IPCC Fourth Assessment Report: Climate Change 2007
Climate Change 2007: Working Group I: The Physical Science Basis Link Between Aerosol Particles and Cloud Microphysics

The local impact of anthropogenic aerosols has been known for a long time. For example, smoke from sugarcane and forest fires was shown to reduce cloud droplet sizes in early case studies utilising in situ aircraft observations (Warner and Twomey, 1967; Eagan et al., 1974). On a regional scale, studies have shown that heavy smoke from forest fires in the Amazon Basin have led to increased cloud droplet number concentrations and to reduced cloud droplet sizes (Reid et al., 1999; Andreae et al., 2004; Mircea et al., 2005). The evidence concerning aerosol modification of clouds provided by the ship track observations reported in the TAR has been further confirmed, to a large extent qualitatively, by results from a number of studies using in situ aircraft and satellite data, covering continental cases and regional studies. Twohy et al. (2005) explored the relationship between aerosols and clouds in nine stratocumulus cases, indicating an inverse relationship between particle number and droplet size, but no correlation was found between albedo and particle concentration in the entire data set. Feingold et al. (2003), Kim et al. (2003) and Penner et al. (2004) presented evidence of an increase in the reflectance in continental stratocumulus cases, utilising remote sensing techniques at specific field sites. The estimates in Feingold et al. (2003) confirm that the relationship between aerosol and cloud droplet number concentrations is nonlinear, that is Nd (Na)b, where Nd is the cloud drop number density and Na is the aerosol number concentration. The parameter b in this relationship can vary widely, with values ranging from 0.06 to 0.48 (low values of b correspond to low hygroscopicity). This range highlights the sensitivity to aerosol characteristics (primarily size distribution), updraft velocity and the usage of aerosol extinction as a proxy for CCN (Feingold, 2003). Disparity in the estimates of b (or equivalent) based on satellite studies (Nakajima et al., 2001; Breon et al., 2002) suggests that a quantitative estimate of the albedo effect from remote sensors is problematic (Rosenfeld and Feingold 2003), particularly since measurements are not considered for similar liquid water paths.

Many recent studies highlight the importance of aerosol particle composition in the activation process and droplet spectral evolution (indicated in the early laboratory work of Gunn and Philips, 1957), but the picture that emerges is not complete. Airborne aerosol mass spectrometers provide firm evidence that ambient aerosols consist mostly of internal mixtures, for example, biomass burning components, organics and soot are mixed with other aerosol components (McFiggans et al., 2006). Mircea et al. (2005) showed the importance of the organic aerosol fraction in the activation of biomass burning aerosol particles. The presence of internal mixtures (e.g., sea salt and organic compounds) can affect the uptake of water and the resulting optical properties compared to a pure sea salt particle (Randles et al., 2004). Furthermore, the varying contents of water-soluble and insoluble substances in internally mixed particles, the vast diversity of organics, and the resultant effects on cloud droplet sizes, makes the situation even more complex. Earlier observations of fog water (Facchini et al., 1999, 2000) suggested that the presence of organic aerosols would reduce surface tension and lead to a significant increase in the cloud droplet number concentration (Nenes et al., 2002; Rissler et al., 2004; Lohmann and Leck, 2005; Ming et al., 2005a; McFiggans et al., 2006). On the other hand, Feingold and Chuang (2002) and Shantz et al. (2003) indicated that organic coating on CCN delayed activation, leading to a reduction in drop number and a broadening of the cloud droplet spectrum, which had not been previously considered. Ervens et al. (2005) addressed numerous composition effects in unison to show that the effect of composition on droplet number concentration is much less than suggested by studies that address individual composition effects, such as surface tension. The different relationships observed between cloud optical depth and liquid water path in clean and polluted stratocumulus clouds (Penner et al., 2004) have been explained by differences in sub-cloud aerosol particle distributions, while some contribution can be attributed to CCN composition (e.g., internally mixed insoluble dust; Asano et al., 2002). Nevertheless, the review by McFiggans et al. (2006) points to the remaining difficulties in quantitatively explaining the relationship between aerosol size and composition and the resulting droplet size distribution. Dusek et al. (2006) concluded that the ability of a particle to act as a CCN is largely controlled by size rather than composition.

The complexity of the aerosol-cloud interactions and local atmospheric conditions where the clouds are developing are factors in the large variation evidenced for this phenomenon. Advances have been made in the understanding of the regional and/or global impact based on observational studies, particularly for low-level stratiform clouds that constitute a simpler cloud system to study than many of the other cloud types. Column aerosol number concentration and column cloud droplet concentration over the oceans from the AVHRR (Nakajima et al., 2001) indicated a positive correlation, and an increase in shortwave reflectance of low-level, warm clouds with increasing cloud optical thickness, while liquid water path (LWP) remained unmodified. While these results are only applicable over the oceans and are based on data for only four months, the positive correlation between an increase in cloud reflectance and an enhanced ambient aerosol concentration has been confirmed by other studies (Brenguier et al., 2000a,b; Rosenfeld et al., 2002). However, other studies highlight the sensitivity to LWP, linking high pollution entrained into clouds to a decrease in LWP and a reduction in the observed cloud reflectance (Jiang et al., 2002; Brenguier et al., 2003; Twohy et al., 2005). Still others (Han et al., 2002, using AVHRR observations) have reported an absence of LWP changes in response to increases in the column-averaged droplet number concentration, this occurred for one-third of the cloud cases studied for which optical depths ranged between 1 and 15. Results of large-eddy simulations of stratocumulus (Jiang et al., 2002; Ackerman et al., 2004; Lu and Seinfeld, 2005) and cumulus clouds (Jiang and Feingold, 2006; Xue and Feingold, 2006) seem to confirm the lack of increase in LWP due to increases in aerosols; they point to a dependence on precipitation rate and relative humidity above the clouds (Ackerman et al., 2004). The studies above highlight the difficulty of devising observational studies that can isolate the albedo effect from other effects (e.g., meteorological variability, cloud dynamics) that influence LWP and therefore cloud RF.

Results from the POLDER satellite instrument, which retrieves both submicron aerosol loading and cloud droplet size, suggest much larger cloud effective radii in remote oceanic regions than in the highly polluted continental source areas and downwind adjacent oceanic areas, namely from a maximum of 14 µm down to 6 µm (Bréon et al., 2002). This confirms earlier studies of hemispheric differences using AVHRR. Further, the POLDER- and AVHRR-derived correlations between aerosol and cloud parameters are consistent with an aerosol indirect effect (Sekiguchi et al., 2003). These results suggest that the impact of aerosols on cloud microphysics is global. Note that the satellite measurements of aerosol loading and cloud droplet size are not coincident, and an aerosol index is not determined in the presence of clouds. Further, there is a lack of simultaneous measurements of LWP, which makes assessment of the cloud albedo RF difficult.

The albedo effect is also estimated from studies that combined satellite retrievals with a CTM, for example, in the case of two pollution episodes over the mid-latitude Atlantic Ocean. Results indicated a brightening of clouds over a time scale of a few days in instances when LWP did not undergo any significant changes (Hashvardhan et al., 2002; Schwartz et al., 2002; Krüger and Graβl, 2002). There have been fewer studies on aerosol-cloud relationships under more complex meteorological conditions (e.g., simultaneous presence of different cloud types).

The presence of insoluble particles within ice crystals constituting clouds formed at cold temperatures has a significant influence on the radiation transfer. The inclusions of scattering and absorbing particles within large ice crystals (Macke et al., 1996) suggest a significant effect. Hence, when soot particles are embedded, there is an increase in the asymmetry parameter and thus forward scattering. In contrast, inclusions of ammonium sulphate or air bubbles lead to a decrease in the asymmetry parameter of ice clouds. Given the recent observations of partially insoluble nuclei in ice crystals (Cziczo et al., 2004) and the presence of small crystal populations, there is a need to further develop the solution for radiative transfer through such systems.