Answers are grounded in this paper and cite their sources.
Hudson et al. [2012, hereafter H12] have lately shown that the standard deviation (σ) of droplet size distributions was inversely related to cloud condensation nuclei (CCN) concentration at 1% supersaturation (NCCN) for the in situ aircraft measurements collected during the Rain in Cumulus over the Ocean (RICO) project. Using adiabatic parcel model simulations for various values of updraft velocity (w), they further reported a tendency for the σ- NCCN correlation to change signs from positive to negative as w increases beyond a certain value. The analysis of σ- NCCN correlation and its dependence on w certainly add new understanding of dispersion effect (e.g., Liu and Daum, 2002, LD02 hereafter; Liu et al., 2006b, LDY06 hereafter) and imply that like the more widely known aerosol effect on cloud droplet number concentration (N), dispersion effect may exhibit aerosol-limited and w-limited regimes such that dispersion effect can either diminish or enhance the cooling of number effect, depending on the regimes. However, there appears to be some misunderstanding/misinterpretation of LD02 and LDY06 concerning the use of relative dispersion as a measure of spectral width. Furthermore, our examination of the data reported in Table 1 of H12 shows that there is a positive correlation between NCCN and w, and thus it cannot be ruled out that the observed negative σ- NCCN correlation is a manifestation of the covariation in w, or arises from the indirect effect of aerosol on cloud dynamics, or a combination of both. Below these points are detailed. Spectral width has been a much debated topic since late 1950s [Squires, 1958; Warner, 1969]; two commonly used measures in literature are standard deviation (σ) and relative dispersion (ε) defined as the ratio of σ to the corresponding mean radius. Note that mean diameter (MD) and the corresponding σ were used in H12. In Section 7 of H12, the authors state "… However, most previous studies presented droplet spectral width in terms of relative dispersion (ɛ = σ/MD) rather than standard deviation(σ). This confused the issue of the effect of input CCN on droplet spectral width because the denominator (mean diameter) of ε is so dependent on NCCN that it overwhelms the effect of NCCN on σ … Liu et al. [2006a, 2006b] admit that much of the increase of ε attributed to higher NCCN is due to the decrease of MD, but they also claim a σ increase with NCCN …." Although we concur with most of this statement from a microphysical perspective, the authors of H12 missed the point as to why ε instead of σ has been used in our studies cited in H12. The same three variables of L, N, and ε also explicitly appear in the state-of-art parameterization of autoconversion rate, which is essential for studying the second aerosol indirect effect [Liu and Daum, 2004; Liu et al., 2006a, 2007; Xie and Liu, 2011; Xie et al., 2013]. It is worth stressing that in bulk microphysical parameterizations for climate models, cloud properties are represented in terms of L (one-moment scheme), or L and N (two moment scheme), and further improvement demands explicit specification of ε, neither σ nor mean radius. Furthermore, although the differences between σ and ε are self-evident from their definitions, the two have often been ambiguously used in literature as a measure of spectral width, including H12, unfortunately. For example, ε was a focus of both LD02 and LDY06, and the author of H12 appeared to be aware of this according to the previous quote. However, there are at least two places of misrepresentation/confusion in H12: In P3 of H12, the authors stated, "Negative R in Figure 1a (smaller σ for higher N1%) is directly opposite of the broader spectra for more polluted clouds displayed in HY97 and LD2" Again in P8, the authors state, "Both observations and theory indicates that droplet spectral width (σ) at low cloud altitudes in RICO shows an opposite trend with NCCN from that observed and predicted by HY97 and presented by LD2 and Liu et al. [2006b]." Although it may be inferred from their Figure 3 that the observed negative ε-N1% correlation does stand in contrast with those reported in LD02 and LDY06, the quoted statements are misleading, because LD2 and LDY06 were mainly concerned with ε and its relationship to aerosols and vertical velocity, not σ, and changes in ε could be caused by changes either in σ or mean radius or both. The P8 statement was especially misleading because H12 only presented the results of σ- NCCN correlation and its variation with varying w obtained with the parcel model, not the simulated ε- NCCN correlation and its variation with w. In fact, additionally showing the simulated ε- NCCN relationship and its variation with w would be extremely valuable in that the result, if confirmed, could suggest the existence of w-limited and aerosol-limited regimes for dispersion effect just as for droplet concentration. Peng et al. [2007] reported similar results using a different parcel model. It is worth mentioning that the preference of ε over σ in improving microphysical parameterization and quantification of aerosol indirect effects in climate models does not diminish the usefulness of σ analysis. In fact, a combined analysis of σ and ε is often more powerful to dissect microphysical mechanisms [Pawlowska et al., 2006; Pandithurai et al., 2012; Prabha et al., 2012] and should be recommended. H12 reported observational evidence for a negative NCCN-σ correlation and further attributed it to aerosol effects at higher w. Although both the empirical and modeling results are well based and provide some new understanding of the aerosol effect on σ, the data provided in their Table 1 suggest other possibilities as well. Our examination of the same data shows a clear covariation of NCCN and w: an increase of NCCN is generally associated with an increase of w (Figure 1). Figure 2 further compares N, mean diameter, σ, and ε as a function of w. Together with the positive NCCN-w correlation, the striking similarities between what is shown in Figure 2 and the corresponding plots in H12 suggest that the negative NCCN-σ correlation could also be a manifestation of the effect of w covarying with NCCN. Both the increase of N and the decrease of ε with increasing w are also consistent with LDY06. It is noteworthy that the contrasting changes of ε with increasing w and increasing aerosols are closely related to the supersaturation: a stronger w leads to a higher supersaturation that increases N but decreases ε whereas a higher aerosol concentration leads to a higher N but lower supersaturation. Also noteworthy is that the negative correlations between ε (σ) and w arise largely from the data points with lower w (< 1 ms−1); for higher w the decreasing trend is not obvious. This result could suggest that the effect of w on ε (σ) starts to level off after w reaches a certain value, or the value of ε (σ) is close to the limit of the probe used, or is due to poor statistics; more studies are needed on this front. Although the covariation of NCCN and w does not necessarily mean a cause-and-effect relationship, one cannot completely rule out an even more interesting possibility that the negative NCCN-σ correlation results from the indirect effect of aerosol on vertical velocity according to some modeling studies [Xue and Feingold, 2006; Lu and Seinfeld, 2006]. These modeling studies suggested that increasing aerosols tend to strengthen vertical motion and entrainment processes in clouds by altering condensation and evaporation rates and latent heating. The potential aerosol effect on entrainment processes adds an additional layer of difficulty to the already vexing problem of untangling aerosol effects from dynamical effects. A positive correlation between NCCN and w was also reported in Berg et al. [2011]. Note that the CO concentration was used as a proxy of aerosol loading. Being coincidence or aerosol dynamical effects, the association between the subcloud NCCN and in-cloud w makes the data interpretation more challenging and calls for further investigation into the separation of aerosol effects from dynamical effects. An examination of w and aerosol properties below clouds would also help resolve the issue: a coexisting positive w-NCCN correlation just below clouds suggests a common mechanism for the increases in w and NCCN. This work is supported by the U.S. Department of Energy's Earth Systems Modeling (ESM) via the FASTER project (www.bnl.gov/faster) and Atmospheric Science Research (ASR) Programs. Lu is also supported by the National Natural Science Foundation of China (41305120), the Natural Science Foundation of Jiangsu Province, China (BK20130988), the Specialized Research Fund for the Doctoral Program of Higher Education (20133228120002), and the Natural Science Foundation of the Higher Education Institutions of Jiangsu Province, China (13KJB170014).
No takes yet. Share an insight, caveat, or question.
Liu et al. (2013) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: