JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 110, D10S11, doi: /2004jd005124, 2005

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 110,, doi: /2004jd005124, 2005 Variability of aerosol and spectral lidar and backscatter and extinction ratios of key aerosol types derived from selected Aerosol Robotic Network locations Christopher Cattrall, 1 John Reagan, 2 Kurt Thome, 1 and Oleg Dubovik 3 Received 13 June 2004; revised 1 February 2005; accepted 3 March 2005; published 3 May [1] The lidar (extinction-to-backscatter) ratios at 0.55 and 1.02 mm and the spectral lidar, extinction, and backscatter ratios of climatically relevant aerosol species are computed on the basis of selected retrievals of aerosol properties from 26 Aerosol Robotic Network (AERONET) sites across the globe. The values, obtained indirectly from sky radiance and solar transmittance measurements, agree very well with values from direct observations. Low mean values of the lidar ratio, S a, at 0.55 mm for maritime (27 sr) aerosols and desert dust (42 sr) are clearly distinguishable from biomass burning (60 sr) and urban/industrial pollution (71 sr). The effects of nonsphericity of mineral dust are shown, demonstrating that particle shape must be taken into account in any spaceborne lidar inversion scheme. A new aerosol model representing pollution over Southeast Asia is introduced since lidar (58 sr), color lidar, and extinction ratios in this region are distinct from those over other urban/industrial centers, owing to a greater number of large particles relative to fine particles. This discrimination promises improved estimates of regional climate forcing by aerosols containing black carbon and is expected to be of utility to climate modeling and remote sensing communities. The observed variability of the lidar parameters, combined with current validated aerosol data products from Moderate Resolution Imaging Spectroradiometer (MODIS), will afford improved accuracy in the inversion of spaceborne lidar data over both land and ocean. Citation: Cattrall, C., J. Reagan, K. Thome, and O. Dubovik (2005), Variability of aerosol and spectral lidar and backscatter and extinction ratios of key aerosol types derived from selected Aerosol Robotic Network locations, J. Geophys. Res., 110,, doi: /2004jd Introduction [2] The geographical and temporal variability of aerosols dictate the use of long-term detailed global measurements from satellites and Earth s surface to accurately assess their impact on the global climate [Kaufman et al., 1997a]. Advances in radiometric measurements from space and ground have now afforded a view of the global aerosol system [Kaufman et al., 2002] and recognition of several key aerosol species that affect global climate (biomass burning, desert dust, sea salt, pollution). Ground-based remote sensing, in particular, has evolved to become a powerful method for characterizing the suspended aerosol [Dubovik and King, 2000], such that a clearer picture of the optical properties of each of these aerosol species is emerging [Dubovik et al., 2002a]. Passive spaceborne measurements presently employ such models in their processing schemes [Tanré etal., 1997; Vermote et al., 1997], 1 Remote Sensing Group, University of Arizona, Tucson, Arizona, USA. 2 Department of Electrical and Computer Engineering, University of Arizona, Tucson, Arizona, USA. 3 NASA Goddard Space Flight Center, Greenbelt, Maryland, USA. Copyright 2005 by the American Geophysical Union /05/2004JD and both ground- and space-based aerosol data products are used to evaluate the performance of general circulation models [Chin et al., 2002; Kinne et al., 2003]. [3] Active lidar systems contribute to the global climate effort through their ability to determine the vertical profiles of aerosol extinction and backscattering, which must be known to reduce uncertainty in the aerosol forcing of climate [Hansen et al., 1997; Haywood and Boucher, 2000]. This is particularly important in the case of lightabsorbing black carbon, where climatic effects of pollution can be felt on a regional scale [Menon et al., 2002; Sato et al., 2003]. Spaceborne lidar, such as the Cloud- Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO) sensor expected to launch in 2005, thus promises to improve our understanding of aerosol properties and distribution on both a global and regional basis. [4] Currently, space-based retrieval of vertical profiles of aerosol extinction and backscatter coefficients necessitates the prescription of their ratio, commonly called the lidar ratio, S a. The accuracy of this prescribed ratio determines the accuracy of the retrieved profiles. While a number of height-resolved lidar ratio observations have been made using special types of ground-based lidar or by combinations of instruments at the surface, these are insufficient in time or space to give a global climatology of S a values. A 1of13

2 measure of the variability in S a of key aerosol types around the world would prove useful to inverting global lidar observations. [5] In this paper, we employ ground-based retrievals of aerosol properties from the global aerosol network AERONET [Holben et al., 1998] for several key aerosol types (biomass burning, mineral dust, oceanic and anthropogenic pollution) to quantify the variability in lidar ratios at 0.55 and 1.02 mm and spectral backscatter and extinction ratios. Careful selection of the geographic location and time of year based upon known seasonal events helped isolate retrievals where a single aerosol species dominated the atmospheric column. The use of 26 worldwide sites provides a fairly robust picture of the natural variability within each aerosol species over a global scale. We then compare these values to those found in the literature and draw some preliminary conclusions on how the use of spectral lidar ratios, combined with backscatter and extinction ratios, can discriminate the aerosol types. The effect of particle nonsphericity upon the lidar parameters is explored through a combination of modeling and comparison with field measurements. Radiometric satellite measurements over the ocean have been shown to enhance the information retrieved by spaceborne lidar [Kaufman et al., 2003; Leon et al., 2003]. We also explore in this paper ways to retrieve aerosol vertical profiles over both land and ocean using validated aerosol data products of MODIS and AERONET. 2. Theory 2.1. Lidar Equation [6] The range- and energy-normalized signal of a returned lidar pulse, X(r), can be expressed by XðÞ¼Cb r ðþt r 2 ðþ; r where C is a calibration constant, which depends upon factors such as transmitted power; receiver cross section; efficiency of the detector and optical system; and corrections for near-range field-of-view problems, b(r) is the volume atmospheric backscattering coefficient, and T 2 (r) represents the total round-trip transmittance to range r, such that ð1þ R T 2 ðþ¼t r a 2 ðþt r R 2 ðþ¼e 2 r ½s a ðr 0 Þþs R ðr 0 ÞŠdr 0 ; ð2þ bðþ¼b r a ðþþb r R ðþ: r The subscripts a and R represent aerosol and molecular (Rayleigh) atmospheric components, respectively, and s represents the unit volume extinction coefficient. The backscatter solution to the above nonlinear equation, obtained using a modeled value for the lidar ratio (S a ), is given by [Fernald, 1984; Fernald et al., 1972], b a ðþ¼ r Xðr c Þ b a ðr c Þþb R ðr c XðÞe r Þ 2S a R 2 ð Sa SR Þ r b Rðr 0 Þdr 0 rc Z r R 2 XðÞe r ð Sa SR Þ r b R r 0 rc r c ð Þdr 0 dr 0 ð3þ b R ðþ; r ð4þ where r c is a Rayleigh reference calibration range; i.e., where b(r c ) b R (r c ), and S a ¼ s aðþ r b a ðþ r and S R ¼ 8p 3 : Hence the extinction solution is also obtained by s a (r) = S a b a (r). [7] The aerosol lidar ratio can vary widely depending on aerosol size distribution and refractive index [Ackermann, 1998; Barnaba and Gobbi, 2004] but, in the absence of auxiliary or lidar self-determined transmittance information, must be specified before the vertical profiles of the extinction and backscattering coefficients can be determined Lidar Parameters Used in the Study [8] To take advantage of the dual-wavelength capability of present and future satellite missions, we use three spectral lidar parameters in this study to further discriminate between aerosol types. These are the spectral ratios of S a, defined as the color ratio, S a (l 1 )/S a (l 2 ); the ratio of extinction coefficient, or the extinction ratio, s a (l 1 )/s a (l 2 ); and the ratio of backscattering coefficients, or the backscattering ratio, b a (l 1 )/b a (l 2 ). For the rest of the paper, the extinction and backscatter coefficients are referred to without using the subscript a, as the molecular contribution is assumed to be treated separately. In this paper, l 1 and l 2 are 550 and 1020 nm; for CALIPSO, they would be 532 and 1064 nm. AERONET measures sky radiance and aerosol optical depth at 440, 670, 870 and 1020 nm; our results at 550 nm are obtained from interpolating AERONET data at 440 and 670 nm. We chose 550 nm rather than 532 nm because it is a common reference wavelength for the remote sensing of aerosol, and there is negligible difference in extinction and backscatter between the two wavelengths. [9] The variability of the aerosol Ångström exponent [ Angström, 1964] for each aerosol species is also explored. This parameter is often used to parameterize the wavelength behavior of aerosol optical depth; i.e., t a ðlþ ¼ k l a ; such that a ¼ ln ð t l 1 =t l2 Þ lnðl 1 =l 2 Þ where l 1 and l 2 represent two reference wavelengths, t the aerosol optical depth, k a turbidity coefficient, and a the Angstrom exponent. It is also often used to provide basic information on the aerosol size distribution (i.e., aerosol type) since its value is related to the relative abundances of coarse and fine modes of the aerosol size distribution. The apparent log linear relationship between two wavelengths does not strictly exist in nature; however, making the selection of reference wavelengths is of some importance if using the parameter to characterize an air mass [Eck et al., 1999]. Specifically, use of longer wavelength pairs, such as 550 and 1020 nm, would furnish less information than wavelength pairs spanning more of the visible spectrum. Accordingly, in this paper we use the Angstrom exponent obtained from the AERONET wavelengths of 440 and 870 nm to explore how this value, expected from passive radiometric observation systems, may be used to refine a ð5þ ð6þ 2of13

3 Figure 1. Map of the AERONET sites used in the study. Oceanic sites are represented by an X. See color version of this figure at back of this issue. prediction of the lidar ratio for the inversion of spaceborne lidar signals from CALIPSO. [10] It is worth noting that the extinction coefficient is related to the column-integrated aerosol optical depth by t a ðlþ ¼ Z 1 0 s a ðl; zþdz; ð7þ such that the height-specific Angstrom exponent is obtained through [Ansmann et al., 2002] aðþ¼ z ln ½ s l 1 ðþ=s z l2 ðþ z Š lnðl 1 =l 2 Þ ln Sa ½ l 1 ðþ=sa z l2 ðþ z Šþln b l1 ðþ=b z l2 ðþ z : ð8þ lnðl 1 =l 2 Þ The height-resolved exponent can be averaged over an aerosol layer to compare with the column-integrated aerosol optical depths and Angstrom exponent measured from ground-based Sun photometers. The expression for the Angstrom exponent containing the extinction coefficient can also be checked for consistency with the expression for the Angstrom exponent containing the backscatter coefficient and lidar ratio. 3. Method [11] AERONET is a internationally federated network of globally distributed sun and sky radiometers that has been in operation for over a decade for the purpose measuring global aerosol optical properties [Holben et al., 1998]. These annually calibrated radiometers measure the direct solar radiance at 8 wavelengths (340, 380, 440, 500, 675, 870, 940 and 1020 nm) and sky radiance at four of these wavelengths (440, 670, 865 and 1020 nm), providing sufficient information to determine the aerosol size distribution and refractive index [Dubovik and King, 2000; Dubovik et al., 2000]. These properties can then be used to compute the aerosol size distribution and absorption and scattering properties at the four sky radiance channels within the accuracy limits defined by Dubovik et al. [2000]. [12] We selected 26 sites with well known meteorological and climatic conditions where, for at least a period of the year, a single aerosol species dominates the atmospheric column. Many of the sites have been previously used to develop aerosol climatologies of aerosol optical depth and optical properties [Dubovik et al., 2002a; Holben et al., 2001; Kinne et al., 2003; Smirnov et al., 2002a, 2002b] and in several aerosol characterization campaigns [Kaufman et al., 1998; Ramanathan et al., 2001; Swap et al., 2003]. Figure 1 displays the geographic location of each selected site. [13] For each AERONET location, cloud-screened, quality assured retrievals of aerosol single-scattering albedo, scattering phase function, refractive index, and size distributions were obtained, when aerosol optical depths were at least 0.4 (except for maritime locations; see Table 1) and solar zenith angle was at least 65, in accordance with the limits suggested for reliable retrievals of aerosol properties [Dubovik et al., 2000]. From these selected retrievals, we used only the months when the atmospheric column was likely dominated by the desired aerosol type. For the most part, these corresponded to the months defined in the work of Dubovik et al. [2002a]; however, we further refined the selection for urban and maritime sites where contamination was possible from long-range transport of Asian or African dust or biomass burning aerosols originating in South Africa. For dust sites, we used only the most arid summer months when dust production is greatest. Monthly statistics of Angstrom exponent for each site corroborated these choices. Table 1 presents a summary of the sites and months used to analyze each aerosol species. [14] Four of the five aerosol species employed were those presented in the work of Dubovik et al [2002a]. We had originally intended to use a single aerosol type to characterize urban pollution as these authors did. During our analysis, however, it quickly became apparent that pollution over SE Asia displayed lidar parameters distinctly different from pollution over other regions, due to a strong increase in the large particle mode relative to the fine particle mode observed in the retrieved size distributions. We concluded that nations in this region, who typically possess less strict 3of13

4 Table 1. Summary of AERONET Sites Used to Construct Aerosol Lidar Parameters Site Name Country Aerosol Type Months Year N meas S a a S a /S a a b/b a s/s a a a Aire Adour France urban/industrial June Sept Creteil France urban/industrial June Sept Lille France urban/industrial June Sept GISS USA urban/industrial June Sept GSFC USA urban/industrial June Sept Mexico City Mexico urban/industrial Y (Aug. Oct.) b Alta Floresta Brazil biomass burning Aug. Oct Brasilia Brazil biomass burning Aug. Oct Cuiaba Brazil biomass burning Aug. Oct Los Fieros Bolivia biomass burning Aug. Oct Mongu Zambia biomass burning Aug. Nov Senanga Zambia biomass burning Aug. Nov Zambezi Zambia biomass burning Aug. Nov Anmyon S. Korea SE Asia Y Chen-Kung Taiwan SE Asia Y(Apr.) b Beijing China SE Asia Y(March May) b Che Ju S. Korea SE Asia Y (March May) b Ascension UK maritime Dec. May Bermuda Bermuda maritime Oct. March Lanai USA maritime Y (March May) b Nauru Nauru maritime Y Tahiti Tahiti maritime Y Bahrain Bahrain dust March July /41 1.8/ / Banizoumbou Nigeria dust March July /35 1.7/ / Capo Verde Spain dust May Oct /38 1.6/ / Solar Village Saudi Arabia dust March July /39 1.7/ / a Mean values for a particular site. Their mean does not necessarily correspond to those in Table 1 using Gaussian fit. b Months in parentheses are months neglected from entire year (Y). Minimum optical depth for retrieval was 0.4 except for oceanic aerosol, where 0.05 < t aer < emission and environmental regulations or who frequently use coal for domestic heating at lower combustion temperatures, would indeed emit large particles not produced in the higher temperature combustion of industrial processes or by countries with stricter emission and environmental regulations [Cooke et al., 1999; Environmental Information Agency (EIA), 2003]. Accordingly, we created a fifth aerosol class entitled SE Asia to describe aerosols observed over this region. [15] The retrieved properties of an aerosol species are expected to display significant variability, due to differences in aerosol source region, production, transport and deposition, and due to uncertainties in the inversion of groundbased remote sensing measurements. Accordingly, we gave equal weight to the retrieved properties from each AERONET site rather giving equal weight to each retrieval. This avoids overwhelming contribution from a site with a much greater number of successful retrievals (e.g., Mongu for biomass burning aerosol) Special Considerations for Oceanic Aerosols [16] The optical depth of the oceanic aerosol rarely exceeds 0.15 in pure maritime conditions, a value somewhat below the minimum optical depth for which most aerosol properties may be reliably retrieved. The size distributions, however, are still reliably retrieved at even low optical depths [Dubovik et al., 2000], so we used the retrieved size distribution with literature values for the refractive index of oceanic aerosols, n = i [d Almeida et al., 1991] to compute the lidar parameters for the oceanic aerosol Special Considerations for Dust Aerosols [17] Mineral dust is flattened and irregular in shape and certainly not spherical, the shape assumed by the normal AERONET retrieval algorithm. Nonspherical particles such as mineral dust produce angular scattering patterns distinctly different from surface-area-equivalent spheres [Mishchenko et al., 1997] and create difficulty in interpreting the remotely sensed scattering of light. Indeed, the effect of nonsphericity has been observed in AERONET retrievals of aerosol properties [Dubovik et al., 2000] in the presence of mineral dust. Recently, the use of polydisperse, randomly oriented spheroids has shown significant improvement in retrieving the size distribution and real refractive index [Dubovik et al., 2002b]. [18] Accordingly, we have dealt with the mineral dust aerosol in the following manner. First, as with the oceanic aerosol, we computed the lidar parameters by combining the size distributions retrieved using the AERONET spheroid model with a real refractive index of 1.53 and a small imaginary index (0.0015i and i for 550 and 1020 nm, respectively) to represent the small absorption by mineral dust observed using various remote sensing methods [Cattrall et al., 2003; Dubovik et al., 2002a; Haywood et al., 2003b; Kaufman et al., 2001; Tanré etal., 2001]. These computations give the lidar parameters of spherical particles. To account for particle nonsphericity in each of these retrievals, a relationship between the lidar parameters of spherical and nonspherical particles had to be determined. To accomplish this, we used for each dust site the mean size distribution and above refractive index to compute the mean lidar parameters for spherical particles. Then, using these same mean size distributions and assumed refractive indices, we employed the spheroid model of Dubovik et al. [2002b] and a shape distribution obtained by inverting recent measurements of the scattering matrix of suspended mineral dust [Volten et al., 2001; O. Dubovik et al., manuscript in preparation, 2005] to compute the lidar parameters for these 4of13

5 Figure 2. Frequency distribution of lidar ratio at 550 nm, S a (550) by aerosol class. spheroids. The ratios of the derived lidar parameters thus provide the desired relationship, which was applied to the individual retrievals described in the first part of this paragraph. This produced the lidar parameters of nonspherical particles for each retrieval at each dust site used to construct the climatology of lidar parameters for nonspherical mineral dust. 4. Results 4.1. Aerosol Lidar Parameters (S a, Ratios of S a, B, S) [19] All the lidar parameters obtained for each aerosol type were normally distributed about a mean value (Figures 2 6). Clearly, the greater the number of total measurements (N meas ), the better the frequency distribution displays a normal distribution. Accordingly, we used a Gaussian fit to these frequency distributions with which to describe the mean and standard deviations of each lidar parameter. The values derived for each aerosol type are given in Table 2. Figure 7 displays a summary of all these results, along with bounds under which 50% of the retrievals fall. A value of 50% seems a reasonable basis on which to draw conclusions of the behavior of the lidar parameters for each aerosol species, given the natural variability of aerosols and errors expected from the inversion itself. Overlap of these bounds would indicate increasing difficulty in using a certain lidar parameter to distinguish between aerosol types, whereas clear separation between aerosol types indicates a utility for that particular lidar parameter to aid in distinguishing between aerosol types. [20] Figure 8 displays a summary of the variability observed in the lidar parameters. Table 2 presents the same data in tabular format. Some items in Figures 7 and 8 may be immediately noted. The lidar ratios of biomass burning and both types of pollution are separated from the oceanic and mineral dust aerosols with little or no overlap. The mean values of S a for urban/industrial aerosols are higher than those for aerosols from biomass burning or SE Asia aerosols, though some overlap is present. Aerosols from SE Asia, while possessing a mean lidar ratio essentially the same as that of a biomass burning aerosol, display a very different color ratio. For the backscattering ratio, while most aerosol species are tightly grouped and thus indistinguishable from one another, the dust aerosol regardless of the assumption of particle shape is strikingly different from the other aerosol species. Indeed, the backscattering ratio has been used to distinguish the presence of dust [Sugimoto et al., 2002]. For the extinction ratio or, equivalently, the Angstrom exponent, each aerosol species is rather well defined, except perhaps between the urban/industrial and biomass burning aerosols. [21] Checks of the mean derived values of the spectral lidar, extinction and backscattering ratios using equation (8) agree well and are close to the mean derived Angstrom exponent, showing the results are self-consistent. [22] In summary, most aerosol species are observed to possess a unique lidar ratio that can be applied, with knowledge of geographic location and season, to the processing of lidar returns. If, as is the case with CALIPSO, two wavelengths are available, use of the color, extinction and backscattering ratios should provide enough information to further constrain the lidar retrievals Uncertainty Estimates in Retrieved Parameters [23] AERONET cannot directly measure aerosol extinction and backscattering coefficients. Rather, the lidar parameters are computed from physical and chemical aerosol 5of13

6 Figure 3. Frequency distribution of color ratio, S a (550)/S a (1020), by aerosol class. properties obtained from inversions of spectral measurements of sky radiance and atmospheric transmittance. The possible errors in instrument and inversion cannot be computed analytically to furnish an uncertainty estimate in the retrieved quantities. Analyses of the sensitivity of the retrieved physical and chemical characteristics to expected instrumental and random errors [Dubovik et al., 2000] are of limited use in assessing errors in the computed phase function at the nonobservable 180 backscatter direction. It is worth noting, however, that recent efforts indicate this error is not Figure 4. Frequency distribution of backscatter ratio, b(550)/b(1020) by aerosol class. 6of13

7 Figure 5. Frequency distribution of extinction ratio, s(550)/s(1020) by aerosol class. high (unpublished data). Airborne in situ measurements of the size distribution, refractive index and single-scattering albedo of biomass burning aerosols agreed very well with those retrieved from concomitant AERONET measurements [Haywood et al., 2003a]. Directly relevant to our study, comparisons of lidar ratios of spherical aerosol types (i.e., not desert dust) obtained from Raman lidar, as well other aerosol physical and chemical characteristics, have agreed very well with those derived from AERONET [Ferrare et al., 2001; Müller et al., 2004]. Further, despite limitations Figure 6. Frequency distribution of Ångström exponent, a, by aerosol class. 7of13

8 Table 2. Summary of Lidar Parameters Retrieved From Selected AERONET Sites Aerosol Type Lidar Ratio (SD) a S a Ratio b Ratio s Ratio a Biomass burning 60 (8) 2.1 (0.3) 1.8 (0.3) 3.8 (0.4) 1.8 (0.2) SE Asia 58 (10) 1.5 (0.3) 1.6 (0.2) 2.4 (0.3) 1.3 (0.2) Urban/industrial 71 (10) 1.9 (0.3) 1.6 (0.2) 3.3 (0.5) 1.7 (0.2) Oceanic 28 (5) 1.0 (0.2) 1.4 (0.1) 1.5 (0.4) 0.7 (0.4) Dust (spheres) 15 (2) 1.6 (0.2) 0.7 (0.1) 1.2 (0.1) 0.1 (0.1) Dust (spheroids) 42 (4) 1.2 (0.1) 0.9 (0.1) 1.2 (0.1) 0.1 (0.1) a SD, standard deviation of Gaussian fit. in measurement methods for nonspherical particles, AERONET retrievals of the size distribution of desert dust aerosols were found to be consistent with in situ measurements [Reid et al., 2003]. More discussion of the uncertainties inherent to accounting for particle nonsphericity may be found in section 5.2. [24] In the absence of systematic errors, experimental uncertainty can be internally estimated from the spread of the measurements when it cannot be externally estimated for a single measurement [Bevington and Robinson, 1992]. Given that aerosol optical depths of 0.2 represent a majority of observations at continental sites, we explored the potential of a systematic bias resulting from our choice of using a minimum aerosol optical depth of 0.4 by performing our analysis using a minimum optical depth of 0.2 for the urban/ industrial, biomass burning and SE Asia aerosol sites. The greatest change in the Ångström exponent and mean colour lidar, backscatter and extinction ratios was 0.1; in 9 of the 12 cases (three aerosol types; four lidar parameters) the mean retrieved lidar parameters were unchanged. Other studies also show very weak relationship between Angstrom exponent and aerosol optical depth [Holben et al., 2001; Mattis et al., 2004]. The mean lidar ratio at 550 nm of the biomass burning, SE Asia and urban/industrial classes was only reduced by 1, 2, and 4 sr 1, respectively. These results indicate that no systematic bias is introduced by using the relatively high minimum aerosol optical depth of 0.4 to isolate a single aerosol species. We believe this is further evidence of the utility of judiciously choosing the geographic region and season with which to make this analysis. [25] Thus, since all the lidar parameters obtained for each aerosol type fell in a Gaussian pattern about a well-defined mean, and since a greater number of measurements always led to a smoother distribution, the standard deviations of the mean lidar parameters found in this study may be interpreted to represent the uncertainty in the mean derived values. Furthermore, these standard deviations are also a measure of the natural variability that may be expected for each aerosol type Comparison With Other Studies [26] Quality measurements of the extinction-to-backscatter ratio by surface measurement or ground-based lidar have accumulated for a variety of aerosol types. Micropulse lidar [Spinhirne, 1993] has been successfully used in several field campaigns to retrieve the lidar ratio for maritime and dust aerosols [Campbell et al., 2003; Powell et al., 2000; Voss et al., 2001; Welton et al., 2000, 2002]. Multiwavelength Raman lidar, capable of separately retrieving the profiles of backscatter and extinction coefficients, has successfully been used to measure air masses influenced by all the aerosol types considered in this paper (see Table 3 for references). Spaceborne lidar has been used to determine values of S a for aerosols of biomass burning origin using Figure 7. Fitted frequency distribution of aerosol classes for each lidar parameter. See color version of this figure at back of this issue. 8of13

9 Figure 8. Contour plot of lidar ratio at 550 nm versus lidar parameters and Angstrom exponent. The darker and lighter shaded regions represent, respectively, the areas within which 25% and 50% of the retrievals fall. See color version of this figure at back of this issue. constraints of physically plausible lidar retrievals and assuming S a was constant over the upper troposphere [Kent et al., 1998]. The combination of backscattering nephelometer [Anderson et al., 1996] and measurements of integrated scattering and absorption has afforded directly measured S a values at two polluted continental sites in Washington and Illinois and aerosols of clean continental and marine origin [Anderson et al., 1999, 2000; Doherty et al., 1999; Masonis et al., 2003]. A combination of extinction nephelometer and backscattersonde has also furnished surface measurements of the extinction-to-backscatter ratio in New Mexico [Rosen et al., 1997]. [27] The literature values presented in Table 3 are from direct measurements that do not rely on assumptions of particle composition, size or shape. These papers contained back-trajectory analyses or source region identification and usually identified the air mass type with respect to altitude. This allowed us to extract the lidar ratio associated with each aerosol type if it was not clearly explicated by the authors. In the case of Raman lidar results, we often obtained the lidar values for a specific aerosol type from the separate extinction and backscatter profiles over the appropriate altitude range identified by the authors. In this way, it was possible to obtain values of S a for all of the aerosol types. A number of papers were not included in Table 3 due to insufficient information from which to elicit a value for a specific aerosol type. Thus Table 3 is not exhaustive, but rather a summary of the literature values we have gleaned to form meaningful comparisons with our own results. [28] The mean values for each of the aerosol species obtained from AERONET compare well with those obtained from the literature (Table 4), in all cases agreeing well within the combined uncertainties of the two bodies of data. The mean lidar ratio of 29 sr we derived for marine aerosols is nearly identical to the mean of those directly observed in the field. Aerosols of polluted origin are generally observed to be between 60 and 70 sr, whereas we have derived a mean value of 71 sr. Our biomass burning mean lidar ratio of 60 sr compares well with literature values which generally fall between 60 and 70. The lower lidar ratios of the SE Asia aerosol is also mirrored in the lower values observed in aerosol originating from these regions. Last, the lidar values for mineral dust observed in field investigations (43 sr) agree very well with those we obtained from using spheroids to correct for the nonspherical shape of mineral dust (45 sr). [29] Some mention should be made of the limitations of both data sets in order to clarify these comparisons. First, some bias may be introduced in the AERONET retrievals because cloudy days (thus, comparatively high aerosol water content) are underrepresented. Second, the AERONET measurements are of the entire atmospheric column, whereas lidar cannot observe particles below 500 or 1000 m, a region 9of13

10 Table 3. Literature Values of Sa at Wavelengths Between 490 and 550 nm Value System Source Region Reference Oceanic Aerosol 25 ± 4 surface measurement Hawaii Masonis et al. [2003] 33 ± 6 micropulse lidar Indian Ocean Welton et al. [2002] 32 ± 6 micropulse lidar N. Atlantic Voss et al. [2001] 39 ± 5 lidar Maritime Young et al. [1993] 25 Raman lidar Indian Ocean Müller et al. [2001] 28 Raman lidar Indian Ocean Franke et al. [2001] 24 ± 5 lidar N. Atlantic Ansmann et al. [2001] 30 ± 3 lidar Australian coast Gras et al. [1991] 25 Raman lidar N. Atlantic Müller et al. [2000b] 25 ± 4 surface measurement Hawaii Masonis et al. [2003] Urban/Industrial Aerosol Raman lidar Portugal Ansmann et al. [2001] 60 ± 8 surface measurement Wash. State, USA Doherty et al. [1999] 64 ± 4 surface measurement Illinois, USA Anderson et al. [2000] 60 Raman lidar India Müller et al. [2001] 70 ± 10 Raman lidar India Ansmann et al. [2000] Raman lidar Portugal Ansmann et al. [2001] Biomass Burning Aerosol 60 ± 10 Raman lidar NW Canada Müller et al. [2000a] 69 ± 9 Raman lidar Central US Peppler et al. [2000] a 59 ± 11 Raman lidar Siberia Mattis et al. [2003] 50 ± 5 Raman lidar Black Sea area Balis et al. [2003] 67 ± 7 Raman lidar India Franke et al. [2001] 63 ± 10 micropulse lidar S. Africa Campbell et al. [2003] 70 ± 10 LITE S. Africa Kent et al. [1998] Dust Aerosol 41 ± 8 lidar Africa Voss et al. [2001] 37 ± 9 micropulse lidar Africa Welton et al. [2000] 42 ± 8 surface measurement New Mexico Rosen et al. [1997] 45 ± 9 lidar E. Africa and ArabiaMüller et al. [2001] 46 ± 5 Raman lidar Gobi Desert Sakai et al. [2002] 46 ± 10 Raman lidar Asia Murayama et al. [2003] 53 ± 5 Raman lidar Sahara Desert Mattis et al. [2002]; Müller et al. [2003] 53 ± 5 in situ measurement Asia Anderson et al. [2003] 51 ± 2 Raman lidar Africa De Tomasi et al. [2003] 47 ± 7 Raman lidar Asia Liu et al. [2002] 35 ± 5 micropulse lidar Africa Powell et al. [2000] 41 ± 8 lidar Africa Voss et al. [2001] SE Asia Aerosol 51 ± 5 Raman lidar SE Asia Franke et al. [2001] 50 ± 5 in situ measurement SE Asia Anderson et al. [2003] a Lidar measurement was actually 90 sr at 355 nm. The authors state the corresponding value at 532 nm is 69 sr. where aerosols different from those measured aloft may be present. Thus measurements by AERONET and lidar do not sample an identical air mass and some bias may result, the sign of which would depend, in each instance, on the size and chemical differences between the boundary layer aerosols and those dominating the aerosol properties of the atmospheric column. Urban/industrial pollution also displays significant variability in absorption characteristics from site to site [Dubovik et al., 2002a], such that the mean values of the lidar parameters from our selected sites may not comprehensively represent those at other locations. It is gratifying that even with these caveats the mean values of these two bodies of data agree within their combined standard deviations. [30] Comparisons of the backscatter ratio are also possible, though we found far fewer published studies in which this value could be gleaned. Values for the backscatter ratio have been observed to fall between 1.7 and 2.0 for pollution [Ansmann et al., 2002; Müller et al., 2001, 2000a], between 0.8 and 1.1 for desert dust [Mattis et al., 2002; Sugimoto et al., 2002] and between 2.1 and 2.2 for biomass burning [Wandinger et al., 2002]. These values agree qualitatively with those found in this study (see Table 2), though our values are somewhat lower, falling about a mean value of for urban/industrial pollution and about 1.8 for biomass burning (Figure 4). These discrepancies are still within the combined uncertainties of the two methods. 5. Discussion 5.1. Discrimination of Pollution by Source Region [31] The apportionment of pollution aerosol properties by source region or season has been incorporated into some processing schemes of satellite aerosol observations [Bellouin et al., 2003; Kaufman et al., 1997b] and is often considered by the climate modeling community [Koch, 2001; Takemura et al., 2002; Yu et al., 2004]. In attempting to account for regional differences in aerosol absorption, for example, Koch [2001] examined organic and black carbon content relative to sulfate production; in effect, examining emission of black carbon with respect to dominant combustion processes. Two recent papers examining climate forcing by black carbon [Menon et al., 2002; Sato et al., 2003] both discuss the importance of knowing the black carbon content in aerosol pollution, which varies greatly depending on the combustion source and method [Streets et al., 2001]. Given the differences in lidar parameters between pollution over SE Asia and those of other regions, we believe inclusion of this regional aerosol model in the processing of spaceborne lidar returns will result in improved retrievals of aerosol extinction and scattering profiles, hence, more accurate assessments of the aerosol forcing of climate Accounting for Particle Nonsphericity in Remote Sensing Problems [32] The real shape of a nonspherical aerosol particle such as mineral dust is very complex and a defining answer has yet to be found. Most studies use microphotographs to guide descriptions of particle shape such as edge sharpness, circularity, aspect ratio, etc. [Hill et al., 1984; Kalashnikova and Sokolik, 2004; Okada et al., 2001] used to compute their optical properties, but the computational intensity required to account for most irregular shapes is currently too great to be of utility in remote sensing problems. At present, only the light scattering properties of spheroids can be modeled over the required range of aerosol size and refractive index with reasonable computational efficiency. Thus there is little alternative in aerosol remote sensing to using spheroids to account for aerosol nonsphericity. With Table 4. Comparison With Literature Values by Type Aerosol Type This Study Literature Oceanic 28 ± 5 29 ± 5 Urban/industrial 71 ± ± 4 Biomass burning 60 ± 8 63 ± 7 Dust 42 ± 4 45 ± 6 SE Asia 58 ± ± 5 10 of 13

11 this limitation, the pragmatic approach has been to vary the shape distribution of these spheroids to best match the observed and computed scattering intensities. The success of this approach can be measured by the fact that predictions of nearly flat phase functions at side scattering angles by size distributions of volume-equivalent spheroids are borne out in laboratory measurements of natural particles [Jaggard et al., 1981; Volten et al., 2001], and in the significant improvements between measured and modeled sky radiance when spheroids are used instead of spheres in the case of mineral dust [Dubovik et al., 2002b]. With these considerations, the agreement between the lidar ratios for mineral dust we predicted using spheroids (i.e., scattering at 180 degrees) and those measured in the field is very encouraging. Thus, while the best method to deal with particle nonsphericity in remote sensing problems remains open to debate, we observe that the use of spheroids with appropriate shape distributions seems effective in predicting the lidar ratio of mineral dust. Furthermore, given the substantial difference in lidar ratios computed assuming spheres and volume-equivalent spheres (see Table 2), it appears that even the most basic lidar inversion algorithm must attempt to account for particle nonsphericity when mineral dust is present. [33] It is possible that lidar itself can help constrain this problem. Presently, we are working on processing data from the special case of a lofted dust layer collected by the Lidar In Space Technology Experiment, LITE [Kent et al., 1998], where the only necessary assumption to reduce data at both wavelengths is the color ratio. Preliminary comparisons of the retrieved lidar, backscattering and extinction ratios all show very good agreement with those computed using the present shape distribution of spheroids, lending support to our choice of shape distribution to describe nonspherical particles. 6. Conclusion [34] Several thousand inversions of AERONET data satisfying stringent quality requirements have been used to determine the variability in lidar, backscattering, and extinction ratios of five distinct aerosol species. The careful selection of geographic location and season in the analysis led to a robust climatology of lidar ratios that agrees extremely well with direct field measurements for all aerosol species considered. This indicates that global distributions of assumed regions of aerosol species, by season, may in itself lead to reasonably accurate predictions of S a for the purpose of inverting lidar observations on a global basis. The wavelengths used in the analysis were near those possessed by the upcoming spaceborne lidar CALIPSO mission, which allows their direct application to active remote sensing from space. Spectral lidar parameters such as the backscattering and extinction ratios promise to further improve the accuracy of the retrievals of aerosol properties from multiwavelength lidar observations such as those anticipated from CALIPSO. [35] Use of simultaneous passive radiometer measurements (e.g., Angstrom exponent) can further refine the choice of lidar ratio used to invert CALIPSO data. The similarity in aerosol types in this study with those of the MODIS aerosols over land suggests that the MODIS data product of aerosol type over land can be used with reasonable confidence to select the lidar ratio over land. With the more precise information on spectral optical depth obtained from radiometric measurements over the ocean, accurate selection of S a should be possible over much of the Earth s surface. [36] The use of spheroids to predict the optical properties of nonspherical particles increases the lidar ratio of mineral dust by a factor of over 2.5 from the lidar ratio computed assuming spheres, and produces results that agree surprisingly well with field and surface measurements. This finding clearly indicates that the nonsphericity of mineral dust must be taken into account in spaceborne lidar retrievals. [37] Last, it is evident that regional consideration of the sources of black carbon aerosols is desirable to more accurately retrieve the vertical profiles of their properties using space-based lidar. This could perhaps be achieved through use of emission inventories or level of industrial development. The results of this study indicate that accounting for the distinct differences in properties of polluting aerosols from different geographic regions would improve the monitoring of pollution sources and estimates of the radiative forcing by black carbon aerosols. [38] Acknowledgments. We would like to thank Brent Holben and the AERONET team members, without whose continual efforts this paper would literally not have been possible. We are grateful to Tatyana Lapyonok, who helped write the software for the spheroid simulations. The authors gratefully acknowledge financial support from NASA through contract NAS awarded to K. Thome. We would also like to thank the AERONET PIs for contributing data to this study. (1) Urban/industrial sites: Creteil, Bernadette Chatenet; Lille, Luc Blarel; GSFC, Wayne Newcomb; and Mexico City, Amando Leyva. (2) Biomass burning sites: Alta Floresta, Edilson Bernardino de Andrade; Los Fieros, Ademar Soto; and Mongu and Senanga, Mukufute Mukulabai. (3) SE Asia sites: Beijing, Zhang Wenxing; Che Ju, Jae-Gwang Won; and Chen-Kung, Po-Hsiung Lin. (4) Oceanic sites: Ascension, George Thomas; Lanai, Dennis Clark; Nauru, Larry Jones; and Tahiti, Keitaupu Maamaatuaihutapu. (5) Desert dust sites: Bahrain, Andreas Goroch; Banizoumbou, Ousmane Manga; Capo Verde, Bernadette Chatenet; and Solar Village, Naif Al-Abbadi. Comments from two anonymous reviewers helped greatly improve the manuscript. References Ackermann, J. (1998), The extinction-to-backscatter ratio of tropospheric aerosol: A numerical study, J. Atmos. Oceanic Technol., 15, Anderson, T. L., et al. (1996), Performance characteristics of a highsensitivity, three-wavelength, total scatter/backscatter nephelometer, J. Atmos. Oceanic Technol., 13, Anderson, T. L., D. S. Covert, J. D. Wheeler, J. M. Harris, K. D. Perry, B. E. Trost, D. J. Jaffe, and J. A. Ogren (1999), Aerosol backscatter fraction and single scattering albedo: Measured values and uncertainties at a coastal station in the Pacific Northwest, J. Geophys. Res., 104(D21), 26,793 26,807. Anderson, T. L., S. J. Masonis, D. S. Covert, R. J. Charlson, and M. J. Rood (2000), In situ measurements of the aerosol extinction-to-backscatter ratio at a polluted continental site, J. Geophys. Res., 105(D22), 26,907 26,915. Anderson, T. L., S. J. Masonis, D. S. Covert, N. C. Ahlquist, S. G. Howell, A. D. Clarke, and C. S. McNaughton (2003), Variability of aerosol optical properties derived from in situ aircraft measurements during ACE-Asia, J. Geophys. Res., 108(D23), 8647, doi: /2002jd Ångström, A. (1964), The parameters of atmospheric turbidity, Tellus, 16(1), Ansmann, A., D. Althausen, U. Wandinger, K. Franke, D. Muller, F. Wagner, and J. Heintzenberg (2000), Vertical profiling of the Indian aerosol plume with six-wavelength lidar during INDOEX: A first case study, Geophys. Res. Lett., 27(7), Ansmann, A., F. Wagner, D. Althausen, D. Müller, A. Herber, and U. Wandinger (2001), European pollution outbreaks during ACE 2: Lofted aerosol plumes observed with Raman lidar at the Portuguese coast, J. Geophys. Res., 106(D18), 20,725 20, of 13

12 Ansmann, A., F. Wagner, D. Muller, D. Althausen, A. Herber, W. von Hoyningen-Huene, and U. Wandinger (2002), European pollution outbreaks during ACE 2: Optical particle properties inferred form multiwavelength lidar and star-sun photometry, J. Geophys. Res., 107(D15), 4259, doi: /2001jd Balis, D. S., V. Amiridis, C. Zerefos, E. Gerasopoulos, M. Andreae, P. Zanis, A. Kazantzidis, S. Kazadzis, and A. Papayannis (2003), Raman lidar and sunphotometric measurements of aerosol optical properties over Thessaloniki, Greece during a biomass burning episode, Atmos. Environ., 37, Barnaba, F., and G. P. Gobbi (2004), Modeling the aerosol extinction versus backscatter relationship for lidar applications: Maritime and continental conditions, J. Atmos. Oceanic Technol., 21, Bellouin, N., O. Boucher, D. Tanré, and O. Dubovik (2003), Aerosol absorption over the clear-sky oceans deduced from POLDER-1 and AERONET observations, Geophys. Res. Lett., 30(14), 1748, doi: /2003gl Bevington, P. R., and D. K. Robinson (1992), Data Reduction and Error Analysis for the Physical Sciences, 328 pp., McGraw-Hill, New York. Campbell, J. R., E. J. Welton, J. D. Spinhirne, Q. Ji, S.-C. Tsay, S. J. Piketh, M. Barenbrug, and B. N. Holben (2003), Micropulse lidar observations of tropospheric aerosols over northeastern South Africa during the ARREX and SAFARI 2000 dry season experiments, J. Geophys. Res., 108(D13), 8497, doi: /2002jd Cattrall, C., K. L. Carder, and H. R. Gordon (2003), Retrieval of columnar aerosol single-scattering albedo and phase function from sky radiance over the ocean: Measurements of African dust, J. Geophys. Res., 108(D9), 4287, doi: /2002jd Chin, M., P. Ginoux, S. Kinne, O. Torres, B. N. Holben, B. N. Duncan, R. V. Martin, J. A. Logan, A. Higurashi, and T. Nakajima (2002), Tropospheric aerosol optical thickness from the GOCART model and comparisons with satellite and sun photometer measurements, J. Atmos. Sci., 59, Cooke, W. F., C. Liousse, H. Cachier, and J. Feichter (1999), Construction of a 1 1 fossil fuel emission data set for carbonaceous aerosol and implementation and radiative impact in the ECHAM4 model, J. Geophys. Res., 104(D18), 22,137 22,162. d Almeida, G. A., P. Koepke, and E. P. Shettle (1991), Atmospheric Aerosols: Global Climatology and Radiative Characteristics, 561 pp., A. Deepak, Hampton, Va. De Tomasi, F., A. Blanco, and M. R. Perrone (2003), Raman lidar monitoring of extinction and backscattering of African dust layers and dust characterization, Appl. Opt., 42(9), Doherty, S. J., T. L. Anderson, and R. J. Charlson (1999), Measurement of the lidar ratio for atmospheric aerosols with a 180 backscatter nephelometer, Appl. Opt., 38, Dubovik, O., and M. D. King (2000), A flexible inversion algorithm for retrieval of aerosol optical properties from Sun and sky radiance measurements, J. Geophys. Res., 105(D16), 20,673 20,696. Dubovik, O., A. Smirnov, B. N. Holben, M. D. King, Y. J. Kaufman, T. F. Eck, and I. Slutsker (2000), Accuracy assessments of aerosol optical properties retrieved from Aerosol Robotic Network (AERONET) Sun and sky radiance measurements, J. Geophys. Res., 105(D8), Dubovik, O., B. N. Holben, T. F. Eck, A. Smirnov, Y. J. Kaufman, M. D. King, D. Tanré, and I. Slutsker (2002a), Variability of absorption and optical properties of key aerosol types observed in worldwide locations, J. Atmos. Sci., 59, Dubovik, O., B. N. Holben, T. Lapyonok, A. Sinyuk, M. I. Mishchenko, P. Yang, and I. Slutsker (2002b), Non-spherical aerosol retrieval method employing light scattering by spheroids, Geophys. Res. Lett., 29(10), 1415, doi: /2001gl Eck, T. F., B. N. Holben, J. S. Reid, O. Dubovik, A. Smirnov, N. T. O Neill, I. Slutsker, and S. Kinme (1999), Wavelength dependence of the optical depth of biomass burning, urban, and desert dust aerosols, J. Geophys. Res., 104(D24), 31,333 31,350. Environmental Information Agency (EIA) (2003), Country analysis briefs, Washington, D. C. Fernald, F. G. (1984), Analysis of atmospheric lidar observations, Appl. Opt., 23, Fernald, F. G., B. M. Herman, and J. A. Reagan (1972), Determination of aerosol height distribution by lidar, J. Appl. Meteorol., 11, Ferrare, R. A., D. Turner, L. Brasseur, W. Feltz, O. Dubovik, and T. Tooman (2001), Raman lidar measurements of the aerosol extinction-tobackscatter ratio over the Southern Great Plains, J. Geophys. Res., 106(D17), 20,333 20,347. Franke, K., A. Ansmann, D. Müller, D. Althausen, F. Wagner, and R. Scheele (2001), One-year observations of particle lidar ratio over the tropical Indian Ocean with Raman lidar, Geophys. Res. Lett., 28(24), Gras, J. L., C. M. R. Platt, W. D. Jones, R. M. Hufaker, S. A. Young, S. M. Banks, and J. J. Booth (1991), Southern Hemisphere tropospheric aerosol backscatter measurements Implications for a laser wind system, J. Geophys. Res., 96(D3), Hansen, J., M. Sato, and R. Ruedy (1997), Radiative forcing and climate response, J. Geophys. Res., 102(D6), Haywood, J., and O. Boucher (2000), Estimates of the direct and indirect radiative forcing due to tropospheric aerosols: A review, Rev. Geophys., 38(4), Haywood, J., P. Francis, O. Dubovik, M. Glew, and B. Holben (2003a), Comparison of aerosol size distributions, radiative properties, and optical depths determined by aircraft observations and Sun photometers during SAFARI 2000, J. Geophys. Res., 108(D13), 8471, doi: / 2002JD Haywood, J., P. Francis, S. Osborne, M. Glew, N. Loeb, E. Highwood, D. Tanré, G. Myhre, P. Formenti, and E. Hirst (2003b), Radiative properties and direct radiative effect of Saharan dust measured by the C-130 aircraft during SHADE: 1. Solar spectrum, J. Geophys. Res., 108(D18), 8577, doi: /2002jd Hill, S. C., A. C. Hill, and P. W. Barber (1984), Light scattering by size/ shape distributions of soil particles and spheroids, Appl. Opt., 23, Holben, B. N., et al. (1998), AERONET A federated instrument network and data archive for aerosol characterization, Remote Sens. Environ., 66, Holben, B. N., et al. (2001), An emerging ground-based aerosol climatology: Aerosol optical depth from AERONET, J. Geophys. Res., 106(D11), 12,067 12,097. Jaggard, D. L., C. Hill, R. W. Shorthill, D. Stuart, M. Glantz, F. Rosswog, B. Taggart, and S. Hammond (1981), Light scattering from particles of regular and irregular shape, Atmos. Environ., 15(12), Kalashnikova, O. V., and I. N. Sokolik (2004), Modeling the radiative properties of nonspherical soil-derived mineral aerosols, J. Quant. Spectrosc. Radiat. Transfer, 87, Kaufman, Y. J., D. Tanré, H. R. Gordon, T. Nakajima, J. Lenoble, R. Frouin, H. Grassl, B. M. Herman, M. D. King, and P. M. Teillet (1997a), Passive remote sensing of tropospheric aerosol and atmospheric correction for the aerosol effect, J. Geophys. Res., 102(D14), 16,815 16,830. Kaufman, Y. J., D. Tanré, L. A. Remer, E. F. Vermote, A. Chu, and B. N. Holben (1997b), Operational remote sensing of tropospheric aerosol over land from EOS moderate resolution imaging spectroradiometer, J. Geophys. Res., 102(D14), 17,051 17,067. Kaufman, Y. J., et al. (1998), Smoke, clouds, and radiation Brazil (SCAR-B) experiment, J. Geophys. Res., 103(D24), 31,783 31,808. Kaufman, Y. J., D. Tanré, O. Dubovik, A. Karnieli, and L. A. Remer (2001), Absorption of sunlight by dust as inferred from satellite and ground-based remote sensing, Geophys. Res. Lett., 28(8), Kaufman, Y. J., D. Tanré, and O. Boucher (2002), A satellite view of aerosols in the climate system, Nature, 419, Kaufman, Y. J., D. Tanré, J.-F. Leon, and J. Pelon (2003), Retrievals of profiles of fine and coarse aerosols using lidar and radiometric space measurements, IEEE Trans. Geosci. Remote Sens., 41(8), Kent, G. S., C. R. Trepte, K. M. Skeens, and D. M. Winker (1998), LITE and SAGE II measurements of aerosols in the Southern Hemisphere upper troposphere, J. Geophys. Res., 103(D15), 19,111 19,127. Kinne, S., et al. (2003), Monthly averages of aerosol properties: A global comparison among models, satellite data, and AERONET ground data, J. Geophys. Res., 108(D20), 4634, doi: /2001jd Koch, D. (2001), Transport and direct radiative forcing of carbonaceous and sulfate aerosols in the GISS GCM, J. Geophys. Res., 106(D17), 20,311 20,332. Leon, J.-F., D. Tanre, J. Pelon, Y. J. Kaufman, J. M. Haywood, and B. Chatenet (2003), Profiling of a Saharan dust outbreak based on a synergy between active and passive remote sensing, J. Geophys. Res., 108(D18), 8575, doi: /2002jd Liu, Z., N. Sugimoto, and T. Murayam (2002), Extinction-to-backscatter ratio of Asian dust observed with high-spectral-resolution lidar and Raman lidar, Appl. Opt., 41(15), Masonis, S. J., T. L. Anderson, D. S. Covert, V. Kapustin, A. D. Clarke, S. Howell, and K. Moore (2003), A study of the extinction-to-backscatter ratio of marine aerosol during the Shoreline Environment Aerosol Study, J. Atmos. Oceanic Technol., 20, Mattis, I., A. Ansmann, D. Müller, U. Wandinger, and D. Althausen (2002), Dual-wavelength Raman lidar observations of the extinction-tobackscatter ratio of Saharan dust, Geophys. Res. Lett., 29(9), 1306, doi: /2002gl Mattis, A., A. Ansmann, U. Wandinger, and D. Müller (2003), Unexpectedly high aerosol load in the free troposphere over central Europe in spring/ summer 2003, Geophys. Res. Lett., 30(22), 2178, doi: / 2003GL of 13

13 Mattis, I., A. Ansmann, D. Müller, U. Wandinger, and D. Althausen (2004), Multiyear aerosol observations with dual-wavelength Raman lidar in the framework of EARLINET, J. Geophys. Res., 109, D13203, doi: / 2004JD Menon, S., J. Hansen, L. Nazarenko, and Y. Luo (2002), Climate effects of black carbon aerosols in China and India, Science, 297, Mishchenko, M. I., L. D. Travis, R. A. Kahn, and R. A. West (1997), Modeling phase functions for dustlike tropospheric aerosols using a shape mixture of randomly oriented polydisperse spheroids, J. Geophys. Res., 102(D14), 16,831 16,848. Müller, D., F. Wagner, D. Althausen, U. Wandinger, and A. Ansmann (2000a), Physical properties of the Indian aerosol plume derived from six-wavelength lidar observations on 25 March 1999 of the Indian Ocean Experiment, Geophys. Res. Lett., 27(9), Müller, D., F. Wagner, U. Wandinger, A. Ansmann, M. Wendisch, D. Althausen, and W. von Hoyningen-Huene (2000b), Microphysical particle parameters from extinction and backscatter lidar data by inversion with regularization: Experiment, Appl. Opt., 39(12), Müller, D., K. Franke, F. Wagner, D. Althausen, A. Ansmann, and J. Heintzenberg (2001), Vertical profiling of optical and physical particle properties over the tropical Indian Ocean with six-wavelength lidar: 2. Case studies, J. Geophys. Res., 106(D22), 28,577 28,595. Müller, D., I. Mattis, U. Wandinger, A. Ansmann, D. Althausen, O. Dubovik, S. Eckhardt, and A. Stohl (2003), Saharan dust over a central European EARLINET-AERONET site: Combined observations with Raman lidar and Sun photometer, J. Geophys. Res., 108(D12), 4345, doi: / 2002JD Müller, D., I. Mattis, A. Ansmann, B. Wehner, D. Althausen, U. Wandinger, and O. Dubovik (2004), Closure study on optical and microphysical properties of a mixed urban and Arctic haze air mass observed with Raman lidar and Sun photometer, J. Geophys. Res., 109, D13206, doi: /2003jd Murayama, T., et al. (2003), An intercomparison of lidar-derived aerosol optical properties with airborne measurements near Tokyo during ACE- Asia, J. Geophys. Res., 108(D23), 8651, doi: /2002jd Okada, K., J. Heintzenberg, K. Kai, and Y. Qin (2001), Shape of atmospheric mineral particles in three Chinese arid regions, Geophys. Res. Lett., 28(16), Peppler, R. A., et al. (2000), ARM Southern Great Plains site observations of the smoke pall associated with the 1998 Central American fires, Bull. Am. Meteorol. Soc., 81(11), Powell, D. M., J. A. Reagan, M. A. Rubio, W. H. Erxleben, and J. D. Spinhirne (2000), ACE-2 multiple angle micro-pulse lidar observations from Las Galletas, Tenerife, Canary Islands, Tellus, Ser. B, 52, Ramanathan, V., et al. (2001), Indian Ocean Experiment: An integrated analysis of the climate forcing and effects of the great Indo-Asian haze, J. Geophys. Res., 106(D22), 28,371 28,398. Reid, J. S., et al. (2003), Comparison of size and morphological measurement of coarse mode dust particles from Africa, J. Geophys. Res., 108(D19), 8593, doi: /2002jd Rosen, J. M., R. G. Pinnick, and D. M. Garvey (1997), Measurement of extinction-to-backscatter ratio for near-surface aerosols, J. Geophys. Res., 102(D5), Sakai, T., et al. (2002), Case study of Raman lidar measurements of Asian dust events in 2000 and 2001 at Nagoya and Tsukuba, Japan, Atmos. Environ., 36, Sato, M., J. Hansen, D. Koch, A. Lacis, R. Ruedy, O. Dubovik, B. Holben, M. Chin, and R. Novakov (2003), Global atmospheric black carbon inferred from AERONET, Proc. Natl. Acad. Sci., 100(11), Smirnov, A., B. N. Holben, O. Dubovik, N. T. O Neill, T. F. Eck, D. L. Westphal, A. K. Goroch, C. Pietras, and I. Slutsker (2002a), Atmospheric aerosol optical properties in the Persian Gulf, J. Atmos. Sci., 59, Smirnov, A., B. N. Holben, Y. J. Kaufman, O. Dubovik, T. F. Eck, I. Slutsker, C. Pietras, and R. N. Halthore (2002b), Optical properties of atmospheric aerosol in maritime environments, J. Atmos. Sci., 59, Spinhirne, J. D. (1993), Micropulse lidar, IEEE Trans. Geosci. Remote Sens., 31, Streets, D. G., S. Gupta, S. T. Walshoff, M. Q. Wang, T. C. Bond, and B. Yiyun (2001), Black carbon emissions in China, Atmos. Environ., 35, Sugimoto,N.,I.Matsui,A.Shimizu,I.Uno,K.Asai,T.Endoh,and T. Nakajima (2002), Observation of dust and anthropogenic aerosol plumes in the Northwest Pacific with a two-wavelength polarization lidar on board the research vessel Mirai, Geophys. Res. Lett., 29(19), 1901, doi: /2002gl Swap, R. J., H. J. Annegarn, J. T. Suttles, M. D. King, S. Platnick, J. L. Privette, and R. J. Scholes (2003), Africa burning: A thematic analysis of the Southern African Regional Science Initiative (SAFARI 2000), J. Geophys. Res., 108(D13), 8465, doi: /2003jd Takemura,T.,T.Nakajima,O.Dubovik,B.N.Holben,andS.Kinne (2002), Single-scattering albedo and radiative forcing of various aerosol species with a global three-dimensional model, J. Clim., 15(4), Tanré, D., Y. J. Kaufman, M. Herman, and S. Mattoo (1997), Remote sensing of aerosol properties over oceans using the MODIS/EOS spectral radiances, J. Geophys. Res., 102(D14), 16,971 16,988. Tanré, D.,Y.J.Kaufman,B.N.Holben,B.Chatenet,A.Karnieli, F. Lavenu, L. Blarel, O. Dubovik, L. A. Remer, and A. Smirnov (2001), Climatology of dust aerosol size distribution and optical properties derived from remotely sensed data in the solar spectrum, J. Geophys. Res., 106(D16), 18,205 18,217. Vermote, E. F., N. El Saleous, C. O. Justice, Y. J. Kaufman, J. L. Privette, L. Remer, J. C. Roger, and D. Tanré (1997), Atmospheric correction of visible to middle-infrared EOS-MODIS data over land surfaces: Background, operational algorithm and validation, J. Geophys. Res., 102(D14), 17,131 17,142. Volten, H., O. Munoz, E. Rol, J. F. de Haan, W. Vassen, J. W. Hovenier, K. Muinonen, and T. Nousiainen (2001), Scattering matrices of mineral aerosol particles at and nm, J. Geophys. Res., 106(D15), 17,375 17,401. Voss, K. J., E. J. Welton, P. K. Quinn, J. Johnson, A. M. Thompson, and H. R. Gordon (2001), Lidar measurements during Aerosol99, J. Geophys. Res., 106(D18), 20,821 20,831. Wandinger, U., et al. (2002), Optical and microphysical characterization of biomass-burning and industrial-pollution aerosols from multiwavelength lidar and aircraft measurements, J. Geophys. Res., 107(D21), 8125, doi: /2000jd Welton, E. J., et al. (2000), Ground-based lidar measurements of aerosols during ACE-2: Instrument description, results, and comparisons with other ground-based and airborne measurements, Tellus, Ser. B, 52, Welton, E. J., K. J. Voss, P. K. Quinn, P. J. Flatau, K. Markowicz, J. R. Campbell, J. D. Spinhirne, H. R. Gordon, and J. E. Johnson (2002), Measurements of aerosol vertical profiles and optical properties during INDOEX 1999 using micropulse lidars, J. Geophys. Res., 107(D19), 8019, doi: /2000jd Young, S. A., D. R. Cutten, M. J. Lunch, and J. E. Davies (1993), Lidarderived variations in the backscatter-to-extinction ratio in Southern Hemisphere coastal maritime aerosols, Atmos. Environ. Part A, 27(10), Yu, H., R. E. Dickinson, M. Chin, Y. J. Kaufman, M. Zhou, L. Zhou, Y. Tian, O. Dubovik, and B. N. Holben (2004), Direct radiative effect of aerosols as determined from a combination of MODIS retrievals and GOCART simulations, J. Geophys. Res., 109, D03206, doi: / 2003JD C. Cattrall and K. Thome, Remote Sensing Group, Optical Sciences, University of Arizona, 630 E. University Boulevard, Tucson, AZ , USA. (chris.cattrall@opt-sci.arizona.edu) O. Dubovik, NASA Goddard Space Flight Center, Code 923, Greenbelt, MD 20771, USA. J. Reagan, Electrical and Computer Engineering Department, University of Arizona, PO Box , 1230 E. Speedway Boulevard, Tucson, AZ , USA. 13 of 13

14 Figure 1. Map of the AERONET sites used in the study. Oceanic sites are represented by an X. Figure 7. Fitted frequency distribution of aerosol classes for each lidar parameter. 3of13and8of13