Stratospheric aerosol size reduction after volcanic eruptions
Abstract. The stratospheric aerosol layer plays an important role in the radiative balance of earth primarily through scattering of solar radiation. The magnitude of this effect depends critically on the size distribution of the aerosols. The aerosol layer is in large part fed by volcanic eruptions strong enough to inject gaseous sulfur species into the stratosphere. The evolution of the stratospheric aerosol size after volcanic eruptions is currently one of the biggest uncertainties in stratospheric aerosol science. We retrieved aerosol particle size information from satellite solar occultation measurements from the Stratospheric Aerosol and Gas Experiment III mounted on the International Space Station (SAGE III/ISS) using a robust spectral method. We show that, surprisingly, some volcanic eruptions can lead to a decrease in average aerosol size, like the 2018 Ambae and the 2021 La Soufrière eruptions. In 2019 an intriguing contrast is observed, where the Raikoke eruption (48° N, 153° E) in 2019 led to the more expected stratospheric aerosol size increase, while the Ulawun eruptions (5° S, 151° E), which followed shortly after, again resulted in a reduction of the median radius and absolute mode width values in the lowermost stratosphere. In addition, the Raikoke and Ulawun eruptions were simulated with the aerosol climate model MAECHAM5-HAM. In these model runs, the evolution of the extinction coefficient as well as of the effective radius could be reproduced well for the first 3 months of volcanic activity. However, the long lifetime of the very small aerosol sizes of many months observed in the satellite retrieval data could not be reproduced.
Felix Wrana et al.
Status: open (until 15 Jun 2023)
- RC1: 'Referee Comment on egusphere-2023-837', Anonymous Referee #1, 30 May 2023 reply
- RC2: 'Comment on egusphere-2023-837', Daniele Visioni, 05 Jun 2023 reply
Felix Wrana et al.
Felix Wrana et al.
Viewed (geographical distribution)
This is a useful contribution on determining stratospheric aerosol size distributions from SAGE III/ISS data and using the size distributions to assess the evolution of aerosol size following several recent small volcanic eruptions. The paper suffers some confusion over the definition and use of effective radius and in the presentations of distributions of dn/dr. Once these are clarified the paper should be published.
Here are detailed comments using the manuscript line numbers.
81 Is σ the mode width or the distribution width? This is a monomodal lognormal so there is only one mode. Mode width is used throughout, but really what is intended is the distribution width, which is a better description.
90 To be clear use : or / notation when writing the ratios. According to the text the ratios used are 449:756 and 1544:756. This seems a bit odd, one ratio is lesser:greater and the other greater:lesser? But this is what is done, Fig. 6, so the text is interpreted correctly as written.
93-95 “The real parts of the refractive indices at the used wavelengths that were necessary for the calculations were taken from Palmer and Williams (1975) and Lorentz-Lorenz-corrections, as described by Steele and Hamill (1981) were applied to them to obtain refractive indices at typical lower stratospheric temperatures” is very confusing.
Try something like: The real parts of the refractive indices at the wavelengths used were calculated from Palmer and Williams (1975) using Lorentz-Lorenz-corrections. This has been described by Steele and Hamill (1981) to obtain refractive indices at typical lower stratospheric temperatures.
102 It should be mentioned here that this formula is just the result of the ratio of volume to surface area from the ratio of the third to second moments of a lognormal size distribution. The authors need to be clear for the readers that this expression does not include the factor 3 traditionally used in the definition of effective radius as V/A * 3, see Eqn. (4) in the discussion of the model results.
189 It is unlikely that a majority of readers will be familiar with the absolute mode width. Perhaps it would be worthwhile to provide some examples of the relationship of omega with sigma for the range of median radii covered.
2020 Delete already.
245 Change “where” to were.
274-275 Isn’t it pretty well established that volcanic eruptions generally lead to changes in the extinction ratios between the longer and shorter wavelengths? In fact that is often the way that particle size following an eruptions is generally assessed. Thus “may” is not quite right here.
Fig. 6a) Is something else being added to this figure besides just the differences in median radii and σ? Using the values in the legend of Fig. 6a), equation (1), and fiddling with No to roughly reproduce the y-axis in Fig. 6a), leads to the attached figure where the color coding, median radii, σs, and axes are the same as Fig. 6a). According to the figure caption only σ is changed in steps of 0.1, but doesn’t each change in sigma also require a change in the median radius. In fact this is explained in the text, but should also be included in the figure caption. Still it seems that the figure should be easily reproduced for the specific sizes and widths quoted, and yet the attempt shown here is not consistent with Fig. 6a).
296 Just to be clear suggest … Each curve corresponds to a single σ value …
310-312 This observation about the distribution width is a bit misleading, since it only appears in distributions of dn/dr. It does not appear in distributions of dn/dln(r), then the widths do appear as expected with σ near 1.0 being narrow and near 2.0 much broader, and each distribution is centered on the median radius, instead of drifting with σ due to the inherent 1/r in the dn/dr distribution. Recall these are lognormal distributions not normal distributions.
370-375 This explanation of how the comparison was conducted between model and measurements is the best way to do it; however, Fig. 7 is not entirely consistent with this explanation, particularly Fig. 7d). Note the differences between model estimates below the tropopause, particularly in the northern hemisphere, compared to the measurements. Including these in the model results distracts the reader from the comparison’s salient points, and the authors make no mention of this region in their discussion.
390-395 The inclusion of m3/m2 is rather superfluous. The important quantities are V and A, however they are calculated. Here the authors also need to be clear that there is a difference of a factor of three between the measurements (which do not include the factor of 3) and the model (which do) when comparing the effective radii using Eqns (2) and (4).
396-424 Based on the comparison in Fig. 8, it appears that the discrepancy of the factor of three has been accounted for by the authors. At least one would hope so. But until this distinction is made clear it is difficult to fully trust this comparison.