of individual species, while CM is derived by the difference in PM₁₀ and PM_2.5 gravimetric mass and represents a mixed coarse mode aerosol. Collinearities occur because the species used to construct soil dust and sea salt mass also exist in coarse mass, therefore the regression may result in inflated values of mass scattering efficiencies for these species. Fine mode soil dust mass scattering efficiencies may also be inflated because of the high accuracy by which some of the individual soil dust species typically are measured, similar to the issue addressed in the previous paragraph.

[24] Total extinction by aerosols is fairly insensitive to the microscopic structure of aerosols (whether they are internally or externally mixed); however, as White [1986] and Malm and Kreidenweis [1997] show, the apportionment of scattering by more than one species to total scattering does depend on whether the aerosols are internally or externally mixed. With an externally mixed aerosol containing more than one species, each species would be assigned its own mass scattering efficiency (α_sp) (as shown in equation (1)) and the fraction of scattering a given species contributes to b_sp is straight forward to compute. However, with an internally mixed aerosol, one specific mass scattering efficiency is applied to all species in a size mode. For example, in equation (1) a single coarse mode specific mass scattering efficiency is applied to coarse mass, rather than apportioned to separate coarse mode species. If the specific mass scattering efficiency is computed as a mass-weighted average of α_sp from each species, under the internally mixed scenario the amount of scattering apportioned to each species (as if it was externally mixed) will not be the same as assuming an externally mixed scenario initially. To avoid this discrepancy, the specific mass scattering efficiency of the internally mixed aerosol is prorated to its chemical species on the basis of their relative densities of the species. This apportionment is based on the assumption that the species have similar size distributions and refractive indices and that their densities are known.

[25] Multilinear regressions can be performed assuming internal or external mixtures. The chemical extinction budget shown in equation (1) is an example of both an external and internal mixture in that the fine mode species are assumed to be externally mixed while the coarse mode species are grouped as an internal mixture. Performing the MLR technique as an external mixture is sensitive to the above mentioned issues regarding data collinearity and measurement uncertainties. Assuming an internal mixture avoids some of these issues by grouping aerosol species together in the regression, making the regression coefficients less sensitive to the uncertainties in the data. However, performing a regression assuming internally mixed aerosols for different size modes [e.g., White et al., 1994; Chow et al., 2002] provides coefficients that correspond only to specific mass scattering efficiencies of mixed aerosol modes. To interpret the efficiencies of species of internally mixed aerosols as if they are externally mixed, the specific mass scattering efficiency can be prorated to its chemical species on the basis of their relative densities [Malm and Kreidenweis, 1997; Malm and Hand, 2007], as discussed in the previous paragraph.

2.5. Interpretation and Application of Derived Mass Scattering Efficiencies

[26] One must be careful when comparing and applying mass scattering efficiencies derived from the various approaches described above. This point is especially true for mass scattering efficiencies derived from bulk measurements where particles are sized into fine and coarse modes (typically, D_p < 1.0 μm and D_p > 1.0 μm, respectively), compared to those that are theoretically calculated using fine size resolution number size distributions that are obtained from DMA, OPC or similar type measurements. All bulk size-selective sampling systems are less than ideal in that a fraction of particles from the coarse mode will be collected in the fine mode and vice versa. An MLR type analysis with b_ext as the dependent variable and fine and coarse mass as independent variables will necessarily be responsive to all particles collected in the fine and coarse modes, whereas theoretical calculations of α_sp from fine size resolution instruments usually include only the mass in the size region of interest.

[27] As an example, consider soil dust approximated with a lognormal mass size distribution with a refractive index of 1.8 and density of 2.3 g cm⁻³. Suppose this ambient aerosol is sampled with a particle sampling system with a 1.0 μm size cut. Figure 1 shows the fine mass scattering efficiency of soil dust as a function of mass mean diameter (D_g) and geometric standard deviation (σ_g). A cyclone collection efficiency associated with the IMPROVE (Interagency Monitoring of Protected Visual Environments [Malm et al., 1994a]) sampler was assumed. Notice that for coarse mode mass size distributions, the fine mass scattering efficiencies range from 0.4–1.2 m² g⁻¹. These low α_sp values are primarily associated with coarse mass collected on the fine mode substrate. An MLR or similar analysis using these types of data would yield fine mass scattering efficiencies near 1 m² g⁻¹ that actually correspond to aerosols in the coarse mode.

Figure 1

Fine mode mass scattering efficiency (m² g⁻¹, 550 nm wavelength) simulated for a lognormal soil size distribution as a function of geometric mass mean diameter and geometric standard deviation (σ_g). A realistic size-resolved cyclone collection efficiency with a 1.0 μm diameter size cut has been applied in the calculation. A soil refractive index of 1.8 and soil density of 2.3 g cm⁻³ is assumed.

[28] Now suppose that the same aerosols were measured with high size resolution instrumentation such as a DMA or OPC. Theoretical estimates of fine and coarse mass scattering efficiencies using these data are typically computed assuming an ideal size differentiation between the fine and coarse mode, in that no coarse mass is included in the fine mode, and vice versa. This type of an assumption yields quite different fine mass scattering efficiencies associated with a coarse mass size distribution compared to the previous example. For example, estimates of α_sp shown in Figure 2 were computed for the same size distributions shown in Figure 1, except an ideal size cut was assumed. For this case, only the mass and aerosol scattering associated with particles with D_p < 1.0 μm were included in the calculation. Notice that the fine particle mass scattering efficiencies associated with a coarse mode mass size distribution are significantly greater, by factors of 2–3, compared to α_sp computed for bulk measurements with a realistic collection efficiency assumed.

Figure 2

Fine mode mass scattering efficiency (m² g⁻¹, 550 nm wavelength) simulated for a lognormal soil size distribution as a function of geometric mass mean diameter and geometric standard deviation (σ_g). An idealized cyclone collection efficiency (perfect step-function) has been applied in the calculation. A soil refractive index of 1.8 and soil density of 2.3 g cm⁻³ is assumed.

[29] It would be inappropriate to apply a theoretically derived mass scattering efficiency in combination with bulk measurements of fine mass when estimating fine mode light scattering coefficients for species that are predominantly found in the coarse mode (e.g., dust and sea salt). The fine mode light scattering coefficient would be overestimated by 200–300%. Similar issues exist for estimating coarse mode light scattering coefficients associated with a fine mode aerosol.

3. Survey of Mass Scattering Efficiencies

[30] A review of mass scattering efficiencies has been organized so that values are separated on the basis of the four different methods described above. Within each method, estimates have been separated by aerosol species (e.g., sulfate, dust) on the basis of reported chemical composition, and size mode (fine, coarse or total mode). A major difficulty in interpreting values of α_sp is that the basis for reporting them varies from study to study, specifically the molecular form of the assumed species (e.g., sulfate ion versus degrees of neutralized sulfate) and associated water associated at a given relative humidity, therefore we performed normalizations to a common mass and RH basis.

[31] The normalizations were performed for sulfate, nitrate, particulate organic matter (POM) and sea salt. Normalizations for sulfate and nitrate included converting α_sp to dry (RH ≈ 0%) fully neutralized ammonium sulfate and ammonium nitrate if they were reported under different conditions. This normalization was only possible if the authors of a given study reported the form of sulfate or nitrate (as ions or ammonium to sulfate molar ratios, for example), which was not always the case. Normalizing to a dry basis required that the relative humidity of the measurement was reported, which was also not always the case. The RH normalization required the assumption of a size distribution in order to compute the f(RH) curves used to “dry out” the efficiency. We assumed an accumulation mode lognormal size distribution with a mass mean diameter of 0.4 μm and geometric standard deviation of 1.9. Water was assumed to be associated with the particle continuously to 0% RH on the basis of the no-solids curve of the AIM model (Aerosol Inorganics Model [Clegg et al., 1998]). The f(RH) curve for ammonium sulfate was also applied to ammonium nitrate [Malm et al., 1994a]. If efficiencies were computed on the basis of a wet mass (water + sulfate), we “dried” out the mass using diameter growth curves (D/D_o), so that both mass and b_sp corresponded to dry conditions.

[32] Sea salt efficiencies were normalized to dry conditions also. NaCl D/D_o curves from Tang [1997] were used and interpolated to RH = 0% below 40% RH. Sea salt size distributions for a given study (when reported) were used to compute an f(RH) assuming a refractive index and density of 1.54 and 2.165 g cm⁻³, respectively [Tang, 1996]. If no sea salt size distributions were reported, a lognormal size distribution with a mass mean diameter and geometric standard deviation of 2.5 μm and 2, respectively, were used [Quinn et al., 1996]. We also normalized wet mass to dry conditions if the efficiencies were reported that way. The designation of sea salt aerosols is based on chemical composition and sea salt markers. We did not convert all values to the same sea salt mass basis as each individual study computed sea salt differently and a multiplicative correction factor was not possible.

[33] Particulate organic material (POM) is computed by multiplying organ carbon concentrations by a molecular weight per carbon weight ratio. Most of the reported values of POM mass scattering efficiencies were computed with multipliers of 1.2, 1.4, 1.6, or 2.1, based on the work by Turpin and Lim [2001] who suggests a value of 1.6 for urban aerosols and 2.1 for nonurban aerosols. We normalized the POM mass scattering efficiencies to a multiplier of 1.8 [Malm and Hand, 2007] so that all values can be directly compared. We did not normalize efficiencies for a given study when the authors did not report the value of the multiplier used to compute POM. POM was assumed to be nonhygroscopic. If a given study assumed POM was hygroscopic, we did not normalize these estimates to a dry value.

[34] No RH normalizations were performed for dust, as it was usually reported for dry conditions (assumed to be nonhygroscopic). Dust aerosols were defined by composition, not sampling location. Each study assumed a different dust mass composition, therefore it was not possible to perform any normalization for mass. Also, mixed aerosol fine, coarse and total mode specific mass scattering efficiencies were not normalized because individual species mass concentrations were not known and RH corrections could not be applied. Fine mode mass scattering efficiencies typically referred to aerosols with aerodynamic diameters less than 1.0, 1.1, 2.5, or 3 μm, and coarse mode efficiencies could correspond to upper diameters of 10 or 15 μm. We did not normalize efficiencies on the basis of the differing size ranges within a mode.

[35] Estimates of coarse or total mode mass scattering efficiencies derived using nephelometer measurements are affected by angular nonidealities such as the truncation of near forward scattering. Because most models of nephelometers do not sense light in the near-forward direction they underestimate the amount of coarse particle scattering by 50% or more [Anderson and Ogren, 1998], depending on particle size. Corrections are sensitive to and require knowledge of the coarse mass size distribution. Many of the authors of the studies reviewed here did apply truncation corrections and we did not attempt to correct for this effect for the studies that did not. Uncorrected estimates of coarse mode α_sp could be artificially low because coarse particle scattering is reduced while coarse particle mass is not.

[36] The results from the survey are reported in several tables (Tables 1–6) on the basis of the method used to compute them. The tables can be read in the following way: the mass scattering efficiencies are listed by species (e.g., sulfate, POM) or as specific mass scattering efficiencies for mixtures of aerosols by size mode (i.e., fine, coarse, or total). The tables also list the study location, study name (if appropriate), aerosol characteristics (e.g., smoke) if noted, the study time period, and the citation. The footnote for each entry provides further details of the measurements, size ranges, relative humidity of the measurements, whether nephelometer truncation corrections were applied, and the wavelength at which the measurements were made, if it was other than 550 nm. When possible, we compute and report the average efficiency (and one standard deviation) if several values have been reported in that study, for example for different air masses (as noted in the footnotes). Except for averages, the original values are listed as reported; however, we also provide the normalized values if computed.

Table 1. Summary of Mass Scattering Efficiencies (α_sp) Using the Theoretical Method^aa Fine, coarse, and total modes refer to a mixed aerosol in those respective size ranges. POM refers to particulate organic material. Species correspond to fine mode unless denoted by the column heading or by a footnote. The “Type” column refers to the type of size distribution measurement: either “I” for impactor, “M” for mobility/aerodynamic, and “A” for apportionment method that uses a combination of both. The reference is given in the last column, and the footnotes attached to the locations give further details of the measurements. All values reported for a wavelength of 550 nm unless otherwise stated in the footnotes. Size ranges, chemical forms and relative humidity (when reported) are stated in the footnotes. In the “sulfate” column, bold values correspond to dry ammonium sulfate as reported in the manuscript. In the “POM” column, bold values are normalized to POM multiplier of R_oc = 1.8 by us or as reported in the manuscript. Unit is m² g⁻¹.

Fine Mode	Coarse Mode	Total Mode	Sulfate	POM	Dust	Sea Salt	Location/Study/Characteristic	Time	Type	Reference
		1.4 ± 0.7					Baltimore, MD, Canadian smoke (Supersite)^bb Dp < 20 μm. Scanning mobility particle sizer and aerodynamic particle sizer used for size distribution measurements. Average and standard deviation of three time periods. Relative humidity not stated. Wavelength of 530 nm.	4, 6, and 10 Jul 2002	M	Adam et al. [2004]
			3.8, 2.4^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.				Great Smoky National Park, TN (SEAVS)^dd Dp < 3 μm. Sulfate is ion. Optical particle counter and apportionment techniques used for size distribution measurements. Dry relative humidity, RH < 15%. Wavelength of 632.8 nm.	Jul–Aug 1995	A	Ames et al. [2000]
					1.07^ee Corresponds to total modes (fine + coarse).	0.52^ee Corresponds to total modes (fine + coarse). ^,^ff Value not included in total average provided in Table 5 with reason listed in footnote for that entry.	Sal Island, Cape Verde^gg Dp < 8.75 μm. EGAI 80 low-volume cascade impactor used for size distribution measurements. Dry relative humidity, value not listed. Dust derived assuming Si is one third the mineral aerosol. Sea salt is 3*Na from marine sources. Sea salt not included in average because RH value not listed. Wavelength of 670 nm.	Dec 1991, Jan 1994, Dec 1994 to Feb 1995	I	Chiapello et al. [1999]
3.4	0.6						Big Bend National Park, TX (BRAVO)^hh Fine mode corresponds to Dp < 1 μm. Differential mobility analyzer and aerodynamic particle sizer used for size distribution measurements. Dry relative humidity (<20%). Wavelength of 530 nm.	Jul–Oct 1999	M	Hand [2001]
			3, 0.9^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.				Santa Maria, Azores^ii Sulfate is ion for 0.18 < Dp < 10 μm. MSP Model 100 MOUDI impactor used for size distribution measurements. Relative humidity = 73%.	10 Jun 1992	I	Howell and Huebert [1998]
			28, 2.4^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.				Santa Maria, Azores^jj Sulfate is ion for 0.18 < Dp < 10 μm. MSP Model 100 MOUDI impactor used for size distribution measurements. Relative humidity = 95%.	13 Jun 1992	I	Howell and Huebert [1998]
			13, 3.3^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.				Santa Maria, Azores^kk Sulfate is ion for 0.18 < Dp < 10 μm. MSP Model 100 MOUDI impactor used for size distribution measurements. Relative humidity = 81%.	19 Jun 1992	I	Howell and Huebert [1998]
			9.1, 2.5^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.				Santa Maria, Azores^ll Sulfate is ion for 0.18 < Dp < 10 μm. MSP Model 100 MOUDI impactor used for size distribution measurements. Relative humidity = 77%.	20 Jun 1992	I	Howell and Huebert [1998]
			5.2, 1.6^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.				Christmas Island^mm Sulfate is ion for 0.18 < Dp < 10 μm. MSP Model 100 MOUDI impactor used for size distribution measurements. Relative humidity = 69%.	3 Aug 1994	I	Howell and Huebert [1998]
			6.9, 6.0^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^ff Value not included in total average provided in Table 5 with reason listed in footnote for that entry.	5.3, 2.9^ff Value not included in total average provided in Table 5 with reason listed in footnote for that entry.	0.64^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	1.3^ff Value not included in total average provided in Table 5 with reason listed in footnote for that entry. ^,^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	Waliguan Mountain, Qinghai Province, China^oo 0.056 < Dp < 18 μm. MOUDI impactor used for size distribution measurements. Sulfate is bisulfate in the accumulation mode. Water soluble organic carbon is in the accumulation mode. Dust (CaCO3) and sea salt (NaCl) correspond to coarse mode. Relative humidity not stated. Roc value not given. Values not included in the average because of unknown RH and Roc.	Oct–Nov 1997, Jan 1998	I	Li et al. [2000]
			2.23				Hopi Point, Grand Canyon National Park, AZ^pp Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be dry ammonium sulfate. Dry relative humidity, RH = 0%. Wavelength not stated.	Jan–Feb 1998	I	Malm and Pitchford [1997]
			2.59				Hopi Point, Grand Canyon National Park, AZ^qq Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be wet ammonium sulfate. Relative humidity of 48.3%. Wavelength not stated.	Jan–Feb 1998	I	Malm and Pitchford [1997]
			4.81				Hopi Point, Grand Canyon National Park, AZ^rr Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass is dry ammonium sulfate, light scattering includes water. Relative humidity of 48.3%. Wavelength not stated.	Jan–Feb 1998	I	Malm and Pitchford [1997]
			2.03				Meadview, AZ^ss Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be dry ammonium sulfate. Dry relative humidity, RH = 0%. Wavelength not stated.	summer 1992	I	Malm and Pitchford [1997]
			2.05				Meadview, AZ^tt Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be wet ammonium sulfate. Relative humidity of 25.3%. Wavelength not stated.	summer 1992	I	Malm and Pitchford [1997]
			2.11				Meadview, AZ^uu Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass is dry ammonium sulfate, light scattering includes water. Relative humidity of 25.3%. Wavelength not stated.	summer 1992	I	Malm and Pitchford [1997]
			2.63				Shenandoah National Park, VA^vv Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be dry ammonium sulfate. Dry relative humidity, RH = 0%. Wavelength not stated.	summer 1990	I	Malm and Pitchford [1997]
			3.81				Shenandoah National Park, VA^ww Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be wet ammonium sulfate. Relative humidity of 82.5%. Wavelength not stated.	summer 1990	I	Malm and Pitchford [1997]
			18.23				Shenandoah National Park, VA^xx Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass is dry ammonium sulfate, light scattering includes water. Relative humidity of 82.5%. Wavelength not stated.	summer 1990	I	Malm and Pitchford [1997]
			2.4				Great Smoky National Park, TN (SEAVS)^yy Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be dry ammonium sulfate. Dry relative humidity, RH = 0%. Wavelength not stated.	Jul–Aug 1995	I	Malm et al. [2000]
			3.58				Great Smoky National Park, TN (SEAVS)^zz Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be wet ammonium sulfate and calculated wet scattering. Relative humidity of 75%. Wavelength not stated.	Jul–Aug 1995	I	Malm et al. [2000]
			9.64				Great Smoky National Park, TN (SEAVS)^aaaa Dp < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be dry ammonium sulfate and calculated wet scattering. Relative humidity of 75%. Wavelength not stated.	Jul–Aug 1995	I	Malm et al. [2000]
			3.15, 2.96^c				Big Bend National Park, TX (BRAVO)^bbbb 0.18 < Dp < 18 μm. MOUDI impactor used for size distribution measurements. Dry ammoniated sulfate (letovicite). RH = 0%. Wavelength not stated.	Jul–Oct 1999	I	Malm et al. [2003]
4.21							Yosemite National Park, CA (YACS)^cccc 0.1 < Dp < 1 μm. Differential mobility analyzer used for size distribution measurements. Dry relative humidity, value not stated. Wavelength not stated.	Jul–Sep 2002	M	Malm et al. [2005]
					0.48^ff Value not included in total average provided in Table 5 with reason listed in footnote for that entry. ^,^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).		Izana, Tenerife, Canary Islands^dddd Dust is coarse mode (Dp > 0.6 μm). Scanning mobility particle sizer and aerodynamic particle sizer used for size distribution measurements. Relative humidity less than 30%. Assumed all of coarse mode is dust. Dust not included in average because coarse mode could include other species.	Jul 1999	M	Maring et al. [2000]
4.3							Yosemite National Park, CA (YACS)^eeee 0.04 < Dp < 2 μm. Differential mobility analyzer used for size distribution measurements. Dry relative humidity (RH < 15%). Wavelength not stated.	Jul–Sep 2002	M	McMeeking et al. [2005a]
3.6	0.87		5,^ee Corresponds to total modes (fine + coarse). 2.8,^ee Corresponds to total modes (fine + coarse). 2.8^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^ee Corresponds to total modes (fine + coarse).				Pacific Ocean and Cheeka Peak, WA (PSI and MAGE)^ffff Dp < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Sulfate is ion and sulfate aerosol (non-sea-salt-SO4+NH4+H2O), and normalized to dry ammonium sulfate aerosol. Mixed fine corresponds to residual mass for Dp < 1 μm and mixed coarse corresponds to residual mass for 1 < Dp < 10 μm. Relative humidity = 30%.	1991 and 1992	A	Quinn et al. [1995]
3.2 ± 0.3			0.84 ± 0.09, 1.5 ± 0.2, 0.8 ± 0.2^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.			4.1 ± 0.4, 3.7 ± 0.4^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	Southern Ocean Region (ACE 1)^gggg Dp < 1 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation of various air masses. Sulfate mass is sulfate aerosol (non-sea-salt-SO4+NH4+H2O), sulfate ion and normalized to dry ammonium sulfate from ion, respectively. Fine mode is non-sea-salt-SO4 plus sea salt. Sea salt is a combination of Na+, K+, Mg+2, Ca+2, and sea salt sulfate and corresponds to the tail of the coarse mode. Relative humidity varies from 30 to 45%.	Nov–Dec 1995	A	Quinn et al. [1998]
4.8 ± 0.5			3.5 ± 0.7; 4.4 ± 0.9, 1.9 ± 0.4^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	6.4 ± 1.1, 7.2 ± 1.8	3.6 ± 0.8	6.0 ± 0.7, 5.4 ± 0.7^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	Norfolk, VA, to Cape Town, South Africa (Aerosols99)^hhhh Dp < 1.1 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sulfate mass is wet ammoniated sulfate mass, sulfate ion and corrected dry ammonium sulfate from ion, respectively. Sea salt is Cl^− + 1.47Na^+, dust is derived from Al assuming Al is 8% of the crustal material in African regions, and the sum of elements in other regions. Roc = 1.6 or 2.1 depending on air mass. Relative humidity is 55%.	Jan–Feb 1999	A	Quinn et al. [2001]
	1.02 ± 0.05			1.7 ± 0.4, 1.9 ± 0.5^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	0.60 ± 0.07^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	1.12 ± 0.16, 0.88 ± 0.13^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	Norfolk, VA, to Cape Town, South Africa (Aerosols99)^iiii 1.1 < Dp < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sea salt is Cl^− + 1.47Na^+, dust is derived from Al assuming Al is 8% of the crustal material in African regions, and the sum of elements in other regions. Roc = 1.6 or 2.1 depending on air mass. Coarse mode is sum of species, including sea salt, dust and POM. Relative humidity is 55%.	Jan–Feb 1999	A	Quinn et al. [2001]
			4,^ee Corresponds to total modes (fine + coarse). 1.7^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^ee Corresponds to total modes (fine + coarse).	3.3^ee Corresponds to total modes (fine + coarse). ^,^ff Value not included in total average provided in Table 5 with reason listed in footnote for that entry.	1.1^ee Corresponds to total modes (fine + coarse).	1.6, 1.5^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^ee Corresponds to total modes (fine + coarse).	Norfolk, VA, to Cape Town, South Africa (Aerosols99)^jjjj Dp < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Sulfate is ion, and normalized to dry ammonium sulfate. Sea salt is Cl^− + 1.47Na^+, dust is derived from Al assuming Al is 8% of the crustal material in African regions, and the sum of elements in other regions. Roc = 1.6 or 2.1 depending on air mass. Relative humidity is 55%. POM (total mode) not included in average because it was reported for all air masses and therefore is not normalized for Roc.	Jan–Feb 1999	A	Quinn et al. [2001]
4.5 ± 0.5			3.9 ± 0.8, 5.0 ± 0.9, 2.1 ± 0.4^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	6 ± 1, 5 ± 1	3.4 ± 0.4	5.5 ± 0.3, 4.7 ± 0.3^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	Indian Ocean (INDOEX)^kkkk Dp < 1.1 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sulfate mass is wet ammoniated sulfate mass, sulfate ion and normalized to dry ammonium sulfate from ion, respectively. Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl^− + 1.47Na^+, fine and coarse modes are mixed aerosols, Roc = 1.6. Relative humidity is 55%.	Feb–Mar 1999	A	Quinn et al. [2002b]
	2 ± 1		3.7 ± 0.6,^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm). 3.9 ± 0.5,^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm). 1.7 ± 0.2^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	4.1 ± 0.9,^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm). 3.6 ± 0.8^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	0.8 ± 0.3^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	1.1 ± 0.7, 0.91 ± 0.06^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	Indian Ocean (INDOEX)^llll 1.1 < Dp < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sulfate mass is wet ammoniated sulfate mass, sulfate ion and normalized dry ammonium sulfate from ion, respectively. Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl^− + 1.47Na^+, fine and coarse modes are mixed aerosols, Roc = 1.6. Relative humidity is 55%.	Feb–Mar 1999	A	Quinn et al. [2002b]
			4.9,^ee Corresponds to total modes (fine + coarse). 2.1^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^ee Corresponds to total modes (fine + coarse).	4.1,^ee Corresponds to total modes (fine + coarse). 3.6^ee Corresponds to total modes (fine + coarse).	1.5^ee Corresponds to total modes (fine + coarse).	1.6, 1.5^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^ee Corresponds to total modes (fine + coarse).	Indian Ocean (INDOEX)^mmmm Dp < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Sulfate is ion. Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl^− + 1.47Na^+, fine and coarse modes are mixed aerosols, Roc = 1.6. Relative humidity is 55%.	Feb–Mar 1999	A	Quinn et al. [2002b]
4.3 ± 0.6			3.8 ± 0.7, 2.5 ± 0.3^c	5.3 ± 0.7, 4.7 ± 0.6	3.4 ± 0.3	5.0 ± 0.5, 4.0 ± 0.4^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	Pacific Ocean, coastal Asia (ACE-ASIA)^nnnn Dp < 1.1 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation of values for various air masses. Sulfate is sulfate aerosol (non-sea-salt-SO4+NH4+H2O) and normalized to dry ammonium sulfate. Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl^− + 1.47Na^+, fine and coarse modes are mixed aerosols, Roc = 2.1 in marine regions and 1.6 elsewhere. Relative humidity is 55%.	spring 2001	A	Quinn et al. [2004]
	1.4 ± 0.6		3.2 ± 0.4^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	2.0 ± 0.3,^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm). 1.8 ± 0.2^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	0.67 ± 0.16^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	1.6 ± 0.2, 1.2 ± 0.2^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^nn Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	Pacific Ocean, coastal Asia (ACE-ASIA)^oooo 1.1 < Dp < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sulfate is sulfate aerosol (non-sea-salt-SO4+NH4+H2O). Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl^− + 1.47Na^+, fine and coarse modes are mixed aerosols, Roc = 2.1 in marine regions and 1.6 elsewhere. Relative humidity is 55%.	spring 2001	A	Quinn et al. [2004]
		3 ± 1^ee Corresponds to total modes (fine + coarse).	3.7 ± 0.6^ee Corresponds to total modes (fine + coarse).	4.4 ± 0.7,^ee Corresponds to total modes (fine + coarse). 3.9 ± 0.6^ee Corresponds to total modes (fine + coarse).	1.2 ± 0.4^ee Corresponds to total modes (fine + coarse).	2.8 ± 0.4, 2.4 ± 0.3^cc We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^ee Corresponds to total modes (fine + coarse).	Pacific Ocean, coastal Asia (ACE-ASIA)^pppp Dp < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sulfate is sulfate aerosol (non-sea-salt-SO4+NH4+H2O). Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl^− + 1.47Na^+, fine and coarse modes are mixed aerosols, Roc = 2.1 in marine regions and 1.6 elsewhere. Relative humidity of 55%.	spring 2001	A	Quinn et al. [2004]

• a Fine, coarse, and total modes refer to a mixed aerosol in those respective size ranges. POM refers to particulate organic material. Species correspond to fine mode unless denoted by the column heading or by a footnote. The “Type” column refers to the type of size distribution measurement: either “I” for impactor, “M” for mobility/aerodynamic, and “A” for apportionment method that uses a combination of both. The reference is given in the last column, and the footnotes attached to the locations give further details of the measurements. All values reported for a wavelength of 550 nm unless otherwise stated in the footnotes. Size ranges, chemical forms and relative humidity (when reported) are stated in the footnotes. In the “sulfate” column, bold values correspond to dry ammonium sulfate as reported in the manuscript. In the “POM” column, bold values are normalized to POM multiplier of R_oc = 1.8 by us or as reported in the manuscript. Unit is m² g⁻¹.
• b D_p < 20 μm. Scanning mobility particle sizer and aerodynamic particle sizer used for size distribution measurements. Average and standard deviation of three time periods. Relative humidity not stated. Wavelength of 530 nm.
• c We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.
• d D_p < 3 μm. Sulfate is ion. Optical particle counter and apportionment techniques used for size distribution measurements. Dry relative humidity, RH < 15%. Wavelength of 632.8 nm.
• e Corresponds to total modes (fine + coarse).
• f Value not included in total average provided in Table 5 with reason listed in footnote for that entry.
• g D_p < 8.75 μm. EGAI 80 low-volume cascade impactor used for size distribution measurements. Dry relative humidity, value not listed. Dust derived assuming Si is one third the mineral aerosol. Sea salt is 3*Na from marine sources. Sea salt not included in average because RH value not listed. Wavelength of 670 nm.
• h Fine mode corresponds to D_p < 1 μm. Differential mobility analyzer and aerodynamic particle sizer used for size distribution measurements. Dry relative humidity (<20%). Wavelength of 530 nm.
• i Sulfate is ion for 0.18 < D_p < 10 μm. MSP Model 100 MOUDI impactor used for size distribution measurements. Relative humidity = 73%.
• j Sulfate is ion for 0.18 < D_p < 10 μm. MSP Model 100 MOUDI impactor used for size distribution measurements. Relative humidity = 95%.
• k Sulfate is ion for 0.18 < D_p < 10 μm. MSP Model 100 MOUDI impactor used for size distribution measurements. Relative humidity = 81%.
• l Sulfate is ion for 0.18 < D_p < 10 μm. MSP Model 100 MOUDI impactor used for size distribution measurements. Relative humidity = 77%.
• m Sulfate is ion for 0.18 < D_p < 10 μm. MSP Model 100 MOUDI impactor used for size distribution measurements. Relative humidity = 69%.
• n Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).
• o 0.056 < D_p < 18 μm. MOUDI impactor used for size distribution measurements. Sulfate is bisulfate in the accumulation mode. Water soluble organic carbon is in the accumulation mode. Dust (CaCO3) and sea salt (NaCl) correspond to coarse mode. Relative humidity not stated. R_oc value not given. Values not included in the average because of unknown RH and R_oc.
• p D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be dry ammonium sulfate. Dry relative humidity, RH = 0%. Wavelength not stated.
• q D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be wet ammonium sulfate. Relative humidity of 48.3%. Wavelength not stated.
• r D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass is dry ammonium sulfate, light scattering includes water. Relative humidity of 48.3%. Wavelength not stated.
• s D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be dry ammonium sulfate. Dry relative humidity, RH = 0%. Wavelength not stated.
• t D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be wet ammonium sulfate. Relative humidity of 25.3%. Wavelength not stated.
• u D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass is dry ammonium sulfate, light scattering includes water. Relative humidity of 25.3%. Wavelength not stated.
• v D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be dry ammonium sulfate. Dry relative humidity, RH = 0%. Wavelength not stated.
• w D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be wet ammonium sulfate. Relative humidity of 82.5%. Wavelength not stated.
• x D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass is dry ammonium sulfate, light scattering includes water. Relative humidity of 82.5%. Wavelength not stated.
• y D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be dry ammonium sulfate. Dry relative humidity, RH = 0%. Wavelength not stated.
• z D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be wet ammonium sulfate and calculated wet scattering. Relative humidity of 75%. Wavelength not stated.
• aa D_p < 2.5 μm. DRUM impactor used for size distribution measurements. Sulfur mass assumed to be dry ammonium sulfate and calculated wet scattering. Relative humidity of 75%. Wavelength not stated.
• bb 0.18 < D_p < 18 μm. MOUDI impactor used for size distribution measurements. Dry ammoniated sulfate (letovicite). RH = 0%. Wavelength not stated.
• cc 0.1 < D_p < 1 μm. Differential mobility analyzer used for size distribution measurements. Dry relative humidity, value not stated. Wavelength not stated.
• dd Dust is coarse mode (D_p > 0.6 μm). Scanning mobility particle sizer and aerodynamic particle sizer used for size distribution measurements. Relative humidity less than 30%. Assumed all of coarse mode is dust. Dust not included in average because coarse mode could include other species.
• ee 0.04 < D_p < 2 μm. Differential mobility analyzer used for size distribution measurements. Dry relative humidity (RH < 15%). Wavelength not stated.
• ff D_p < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Sulfate is ion and sulfate aerosol (non-sea-salt-SO₄+NH₄+H₂O), and normalized to dry ammonium sulfate aerosol. Mixed fine corresponds to residual mass for D_p < 1 μm and mixed coarse corresponds to residual mass for 1 < D_p < 10 μm. Relative humidity = 30%.
• gg D_p < 1 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation of various air masses. Sulfate mass is sulfate aerosol (non-sea-salt-SO₄+NH₄+H₂O), sulfate ion and normalized to dry ammonium sulfate from ion, respectively. Fine mode is non-sea-salt-SO4 plus sea salt. Sea salt is a combination of Na+, K+, Mg+2, Ca+2, and sea salt sulfate and corresponds to the tail of the coarse mode. Relative humidity varies from 30 to 45%.
• hh D_p < 1.1 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sulfate mass is wet ammoniated sulfate mass, sulfate ion and corrected dry ammonium sulfate from ion, respectively. Sea salt is Cl⁻ + 1.47Na⁺, dust is derived from Al assuming Al is 8% of the crustal material in African regions, and the sum of elements in other regions. R_oc = 1.6 or 2.1 depending on air mass. Relative humidity is 55%.
• ii 1.1 < D_p < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sea salt is Cl⁻ + 1.47Na⁺, dust is derived from Al assuming Al is 8% of the crustal material in African regions, and the sum of elements in other regions. R_oc = 1.6 or 2.1 depending on air mass. Coarse mode is sum of species, including sea salt, dust and POM. Relative humidity is 55%.
• jj D_p < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Sulfate is ion, and normalized to dry ammonium sulfate. Sea salt is Cl⁻ + 1.47Na⁺, dust is derived from Al assuming Al is 8% of the crustal material in African regions, and the sum of elements in other regions. R_oc = 1.6 or 2.1 depending on air mass. Relative humidity is 55%. POM (total mode) not included in average because it was reported for all air masses and therefore is not normalized for R_oc.
• kk D_p < 1.1 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sulfate mass is wet ammoniated sulfate mass, sulfate ion and normalized to dry ammonium sulfate from ion, respectively. Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl⁻ + 1.47Na⁺, fine and coarse modes are mixed aerosols, R_oc = 1.6. Relative humidity is 55%.
• ll 1.1 < D_p < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sulfate mass is wet ammoniated sulfate mass, sulfate ion and normalized dry ammonium sulfate from ion, respectively. Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl⁻ + 1.47Na⁺, fine and coarse modes are mixed aerosols, R_oc = 1.6. Relative humidity is 55%.
• mm D_p < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Sulfate is ion. Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl⁻ + 1.47Na⁺, fine and coarse modes are mixed aerosols, R_oc = 1.6. Relative humidity is 55%.
• nn D_p < 1.1 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation of values for various air masses. Sulfate is sulfate aerosol (non-sea-salt-SO₄+NH₄+H₂O) and normalized to dry ammonium sulfate. Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl⁻ + 1.47Na⁺, fine and coarse modes are mixed aerosols, R_oc = 2.1 in marine regions and 1.6 elsewhere. Relative humidity is 55%.
• oo 1.1 < D_p < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sulfate is sulfate aerosol (non-sea-salt-SO₄+NH₄+H₂O). Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl⁻ + 1.47Na⁺, fine and coarse modes are mixed aerosols, R_oc = 2.1 in marine regions and 1.6 elsewhere. Relative humidity is 55%.
• pp D_p < 10 μm. Berner impactor data apportioned to differential mobility analyzer and aerodynamic particle sizer data for size distributions. Average and standard deviation for various air masses. Sulfate is sulfate aerosol (non-sea-salt-SO₄+NH₄+H₂O). Dust is sum of oxides of elements using IMPROVE formula. Sea salt is Cl⁻ + 1.47Na⁺, fine and coarse modes are mixed aerosols, R_oc = 2.1 in marine regions and 1.6 elsewhere. Relative humidity of 55%.

Table 2. Summary of a Literature Survey of Mass Scattering Efficiencies (α_sp) Using the Measurement Method^aa Fine, coarse and total modes correspond to mixed aerosols in those respective size ranges. Sulfate corresponds to fine mode sulfate and dust refers to total mode dust. The reference is given in the last column, and the footnotes attached to the locations give further details of the measurements. All values reported for a wavelength of 550 nm unless otherwise stated in the note table. Size ranges, chemical forms and relative humidity (when reported) are stated in the footnotes. In the “sulfate” column, bold values correspond to dry ammonium sulfate. Unit is m² g⁻¹.

Fine Mode	Coarse Mode	Total Mode	Sulfate	Total Dust	Location/Characteristics/Study	Time	Reference
		1.2 ± 0.9			Baltimore, MD, Canadian smoke^bb 0.00931 < Dp < 20.535 μm. Average and standard deviation of three time periods. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.	4, 6, and 10 Jul 2002	Adam et al. [2004]
		1.84 ± 0.18			Yulin, China, polluted periods (ACE-Asia)^cc Dp < 9 μm. Average and standard deviation of three polluted periods. Relative humidity not stated. Corrected for nephelometer truncation effects. Wavelength of 450 nm.	Apr 2002	Alfaro et al. [2003]
		1.03 ± 0.13			Yulin, China, dust periods (ACE-Asia)^dd Dp < 9 μm. Average and standard deviation of three dust periods. Relative humidity not stated. Corrected for nephelometer truncation effects. Wavelength of 450 nm.	Apr 2002	Alfaro et al. [2003]
3.0 ± 0.6		2.3			Beijing, China^ee Average and standard deviation of Dp < 1.8 μm and Dp < 2.5 μm. Total corresponds to Dp < 10 μm. Relative humidity <40%. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.	Jun 1999	Bergin et al. [2001]
3.9 ± 0.5					Atlanta, GA (Supersite)^ff Average and standard deviation of Dp < 1.8 μm and Dp < 2.5 μm. Relative humidity is 48%. Nephelometer truncation effects mentioned but unclear whether corrections applied. Wavelength of 530 nm.	Jul–Sep 1999	Carrico et al. [2003]
1.65					Brisbane, Australia^gg Dp < 2.5 μm. Relative humidity is dry, value not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.	Apr–Jun 1999	Chan et al. [2002]
7.6					Fort Meade, MD (MARCH-Atlantic study)^hh Dp < 2.5 μm. Relative humidity is ambient but <60%. No mention of corrections for nephelometer truncation effects. Wavelength of 515 nm.	Jul 1999	Chen et al. [2003]
2.1 ± 0.2					Phoenix, AZ^ii Dp < 2.5 μm. Average and standard deviation of unweighted and effective-variance weighted estimates. Relative humidity not stated. Corrected for nephelometer truncation effects. Wavelength of 450 nm.	winter 1989–1990	Chow et al. [2002]
1.720 ± 0.014					Meadview, AZ (MOHAVE study)^jj Dp < 2.5 μm. Average and standard deviation of unweighted and effective-variance weighted estimates. Relative humidity <60%. Corrected for nephelometer truncation effects. Wavelength of 525 nm.	summer 1992	Chow et al. [2002]
1.89 ± 0.04					Buffalo Pass, CO (Mt. Zirkel visibility study)^kk Dp < 2.5 μm. Average and standard deviation of unweighted and effective-variance weighted estimates. Corrected for nephelometer truncation effects. Relative humidity <60%.	summer/fall 1995	Chow et al. [2002]
2.2 ± 0.4					Brighton, CO (NFRAQS study)^ll Dp < 2.5 μm. Average and standard deviation of unweighted and effective-variance weighted estimates. Corrected for nephelometer truncation effects. Relative humidity not stated.	winter 1997	Chow et al. [2002]
2.0 ± 0.5					Welby, CO (NFRAQS study)^ll Dp < 2.5 μm. Average and standard deviation of unweighted and effective-variance weighted estimates. Corrected for nephelometer truncation effects. Relative humidity not stated.	winter 1997	Chow et al. [2002]
3.45 ± 0.07					Bakersfield, CA (IMS95 study)^ll Dp < 2.5 μm. Average and standard deviation of unweighted and effective-variance weighted estimates. Corrected for nephelometer truncation effects. Relative humidity not stated.	winter 1995	Chow et al. [2002]
5.1 ± 0.4					Mexico City, Mexico^ll Dp < 2.5 μm. Average and standard deviation of unweighted and effective-variance weighted estimates. Corrected for nephelometer truncation effects. Relative humidity not stated.	winter 1997	Chow et al. [2002]
3.7, 4.1, 5.2					Maldives (INDOEX) (bsp < 25 Mm^−1)^mm Dp < 1 μm. Values correspond to estimates from gravimetric mass from Caltech, R/V Ronald Brown and the sum of chemically analyzed mass from R/V Ronald Brown, respectively. Corrected for nephelometer truncation effects. Relative humidity of 33%.	Feb–Mar 1999	Clarke et al. [2002]
3.4, 2.2, 3.9, 4.7					Maldives (INDOEX) (25 < bsp < 55 Mm^−1)^nn Dp < 1 μm. Values correspond to estimates from gravimetric mass from Caltech, sum of chemically analyzed mass from U. of Miami, gravimetric mass from R/V Ronald Brown and the sum of chemically analyzed mass from R/V Ronald Brown, respectively. Corrected for nephelometer truncation effects. Relative humidity of 33%.	Feb–Mar 1999	Clarke et al. [2002]
3.5, 2.9, 4.1, 4.4					Maldives (INDOEX) (bsp > 55 Mm^−1)^oo Dp < 1 μm. Values correspond to estimates from gravimetric mass from Caltech, sum of chemically analyzed mass from U. of Miami, gravimetric mass from R/V Ronald Brown and the sum of chemically analyzed mass from R/V Ronald Brown, respectively. Corrected for nephelometer truncation effects. Relative humidity of 33%.	Feb–Mar 1999	Clarke et al. [2002]
3.5, 3.1, 3.8, 4.5					Maldives, Bay of Bengal trajectory (INDOEX study)^oo Dp < 1 μm. Values correspond to estimates from gravimetric mass from Caltech, sum of chemically analyzed mass from U. of Miami, gravimetric mass from R/V Ronald Brown and the sum of chemically analyzed mass from R/V Ronald Brown, respectively. Corrected for nephelometer truncation effects. Relative humidity of 33%.	Feb–Mar 1999	Clarke et al. [2002]
2.7, 3.9, 4.3					Maldives, Arabian trajectory (INDOEX study)^pp Dp < 1 μm. Values correspond to estimates from sum of chemically analyzed mass from U. of Miami, gravimetric mass from R/V Ronald Brown and the sum of chemically analyzed mass from R/V Ronald Brown, respectively. Corrected for nephelometer truncation effects. Relative humidity of 33%.	Feb–Mar 1999	Clarke et al. [2002]
		4.5		2.2^qq Value not included in total average provided in Table 5 with reason listed in footnote for that entry.	Gosan, Korea (ACE-ASIA)^rr Dp < 10 μm. Periods correspond to dust dominated and pollution dominated. Composition uncertain. No mention of corrections for nephelometer truncation effects. Relative humidity varied from 50 to 95%.	Apr and Nov 2001	Kim et al. [2005]
4.1					Bondville, IL^ss Dp < 1 μm. Corrected for nephelometer truncation effects. Relative humidity <40%.	Jan–Dec 1995	Koloutsou-Vakakis et al. [2001]
				0.77^qq Value not included in total average provided in Table 5 with reason listed in footnote for that entry.	Barbados^tt Dp < 10 μm. Relative humidity <40%. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.	Apr–May 1994	Li et al. [1996]
4.19					Mt. Rainier National Park, WA (PREVENT study)^uu Dp < 2.5 μm. Efficiency computed by us. Relative humidity of 83.2%. Nephelometer truncation effects mentioned but unclear whether corrections applied. Wavelength not stated.	summer 1990	Malm et al. [1994b]
4.26					North Cascades National Park, WA (PREVENT study)^vv Dp < 2.5 μm. Efficiency computed by us. Relative humidity of 79.3%. Nephelometer truncation effects mentioned but unclear whether corrections applied. Wavelength not stated.	summer 1990	Malm et al. [1994b]
			3.70 ± 0.14^qq Value not included in total average provided in Table 5 with reason listed in footnote for that entry.	0.52^qq Value not included in total average provided in Table 5 with reason listed in footnote for that entry.	Tenerife, Canary Islands^ww Dp < 0.6 μm for ammonium sulfate and Dp > 0.6 μm for dust. Average and standard deviation of values from a dusty and nondusty period. Assumes scattering not attributed to dust is due to sulfate but composition uncertain. Corrected for nephelometer truncation effects. Relative humidity not stated.	Jul 1995	Maring et al. [2000]
3.8					Dallas–Fort Worth, TX (DFWWHP study)^xx Dp < 2.5 μm. Ratio of open-air bsp and dry+water mass at a relative humidity of 71%. No mention of corrections for nephelometer truncation effects. Wavelength not stated.	30 Dec 1994	McDade et al. [2000]
3.7					Dallas–Fort Worth, TX (DFWWHP study)^yy Dp < 2.5 μm. Ratio of open-air bsp and dry+water mass at a relative humidity of 73%. No mention of corrections for nephelometer truncation effects. Wavelength not stated.	23 Feb 1995	McDade et al. [2000]
3.1					Dallas–Fort Worth, TX (DFWWHP study)^zz Dp < 2.5 μm. Ratio of open-air bsp and dry+water mass at a relative humidity of 44%. No mention of corrections for nephelometer truncation effects. Wavelength not stated.	25 Feb 1995	McDade et al. [2000]
6					New England and mid-Atlantic smoke aerosol^aaaa Dp < 2.5 μm. Stated as smoke scattering to mass ratio. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength not stated.	Jul 2002	Poirot and Husar [2004]
2.5					Barrow, AK^bbbb Dp < 1 μm. Corrected for nephelometer truncation effects. Relative humidity <40%.	1997–2000	Quinn et al. [2002a]
3.67					San Joaquin Valley, CA (IMS95 study)^cccc Dp < 2.5 μm. Relative humidity <60%. No mention of corrections for nephelometer truncation effects. Wavelength of 475 nm.	Dec 1995 to Jan 1996	Richards et al. [1999]
			5.9, 4.3^qq Value not included in total average provided in Table 5 with reason listed in footnote for that entry. ^,^dddd We converted value to ammonium sulfate.	0.21^qq Value not included in total average provided in Table 5 with reason listed in footnote for that entry.	Finokalia, Greece^eeee Fine mode corresponds to Dp < 1.2 μm. Sulfate as non-sea-salt sulfate ion, and normalized to ammonium sulfate (not dry). Composition assumed. Dust computed from Ca ion and Ca/Al ratios. Ambient relative humidity, value not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 532 nm.	Mar 2001 to Jun 2002	Vrekoussis et al. [2005]
			5.7, 4.2^qq Value not included in total average provided in Table 5 with reason listed in footnote for that entry. ^,^dddd We converted value to ammonium sulfate.	0.96^qq Value not included in total average provided in Table 5 with reason listed in footnote for that entry.	Erdemli, Turkey^eeee Fine mode corresponds to Dp < 1.2 μm. Sulfate as non-sea-salt sulfate ion, and normalized to ammonium sulfate (not dry). Composition assumed. Dust computed from Ca ion and Ca/Al ratios. Ambient relative humidity, value not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 532 nm.	Jul 1999 to Jun 2000	Vrekoussis et al. [2005]
1.91					Buffalo Pass, CO (Mt. Zirkel visibility study)^ffff Dp < 2.5 μm. Nephelometer truncation effects mentioned but unclear whether corrections applied. Relative humidity of 31%.	Aug, Sep, Oct 1995	Watson et al. [2001]
2.37	0.27				Spirit Mountain, NV (SCENES program)^gggg Dp < 2.5 μm for fine and 2.5 < Dp < 15 μm for coarse. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 525 nm. Ratio computed by us.	Apr–Sep 1989	White et al. [1994]
2.53	0.28				Meadview, AZ (SCENES program)^gggg Dp < 2.5 μm for fine and 2.5 < Dp < 15 μm for coarse. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 525 nm. Ratio computed by us.	Apr–Sep 1989	White et al. [1994]
2.5	0.54				Meadview, AZ (SCENES program)^hhhh Dp < 2.5 μm for fine and 2.5 < Dp < 15 μm for coarse. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 525 nm. As reported in manuscript.	Apr–Sep 1989	White et al. [1994]
2.4	0.52				Spirit Mountain, NV (SCENES program)^hhhh Dp < 2.5 μm for fine and 2.5 < Dp < 15 μm for coarse. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 525 nm. As reported in manuscript.	Apr–Sep 1989	White et al. [1994]
4					Linan, China^iiii Dp < 2.5 μm. Relative humidity <40%. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.	Nov 1999	Xu et al. [2002]

• a Fine, coarse and total modes correspond to mixed aerosols in those respective size ranges. Sulfate corresponds to fine mode sulfate and dust refers to total mode dust. The reference is given in the last column, and the footnotes attached to the locations give further details of the measurements. All values reported for a wavelength of 550 nm unless otherwise stated in the note table. Size ranges, chemical forms and relative humidity (when reported) are stated in the footnotes. In the “sulfate” column, bold values correspond to dry ammonium sulfate. Unit is m² g⁻¹.
• b 0.00931 < D_p < 20.535 μm. Average and standard deviation of three time periods. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.
• c D_p < 9 μm. Average and standard deviation of three polluted periods. Relative humidity not stated. Corrected for nephelometer truncation effects. Wavelength of 450 nm.
• d D_p < 9 μm. Average and standard deviation of three dust periods. Relative humidity not stated. Corrected for nephelometer truncation effects. Wavelength of 450 nm.
• e Average and standard deviation of D_p < 1.8 μm and D_p < 2.5 μm. Total corresponds to D_p < 10 μm. Relative humidity <40%. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.
• f Average and standard deviation of D_p < 1.8 μm and D_p < 2.5 μm. Relative humidity is 48%. Nephelometer truncation effects mentioned but unclear whether corrections applied. Wavelength of 530 nm.
• g D_p < 2.5 μm. Relative humidity is dry, value not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.
• h D_p < 2.5 μm. Relative humidity is ambient but <60%. No mention of corrections for nephelometer truncation effects. Wavelength of 515 nm.
• i D_p < 2.5 μm. Average and standard deviation of unweighted and effective-variance weighted estimates. Relative humidity not stated. Corrected for nephelometer truncation effects. Wavelength of 450 nm.
• j D_p < 2.5 μm. Average and standard deviation of unweighted and effective-variance weighted estimates. Relative humidity <60%. Corrected for nephelometer truncation effects. Wavelength of 525 nm.
• k D_p < 2.5 μm. Average and standard deviation of unweighted and effective-variance weighted estimates. Corrected for nephelometer truncation effects. Relative humidity <60%.
• l D_p < 2.5 μm. Average and standard deviation of unweighted and effective-variance weighted estimates. Corrected for nephelometer truncation effects. Relative humidity not stated.
• m D_p < 1 μm. Values correspond to estimates from gravimetric mass from Caltech, R/V Ronald Brown and the sum of chemically analyzed mass from R/V Ronald Brown, respectively. Corrected for nephelometer truncation effects. Relative humidity of 33%.
• n D_p < 1 μm. Values correspond to estimates from gravimetric mass from Caltech, sum of chemically analyzed mass from U. of Miami, gravimetric mass from R/V Ronald Brown and the sum of chemically analyzed mass from R/V Ronald Brown, respectively. Corrected for nephelometer truncation effects. Relative humidity of 33%.
• o D_p < 1 μm. Values correspond to estimates from gravimetric mass from Caltech, sum of chemically analyzed mass from U. of Miami, gravimetric mass from R/V Ronald Brown and the sum of chemically analyzed mass from R/V Ronald Brown, respectively. Corrected for nephelometer truncation effects. Relative humidity of 33%.
• p D_p < 1 μm. Values correspond to estimates from sum of chemically analyzed mass from U. of Miami, gravimetric mass from R/V Ronald Brown and the sum of chemically analyzed mass from R/V Ronald Brown, respectively. Corrected for nephelometer truncation effects. Relative humidity of 33%.
• q Value not included in total average provided in Table 5 with reason listed in footnote for that entry.
• r D_p < 10 μm. Periods correspond to dust dominated and pollution dominated. Composition uncertain. No mention of corrections for nephelometer truncation effects. Relative humidity varied from 50 to 95%.
• s D_p < 1 μm. Corrected for nephelometer truncation effects. Relative humidity <40%.
• t D_p < 10 μm. Relative humidity <40%. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.
• u D_p < 2.5 μm. Efficiency computed by us. Relative humidity of 83.2%. Nephelometer truncation effects mentioned but unclear whether corrections applied. Wavelength not stated.
• v D_p < 2.5 μm. Efficiency computed by us. Relative humidity of 79.3%. Nephelometer truncation effects mentioned but unclear whether corrections applied. Wavelength not stated.
• w D_p < 0.6 μm for ammonium sulfate and D_p > 0.6 μm for dust. Average and standard deviation of values from a dusty and nondusty period. Assumes scattering not attributed to dust is due to sulfate but composition uncertain. Corrected for nephelometer truncation effects. Relative humidity not stated.
• x D_p < 2.5 μm. Ratio of open-air b_sp and dry+water mass at a relative humidity of 71%. No mention of corrections for nephelometer truncation effects. Wavelength not stated.
• y D_p < 2.5 μm. Ratio of open-air b_sp and dry+water mass at a relative humidity of 73%. No mention of corrections for nephelometer truncation effects. Wavelength not stated.
• z D_p < 2.5 μm. Ratio of open-air b_sp and dry+water mass at a relative humidity of 44%. No mention of corrections for nephelometer truncation effects. Wavelength not stated.
• aa D_p < 2.5 μm. Stated as smoke scattering to mass ratio. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength not stated.
• bb D_p < 1 μm. Corrected for nephelometer truncation effects. Relative humidity <40%.
• cc D_p < 2.5 μm. Relative humidity <60%. No mention of corrections for nephelometer truncation effects. Wavelength of 475 nm.
• dd We converted value to ammonium sulfate.
• ee Fine mode corresponds to D_p < 1.2 μm. Sulfate as non-sea-salt sulfate ion, and normalized to ammonium sulfate (not dry). Composition assumed. Dust computed from Ca ion and Ca/Al ratios. Ambient relative humidity, value not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 532 nm.
• ff D_p < 2.5 μm. Nephelometer truncation effects mentioned but unclear whether corrections applied. Relative humidity of 31%.
• gg D_p < 2.5 μm for fine and 2.5 < D_p < 15 μm for coarse. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 525 nm. Ratio computed by us.
• hh D_p < 2.5 μm for fine and 2.5 < D_p < 15 μm for coarse. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 525 nm. As reported in manuscript.
• ii D_p < 2.5 μm. Relative humidity <40%. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.

Table 3. Summary of a Literature Survey of Mass Scattering Efficiencies (α_sp) Using the Multilinear Regression (MLR) Method^aa Fine and coarse modes correspond to mixed aerosols in those respective size ranges. Sulfate, POM (particulate organic material), and nitrate (ammonium nitrate) correspond to the fine mode. Dust and sea salt size ranges are assumed to be fine mode unless denoted with a footnote for coarse mode and total mode. The reference is given in the last column, and the footnotes attached to the locations give further details of the measurements. All values reported for a wavelength of 550 nm unless otherwise stated in the footnotes. Values correspond to fine mode unless otherwise noted in the column header or by footnote. Size ranges, chemical forms and relative humidity (when reported) are stated in the footnotes. In the “sulfate” column, bold values correspond to dry ammonium sulfate as reported in the manuscript. In the “POM” column, bold values are normalized to POM multiplier of R_oc = 1.8 by us or as reported in the manuscript. Unit is m² g⁻¹.

Fine Mode	Coarse Mode	Sulfate	POM	Nitrate	Dust	Sea Salt	Location/Study	Time	Reference
5.8	0.5	5.6, 2.9^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.			0.71^cc Corresponds to total modes (fine + coarse).		Sde Boker, Israel (ARACHNE study)^dd Two separate MLR runs were performed, one with fine mass (Dp < 2 μm) and coarse mass (2 < Dp < 10 μm). The second was with total dust (Dp < 10 μm) and fine sulfate ion (Dp < 2 μm). Dust is derived assuming Al is 8.3% of crustal material plus CaCO3. Corrected for nephelometer truncation effects. Relative humidity of 38%.	Dec 1995 to Oct 1997	Andreae et al. [2002]
		10.9^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.			2.73		Brisbane, Australia^ff Dp < 2.5 μm. Relative humidity stated as dry. Sulfate as ammonium sulfate. Dust computed from IMPROVE formula. Wavelength of 530 nm. No mention of corrections for nephelometer truncation effects. Sulfate not included in average because RH value not stated.	Apr–Jun 1999	Chan et al. [2002]
2.5 ± 0.6	0.62 ± 0.16						Tucson, AZ^gg Dp < 2.5 μm for fine and 2.5 μm Dp < 10 μm for coarse. Average and standard deviation of unweighted and effective-variance weighted estimates. Relative humidity not stated. Corrected for nephelometer truncation effects. Wavelength of 530 nm.	winter 1989–1990	Chow et al. [2002]
1.5 ± 0.4	0.8 ± 0.3						Welby, CO (NFRAQS study)^hh Dp < 2.5 μm for fine and 2.5 μm Dp < 10 μm for coarse. Average and standard deviation of unweighted and effective-variance weighted estimates. Relative humidity not stated. Corrected for nephelometer truncation effects. Wavelength not stated.	summer 1996	Chow et al. [2002]
3.0 ± 0.2	0.275 ± 0.007						Welby, CO (NFRAQS study)^hh Dp < 2.5 μm for fine and 2.5 μm Dp < 10 μm for coarse. Average and standard deviation of unweighted and effective-variance weighted estimates. Relative humidity not stated. Corrected for nephelometer truncation effects. Wavelength not stated.	winter 1996	Chow et al. [2002]
	0.07	2.08^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.	1.05, 0.82^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.	1.77^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.	2.06		Canyonlands National Park, UT^ii Dp < 2.5 μm for mass and bext measured with transmissometer. Sulfate is ammonium sulfate and nitrate is ammonium nitrate. Dry relative humidity, value not listed. Dust computed as an average assuming Si is 28.2% and Fe is 5.6% crustal material. Roc = 1.4. Reporting annual values only. Values not included in the average because RH not listed.	1990–1991	Eatough et al. [1996]
			0.9, 3.11; 0.8, 2.83^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.	3.44^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.			Salt Lake City, UT^jj Dp < 2.5 μm. Organic carbon values are for nonvolatile organics and semivolatile organics, respectively. Roc = 1.64. Relatively humidity not stated, assumed dry. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm. Values not included in the average because RH not listed.	Jan–Feb 2001	Eatough et al. [2004]
3		7.4, 3.3^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.					Sde Boker, Israel (ARACHNE study)^kk Dp < 2 μm. Mixed fine mode refers to fine mass with non-sea-salt-SO4 ion subtracted. Sulfate is sulfate ion. No mention of corrections for nephelometer truncation effects. Relative humidity <50%.	Jun–Jul 1996	Formenti et al. [2001]
		2.02, 1.47^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.					Big Bend National Park, TX (BRAVO study)^ll Corresponds to RH = 0% from a curve fit to scattering efficiency as a function of RH for molar ratio of 1.56 and mass of sulfate ion. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.	Jul–Oct 1999	Hering et al. [2003]
3.669	0.121						Sde Boker, Israel (ARACHNE study)^mm Dp < 2 μm for fine mode and 2 < Dp < 10 μm for coarse mode. No mention of corrections for nephelometer truncation effects. Relative humidity <50%.	Feb–Mar 1997	Ichoku et al. [1999]
		4.57, 3.2^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	4.11^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.				Bondville, IL^nn Dp < 1 μm. Organic carbon value refers to carbon-containing aerosols. Sulfate is generally ammonium sulfate. Corrected for nephelometer truncation effects. Relative humidity <40%. POM not in average because value is for carbon containing aerosols.	Jan–Dec 1995	Koloutsou–Vakakis et al. [2001]
		3.2 ± 0.3, 2.5 ± 0.2^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	3.10 ± 0.10, 2.07 ± 0.07^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.	3.47 ± 0.12, 2.75 ± 0.09	0.7 ± 0.6^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.		Meadview, AZ (MOHAVE study)^oo Dp < 2.5 μm. Average and standard deviation of effective variance weighting with and without intercepts, and no weighting. Relative humidity of 26.1%. Dust is derived using IMPROVE formula. Roc = 1.2. POM and dust not included in average because they are considered partly soluble. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.	summer 1992	Lowenthal et al. [1995]
		2.9 ± 0.4, 1.9 ± 0.3^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	3.7 ± 0.5, 2.5 ± 0.3^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.	5.0 ± 0.9, 3.3 ± 0.6^b	1.5 ± 1.8^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.		Meadview, AZ (MOHAVE study)^pp Dp < 2.5 μm. Samples with average absolute difference <20% effective variance weighting. Relative humidity of 45.1%. Dust is derived using IMPROVE formula. Roc = 1.2. POM and dust not included in average because they are considered partly soluble. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.	winter 1992	Lowenthal et al. [1995]
		4.0 ± 1.8, 3.0 ± 1.4^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	3.9 ± 0.4, 2.6 ± 0.3^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.	3.1 ± 0.4, 2.3 ± 0.3^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	0.84 ± 0.99^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.		Phoenix, AZ^qq Dp < 2.5 μm. Samples with average absolute difference <20% effective variance weighting. Relative humidity of 31.8%. Dust is derived using IMPROVE formula. Roc = 1.2. POM and dust not included in average because they are considered partly soluble. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.	winter 1989–1990	Lowenthal et al. [1995]
		7.3 ± 0.3, 3.05 ± 0.11^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	12 ± 8, 8 ± 5^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.		1.8 ± 0.9^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.		Uniontown, PA^rr Dp < 2.5 μm. Average and standard deviation of effective variance weighting with and without intercepts, and no weighting. Relative humidity of 66.7%. Roc = 1.2. Dust is derived using IMPROVE formula. POM and dust not included in average because they are considered partly soluble. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.	summer 1990	Lowenthal et al. [1995]
5.2	0.34						Sde Boker, Israel^rr Dp < 2.5 μm. Average and standard deviation of effective variance weighting with and without intercepts, and no weighting. Relative humidity of 66.7%. Roc = 1.2. Dust is derived using IMPROVE formula. POM and dust not included in average because they are considered partly soluble. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.	1995 to Feb 1997	Maenhaut et al. [1997]
		2.2^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.	1.8, 1.4^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.	2.2^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.	0.79^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.		Grand Canyon National Park, AZ^tt Dp < 2.5 μm. Dry relative humidity, RH not listed. Roc = 1.4. Dust coefficient is not significant. Nephelometer truncation effects discussed but unclear whether corrections applied. Wavelength of 530 nm. Not averaged because no RH listed.	Jul–Aug 1998	Malm and Day [2000]
		2.51	3.18	2.51			Acadia National Park, ME^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1998 to Dec 2003	Malm and Hand [2007]
		2.48	3.15	2.48			Big Bend National Park, TX^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1999 to Dec 2003	Malm and Hand [2007]
		2.87	3.65	2.87			Boundary Waters Canoe Area, MN^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1994 to Dec 1997	Malm and Hand [2007]
		2.33	2.96	2.33			Columbia River Gorge, WA^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1996 to Dec 2003	Malm and Hand [2007]
		2.71	3.44	2.71			Dolly Sods Wilderness, WV^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1994 to Dec 1997	Malm and Hand [2007]
		2.51	3.19	2.51			Gila Wilderness, NM^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1995 to Dec 2003	Malm and Hand [2007]
		2.77	3.52	2.77			Grand Canyon National Park, AZ^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1998 to Dec 2003	Malm and Hand [2007]
		2.58	3.27	2.58			Great Smoky Mountains National Park, TN^u	Jan 1994 to Dec 2003	Malm and Hand [2007]
		2.25	2.85	2.25			Ike's Backbone, AZ^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 2002 to Dec 2003	Malm and Hand [2007]
		2.84	3.61	2.84			Jarbidge Wilderness, NV^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1994 to Dec 1997	Malm and Hand [2007]
		2.51	3.19	2.51			Lone Peak Wilderness Area, UT^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1994 to Dec 2000	Malm and Hand [2007]
		3.03	3.85	3.03			Mammoth Cave National Park, KY^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1993 to Dec 2003	Malm and Hand [2007]
		2.58	3.27	2.58			Mount Rainier National Park, WA^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1993 to Dec 2003	Malm and Hand [2007]
		3.00	3.81	3.00			Okefenokee National Wildlife, GA^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1993 to Dec 1996	Malm and Hand [2007]
		1.94	2.47	1.94			Phoenix, AZ^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1999 to Dec 2003	Malm and Hand [2007]
		3.14	3.99	3.14			Seney National Wildlife Refuge, MI^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 2002 to Dec 2003	Malm and Hand [2007]
		2.55	3.23	2.55			Shenandoah National Park, VA^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1997 to Dec 2000	Malm and Hand [2007]
		2.54	3.22	2.54			Snoqualamie Pass, WA^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1994 to Dec 2000	Malm and Hand [2007]
		2.25	2.86	2.25			Sycamore Canyon, AZ^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1999 to Dec 2003	Malm and Hand [2007]
		2.44	3.10	2.44			Three Sisters Wilderness, OR^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1994 to Dec 2000	Malm and Hand [2007]
		2.83	3.59	2.83			Upper Buffalo Wilderness Area, AR^uu Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).	Jan 1993 to Dec 1997	Malm and Hand [2007]
	0.5	3.1	4.0	3.1			Badlands National Park, SD^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1989 to Dec 2003	Malm and Hand [2007]
	0.9	2.7	3.4	2.7			Bandalier National Monument, NM^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1989 to Dec 2003	Malm and Hand [2007]
	0.7	3.1	3.9	3.1			Bridger Wilderness, WY^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1989 to Dec 2003	Malm and Hand [2007]
	0.4	2.8	3.5	2.8			Canyonlands National Park, UT^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1987 to Dec 2003	Malm and Hand [2007]
	0.8	2.7	3.4	2.7			Chiricahua National Monument, AZ^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1989 to Feb 1998	Malm and Hand [2007]
	0.3	3.2	4.1	3.2			Glacier National Park, MT^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1989 to Dec 2003	Malm and Hand [2007]
	0.8	2.3	3.0	2.3			Great Basin National Park, NV^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1993 to Dec 2003	Malm and Hand [2007]
	0.7	2.5	3.1	2.5			Grand Canyon National Park, AZ^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1987 to Dec 2003	Malm and Hand [2007]
	0.4	2.9	3.7	2.9			Guadalupe Mountains National Park, TX^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1989 to Dec 2003	Malm and Hand [2007]
	0.5	2.6	3.4	2.6			Petrified Forest National Park, AZ^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1988 to Dec 2003	Malm and Hand [2007]
	1.1	2.4	3.1	2.4			Pinnacles National Monument, CA^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1988 to Dec 1992	Malm and Hand [2007]
	0.5	1.9	2.4	1.9			Rocky Mountain National Park, CO^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1988 to Dec 1996	Malm and Hand [2007]
	1.2	3.0	3.8	3.0			Yosemite National Park, CA^vv Dp < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. Roc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm^−3) to POM density (1.4 g cm^−3). Coarse mass is gravimetric (PM10 – PM2.5).	Jan 1995 to Dec 2003	Malm and Hand [2007]
		6.3, 3.0^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.				2.9,^ww Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm). 0.56^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^ee Value not included in total average in Table 5 with reason listed in footnote for that entry. ^,^ww Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	Tenerife, Canary Islands^xx Dp < 1 μm for fine mode and 1 < Dp < 10 μm for coarse mode. Relative humidity of 45%. Sea salt efficiency for Na^+ only. Sulfate is non-sea-salt-SO4 ion. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm. Not averaged because fine mode is assumed to be all sulfate and coarse mode assumed to be all sea salt.	Jul 1995	McGovern et al. [1999]
		8.3, 4.6^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.					rural Great Hungarian Plain^yy Dp < 1 μm. Relative humidity of 30%. No mention of corrections for nephelometer truncation effects. Not averaged because fine mode assumed to be all sulfate.	Oct 1994 to Mar 1995	Mészáros et al. [1998]
4.3	0.6	3.6,^cc Corresponds to total modes (fine + coarse). 2.0^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^cc Corresponds to total modes (fine + coarse).					Pacific Ocean and Cheeka Peak, WA (PSI91 and MAGE92 studies)^zz Sulfate corresponds to sulfate ion for Dp < 10 μm. Fine mixed is residual mass for Dp < 1 μm and coarse mixed is for residual mass for 1 < Dp < 10 μm. Residual mass is sea salt+NO3+MSA. Corrected for nephelometer truncation effects. Relative humidity of 30%.	1991 and 1992	Quinn et al. [1995]
		4.5 ± 0.9,^cc Corresponds to total modes (fine + coarse). 2.5 ± 0.5^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^cc Corresponds to total modes (fine + coarse).				4.95 ± 0.07, 4.71 ± 0.07; 0.9 ± 0.3,^ww Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm). 0.72 ± 0.02^ww Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).	central Pacific Ocean (RITS 93 and RITS 94)^aaaa Average and standard deviation for two cruises. Sulfate is non-sea-salt-SO4 ion for Dp < 10 μm. Sea salt is listed for Dp < 1 μm and 1 < Dp < 10 μm, respectively. Sea salt is a combination of Na+, K+, Mg+2, Ca+2, Cl−, Br−, NO3 and sea salt SO4. Relative humidity of 30%. No mention of corrections for nephelometer truncation effects.	1993 and 1994	Quinn et al. [1996]
		5.7,^cc Corresponds to total modes (fine + coarse). 2.4^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^cc Corresponds to total modes (fine + coarse).	2.7^cc Corresponds to total modes (fine + coarse). ^,^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.		0.9^cc Corresponds to total modes (fine + coarse).	2.2,^cc Corresponds to total modes (fine + coarse). 2.0^cc Corresponds to total modes (fine + coarse).	Norfolk, VA, to Cape Town, South Africa (Aerosols99)^bbbb Dp < 10 μm. Sulfate is non-sea-salt-SO4 ion. Sea salt is Cl^− + 1.47Na. Dust is derived assuming Al is 8% of crustal material in some regions, sum of elements in others. Roc = 1.6 and 2.1. POM not included in average because of varying Roc factors. Corrected for nephelometer truncation effects. Relative humidity of 55%.	Jan–Feb 1999	Quinn et al. [2001]
0.82		5.3, 2.7^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.				2.2, 1.8	Barrow, AK^cccc Dp < 1 μm. Sulfate is sulfate ion. Sea salt is Cl^− + 1.47Na. Residual is gravimetric-ions-water. Relative humidity <40%. Corrected for nephelometer truncation effects.	1997–2000	Quinn et al. [2002a]
		2,^c0.83^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^cc Corresponds to total modes (fine + coarse).	1.6, 1.4^cc Corresponds to total modes (fine + coarse).		0.51^cc Corresponds to total modes (fine + coarse).	1.7,^cc Corresponds to total modes (fine + coarse). 1.6^cc Corresponds to total modes (fine + coarse).	Indian Ocean (INDOEX)^dddd Dp < 10 μm. Sulfate is non-sea-salt-SO4 ion. Roc = 1.6. Sea salt is Cl^− + 1.47Na. Dust is computed using IMPROVE formula. Relative humidity of 55%. Corrected for nephelometer truncation effects.	Feb–Mar 1999	Quinn et al. [2002b]
3.9 ± 0.3		4.0 ± 0.2, 2.12 ± 0.11^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt. ^,^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.	1.95^ee Value not included in total average in Table 5 with reason listed in footnote for that entry.				San Joaquin Valley, CA (IMS95)^eeee Dp < 2.5 μm. Sulfate corresponds to ammonium salts (SO4+NO3+NH4). Mixed fine mode is everything besides salts. Average and standard deviation for OC back up filter added back and subtracted. No Roc given. Relative humidity <60%. No mention of corrections for nephelometer truncation effects. Wavelength of 475 nm. Not included in average because no Roc is given and sulfate includes contribution of nitrate.	Dec 1995 to Jan 1996	Richards et al. [1999]
		5.2 ± 0.3, 3.2 ± 0.2^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	1.9 ± 0.4, 1.8 ± 0.4				Cranmore Mountain, NH^ffff Dp < 2.5 μm. Average and standard deviation various air mass types. Sulfate as ammonium sulfate. Roc = 1.7. Relative humidity typically <50%. Effects of nephelometer truncation included as uncertainty. Wavelength of 530 nm.	Apr–May 2000	Slater et al. [2002]
		6.54, 4.28^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	3.36, 2.61				southern New Hampshire^gggg Dp < 2.5 μm. Average and standard deviation various air mass types. Sulfate as ammonium sulfate. Relative humidity typically <45%. Roc = 1.4. Effects of nephelometer truncation included as uncertainty. Wavelength of 530 nm.	Jul 2000 to Sep 2001	Slater and Dibb [2004]
		5.3 ± 0.3, 3.7 ± 0.2^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.	0.9 ± 0.3, 0.7 ± 0.2	6.2 ± 0.4, 4.3 ± 0.3^bb We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.			Mt Zirkel Wilderness Area, CO (MZVS study)^hhhh Dp < 2.5 μm. Nitrate is ammonium nitrate. Average and standard deviation for uncorrected OC, OC corrected for positive organic artifact, and corrected for negative organic artifact, respectively. Roc = 1.4. Only statistically significant values are included. Nephelometer truncation effects discussed but unclear whether corrections applied. Relative humidity <40%.	Aug–Oct 1995	Watson et al. [2001]
2.4	1.7				2.6, 0.45^ww Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).		Meadview, AZ^iiii Dp < 2.5 μm for mixed fine mode and 2.5 < Dp < 15 μm for coarse mode. Dust is derived from Si. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 525 nm.	Apr–Sep 1989	White et al. [1994]
2.2	1.9				3.1, 0.34^ww Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).		Spirit Mountain, NV^iiii Dp < 2.5 μm for mixed fine mode and 2.5 < Dp < 15 μm for coarse mode. Dust is derived from Si. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 525 nm.	Apr–Sep 1989	White et al. [1994]

• a Fine and coarse modes correspond to mixed aerosols in those respective size ranges. Sulfate, POM (particulate organic material), and nitrate (ammonium nitrate) correspond to the fine mode. Dust and sea salt size ranges are assumed to be fine mode unless denoted with a footnote for coarse mode and total mode. The reference is given in the last column, and the footnotes attached to the locations give further details of the measurements. All values reported for a wavelength of 550 nm unless otherwise stated in the footnotes. Values correspond to fine mode unless otherwise noted in the column header or by footnote. Size ranges, chemical forms and relative humidity (when reported) are stated in the footnotes. In the “sulfate” column, bold values correspond to dry ammonium sulfate as reported in the manuscript. In the “POM” column, bold values are normalized to POM multiplier of R_oc = 1.8 by us or as reported in the manuscript. Unit is m² g⁻¹.
• b We normalized value to dry ammonium sulfate, dry ammonium nitrate, or dry sea salt.
• c Corresponds to total modes (fine + coarse).
• d Two separate MLR runs were performed, one with fine mass (D_p < 2 μm) and coarse mass (2 < D_p < 10 μm). The second was with total dust (D_p < 10 μm) and fine sulfate ion (D_p < 2 μm). Dust is derived assuming Al is 8.3% of crustal material plus CaCO3. Corrected for nephelometer truncation effects. Relative humidity of 38%.
• e Value not included in total average in Table 5 with reason listed in footnote for that entry.
• f D_p < 2.5 μm. Relative humidity stated as dry. Sulfate as ammonium sulfate. Dust computed from IMPROVE formula. Wavelength of 530 nm. No mention of corrections for nephelometer truncation effects. Sulfate not included in average because RH value not stated.
• g D_p < 2.5 μm for fine and 2.5 μm D_p < 10 μm for coarse. Average and standard deviation of unweighted and effective-variance weighted estimates. Relative humidity not stated. Corrected for nephelometer truncation effects. Wavelength of 530 nm.
• h D_p < 2.5 μm for fine and 2.5 μm D_p < 10 μm for coarse. Average and standard deviation of unweighted and effective-variance weighted estimates. Relative humidity not stated. Corrected for nephelometer truncation effects. Wavelength not stated.
• i D_p < 2.5 μm for mass and b_ext measured with transmissometer. Sulfate is ammonium sulfate and nitrate is ammonium nitrate. Dry relative humidity, value not listed. Dust computed as an average assuming Si is 28.2% and Fe is 5.6% crustal material. R_oc = 1.4. Reporting annual values only. Values not included in the average because RH not listed.
• j D_p < 2.5 μm. Organic carbon values are for nonvolatile organics and semivolatile organics, respectively. R_oc = 1.64. Relatively humidity not stated, assumed dry. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm. Values not included in the average because RH not listed.
• k D_p < 2 μm. Mixed fine mode refers to fine mass with non-sea-salt-SO₄ ion subtracted. Sulfate is sulfate ion. No mention of corrections for nephelometer truncation effects. Relative humidity <50%.
• l Corresponds to RH = 0% from a curve fit to scattering efficiency as a function of RH for molar ratio of 1.56 and mass of sulfate ion. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.
• m D_p < 2 μm for fine mode and 2 < D_p < 10 μm for coarse mode. No mention of corrections for nephelometer truncation effects. Relative humidity <50%.
• n D_p < 1 μm. Organic carbon value refers to carbon-containing aerosols. Sulfate is generally ammonium sulfate. Corrected for nephelometer truncation effects. Relative humidity <40%. POM not in average because value is for carbon containing aerosols.
• o D_p < 2.5 μm. Average and standard deviation of effective variance weighting with and without intercepts, and no weighting. Relative humidity of 26.1%. Dust is derived using IMPROVE formula. R_oc = 1.2. POM and dust not included in average because they are considered partly soluble. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.
• p D_p < 2.5 μm. Samples with average absolute difference <20% effective variance weighting. Relative humidity of 45.1%. Dust is derived using IMPROVE formula. R_oc = 1.2. POM and dust not included in average because they are considered partly soluble. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.
• q D_p < 2.5 μm. Samples with average absolute difference <20% effective variance weighting. Relative humidity of 31.8%. Dust is derived using IMPROVE formula. R_oc = 1.2. POM and dust not included in average because they are considered partly soluble. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.
• r D_p < 2.5 μm. Average and standard deviation of effective variance weighting with and without intercepts, and no weighting. Relative humidity of 66.7%. R_oc = 1.2. Dust is derived using IMPROVE formula. POM and dust not included in average because they are considered partly soluble. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm.
• s D_p < 2 μm for fine mode and 2 < D_p < 10 μm for coarse mode. No mention of corrections for nephelometer truncation effects. Relative humidity not stated.
• t D_p < 2.5 μm. Dry relative humidity, RH not listed. R_oc = 1.4. Dust coefficient is not significant. Nephelometer truncation effects discussed but unclear whether corrections applied. Wavelength of 530 nm. Not averaged because no RH listed.
• u D_p < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. R_oc = 1.8. Optical data from nephelometer. Mass-adjusted efficiencies. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm⁻³) to POM density (1.4 g cm⁻³). Nephelometer truncation effects discussed but not applied (results only refer to fine mass).
• v D_p < 2.5 μm. Dry relative humidity, RH = 0%. Sulfate is dry ammonium sulfate. R_oc = 1.8. Optical data from transmissometer. POM efficiency prorated by ratio of ammonium sulfate density (1.78 g cm⁻³) to POM density (1.4 g cm⁻³). Coarse mass is gravimetric (PM₁₀ – PM_2.5).
• w Corresponds to coarse mode (supermicron, usually greater than 1 μm or 2.5 μm).
• x D_p < 1 μm for fine mode and 1 < D_p < 10 μm for coarse mode. Relative humidity of 45%. Sea salt efficiency for Na⁺ only. Sulfate is non-sea-salt-SO₄ ion. No mention of corrections for nephelometer truncation effects. Wavelength of 530 nm. Not averaged because fine mode is assumed to be all sulfate and coarse mode assumed to be all sea salt.
• y D_p < 1 μm. Relative humidity of 30%. No mention of corrections for nephelometer truncation effects. Not averaged because fine mode assumed to be all sulfate.
• z Sulfate corresponds to sulfate ion for D_p < 10 μm. Fine mixed is residual mass for D_p < 1 μm and coarse mixed is for residual mass for 1 < D_p < 10 μm. Residual mass is sea salt+NO3+MSA. Corrected for nephelometer truncation effects. Relative humidity of 30%.
• aa Average and standard deviation for two cruises. Sulfate is non-sea-salt-SO₄ ion for D_p < 10 μm. Sea salt is listed for D_p < 1 μm and 1 < D_p < 10 μm, respectively. Sea salt is a combination of Na+, K+, Mg+2, Ca+2, Cl−, Br−, NO3 and sea salt SO4. Relative humidity of 30%. No mention of corrections for nephelometer truncation effects.
• bb D_p < 10 μm. Sulfate is non-sea-salt-SO₄ ion. Sea salt is Cl⁻ + 1.47Na. Dust is derived assuming Al is 8% of crustal material in some regions, sum of elements in others. R_oc = 1.6 and 2.1. POM not included in average because of varying R_oc factors. Corrected for nephelometer truncation effects. Relative humidity of 55%.
• cc D_p < 1 μm. Sulfate is sulfate ion. Sea salt is Cl⁻ + 1.47Na. Residual is gravimetric-ions-water. Relative humidity <40%. Corrected for nephelometer truncation effects.
• dd D_p < 10 μm. Sulfate is non-sea-salt-SO₄ ion. R_oc = 1.6. Sea salt is Cl⁻ + 1.47Na. Dust is computed using IMPROVE formula. Relative humidity of 55%. Corrected for nephelometer truncation effects.
• ee D_p < 2.5 μm. Sulfate corresponds to ammonium salts (SO₄+NO₃+NH₄). Mixed fine mode is everything besides salts. Average and standard deviation for OC back up filter added back and subtracted. No R_oc given. Relative humidity <60%. No mention of corrections for nephelometer truncation effects. Wavelength of 475 nm. Not included in average because no R_oc is given and sulfate includes contribution of nitrate.
• ff D_p < 2.5 μm. Average and standard deviation various air mass types. Sulfate as ammonium sulfate. R_oc = 1.7. Relative humidity typically <50%. Effects of nephelometer truncation included as uncertainty. Wavelength of 530 nm.
• gg D_p < 2.5 μm. Average and standard deviation various air mass types. Sulfate as ammonium sulfate. Relative humidity typically <45%. R_oc = 1.4. Effects of nephelometer truncation included as uncertainty. Wavelength of 530 nm.
• hh D_p < 2.5 μm. Nitrate is ammonium nitrate. Average and standard deviation for uncorrected OC, OC corrected for positive organic artifact, and corrected for negative organic artifact, respectively. R_oc = 1.4. Only statistically significant values are included. Nephelometer truncation effects discussed but unclear whether corrections applied. Relative humidity <40%.
• ii D_p < 2.5 μm for mixed fine mode and 2.5 < D_p < 15 μm for coarse mode. Dust is derived from Si. Relative humidity not stated. No mention of corrections for nephelometer truncation effects. Wavelength of 525 nm.

Table 4. Summary of a Literature Survey of Mass Scattering Efficiencies (α_sp) Using the Partial Scattering Method^aa Fine corresponds to mixed aerosol in that size range. Sulfate, POM (particulate organic matter), nitrate (ammonium nitrate) and dust correspond to the fine mode. Values in bold refer to estimates computed assuming size is constant. The reference is given in the eighth column, and the footnotes attached to the locations give further details of the measurements. All values reported for a wavelength of 550 nm unless otherwise stated in the footnotes. Size ranges, chemical forms and relative humidity (when reported) are stated in the footnotes. In the “sulfate” column, italicized values correspond to dry ammonium sulfate as reported in the manuscript. In the “POM” column, italicized values are normalized to POM multiplier of R_oc = 1.8 by us or as reported in the manuscript. Unit is m² g⁻¹.

Fine Mode	Sulfate	POM	Nitrate	Dust	Location (Study)	Time	Reference
	3.5, 2.8^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	4.8, 3.2	3.8, 3.0^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	2.4	Meadview, AZ (MOHAVE)^cc Dp < 2.5 μm. Core/shell scenario with constant particle number. Sulfate is ammonium sulfate. Roc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 26.1%. Wavelength of 530 nm.	summer 1992	Lowenthal et al. [1995]
	2.7, 2.1^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	3.6, 2.4^dd Value not included in total average in Table 5 with reason listed in footnote for that entry.	3.3, 2.6^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	2.3^dd Value not included in total average in Table 5 with reason listed in footnote for that entry.	Meadview, AZ (MOHAVE)^ee Dp < 2.5 μm. Homogeneous composition with constant particle size. Sulfate is ammonium sulfate. Roc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 26.1%. Wavelength of 530 nm. Dust and POM not included in average because of solubility.	summer 1992	Lowenthal et al. [1995]
	4, 2.6^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	5, 3.3	4.1, 2.7^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	2.1	Meadview, AZ (MOHAVE)^ff Dp < 2.5 μm. Core/shell scenario with constant particle number. Sulfate is ammonium sulfate. Roc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 45.1%. Wavelength of 530 nm.	winter 1992	Lowenthal et al. [1995]
	3.1, 2.0^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	4, 2.7^dd Value not included in total average in Table 5 with reason listed in footnote for that entry.	3.7, 2.4^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	2.4^dd Value not included in total average in Table 5 with reason listed in footnote for that entry.	Meadview, AZ (MOHAVE)^gg Dp < 2.5 μm. Homogeneous composition with constant particle size. Sulfate is ammonium sulfate. Roc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 45.1%. Wavelength of 530 nm. Dust and POM not included in average because of solubility.	winter 1992	Lowenthal et al. [1995]
	2.7, 2.0^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	3.6, 2.4	3.8, 2.9^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	2	Phoenix, AZ^hh Dp < 2.5 μm. Core/shell scenario with constant particle number. Sulfate is ammonium sulfate. Roc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 31.8%. Wavelength of 530 nm.	winter 1989–1990	Lowenthal et al. [1995]
	2.2, 1.7^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	3, 2^dd Value not included in total average in Table 5 with reason listed in footnote for that entry.	3.1, 2.3^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	2.1^dd Value not included in total average in Table 5 with reason listed in footnote for that entry.	Phoenix, AZ^ii Dp < 2.5 μm. Homogeneous composition with constant particle size. Sulfate is ammonium sulfate. Roc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 31.8%. Wavelength of 530 nm. Dust and POM not included in average because of solubility.	winter 1989–1990	Lowenthal et al. [1995]
	6.9, 2.9^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	8.5, 5.7	1.8, 0.9^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	4.3	Uniontown, PA^jj Dp < 2.5 μm. Core/shell scenario with constant particle number. Sulfate is ammonium bisulfate. Roc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 66.7%. Wavelength of 530 nm.	summer 1990	Lowenthal et al. [1995]
	7, 2.9^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	8.8, 5.9^dd Value not included in total average in Table 5 with reason listed in footnote for that entry.	3.7, 1.8^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	4.2^dd Value not included in total average in Table 5 with reason listed in footnote for that entry.	Uniontown, PA^kk Dp < 2.5 μm. Homogeneous composition with constant particle size. Sulfate is ammonium bisulfate. Roc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 66.7%. Wavelength of 530 nm. Dust and POM not included in average because of solubility.	summer 1990	Lowenthal et al. [1995]
	1.23, 0.79^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	3.81, 2.5			Bondville, IL^ll Dp < 2.5 μm. Core/shell scenario assumed. Low relative humidity (30–60%). Roc = 1.2.	May–Dec 1994	Omar et al. [1999]
	5.78, 2.33^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	6.9, 4.6			Bondville, IL^mm Dp < 2.5 μm. Core/shell scenario assumed. High relative humidity (RH > 75%). Roc = 1.2.	May–Dec 1994	Omar et al. [1999]
4.5	7.1, 3.1^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	5.6, 3.11^dd Value not included in total average in Table 5 with reason listed in footnote for that entry.	6.6, 2.9^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.		Denver, CO^nn Dp < 2.5 μm. Noninteractive efficiency, particle size constant. Water soluble organic mass is 25% of total carbon. Relative humidity of 74%. Roc not stated. Wavelength not stated. POM not included in average because of OC solubility and no Roc.	5 Jan 1998	Sloane et al. [1991]
2.3	2.1, 1.5^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	4.0, 2.2^dd Value not included in total average in Table 5 with reason listed in footnote for that entry.	2.3, 1.6^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.		Denver, CO^oo Dp < 2.5 μm. Noninteractive efficiency, particle size constant. Water soluble organic mass is 25% of total carbon. Relative humidity of 38%. Roc not stated. Wavelength not stated. POM not included in average because of OC solubility and no Roc.	13 Jan 1998	Sloane et al. [1991]
	2.47, 1.87^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	3.61, 2.8	2.63, 1.99^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	1.76	Mt Zirkel Wilderness Area, CO (MZVS)^pp Dp < 2.5 μm. Constant composition as a function of size assumed, unclear if size is kept constant. Roc = 1.4. Relative humidity of 31%.	Aug, Sep, Oct 1995	Watson et al. [2001]
	16, 2.7^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	6.7, 5.2			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^qq Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 92.2%. Wavelength of 525 nm.	18 Jan 1990	Zhang et al. [1994]
	6.3, 3.2^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	5.3, 4.12			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^rr Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 63.3%. Wavelength of 525 nm.	20 Jan 1990	Zhang et al. [1994]
	4.6, 3.0^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	4.5, 3.5			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^ss Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 44.8%. Wavelength of 525 nm.	21 Jan 1990	Zhang et al. [1994]
	3.4, 2.5^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	5.7, 4.4			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^tt Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 32.9%. Wavelength of 525 nm.	22 Jan 1990	Zhang et al. [1994]
	5.6, 2.4^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	6.3, 4.9			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^uu Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 72.2%. Wavelength of 525 nm.	30 Jan 1990	Zhang et al. [1994]
	3.5, 2.0^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	8.1, 6.3			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^vv Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 53.6%. Wavelength of 525 nm.	12 Feb 1990	Zhang et al. [1994]
	8, 2.7^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	5.6, 4.9			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^ww Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 81.8%. Wavelength of 525 nm.	13 Feb 1990	Zhang et al. [1994]
	15.4, 2.2^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	6.7, 5.2			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^xx Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 92.6%. Wavelength of 525 nm.	19 Feb 1990	Zhang et al. [1994]
	5.6, 3.2^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	5.6, 4.4			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^yy Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 55.2%. Wavelength of 525 nm.	28 Feb 1990	Zhang et al. [1994]
	3, 1.9^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	7, 5.4			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^zz Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 48.0%. Wavelength of 525 nm.	1 Mar 1990	Zhang et al. [1994]
	5.4, 2.6^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	4.2, 3.3			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^aaaa Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 67.9%. Wavelength of 525 nm.	2 Mar 1990	Zhang et al. [1994]
	5.4, 2.7^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	5.1, 4.0			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^bbbb Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 63.5%. Wavelength of 525 nm.	3 Mar 1990	Zhang et al. [1994]
	5.2, 1.93^bb We normalized value to dry ammonium sulfate or dry ammonium nitrate.	1.3, 1.01			Hopi Point, Grand Canyon National Park, AZ (NGSVS)^cccc Dp < 2.5 μm. Total number remains constant but peak size shifts. Roc = 1.4. Relative humidity of 76.5%. Wavelength of 525 nm.	4 Mar 1990	Zhang et al. [1994]

• a Fine corresponds to mixed aerosol in that size range. Sulfate, POM (particulate organic matter), nitrate (ammonium nitrate) and dust correspond to the fine mode. Values in bold refer to estimates computed assuming size is constant. The reference is given in the eighth column, and the footnotes attached to the locations give further details of the measurements. All values reported for a wavelength of 550 nm unless otherwise stated in the footnotes. Size ranges, chemical forms and relative humidity (when reported) are stated in the footnotes. In the “sulfate” column, italicized values correspond to dry ammonium sulfate as reported in the manuscript. In the “POM” column, italicized values are normalized to POM multiplier of R_oc = 1.8 by us or as reported in the manuscript. Unit is m² g⁻¹.
• b We normalized value to dry ammonium sulfate or dry ammonium nitrate.
• c D_p < 2.5 μm. Core/shell scenario with constant particle number. Sulfate is ammonium sulfate. R_oc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 26.1%. Wavelength of 530 nm.
• d Value not included in total average in Table 5 with reason listed in footnote for that entry.
• e D_p < 2.5 μm. Homogeneous composition with constant particle size. Sulfate is ammonium sulfate. R_oc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 26.1%. Wavelength of 530 nm. Dust and POM not included in average because of solubility.
• f D_p < 2.5 μm. Core/shell scenario with constant particle number. Sulfate is ammonium sulfate. R_oc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 45.1%. Wavelength of 530 nm.
• g D_p < 2.5 μm. Homogeneous composition with constant particle size. Sulfate is ammonium sulfate. R_oc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 45.1%. Wavelength of 530 nm. Dust and POM not included in average because of solubility.
• h D_p < 2.5 μm. Core/shell scenario with constant particle number. Sulfate is ammonium sulfate. R_oc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 31.8%. Wavelength of 530 nm.
• i D_p < 2.5 μm. Homogeneous composition with constant particle size. Sulfate is ammonium sulfate. R_oc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 31.8%. Wavelength of 530 nm. Dust and POM not included in average because of solubility.
• j D_p < 2.5 μm. Core/shell scenario with constant particle number. Sulfate is ammonium bisulfate. R_oc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 66.7%. Wavelength of 530 nm.
• k D_p < 2.5 μm. Homogeneous composition with constant particle size. Sulfate is ammonium bisulfate. R_oc = 1.2. 60% of OC and 20% of dust is considered soluble. Dust computed as the difference between the PM2.5 mass and the sum of ionic and carbonaceous compounds. Relative humidity of 66.7%. Wavelength of 530 nm. Dust and POM not included in average because of solubility.
• l D_p < 2.5 μm. Core/shell scenario assumed. Low relative humidity (30–60%). R_oc = 1.2.
• m D_p < 2.5 μm. Core/shell scenario assumed. High relative humidity (RH > 75%). R_oc = 1.2.
• n D_p < 2.5 μm. Noninteractive efficiency, particle size constant. Water soluble organic mass is 25% of total carbon. Relative humidity of 74%. R_oc not stated. Wavelength not stated. POM not included in average because of OC solubility and no R_oc.
• o D_p < 2.5 μm. Noninteractive efficiency, particle size constant. Water soluble organic mass is 25% of total carbon. Relative humidity of 38%. R_oc not stated. Wavelength not stated. POM not included in average because of OC solubility and no R_oc.
• p D_p < 2.5 μm. Constant composition as a function of size assumed, unclear if size is kept constant. R_oc = 1.4. Relative humidity of 31%.
• q D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 92.2%. Wavelength of 525 nm.
• r D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 63.3%. Wavelength of 525 nm.
• s D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 44.8%. Wavelength of 525 nm.
• t D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 32.9%. Wavelength of 525 nm.
• u D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 72.2%. Wavelength of 525 nm.
• v D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 53.6%. Wavelength of 525 nm.
• w D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 81.8%. Wavelength of 525 nm.
• x D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 92.6%. Wavelength of 525 nm.
• y D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 55.2%. Wavelength of 525 nm.
• z D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 48.0%. Wavelength of 525 nm.
• aa D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 67.9%. Wavelength of 525 nm.
• bb D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 63.5%. Wavelength of 525 nm.
• cc D_p < 2.5 μm. Total number remains constant but peak size shifts. R_oc = 1.4. Relative humidity of 76.5%. Wavelength of 525 nm.

Table 5. Average and One Standard Deviation of Mass Scattering Efficiencies Reported in Tables 1–4^aa Fine, coarse, and total correspond to the size range mode of the aerosol. “Mixed” refers to a mixed composition aerosol. Sulfate efficiencies correspond to dry ammonium sulfate, nitrate entries correspond to dry ammonium nitrate, POM efficiencies have been normalized to an R_oc value of 1.8, and sea salt efficiencies have been adjusted to a dry state. The sixth column gives the overall average for all methods for the mixed composition aerosols and the average of three methods (theoretical, MLR and partial) for the remaining species. The number of observations is given in parentheses. Estimates are for visible wavelengths (near 550 nm).

Species/Mode	Theoretical, m^2 g^−1	Measurement, m^2 g^−1	MLR, m^2 g^−1	Partial, m^2 g^−1	All Methods, m^2 g^−1
Fine mixed	4.3 ± 0.7 (26)	3.4 ± 1.2 (54)	3.1 ± 1.4 (16)	3.4 ± 1.6 (2)	3.6 ± 1.2 (98)
Coarse mixed	1.6 ± 1.0 (21)	0.40 ± 0.15 (4)	0.7 ± 0.4 (26)		1.0 ± 0.9 (51)
Total mixed	2.2 ± 1.0 (9)	1.7 ± 1.1 (11)			1.9 ± 1.1 (20)
Fine sulfate	2.1 ± 0.7 (34)		2.8 ± 0.5 (53)	2.2 ± 0.7 (6)	2.5 ± 0.6 (93)
Fine nitrate			2.8 ± 0.5 (42)	2.3 ± 0.5 (6)	2.7 ± 0.5 (48)
Fine POM	5.6 ± 1.5 (19)		3.1 ± 0.8 (39)		3.9 ± 1.5 (58)
Coarse POM	2.6 ± 1.1 (19)				2.6 ± 1.1 (19)
Total POM	3.8 ± 0.5 (7)		1.4 (1)		3.5 ± 1.0 (8)
Fine dust	3.4 ± 0.5 (19)		2.6 ± 0.4 (4)		3.3 ± 0.6 (23)
Coarse dust	0.7 ± 0.2 (20)		0.40 ± 0.08 (2)		0.7 ± 0.2 (22)
Total dust	1.2 ± 0.3 (9)	0.9 ± 0.8 (5)	0.7 ± 0.2 (3)		1.1 ± 0.4 (12)
Fine sea salt	4.5 ± 0.7 (22)		3.7 ± 1.7 (3)		4.5 ± 0.9 (25)
Coarse sea salt	1.0 ± 0.2 (19)		0.72 ± 0.02 (2)		1.0 ± 0.2 (21)
Total sea salt	2.2 ± 0.5 (8)		1.8 ± 0.3 (2)		2.1 ± 0.5 (10)

• a Fine, coarse, and total correspond to the size range mode of the aerosol. “Mixed” refers to a mixed composition aerosol. Sulfate efficiencies correspond to dry ammonium sulfate, nitrate entries correspond to dry ammonium nitrate, POM efficiencies have been normalized to an R_oc value of 1.8, and sea salt efficiencies have been adjusted to a dry state. The sixth column gives the overall average for all methods for the mixed composition aerosols and the average of three methods (theoretical, MLR and partial) for the remaining species. The number of observations is given in parentheses. Estimates are for visible wavelengths (near 550 nm).

Table 6. Mass Scattering Efficiencies From Earlier Tables Separated Into Urban, Remote and Ocean Regions^aa Mixed fine, coarse, and total mode specific mass scattering efficiencies are averages of all four methods for both dry and ambient conditions. The mass scattering efficiencies for individual species are averages of only the theoretical, MLR and partial scattering methods under dry conditions (RH = 0%). The average for a specific method is listed for the theoretical (T), measurement (M), multilinear regression (MLR) or partial (P) methods. Estimates are for visible wavelengths (near 550 nm but could vary depending on individual study). The number of observations in the average is in parentheses. Unit is m² g⁻¹.

Species	Urban	Remote/Rural Continental	Ocean/Marine
Fine mode mixed	3.2 ± 1.3 (32), (M) 3.4 ± 1.4 (22), (MLR) 2.3 ± 0.8 (6), (P) 3.4 ± 1.6 (2)	3.1 ± 1.4 (24), (T) 4.0 ± 0.5 (3), (M) 2.9 ± 1.3 (14), (MLR) 3.3 ± 1.7 (7)	4.1 ± 0.8 (42), (T) 4.3 ± 0.7 (23), (M) 3.8 ± 0.8 (18), (MLR) 4.3 (1)
Coarse mode mixed	(MLR) 0.6 ± 0.3 (6)	0.7 ± 0.4 (24), (T) 0.6 (1), (M) 0.40 ± 0.15 (4), (MLR) 0.7 ± 0.5 (19)	1.6 ± 1.0 (21), (T) 1.7 ± 1.0 (20), (MLR) 0.6 (1)
Total mode mixed	1.7 ± 1.0 (14), (T) 1.4 ± 0.7 (3), (MLR) 1.7 ± 1.1 (11)		(T) 2.5 ± 1.0 (6)
Fine mode sulfate	2.6 ± 0.7 (9), (MLR) 2.8 ± 0.5 (5), (P) 2.3 ± 0.8 (4)	2.7 ± 0.5 (56), (T) 2.5 ± 0.3 (6), (MLR) 2.8 ± 0.5 (48), (P) 2.05 ± 0.07 (2)	(T) 2.0 ± 0.7 (28)
Fine mode nitrate	2.2 ± 0.5 (6), (MLR) 2.1 ± 0.3 (2), (P) 2.2 ± 0.6 (4)	2.8 ± 0.5 (42), (MLR) 2.8 ± 0.5 (40), (P) 2.50 ± 0.14 (2)
Fine mode POM	(MLR) 2.5 (1)	(MLR) 3.1 ± 0.8 (38)	(T) 5.6 ± 1.5 (19)
Coarse mode POM			(T) 2.6 ± 1.1 (19)
Total mode POM			3.5 ± 0.9 (8), (T) 3.8 ± 0.5 (7), (MLR) 1.4 (1)
Fine mode dust		(MLR) 2.6 ± 0.4 (4)	(T) 3.4 ± 0.5 (19)
Coarse mode dust		(MLR) 0.5 ± 0.2 (3)	(T) 0.7 ± 0.2 (19)
Total mode dust		(MLR) 0.71 (1)	1.1 ± 0.4 (11), (T) 1.2 ± 0.3 (9), (MLR) 0.7 ± 0.3 (2)
Fine mode sea salt		(MLR) 1.8 (1)	4.6 ± 0.7 (24), (T) 4.5 ± 0.7 (22), (MLR) 4.71 ± 0.07 (2)
Coarse mode sea salt			0.96 ± 0.18 (21), (T) 0.99 ± 0.17 (19), (MLR) 0.72 ± 0.02 (2)
Total mode sea salt			2.1 ± 0.5 (10), (T) 2.2 ± 0.5 (8), (MLR) 1.8 ± 0.3 (2)

• a Mixed fine, coarse, and total mode specific mass scattering efficiencies are averages of all four methods for both dry and ambient conditions. The mass scattering efficiencies for individual species are averages of only the theoretical, MLR and partial scattering methods under dry conditions (RH = 0%). The average for a specific method is listed for the theoretical (T), measurement (M), multilinear regression (MLR) or partial (P) methods. Estimates are for visible wavelengths (near 550 nm but could vary depending on individual study). The number of observations in the average is in parentheses. Unit is m² g⁻¹.

[37] The normalized efficiencies listed in the tables are in bold and noted with a footnote if we performed the normalization, and with no footnote if the value was reported under normalized conditions. We also list the original value. For example, a given study may report a sulfate ion mass scattering efficiency at 60% RH. That estimate will be listed, but beside it the dry ammonium sulfate value will be listed in bold and with a footnote. Similar notation is applied for POM, nitrate and sea salt. Estimates correspond to the fine mode unless noted with a footnote, denoting whether the efficiency corresponds to the coarse, or total mode.

[38] A summary of mass scattering efficiencies for all methods is presented in Table 5 as a function of species and size range. Estimates in Table 5 correspond to dry fully neutralized fine mode ammonium sulfate, dry fine mode ammonium nitrate, POM with a multiplier of 1.8, and dry sea salt. These estimates are averages (and one standard deviation) of the individual values reported in Tables 1–4 for each method. Values included in the average for each method have met the following criteria. The mass scattering efficiency for a given species is normalized (either by the authors or by us) to the above stated conditions. A given study had to clearly state the assumptions and methodology used to compute a reported efficiency. POM and dust are assumed to be nonhygroscopic. Estimates from the partial scattering method are computed assuming constant particle size. If an estimate in Tables 1–4 is not included in the overall average, it is marked by a footnote listing the reason. Some studies report estimates obtained using more than one method and therefore might be cited in more than one method table. In the following sections the efficiencies for each method will be discussed.

3.1. Theoretical Method

[39] Of the 17 studies listed for the theoretical method in Table 1, 7 of them occurred in ocean or marine regions, 9 of them in rural or remote continental regions, and only 1 in an urban location. Although the estimates of α_sp reported for the theoretical method are more representative of remote regions, it is impossible to tell the origin or degree of processing of the measured aerosols, especially since long-range transport and aging of aerosols could have occurred, especially for many of the ocean (cruise-based) measurements. Of these 17 studies, 6 of them used size-resolved impactor data, 5 of them used finer resolution DMA, APS or OPC data, and 6 of them used apportionment methods, combining impactor data with finely resolved DMA, APS or OPC data. The third to last column in Table 1 denotes the type of data used.

[40] The average (and one standard deviation) fine mode dry ammonium sulfate mass scattering efficiency for all the estimates determined with the theoretical method is 2.1 ± 0.7 m² g⁻¹. Of the values included in the average (see Table 1), the highest value is 3.3 m² g⁻¹ at the Azores during a period influenced by a pollution plume from Europe [Howell and Huebert, 1998]. The lowest value of 0.8 m² g⁻¹ was observed in the Southern Ocean region south of Australia during the ACE-1 (Aerosol Characterization Experiment) study [Quinn et al., 1998], corresponding to a small average fine mode surface mean diameter of 0.20 ± 0.04 μm. The range in fine mode dry ammonium sulfate mass scattering efficiencies is most likely due to variations in its size distribution due to aerosol processing, as suggested from the conditions under which the lowest and highest values were observed. For example, values tend to be lower under clean conditions, such as some periods in the Azores [Howell and Huebert, 1998], near the Grand Canyon National Park AZ [Malm and Pitchford, 1997], and remote ocean regions [Quinn et al., 1998, 2002b]. The effects of high RH on sulfate mass scattering efficiencies are dramatic. For example, wet specific mass scattering efficiencies (wet b_sp/dry mass) reported for Hopi Point AZ, Meadview AZ, Shenandoah VA, and Great Smoky Mountains TN [Malm and Pitchford, 1997] are significantly higher than the dry case, especially at Shenandoah where the ambient humidity is high in summer and the wet ammonium sulfate specific mass scattering efficiency was 18.23 m² g⁻¹ for RH = 82.5%. It is notable that given the range in time periods and the global distribution in the measurements, a fairly low variability in dry ammonium sulfate efficiency is observed, perhaps a reflection of the robustness of this method for computing efficiencies, especially when the optical and hygroscopic properties are well understood. This low variability is in contrast to the range reported by Charlson et al. [1999] (2.6–5.5 m² g⁻¹ for ambient ammonium sulfate), reflecting the importance for normalizing estimates to the same RH conditions.

[41] All of the POM mass scattering efficiencies are from cruise-based measurements. Some of these estimates were computed assuming that POM size distributions are similar to that of apportioned impactor measurements of sulfate distributions [e.g., Quinn et al., 2001], but others are derived from POM seven-stage impactor mass distributions apportioned to finer resolution size distributions [Quinn et al., 2002b, 2004]. The average fine mode POM mass scattering efficiency normalized for R_oc = 1.8 is 5.6 ± 1.5 m² g⁻¹. The normalization did not significantly change most of the estimates (see bolded values in Table 1) since most authors in these studies applied R_oc values consistent with Turpin and Lim's [2001] suggested values that vary according to aerosol type. For example, during the Aerosols99 study, Quinn et al. [2001] reported values for locations ranging from the Atlantic Ocean off the coast of Virginia to Cape Town, South Africa, with aerosol origins ranging from remote marine to African dust and biomass burning. They applied a POM multiplier of 1.6 for North American air masses (corresponding to more urban aerosols) and 2.1 for all other air masses. The minimum average fine mode value of 4.7 ± 0.6 m² g⁻¹ was reported for several air masses during ACE-Asia [Quinn et al., 2004], especially during air masses influenced by dust outbreaks. The maximum average value (7.2 ± 1.5 m² g⁻¹) was reported during the Aerosols99 study [Quinn et al., 2001] corresponding to air masses influenced by African dust and biomass burning off the coast of Africa. Higher values of POM α_sp corresponding to biomass burning aerosols are consistent with recent results by Malm et al. [2005] and McMeeking et al. [2005a] who obtained values of mixed fine mode α_sp near 6 m² g⁻¹ in Yosemite National Park CA during periods dominated by biomass smoke. A POM mass scattering efficiency near 7.2 m² g⁻¹ would require specific conditions of a fairly narrow lognormal size distribution with a mass mean diameter near 0.4 μm, geometric standard deviation of 1.5, an organic aerosol density of 1.2 and refractive index of 1.55. The average coarse mode and total mode POM mass scattering efficiencies are 2.6 ± 1.1 m² g⁻¹ and 3.8 ± 0.5 m² g⁻¹. Unlike sulfate aerosols, the physical and chemical properties of organic carbon aerosols are not well understood and can vary considerably depending on whether they are of primary or secondary origin. The range of values and degree of variability observed for fine mode POM reflects the degree of uncertainty surrounding the optical properties of this aerosol type.

[42] All of the dust mass scattering efficiency estimates from the theoretical method are derived from measurements near marine or ocean regions and most occur near large dust sources, such as North Africa or Asia. Average fine, coarse, and total (fine + coarse) dust mass scattering efficiencies are 3.4 ± 0.5, 0.7 ± 0.2, and 1.2 ± 0.3 m² g⁻¹, respectively, with the individual values reported in Table 1. The fine dust mode mass scattering efficiency most likely corresponds to the tail of the coarse mode dust distribution extending into the fine mode size range. The high fine mode scattering efficiencies (compared to the coarse mode) are a result of assuming an ideal collection efficiency that includes only the particles with diameters less than the fine mode size cut (see section 2.5). The lower estimates associated with the coarse mode reflect less efficient light-scattering particles and larger mass. Midrange values corresponding to the total aerosol mode reflect the integration over all sizes. The variability for each of these estimates is fairly low, considering differences in dust composition and size distributions, the study locations and long-range transport, and dust origins, which include Asia, the Indian Ocean, and off the coast of North Africa [Chiapello et al., 1999; Li et al., 2000; Quinn et al., 2001, 2002b, 2004]. However, fairly well understood dust properties also contribute to lower uncertainties, such as its nonhygroscopicity and fairly well known refractive index and density.

[43] The measurements used to derive average dry sea salt mass scattering efficiencies occurred on cruise platforms around the globe and again were derived assuming ideal collection efficiencies and apportioned size distributions. The average dry fine, coarse, and total sea salt specific mass scattering efficiencies are 4.5 ± 0.7, 1.0 ± 0.2, and 2.2 ± 0.5 m² g⁻¹, respectively. Similar to dust, the fine mode sea salt α_sp values correspond to the tail of the coarse mode sea salt size distribution extending into the fine mode size range. The lower values associated with the coarse and total modes reflect the dominance of larger particles that are less efficient at light scattering. These values have some uncertainty associated with the assumptions used to normalize the values to a dry basis, and from the different assumptions of sea salt composition. However, given that the measurements were performed all over the globe, from clean regions in the Southern Ocean [Quinn et al., 2001], to more polluted regions off of India and Asia [Quinn et al., 2002b, 2004], the low variability is notable.

[44] The average mixed aerosol fine, coarse, and total (fine + coarse) mode specific mass scattering efficiencies are 4.3 ± 0.7, 1.6 ± 1.0, and 2.2 ± 1.0 m² g⁻¹, respectively. The fine mode values could include ammoniated salts, carbonaceous aerosols, and mineral aerosols. These efficiencies were reported for a range of RH values, and no normalizations to a dry aerosol or common mass basis were performed because of the lack of information on the fine mode composition. The cutoff diameter for these modes may also vary from study to study, ranging from 1 to 3.0 μm. Given these differences, there is still a fairly narrow range of values reported for fine mode specific efficiencies. A larger variation in the coarse mode efficiencies was observed, with a minimum value 0.6 m² g⁻¹ observed at Big Bend National Park [Hand, 2001] where soil was ∼40% of the coarse aerosol composition on average. The maximum value of 2 m² g⁻¹ was observed in the Indian Ocean [Quinn et al., 2002b]. Mixed coarse mode α_sp values derived from many cruise-based measurements [Quinn et al., 2001, 2002b, 2004] were typically between 1–2 m² g⁻¹, reflecting the influence of other species besides dust, such as hygroscopic sea salt (and associated water) and POM (recall the average coarse mode POM α_sp was 2.6 ± 1.1 m² g⁻¹). The composition, size distribution, and optical properties of the coarse mode mixed aerosol vary considerably on the basis of the studies surveyed here.

3.2. Measurement Method

[45] Of the 21 studies listed for the measurement method, most of them occurred in rural, remote continental regions and urban regions, with only 4 studies occurring in ocean or marine areas. Estimates of specific mass scattering efficiencies for the measurement method are reported in Table 2. Normalizing these values to a common RH or composition is not possible because of the range of RH values at which the measurements are made and the unknown mixed aerosol composition. Specific mass scattering efficiencies for the fine, coarse and total modes listed in Table 2 are averaged and reported in Table 5. Some studies in Table 2 report mass scattering efficiencies for sulfate or dust; however, these estimates are based on assumptions that make them incomparable to other values reported in Table 5. For example, some authors make the assumption that a given polluted air mass is associated with one specific aerosol type [e.g., Vrekoussis et al., 2005], thereby attributing all the measured light scattering to that aerosol species. However, realistically the aerosol mixture is not associated only with one aerosol species, nor can the assumption be justified from the amount of information available; therefore we do not include these estimates in the summary in Table 5.

[46] The average mixed aerosol fine mode specific mass scattering efficiency is 3.4 ± 1.2 m² g⁻¹. The lowest value (1.65 m² g⁻¹) was obtained by Chan et al. [2002] in the city of Brisbane, Australia. However, other urban regions are associated with much higher values (>5 m² g⁻¹) such as those reported in Mexico City, Mexico [Chow et al., 2002] and Fort Meade MD [Chen et al., 2003]. These values could be high because of relative humidity effects (especially in the eastern United States), or because of the aerosol size distribution or composition. With the exception of sites in Phoenix AZ and sites near Denver CO, the urban (e.g., Beijing, China, Atlanta GA, Dallas TX) fine mode specific mass scattering efficiencies are greater than 3 m² g⁻¹. One of the highest values observed (6 m² g⁻¹) was in rural New England-mid-Atlantic region [Poirot and Husar, 2004] and corresponds to biomass smoke aerosol that appear to be associated with higher efficiencies, as suggested in the previous section. Most of the rural sites in the southwestern United States correspond to lower fine mode specific efficiencies (less than ∼2.5 m² g⁻¹), especially in Meadview AZ, Mount Zirkel CO, and Spirit Mountain NV, which are close to the theoretical mass scattering efficiencies computed for dry ammonium sulfate. The exceptions to these low values are those reported for the northwestern United States (Mount Rainier WA and North Cascades WA) and in Bondville IL with efficiencies greater than 4 m² g⁻¹. Higher specific mass scattering efficiencies (usually greater than 3 m² g⁻¹) were observed during the INDOEX experiment (Indian Ocean Experiment) in the Maldives [Clarke et al., 2002]. Specific mass scattering efficiencies for this study are ordered as a function of light scattering coefficients and no significant difference was observed for low scattering (b_sp < 25 Mm⁻¹) compared to high scattering periods (b_sp > 55 Mm⁻¹), in contrast to the relationship Lowenthal and Kumar [2004] and Malm and Hand [2007] observed for data from the IMPROVE network [Malm et al., 1994a]. The average mixed coarse and total mode specific mass scattering efficiencies are 0.40 ± 0.15 and 1.7 ± 1.1 m² g⁻¹, respectively. Specific mass scattering efficiencies corresponding to the total aerosol mode were reported for Yulin, China [Alfaro et al., 2003], during the ACE-Asia study. Values were higher during polluted periods compared to periods dominated by dust aerosols, probably because of the greater influence of more efficient light scattering fine mode particles during the polluted period. However, much of the variability seen in all of the measurement method estimates could be related to the effects of RH on the efficiencies. Also, although many authors accounted for truncation errors from nephelometer measurements, this issue could still contribute some uncertainties to estimates derived using those data (see footnotes in Tables 2 and 3).

3.3. Multilinear Regression (MLR) Method

[47] The multilinear regression method table includes 25 studies, the largest contribution of any of the methods. Most of these studies were performed in rural, remote continental regions, followed by urban regions, and the fewest in ocean/marine regions. Estimates are reported for mixed fine and coarse mode aerosols, sulfate, POM, nitrate, dust and sea salt. These values are subject to variability based on how the regressions were performed (with or without intercepts, assuming an internal or external mixture, or accounting for measurement uncertainties) and the types of data used. Estimates of mass scattering efficiencies obtained using the MLR method are listed in Table 3, and summarized in Table 5. Fine mode mass scattering efficiencies associated with coarse mode species tend to be lower than those derived by the theoretical method because of the effects of the realistic collection efficiencies on the data, as discussed in section 2.5.

[48] MLR analyses are typically performed assuming each species forms an independent variable and the regression coefficients are interpreted as mass scattering or absorption efficiencies. However, Malm and Hand [2007] performed an MLR analysis where the mass scattering efficiency estimates were obtained by assuming inorganic and organic species were internally mixed, forming a single variable in the regression equation and thus reducing or eliminating collinearity issues discussed in section 2.4. The POM mass scattering efficiency then was estimated by prorating the inorganic specific mass scattering efficiency by the ratio of the ammonium sulfate density to organic carbon density, using an organic carbon density of 1.4 g cm⁻³ and assuming POM and ammonium sulfate have similar size distributions and refractive indices (see section 2.4).

[49] The average fine mode dry ammonium sulfate mass scattering efficiency is 2.8 ± 0.5 m² g⁻¹, somewhat higher than the estimate from the theoretical method (2.1 ± 0.7 m² g⁻¹). Of the averaged values, the lowest estimate (1.47 m² g⁻¹) was observed in Big Bend National Park TX [Hering et al., 2003] and was obtained from a curve fit of sulfate efficiency as a function of RH. It is considerably lower than the estimate of 2.96 m² g⁻¹ for the same time period reported by Malm et al. [2003] using the theoretical method during the same study, and the estimate of 2.48 m² g⁻¹ reported by Malm and Hand [2007] using the MLR method, although for this case over a much longer time period. The highest value (4.28 m² g⁻¹) was obtained in southern New Hampshire [Slater and Dibb, 2004]. To achieve a value of α_sp this high for dry ammonium sulfate requires a size distribution shifted to larger sizes, requiring, for example, a lognormal sulfate size distribution with a mass mean size of 0.4 μm and geometric standard deviation of 1.7. However, it is also possible that the higher mass scattering efficiency is associated with the known bias in the MLR method that tends to give more weight to those variables that are more accurately measured. In addition, the normalization procedure could contribute some uncertainty in these estimates. Variations in relative humidity or aerosol acidity could result in uncertainties in the corrections as these were applied in an average sense.

[50] The average normalized value of fine mode POM mass scattering efficiency is 3.1 ± 0.8 m² g⁻¹. This estimate is considerably lower than reported for the theoretical method (5.6 ± 1.5 m² g⁻¹) and possibly artificially low because of artifacts of the MLR technique that result in lower coefficients for data with greater measurement uncertainties. However, these artifacts would not affect results from Malm and Hand [2007] who analyzed data from 34 of the 39 estimates reported in the average in Table 5 by treating the fine mode as an internal mixture, thus forming only one variable in the regression model. The lowest value is 1.8 ± 0.4 m² g⁻¹, obtained at Cranmore Mountain NH [Slater et al., 2002], and the highest values (4.2 and 4.1 m² g⁻¹) were observed at Okefenokee National Wildlife Refuge, GA and Glacier National Park MT [Malm and Hand, 2007].

[51] The average dry fine mode ammonium nitrate mass scattering efficiency is 2.8 ± 0.5 m² g⁻¹. The lowest value (1.9 m² g⁻¹) was observed in Rocky Mountain National Park CO [Malm and Hand, 2007], and the highest estimate (4.3 ± 0.3 m² g⁻¹) was reported by Watson et al. [2001] at Mount Zirkel CO. Some of the range of variability observed with nitrate mass scattering efficiencies could be associated with assumptions applied in the normalization to dry ammonium nitrate. The main assumption is that fine mode nitrate is in the molecular form of ammonium nitrate. In fact, size distribution measurements at many remote locations by Lee et al. [2004] have demonstrated that fine mode nitrate can often be associated with the tail of coarse mode nitrate in the form of Ca(NO₃)₂ or NaNO₃, not fine mode ammonium nitrate. The f(RH) curves and assumed size distributions applied in the normalization are appropriate for ammonium nitrate; if the measured nitrate is in a different form, these physical, hygroscopic and optical properties are inappropriate. Without this additional size distribution and composition information, however, we cannot refine the normalization.

[52] The average fine mode dust mass scattering efficiency is 2.6 ± 0.4 m² g⁻¹, lower than the theoretical estimate (see Table 1) and with fewer number of observations. The minimum value (2.06 m² g⁻¹) was observed at Canyonlands National Park UT [Eatough et al., 1996] and the maximum value (2.73 m² g⁻¹) was observed at the urban location of Brisbane, Australia [Chan et al., 2002]. These estimates are probably also associated with the fine tail of the coarse mode dust distribution extending into the fine mode size range, and both may be high because of collinearity issues as discussed in section 2.4. The average coarse and total mode dust mass scattering efficiencies are 0.40 ± 0.08 and 0.7 ± 0.2 m² g⁻¹, although only a few observations are available (see Table 5). The estimates may be low because of truncation errors associated with the nephelometer measurements. In addition, for the most part these estimates were obtained mainly from remote and rural regions (not near any large dust sources), in contrast to the regions where the majority of the values derived with the theoretical methods were obtained.

[53] The average dry fine mode sea salt mass scattering efficiencies is 3.7 ± 1.7 m² g⁻¹, from three observations. The lowest value (1.8 m² g⁻¹) was obtained in Barrow AK [Quinn et al., 2002a] and was obtained using filter mass measurements and measured scattering. The highest value (4.71 ± 0.7 m² g⁻¹) was obtained from cruise-based measurements in the central Pacific Ocean [Quinn et al., 1996], using apportioned mass from high-resolution size distribution measurements and calculated scattering. The range of estimates may reflect the variable size distribution of coarse mode sea salt in these areas. However, as discussed in section 2.5, the variability is most likely due to differences in the mass measurement and the effects of collection efficiencies on the sampling method, especially since the high values are consistent with those obtained from the theoretical method and the assumption of an ideal collection efficiency, whereas the low value is consistent with predictions that apply a nonideal collection efficiency that includes more coarse particles in the fine mode. Collinearity issues with the MLR method as discussed with respect to dust α_sp values may also affect these sea salt α_sp estimates, as would truncation issues with the nephelometer measurements. The average coarse mode and total mass sea salt scattering efficiencies are 0.72 ± 0.02, and 1.8 ± 0.3 m² g⁻¹, respectively, and are derived from only a few cruise-based measurements in the central Pacific Ocean [Quinn et al., 1996], the Atlantic Ocean between North America and South Africa [Quinn et al., 2001] and the Indian Ocean [Quinn et al., 2002b]. All of these values are within range of those reported using the theoretical method, but lower.

[54] The average fine mode specific mass scattering efficiency is 3.1 ± 1.4 m² g⁻¹, somewhat lower than the average estimate derived for the theoretical method, and with much greater variability. These estimates are probably also affected by nonideal collection efficiencies. The majority of the mixed fine mode estimates are associated with rural or remote continental environments. The lowest value (0.82 m² g⁻¹) was observed in Barrow AK [Quinn et al., 2002a]; however, in this case the mixed fine mode aerosol was computed as gravimetric mass minus the mass of ionic species and water, so its composition (and size distribution) might be very different than some of the other estimates. The highest value (5.8 m² g⁻¹) was reported at a remote site in the Negev desert, Israel [Andreae et al., 2002] but probably is mostly associated with anthropogenic aerosols. Average coarse mode specific mass scattering efficiencies are 0.7 ± 0.4 m² g⁻¹, ranging from values greater than 1 m² g⁻¹ in Nevada and Arizona [White et al., 1994] and California [Malm and Hand, 2007] to values less than 0.5 m² g⁻¹ in Colorado, Arizona, Utah, Montana, Texas and Israel. The variability of coarse mode composition and size distributions could result in large variability in derived efficiencies, as seen when comparing these estimates to those derived with the theoretical method.

3.4. Partial Scattering Method

[55] Values of mass scattering efficiency derived from the partial scattering approach are reported in Table 4 and summarized in Table 5. Most of these studies employed the ELSIE model, as described previously, and correspond to both rural and urban locations. Species included in the model are sulfate, POM, nitrate, dust, elemental carbon, and fine mode aerosols. The two entries listed in the table per site for the first eight entries in Table 4 correspond to model results using two scenarios: (1) assuming a homogeneous composition with constant particle size and (2) a core/shell scenario with constant particle number. All estimates reported in the table using the constant size assumption are in bold, and only estimates derived with this scenario are appropriate to compare with other methods because the second scenario corresponds to changing size distributions that would also change the efficiency. Efficiencies for POM and dust were not included in the overall average when they were considered partly soluble (e.g., 60% of POM and 20% of dust is considered soluble in some studies [Lowenthal et al., 1995]). Finally, the many entries listed for Hopi Point toward the end of the table correspond to estimates made at different RH values [Zhang et al., 1994] assuming particle number is constant. As with the other methods, sulfate, POM and nitrate have been normalized.

[56] The average fine mode dry ammonium sulfate partial scattering efficiency for the constant size assumption is 2.2 ± 0.7 m² g⁻¹, similar to the results from the theoretical method (as is expected because the theoretical basis is the same). Most of the studies included in this table occurred in the western or southwestern United States with the exception of Bondville IL and Uniontown PA. The sites in the Southwest are mostly rural with the exception of Phoenix AZ. No strong spatial variation is noticeable. Both the minimum and maximum values were obtained in the urban location of Denver CO (1.5 and 3.1 m² g⁻¹, respectively [Sloane et al., 1991]). The average ammonium nitrate partial scattering efficiency is 2.3 ± 0.5 m² g⁻¹, with the lowest values in the urban locations of Uniontown PA (1.8 m² g⁻¹ [Lowenthal et al., 1995]) and Denver CO (1.6 m² g⁻¹ [Sloane et al., 1991]). The highest value (2.6 m² g⁻¹) was observed in Meadview AZ during the MOHAVE study [Lowenthal et al., 1995]. The average nitrate mass scattering efficiency derived using this method was lower than that derived by the MLR method; however, as discussed previously, some of the assumptions applied in normalizing these values could contribute to its uncertainty.

4. Discussion and Summary

[57] A survey of mass scattering efficiencies from ground-based measurements reported since 1990 was performed to examine the variability of mass scattering efficiencies as a function of aerosol species and region. The mass scattering values were summarized for different aerosol species (as defined by chemical composition) and by the methods used to derive them. These methods were separated into four techniques: theoretical method using aerosol size distribution data and Mie theory, measurement method using mass concentration data and measured aerosol optical properties, multilinear regression (MLR) method using regression techniques with mass and optical data, and partial scattering method using size distribution data and a model to apportion scattering based on removal or addition of species. Estimates derived from all four methods were averaged for mixed fine mode, coarse mode, and total mode specific mass scattering efficiencies under both dry and ambient conditions, depending on what was reported. For species-dependent mass scattering efficiencies, only three methods were averaged (the measurement method is not included) and correspond to dry conditions (RH ≈ 0%). Only values that were included in the overall averages presented in Table 1 are included in this analysis. Estimates of mass scattering efficiencies (at visible wavelengths of 530–550 nm) were normalized to the same dry mass basis for a given species.

[58] The estimates of α_sp from all the various methods were separated into geographical regions (urban, remote/rural continental and ocean/marine) and averaged as a function of species. The study site location was used to determine the geographical region; however, it is impossible to determine the degree of aging and length and time of transport aerosols underwent that were measured at these locations, so these distinctions are rough. Also, remote/rural and ocean regions may be associated with both pristine and polluted (or a combination of) conditions. For example, it is possible that aerosols measured in remote continental locations were influenced by urban sources or regional hazes, and measurements performed on board cruise ships were influenced by continental outflow.

[59] A summary of the survey of α_sp as a function of aerosol species for the four methods is reported in Table 5, while average values of α_sp for each aerosol species as a function of region are summarized in Table 6. In Table 6 the number of observations in each average is noted in parentheses. The value listed first is the average α_sp from all the methods, followed by α_sp corresponding to an individual method, noted with a (T), (MLR), (M), or (P) for the theoretical, MLR, measurement or partial scattering method, respectively. The number of observations varies considerably for each geographical region, and estimates for a given region can be dominated by values derived by a single method. The range of study locations and periods varies, obviously the same sites and measurement time periods are not available for every method.

[60] Fine mode dry ammonium sulfate mass scattering efficiencies are surprisingly similar when compared over method and region. The overall grand average of 93 observations is 2.5 ± 0.6 m² g⁻¹ where the variability is one standard deviation of the mean. This fairly low variability is notable given the variation in seasons, locations, and different experimental and analysis methods used to derive these values, not to mention differences in the aerosol microphysical properties. This narrow range in values may reflect the high level of understanding of the optical and hygroscopic properties of sulfate aerosols as well as the lower uncertainties associated with sulfate measurements.

[61] Some differences are observed in dry ammonium sulfate mass scattering efficiencies as a function of location. Estimates in ocean regions are somewhat lower (2 m² g⁻¹) compared to remote/rural continental (2.7 m² g⁻¹) or urban locations (2.6 m² g⁻¹), and lower estimates (as low as 2.0 m² g⁻¹) are also observed in fairly clean and arid locations (e.g., southwest U.S.) compared to more humid regions, suggesting that aqueous aerosol processing could result in larger size distributions and higher mass scattering efficiencies.

[62] The average theoretical estimate of fine mode dry ammonium sulfate mass scattering efficiency is 2.1 ± 0.7 m² g⁻¹ while the MLR average is 2.8 ± 0.5 m² g⁻¹. This difference between methods may in part be due to the MLR technique tending to overestimate regression coefficients and thus mass scattering efficiencies for variables that are measured with more accuracy relative to others. However, 34 of the 53 estimates from the MLR method were carried out assuming that the fine mode was internally mixed and formed one variable in the regression equation, thus minimizing or eliminating this problem.

[63] Fine mode POM mass scattering efficiencies range from 3.1 ± 0.8 to 5.6 ± 1.5 m² g⁻¹, with an average of the three methods of 3.9 ± 1.5 m² g⁻¹ for 58 observations. POM efficiencies were normalized assuming a molecular weight per carbon weight ratio of 1.8 and were assumed to be nonhygroscopic. The high value was derived by the theoretical method and the low value was computed using the MLR method. Again, it is possible that the MLR estimates of POM efficiencies are artificially low because of biases in the MLR technique for data with higher uncertainties. Assuming an internal mixture by combining POM data with other data of less uncertainty (i.e., sulfate data) in the regression may help in avoiding this issue. Much higher values of POM efficiencies were obtained over the ocean (5.6 ± 1.5 m² g⁻¹) compared to rural areas (3.1 ± 0.8 m² g⁻¹); however, many of the ocean measurements were influenced by large continental outflow that probably included combustion sources. All of the ocean estimates were computed with the theoretical method compared to the MLR method for the more rural regions. Perhaps the sources and processing of the organic aerosols contribute to these differences also, especially since higher POM efficiencies have been observed in regions influenced by biomass burning. The much larger range in POM efficiencies (compared to sulfate) may reflect the greater uncertainty in organic aerosol chemical and optical properties, size distributions, possible hygroscopic properties and varying sources and atmospheric processes, as well as greater uncertainties in the measurements themselves.

[64] Fine mode dry ammonium nitrate mass scattering efficiencies range from 2.3 ± 0.5 to 2.8 ± 0.5 m² g⁻¹, with an average value of 2.7 ± 0.5 m² g⁻¹ for 48 observations. These values are similar to the estimates for dry ammonium sulfate and in fact are often assumed to be the same [Malm et al., 1994a; Malm and Hand, 2007]. These estimates were derived with only two methods, with the high value associated with the MLR technique, and the low value associated with the partial scattering method. Estimates are somewhat higher in remote/rural locations (2.8 ± 0.5 m² g⁻¹) compared to urban regions (2.2 ± 0.5 m² g⁻¹). The estimates for the urban locations were computed predominantly with the partial scattering method while the rural estimates were computed predominantly by the MLR method. Because the MLR technique depends on bulk mass measurements, it is assumed that nitrate is associated with the fine mode. The partial scattering method relies on size distributions so the assumption of fine mode nitrate is more easily justified. It is possible that for a given observation that nitrate is not associated with fine mode ammonium nitrate, as we have assumed in our normalization, but could instead it could be associated with coarse mode nitrate species.

[65] Fine mode mixed aerosol specific mass scattering efficiencies range from 3.1 ± 1.4 to 4.3 ± 0.7 m² g⁻¹, with an average value of 3.6 ± 1.2 m² g⁻¹ for 98 observations using all four methods. These estimates correspond to a mixture of aerosol species and have not been normalized to the same composition or to dry conditions so they undoubtedly include some water mass. The high value was obtained with the theoretical method and the low value was obtained with the MLR method. The range of estimates is still fairly low given the range of aerosol physicochemical properties and variety of conditions and under which these measurements were performed. The lowest estimates (∼3.2 m² g⁻¹) are observed in urban and rural/remote regions compared to the highest estimates obtained over the ocean (4.3 ± 0.7 m² g⁻¹), most likely because of water associated with sea salt in the fine mode. Mixed aerosol coarse mode specific mass scattering efficiencies ranged from 0.40 ± 0.15 to 1.6 ± 1.0 m² g⁻¹, with an average value of 1.0 ± 0.9 m² g⁻¹ for 51 observations using all methods. As with the fine mode specific mass scattering efficiencies, these estimates correspond to a range of composition and RH values. The high estimates were derived using the theoretical method, and most of these observations were made over ocean so it is likely that water due to hygroscopic sea salt is included in these values. Lower estimates of coarse mode α_sp were derived using the MLR and measurement methods and were associated mostly with rural/remote regions and probably dust aerosols. Variations in coarse mode size distributions and composition can have a significant impact on coarse aerosol specific mass scattering efficiencies, more so than with the fine mode specific mass scattering efficiencies, as is evidenced by the range in variability.

[66] Fine mode dust mass scattering efficiencies range from 2.6 ± 0.4 to 3.4 ± 0.5 m² g⁻¹, with an average value of 3.3 ± 0.6 m² g⁻¹ for 23 values. Estimates were obtained using two methods, with the theoretical method corresponding to the high value (all obtained over the ocean) and the MLR corresponding to the low value (in remote/rural regions). It is likely that the fine dust estimates correspond to the fine tail of a coarse mode size distribution extending into the fine mode size range. Given the different assumptions in computing dust composition and the variety of different source regions that contributed to these estimates, the range in values is fairly narrow. Coarse mode dust mass scattering efficiencies range from 0.40 ± 0.08 (2 observations with MLR method) to 0.7 ± 0.2 m² g⁻¹ (20 observations), with the majority of estimates computed using the theoretical method over ocean regions. The lower values compared to fine dust estimates reflect the less efficient light scattering capability of the larger particles at visible wavelengths.

[67] Dry fine mode sea salt mass scattering efficiencies range from 3.7 ± 1.7 for the MLR method (three observations) to 4.5 ± 0.7 m² g⁻¹ derived with the theoretical method (22 observations) with an average value of 4.5 ± 0.9 m² g⁻¹, predominantly obtained over the ocean. Similar to dust, these estimates may correspond to the fine tail of the coarse mode size distribution. The coarse mode sea salt mass scattering efficiencies range from 0.72 ± 0.02 (2 observations) to 1.0 ± 0.2 m² g⁻¹ (19 observations, all obtained over the ocean). The data used in these estimates were collected in a variety of conditions all over the globe and so reflect average global sea salt mass scattering efficiencies. Our normalization to dry conditions decreased the variability originally observed in the estimates.

5. Conclusions and Recommendations

[68] Part of the difficulty in assessing reported mass scattering efficiencies is the wide variety of data and methods used to derive them. While microphysical and chemical differences undoubtedly exist between these aerosols, it is more likely that the methods themselves contribute to the large range of variability in α_sp as seen in the literature, especially given what is known about biases associated with different methods. Accurate interpretations of reported values require that the authors of the study carefully describe the many assumptions required in computing α_sp. For instance, several of the values reported in Tables 2–5 could not be included in the overall average reported in Tables 5 and 6 because basic assumptions were not stated in the manuscript and therefore we were unable to normalize the efficiency.

[69] This review of dry (RH ≈ 0%) mass scattering efficiencies in the visible wavelength range provides a basis for recommending values to be used for visibility regulation and climate forcing calculations.

[70] 1. For fine mode ammonium sulfate, an α_sp of 2.5 m² g⁻¹ is appropriate for average conditions; however, we observed lower values (∼2 m² g⁻¹) in dry, more pristine environments compared to higher values (∼3 m² g⁻¹) in more polluted environments.

[71] 2. For fine mode ammonium nitrate, an α_sp value of 2.7 m² g⁻¹ is appropriate for remote/rural locations but may be lower for urban regions. We caution that the chemical forms (and size distribution) of nitrate may vary considerably, which would affect its mass scattering efficiency.

[72] 3. For fine mode POM, an appropriate average α_sp for POM (assuming a molecular weight to carbon weight ratio of 1.8) is 3.9 m² g⁻¹, but much higher values (∼6 m² g⁻¹ or greater) were observed for aerosols influenced by industrial and biomass combustion sources.

[73] 4. For mixed fine mode, an α_sp of 3.6 m² g⁻¹ is reasonable; however, this value can vary considerably because of RH conditions, aerosol composition and size distribution.

[74] 5. For mixed coarse mode, an α_sp of 1.0 m² g⁻¹ falls within the range of observed values but can vary because of RH conditions, aerosol composition and size distribution.

[75] 6. For fine mode dust, an α_sp estimate of 3.3 m² g⁻¹ is appropriate if the application takes into account only fine mode aerosols. For fine mode data influenced by coarse mass (i.e., nonideal collection efficiencies, see section 2.5), we recommend a much lower value near 1.0 m² g⁻¹.

[76] 7. For coarse mode dust, an α_sp near 0.7 m² g⁻¹ is recommended.

[77] 8. For fine mode sea salt, a mass scattering efficiency of 4.5 m² g⁻¹ is recommended; however, as discussed above for fine dust, we caution that this value only be applied to data that include only fine mass. A much lower value of about 1.0–1.3 m² g⁻¹ is recommended if fine sea salt mass is derived from a nonideal sampler.

[78] For coarse mode sea salt, an α_sp of 1.0 m² g⁻¹ is recommended.

[79] We further recommend that the following conditions always be reported: the assumed chemical form of the species, including the refractive index and density (if used in the calculation), the wavelength at which the nephelometer measurement or Mie calculations were performed, the relative humidity of the measurement, including any assumptions regarding hygroscopic growth, and a clear description of the method used to calculate efficiency.

[80] On the basis of what we have learned from this survey we offer several recommendations for deriving and interpreting mass scattering efficiencies:

[81] 1. Obtaining representative estimates of mass scattering efficiencies requires performing measurements over an extended period of time. Many field campaigns are temporary and therefore not representative of the inherent variability in mass scattering efficiencies that can occur temporally, under varied transport and mixing conditions as well as source strength of emissions that may contribute to aerosol formation. We recommend performing measurements over extended time periods to capture some of this variability.

[82] 2. Although labor and analytically intensive, we recommend performing fine size resolution aerosol measurements that are required to make theoretical estimates of mass scattering efficiencies. This method allows for the dependence of relative humidity effects, composition, size and mass to be included. This method should be used in more geographical regions, for instance, currently there are few to no estimates in urban regions.

[83] 3. The MLR technique is useful and less data-intensive than the theoretical method but may produce artificially high or low estimates of α_sp depending on the species. We recommend caution in interpreting these estimates and suggest that grouping data for more than one species together in the regression followed by prorating the efficiencies as a technique for avoiding the possible biases.

[84] 4. The measurement technique is obviously the simplest method to employ but we recommend that it only be interpreted as providing an average specific mass scattering efficiency for a mixed aerosol because of the changing relative humidity and aerosol properties during the measurement period.

[85] 5. We recommend applying more than one method to data from a single study (if possible) to provide closure in estimates of α_sp and assist in determining whether the source of variability arises from the method or the data.

[86] 6. Mass measurements affected by size-resolved collection efficiencies can have a large impact on derived mass scattering efficiencies, especially when comparing estimates from MLR and theoretical methods. Caution should be used when interpreting and applying mass scattering efficiencies in chemical extinction budgets and global models, especially for fine mode mass scattering efficiencies associated with coarse mode species or vice versa.

[87] 7. We recommend, when possible, that authors report mass scattering efficiencies on a dry relative humidity basis. Normalization of α_sp to dry conditions is best suited to the authors who perform the measurements, as they have the most thorough understanding of their own data. Comparisons of dry α_sp are simpler and result in lower variability, and α_sp can be modified to humid conditions when the application requires it.

[88] 8. Spurious interpretations of mass scattering efficiencies can arise when values are reported without a clear description of the measurement and analysis method used. We recommend reporting all measurements details (including relative humidity and wavelength) and all assumptions of aerosol composition, size and hygroscopicity so that the values can be properly interpreted.

Acknowledgments

[89] The assumptions, findings, conclusions, judgments, and views presented herein are those of the authors and should not be interpreted as necessarily representing the National Park Service policies.