Use of In Situ Data to Test a Raman Lidar Based Cloud Condensation Nuclei Remote Sensing Method

Transcription

1 FEBRUARY 2004 NOTES AND CORRESPONDENCE 387 Use of In Situ Data to Test a Raman Lidar Based Cloud Condensation Nuclei Remote Sensing Method STEVEN J. GHAN Pacific Northwest National Laboratory, Richland, Washington DONALD R. COLLINS Texas A&M University, College Station, Texas 23 May 2003 and 25 August 2003 ABSTRACT A method of retrieving vertical profiles of cloud condensation nuclei (CCN) concentration from surface measurements is described. Surface measurements of the CCN concentration are scaled by the ratio of the backscatter (or extinction) vertical profile to the backscatter (or extinction) at or near the surface. The backscatter (or extinction) profile is measured by Raman lidar and is corrected to dry conditions using the vertical profile of relative humidity (also measured by Raman lidar) and surface measurements of the dependence of backscatter (or extinction) on relative humidity. The method assumes that the aerosol composition and the shape of the aerosol size distribution at the surface are representative of the vertical column. Aircraft measurements of aerosol size distribution are used to test the dependence of the retrieval on the uniformity of the shape of the aerosol size distribution. The retrieval is found to be robust for supersaturations less than 0.02% but breaks down at higher supersaturations if the vertical profile of the shape of the aerosol size distribution differs markedly from the shape of the distribution at the surface. Such conditions can be detected from the extinction/backscatter ratio. 1. Introduction Cloud condensation nuclei (CCN) are the particles on which cloud droplets form when the supersaturation in a cloud is high enough for the particles to grow by water condensation until they reach a critical radius, beyond which condensational growth continues spontaneously unless supersaturation decreases rapidly (Nenes et al. 2001b). The CCN concentration is a function of supersaturation, increasing monotonically with supersaturation as more and more particles are activated at higher supersaturations. If the maximum supersaturation in an updraft is known, the CCN concentration at that supersaturation provides a close approximation to the number of droplets that will be formed. Thus, measurements of CCN concentration as a function of supersaturation provide important information about the relationship between aerosols and clouds. Such information is essential for improving estimates of the indirect effects of aerosols on climate. CCN measurements are difficult to obtain. Instruments designed to measure CCN concentration must reproduce the supersaturation conditions within a cloud. Corresponding author address: Dr. Steven J. Ghan, Pacific Northwest National Laboratory, Mail Stop K9-30, Richland, WA steve.ghan@pnl.gov Typically these instruments use artificial cloud chambers or continuous flow diffusion chambers (Hudson 1989; Nenes et al. 2001a) that are several tens of centimeters in size with masses of several tens of kilograms. Almost always they are deployed either on the ground (Menon et al. 1998) or, more commonly, in aircraft (Hudson and Li 1995; Hudson et al. 1998; Yum et al. 1998; Hudson and Xie 1999; Hudson and Yum 2001; Yum and Hudson 2001). Studies of the influence of aerosols on clouds are complicated by the dependence of the influence on updraft velocity and cloud thickness. Updraft velocity determines the rate of adiabatic cooling and hence affects the maximum supersaturation and the number of aerosols activated (Abdul-Razzak and Ghan 2002). Cloud thickness is the most important factor determining the liquid water path and optical depth of clouds (Brenguier et al. 2000). Both updraft velocity and cloud thickness are highly variable, so isolating the aerosol influence on cloud optical depth generally requires many thousands of independent samples. Given the high cost of aircraft measurements, isolating the aerosol influence on clouds can be quite expensive. What is clearly needed is a means of measuring CCN concentration near cloud base from the ground. Although mountain sites can provide CCN measurements near cloud base when the cloud base is near the site 2004 American Meteorological Society

2 388 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 21 FIG. 1. Flowchart of the CCN retrieval scheme. elevation (Menon and Saxena 1998; Menon et al. 2002), the cloud base may not be near the site elevation sufficiently frequently to compile enough independent samples. But if CCN concentration near cloud base could be retrieved from remote sensing, a ground site could collect a large number of independent samples of CCN measurements near cloud base, without the high cost of aircraft. Under certain conditions it should be possible to retrieve vertical profiles of CCN spectra from groundbased measurements. One retrieval method was explored by Feingold et al. (1998). However, that retrieval relies on a parameterization of droplet nucleation. To test droplet nucleation parameterizations, an independent retrieval method is required. Here we describe an alternate CCN retrieval method and present an evaluation of the dependence of its accuracy on the validity of one key assumption. 2. CCN retrieval method The CCN retrieval is outlined in Fig. 1. In this retrieval surface measurements of the CCN spectrum CCN(z 0 ) are scaled by the ratio of the 180 backscatter (or extinction) profile E d (z) to the 180 backscatter (or extinction) at or near the surface, E d (z 0 ): CCN(z) CCN(z 0)E d(z)/e d(z 0 ). (1) The 180 backscatter (or extinction) profile E(z) is measured by Raman lidar (Ferrare et al. 2001) and is corrected to dry conditions using the vertical profile of relative humidity RH [estimated from retrievals of absolute humidity q from Raman lidar and temperature T from an atmospheric emitted radiance interferometer (AERI), a radio acoustic sounding system (RASS), or from radiosondes] and surface measurements of the dependence of 180 backscatter (or extinction) on relative humidity, f [RH(z)]: E d (z) E(z)/ f[rh(z)]. (2) The dependence of extinction on relative humidity can be approximated by the dependence of scattering on relative humidity, which can be measured by a humidified nephelometer (the dependence of absorption on humidity is typically weaker than the dependence of scattering, so this approximation will overestimate the dependence of extinction on humidity somewhat). The dependence of 180 backscatter on relative humidity can be measured by a humidified backscatter nephelometer. Although the latter has not been constructed, Anderson et al. (2000) have produced a 180 backscatter nephelometer that operates at ambient humidity; it can, in principle, be connected to a humidity scanning system. Anderson et al. (2000) and Masonis et al. (2002) have shown that the humidity dependence of 180 backscatter is very similar to the humidity dependence of extinction, so that measurements of the humidification factor for extinction can be used for 180 backscatter as well. As an alternative to the surface measurement of f(rh), the dependence of extinction and backscatter on relative humidity can be determined from Raman lidar extinction and backscatter measurements in the range of relative humidities encountered in the vertical profile (Feingold and Morley 2003). However, that method relies on the assumption that the atmosphere is well mixed (so that the aerosol composition and size distribution are uniform), which is a more restrictive condition than that required if f(rh) is measured at the surface. The method assumes that the humidification factor measured at the surface is representative of the humidification factor measured at altitude, and that the vertical structure of CCN concentration is identical to the vertical structure of dry extinction or backscatter. Since both extinction/backscatter and activation at a given supersaturation are determined entirely by the size distribution of number, composition, and shape, both of these assumptions are valid if (i) the shape of the aerosol size distribution (but not the total aerosol number) is independent of altitude, and (ii) the aerosol composition and shape are independent of altitude. Thus, the impact of both of these assumptions on the CCN retrieval needs to be tested. This paper focuses on the first assumption. We test the impact of measured variations in aerosol size distribution on the humidification factor and on the proportionality of CCN concentration to dry backscatter and extinction. 3. Testing the retrieval The full suite of measurements needed to fully test this retrieval is not yet available. In the meantime, assumption (i) can be tested using existing in situ measurements of vertical profiles of aerosol size distribution. Vertical profiles of dry backscatter, dry extinction, humidification factor, and CCN concentration are calculated from the dry aerosol size distribution estimated from aircraft measurements during the 1997 second Aerosol Characterization Experiment (ACE-2) near the Canary Islands (Collins et al. 2000). The dry size distribution was calculated from the size distribution measured at, or slightly below, ambient humidity by a dif-

3 FEBRUARY 2004 NOTES AND CORRESPONDENCE 389 FIG. 2. Extinction humidification factor as a function of relative humidity for each 1-min average of ACE-2 data for 17 Jul ferential mobility analyzer, a passive cavity aerosol spectrometer probe (PCASP-100X), and a forward scattering spectrometer probe, all mounted on a Cessna Skymaster aircraft. These instruments provide measurements of the size distribution at 124 diameters ranging between 10 and 6000 nm. The size distributions were adjusted to dry conditions using Köhler theory (Pruppacher and Klett 1997) with empirical relationships for solution activity, the ambient relative humidity, and the composition measured at two island sites (Collins et al. 2000). The full variability of the dry size distributions on four days during ACE-2 is illustrated in Collins et al. (2000). The backscatter and extinction profiles are calculated from the Mie theory using Wiscombe s (1979) Mie code, assuming a refractive index of (1.53, 10 8 ) [which is appropriate for ammonium sulfate (Kent et al. 1983)] and a wavelength of 355 nm [the wavelength of the Raman lidar at the Atmospheric Radiation Measurement (ARM) Program site in Oklahoma]. To improve the accuracy of the backscatter calculation, the 124 particle size bins in the ACE-2 data are split into 493 bins. The humidification factor for backscatter and extinction was calculated from the dry backscatter and extinction and from the backscatter and extinction at selected relative humidities, with the size distribution at the selected relative humidities calculated from the dry size distribution, the selected relative humidity, and Köhler theory, assuming all particles are composed of ammonium sulfate. The sensitivity of the humidification factor to the size distribution is illustrated in Fig. 2, which shows the humidification factor for extinction as a function of relative humidity for each 1-min sample between the surface and 4 km on one day (17 July 1997). The humidification factor varies between samples as a result of variations in the size distribution. The variation is greater at higher relative humidity, with a twofold range when relative humidity is 80%. This suggests that the approximation of the humidification factor profile using surface measurements can introduce twofold or larger errors in the CCN retrieval near cloud base, where FIG. 3. Extinction and backscatter humidification factors at ambient relative humidity for each 1-min average on 10 and 17 Jul. the relative humidity is likely to exceed 80%. For a well-mixed atmosphere, that corresponds to about 300 m below cloud base. Thus, if errors less than twofold are required, then the retrieval should be restricted to altitudes at least 300 m below cloud base. Although humidified nephelometers for measuring the humidification factor for extinction are quite common, 180 backscatter nephelometers are not. How large an error in the humidification factor for backscatter is introduced by using an extinction nephelometer? As noted above, Anderson et al. (2000) and Masonis et al. (2003) have shown that the humidity dependence of backscatter is very similar to the humidity dependence of extinction. We further address this question with ACE-2 data by comparing the humidification factor for extinction and backscatter at ambient humidity. Figure 3 compares the humidification factor on two different days, 10 and 17 July The extinction and backscatter humidification factors agree quite well on 17 July, but on 10 July the backscatter humidification factor exceeds the extinction humidification factor by nearly 20%. Such an error is much smaller than the error due to the spatial variability in the size distribution. If the shape of the aerosol size distribution is independent of height, then the vertical profile of CCN concentration should be proportional to the profiles of extinction and backscatter. Even if the shape of the size distribution is not uniform, the CCN concentration will be proportional to the extinction and backscatter if the CCN concentration is controlled by the same particle sizes that control extinction and backscatter. We therefore consider CCN concentrations at a variety of supersaturations to determine whether CCN at some supersaturations are more closely related to extinction and backscatter than at other supersaturations. The CCN concentration is calculated at seven different supersaturations S between 0.01% and 1% from the Köhler theory using the hygroscopic properties of ammonium sul-

4 390 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 21 FIG. 4. Extinction and backscatter cross sections at wavelength 355 nm as functions of particle radius. fate (insoluble material will increase particle size and hence improve the retrieval at high supersaturations). Which particle sizes control extinction and backscatter? Figure 4 shows the extinction and backscatter cross sections (per particle mass) as functions of particle radius. Extinction at wavelength 355 nm is most sensitive to particles with radii between 100 and 200 nm; backscatter is most sensitive to particles with radii between 300 and 500 nm. It is not surprising then that extinction and backscatter are most highly correlated with CCN concentration at supersaturations sufficiently low that the CCN concentration is dominated by particles with radii larger than nm. Figure 5 shows the CCN concentration at S 0.01%, 0.1%, and 1% plotted versus dry backscatter for each of two days (10 and 17 July 1997) during ACE-2. Each point represents a 1-min average of samples taken between the surface and about 4-km altitude. On both days the CCN concentration at a supersaturation S of 0.01% (which for ammonium sulfate represents the number of particles with radii larger than 300 nm) is highly correlated with dry backscatter over a wide range in CCN concentration and backscatter. The high correlation results from the fact that both backscatter and CCN at S 0.01% are most sensitive to particles with radii between 300 and 500 nm. The CCN at higher supersaturation is not as well correlated with backscatter, particularly on 17 July. At S 0.1% the relationship between CCN concentration is split, suggesting two different aerosol populations. At S 1% (particles with radii larger than 14 nm) most of the variability in backscatter is almost completely unrelated to the variability in CCN concentration. At such supersaturations the CCN concentration is dominated by particles too small to influence the backscatter. However, if the aerosol population contains a significant amount of insoluble material (which is often measured in the field), particles that activate at a particular supersaturation (say, 1%) will be larger and hence more likely to produce an extinction or backscatter signal. FIG. 5. Scatterplot of CCN concentration at supersaturations of (top) 1%, (middle) 0.01%, and (bottom) 0.1% plotted vs dry backscatter at wavelength 355 nm for 10 and 17 Jul during ACE-2. Each point represents a 1-min average of samples taken between the surface and about 4-km altitude. Figure 6 shows the same CCN concentrations plotted versus dry extinction for the same days. The relationship with CCN concentration is similar to that for backscatter, with a tendency for a stronger correlation at lower supersaturation. However, the correlation is not as strong at S 0.01% as it is for backscatter, because extinction is sensitive to particles with radii nm, which is too small to contribute to the CCN concentration at S 0.01%. As in the case for backscatter, the relationship between extinction and CCN concentration at S 1% is much worse for 17 July than 10 July. It is important to understand why the relationship between CCN concentration at S 1% and extinction and backscatter is worse on 17 July than on 10 July. Supersaturations in clouds can approach 1% when the air is relatively clean and updrafts exceed 1ms 1. Although supersaturations are more typically less than 1%

5 FEBRUARY 2004 NOTES AND CORRESPONDENCE 391 FIG. 7. Dry backscatter plotted vs altitude for 10 and 17 Jul. at S 0.01% and 0.02%, while extinction is poorly correlated on some days. At higher supersaturations backscatter is usually poorly correlated with CCN concentration, but extinction is still well correlated with CCN concentrations on a significant fraction of the days. At S 0.1% extinction explains at least 80% of the variance on 4 and at least 65% of the variance on 7 of the 11 days, and at S 1% it explains more than 80% of the variance on 2 and more than 55% 65% of the variance on 4 of the 11 days. Thus, on many days the retrieval can provide CCN profiles that are significantly more accurate than estimates from the surface CCN concentration alone, but on other days it cannot. Given the inconsistent correlation between CCN concentration and extinction or backscatter, as a practical FIG. 6. As in Fig. 5, but for dry extinction. in most stratiform clouds, the relationship between CCN concentration and scattering at S 0.1% exhibits many of the features evident at S 1%. To understand the relationship, Fig. 7 shows vertical profiles of backscatter on each day. Although the backscatter profiles are very similar below 1 km, they differ markedly above. Primarily because of the presence of a dust layer above the planetary boundary layer, July 17 differs from July 10 (Powell et al. 2000). The dust layer contributes to a coarse aerosol mode (Collins et al. 2000) that enhances the backscatter (and extinction) but has little effect on the CCN concentration at S 1%. The results for 10 and 17 July were chosen to illustrate two extremes between conditions favorable for retrieving CCN concentration and conditions unfavorable. Figure 8 summarizes the relationship between CCN concentration and backscatter or extinction on 11 days during ACE-2. The correlation between CCN concentration and backscatter or extinction is plotted for each day and for seven supersaturations between 0.01% and 1%. Backscatter is consistently highly correlated with CCN FIG. 8. Correlation between CCN concentration and (top) backscatter and (bottom) extinction as functions of supersaturation for 11 days during ACE-2.

6 392 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 21 FIG. 9. Extinction/backscatter ratio plotted vs altitude for 10 and 17 Jul. matter it would be helpful to have a remotely sensed measure of the vertical structure of particle size. The extinction/backscatter ratio can provide such information. Figure 9 shows the extinction/backscatter ratios calculated for 10 and 17 July. Although both days exhibit variability with altitude, the variability on 10 July is much less than on 17 July, which clearly exhibits two distinct aerosol populations within and above the boundary layer. From Fig. 9 one can conclude that on 17 July retrieval of CCN at S 0.02% will only by reliable below an altitude of 1.7 km. However, it must be noted that the extremely low extinction/backscatter ratio above the boundary layer on 17 July is much lower than would be observed, because the Mie theory underestimates the ratio for dust particles (Liu et al. 2002). Moreover, the refractive index for dust differs from that for ammonium sulfate; this can produce a substantial impact on the extinction, backscatter, and extinction/backscatter ratio. If one knows that the aerosol population above the boundary layer differs from that within the boundary layer, one can focus the retrieval on the boundary layer. Although one might expect the CCN concentration to be uniform within the boundary layer, Fig. 10 shows that CCN concentrations can vary tenfold within the boundary layer, and that dry 180 backscatter explains most of the variability at S 0.1%. However, the retrieval at S 1.0% is still poor on most days. Similar conclusions hold for a retrieval based on extinction. The impact of the combination of the error in the humidification factor and in the proportionality between CCN and dry extinction can be estimated by expressing the retrieval C(z) ae(z)/ f[rh(z)], (3) where a C(z 0 )/E d (z 0 ) is the proportionality constant in the relationship between CCN concentration and dry extinction (or backscatter). Taylor series expansion of (3) about central values of the parameters leads to an expression for the error in the concentration, FIG. 10. As in Fig. 5, except for points below 1.7-km altitude only. 2 2 C C C C a E f a E f C C C C cov(a, E) cov(e, f ) a E E f C C cov(a, f ), (4) a f where denotes the uncertainty in each input parameter. The first term in (4) can be related to the correlation between CCN concentration and dry extinction considered in Figs. 5, 6, and 8. The second term, which accounts for the uncertainty in the ambient extinction and backscatter retrieval by Raman lidar, has not been considered here. The third term accounts for the uncertainty in the humidification factor illustrated in Fig. 2; it depends on uncertainty in relative humidity as well as how well the surface measurement of the dependence of backscatter and extinction on humidity applies at altitude. The last terms depend on the correlation between

7 FEBRUARY 2004 NOTES AND CORRESPONDENCE 393 the parameters. For example, the humidification factor f and the proportionality factor a both depend on composition and size distribution in ways that can produce a positive or negative correlation. A full consideration of how these terms contribute to the overall uncertainty in the retrieval requires measurements of vertical profiles of CCN concentration, humidification factor, and relative humidity. Such an analysis will be performed in a subsequent paper, when the measurements from the ARM Program s recent Aerosol Intensive Observation Period are available. 4. Conclusions The CCN retrieval method relies on two assumptions. We have tested the dependence of the retrieval on the validity of one assumption, namely that the vertical profile of the shape of the aerosol size distribution does not differ from the distribution at the surface. We have found that variations in the shape of the size distribution can cause twofold errors in the humidification factor when relative humidity exceeds 80%, which is expected near cloud base. Thus, the retrieval methods should be applied far enough below the cloud that the relative humidity is less than 80%. For a well-mixed atmosphere, that corresponds to about 300 m below cloud base. We have also found that for supersaturations less than 0.02% the retrieval is insensitive to large variations in the shape of the aerosol size distribution. For supersaturations larger than 0.1% the retrieval breaks down when a coarse dust layer overlays an accumulationmode boundary layer (the retrieval within the boundary layer would still be reliable). However, we have found that such conditions can be detected from the vertical profile of the extinction/backscatter ratio. Thus, it may be possible to remotely determine when the retrieval is not compromised by vertical variations in aerosol size distribution. On days when the shape of the size distribution is invariant with height, dry extinction can explain considerable variance in the CCN concentration at supersaturations as high as 1%. Thus, a well-mixed boundary layer is not necessary for the retrieval to provide CCN profiles that are significantly more accurate than estimates from the surface CCN concentration alone. For ACE-2 conditions, dry extinction provides a useful scaling of CCN concentration on most days for S 0.1% and on 20% 40% of the days at S 1%. Further work is required to test the dependence of the retrieval on the validity of the second assumption, namely, that the vertical profile of the aerosol composition does not differ from the composition at the surface. Such work is awaiting coincident measurements of vertical profiles of CCN, f(rh), and relative humidity from aircraft; vertical profiles of extinction, backscatter, and absolute humidity from Raman lidar; vertical profile of temperature from AERI and RASS; CCN at the surface; and f(rh) at the surface. In addition, the accuracy of the relative humidity retrieval, the humidification factor retrieval, and the extinction and backscatter retrieval (particularly near the surface, where the Raman lidar suffers from an overlap problem) need to be determined using in situ measurements. The U.S. Department of Energy (DOE) ARM Program conducted a field experiment in Oklahoma during May 2003 to collect the data needed to assess the accuracy of these retrievals. This will permit an evaluation of the impact of the second assumption on the CCN retrieval. Acknowledgments. John Ogren suggested the use of extinction/backscatter ratio as a measure of particle size. Rich Barchet, Donna Flynn, Rich Ferrare, and Tad Anderson provided helpful suggestions. This study was primarily supported by the ARM Program, which is part of the DOE Biological and Environmental Research Program. Support was also provided by NASA Earth Science Enterprise Grant NAG The Pacific Northwest National Laboratory is operated for the DOE by Battelle Memorial Institute under Contract DE- AC06-76RLO REFERENCES Abdul-Razzak, H., and S. J. Ghan, 2002: A parameterization of aerosol activation. 3. Sectional representation. J. Geophys. Res., 107, 4026, doi: /2001jd Anderson, T. L., S. J. Masonis, D. S. Covert, R. J. Charlson, and M. J. Rood, 2000: In-situ measurement of the aerosol extinction-tobackscatter ratio at a polluted, continental site. J. Geophys. Res., 105, Brenguier, J.-L., H. Pawlowska, L. Schuller, R. Preusker, J. Fischer, and Y. Fouquart, 2000: Radiative properties of boundary layer clouds: Droplet effective radius versus number concentration. J. Atmos. Sci., 57, Collins, D. R., and Coauthors, 2000: In situ aerosol-size distributions and clear-column radiative closure during ACE-2. Tellus, 52B, Feingold, G., and B. Morley, 2003: Aerosol hygroscopic properties as measured by lidar and comparison with in situ measurements. J. Geophys. Res., 108, 4327, doi: /2002jd , S. Yang, R. M. Hardesty, and W. R. Cotton, 1998: Retrieving cloud condensation nucleus properties from Doppler cloud radar, microwave radiometer, and lidar. J. Atmos. Oceanic Technol., 15, Ferrare, R. A., D. D. Turner, L. H. Brasseur, W. F. Feltz, O. Dubovik, and T. P. Tooman, 2001: Raman lidar measurements of the aerosol extinction-to-backscatter ratio over the southern Great Plains. J. Geophys. Res., 106, Hudson, J. G., 1989: An instantaneous CCN spectrometer. J. Atmos. Oceanic Technol., 6, , and H. Li, 1995: Microphysical contrasts in Atlantic stratus. J. Atmos. Sci., 52, , and Y. Xie, 1999: Vertical distributions of cloud condensation nuclei spectra over the summertime northeast Pacific and Atlantic Oceans. J. Geophys. Res., 104, , and S. S. Yum, 2001: Maritime-continental drizzle contrasts in small cumuli. J. Atmos. Sci., 58, , Y. Xie, and S. S. Yum, 1998: Vertical distributions of cloud condensation nuclei spectra over the summertime Southern Ocean. J. Geophys. Res., 103, Kent, G. S., G. K. Yue, U. O. Farrukj, and A. Deepak, 1983: Modeling atmospheric aerosol backscatter at CO 2 laser wavelengths. 1:

8 394 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME 21 Aerosol properties, modeling techniques, and associated problems. Appl. Opt., 22, Liu, Z., N. Sugimoto, and T. Murayajama, 2002: Extinction-to-backscatter ratio of Asian dust observed with high-resolution lidar and Raman lidar. Appl. Opt., 41, Masonis, S. J., T. L. Anderson, D. S. Covert, V. Kapustin, A. D. Clarke, S. Howell, 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, Menon, S., and V. K. Saxena, 1998: Role of sulfates in cloud climate feedback mechanisms. Atmos. Res., 48 49, ,, P. Durkee, B. N. Wenny, and K. Nielsen, 2002: Role of sulfate aerosols in modifying the cloud albedo: A closure experiment. Atmos. Res., 61, Nenes, A., P. Y. Chuang, R. C. Flagan, and J. H. Seineld, 2001a: A theoretical analysis of cloud condensation nucleus (CCN) instruments. J. Geophys. Res., 106, , S. J. Ghan, H. Abdul-Razzak, P. Chuang, and J. Seinfeld, 2001b: Kinetic limitations on droplet formation. Tellus, 53, 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, 52B, Pruppacher, H. R., and J. D. Klett, 1997: Microphysics of Clouds and Precipitation. Dordrecht, 954 pp. Wiscombe, W. J., 1979: Mie scattering calculations: Advances in technique and fast, vector-speed computer codes. NCAR Tech. Note TN-140 STR, 62 pp. [Available from NTIS, 5285 Port Royal Rd., Springfield, VA ] Yum, S. S., and J. G. Hudson, 2001: Vertical distributions of cloud condensation nuclei spectra over the springtime Arctic Ocean. J. Geophys. Res., 106, ,, and Y. Xie, 1998: Comparisons of cloud microphysics with cloud condensation nuclei spectra over the summertime Southern Ocean. J. Geophys. Res., 103,