the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Volcanic eruptions alter sky colours in a geoengineered world
Abstract. This study examines sky colour changes in a geoengineered world following a volcanic eruption, combining CESM2-(WACCM6) climate model simulations, SCIATRAN radiative transfer calculations, and colour modelling. Bridging a gap in existing research, this is the first study to analyse how sky colours respond when a volcanic eruption occurs in the context of an ongoing stratospheric aerosol injection (SAI) deployment. The SAI experiments consist of simulations with feedback-controlled sulphur dioxide injections amounting to 7.5 Tg per year at the time of the volcanic eruption, with eruption magnitudes of 10 and 50 Tg sulphur dioxide. The results show several optical phenomena, including Bishop's Rings near the solar horizon, anomalous greenish twilight colouration, and pronounced purple light. The latter is consistent with prior findings linking vivid purple colours to elevated aerosol loading. Despite variations in SAI scenario design, with Northern Hemisphere suspension in one scenario and complete suspension in the other, no significant differences in sky colour were found between the scenarios, months, and latitude investigated following the volcanic eruption with a magnitude of 10 Tg sulphur dioxide.
- Preprint
(14435 KB) - Metadata XML
- BibTeX
- EndNote
Status: final response (author comments only)
- CC1: 'Comment on egusphere-2026-3347', Alan Robock, 11 Aug 2026
-
RC1: 'Comment on egusphere-2026-3347', Anonymous Referee #2, 30 Oct 2026
This paper aims to simulate the color of the twilight sky when volcanic eruptions occur during ongoing geoengineering by stratospheric aerosol injection. Climate model runs with both SAI and volcanic eruptions are used to create aerosol inputs to a radiative transfer model to simulate radiance spectra that are then transformed to sky colors. The paper is interesting but there are gaps in the study design and the radiative transfer calculations.
I have several concerns regarding the radiative transfer calculations.
- I think it is really critical that the SCIATRAN convergence in approximate spherical mode is verified for the cases studied here. Referencing the previous study is not sufficient as the scenarios are different and this geometry is known to be quite sensitive to the radiative transfer assumptions. In the cases studied here, the stratosphere is optical depth is significant and the geometry goes deep into twilight where multiple scattering and strong aerosol forward scattering dominates the analyzed signals. The authors should at least check against a fully spherical model, either SCIATRAN’s full spherical mode or a Monte Carlo model.
- The description of how the aerosol is included in SCIATRAN is incomplete. Please clarify the assumed refractive index (and/or weight fraction) and exactly how the geometric width of the distribution is set.
- The use of an ozone climatology might be influencing the results. Can the ozone from WACCM be used in SCIATRAN? Ozone depletion from SAI/volcanic aerosol might cause changes to the observed spectrum through Chappuis band absorption and influence the perceived green and purple colors.
In terms of the overall study, I think that to make the claims robust, the authors need to add two control cases in a self consistent way: (1) no geoengineering or volcanic eruption, and (2) with a volcanic eruption but no geoengineering. This is really needed to show that the impacts on sky color are really from the combination of the eruption and the geoengineering.
I am also concerned that the study chooses only months 1 and 5 and a single latitude for the analysis. The residence time of the aerosol (both geoengineering and volcanic) is much longer than 5 months so I think the study needs to look at months something like 12-24 and at higher latitudes to really see any difference in the scenarios.  And is a monthly mean for the first month after the eruption really meaningful due to the ongoing transport and dispersion of the plume? It might be that the limited choice of analysis points is driving the conclusion that there are no significant differences between the scenarios, months, and latitude. (Only one latitude is studied!)
Finally I don’t quite appreciate the function of the Reinhard tone mapping operator here. Fig 4 shows that most of the Y values are << 1, so this means that Y’ is essentially equal to Y and the operator isn’t doing anything.Â
Citation: https://doi.org/10.5194/egusphere-2026-3347-RC1 -
RC2: 'Comment on egusphere-2026-3347', Anonymous Referee #1, 02 Nov 2026
Â
Â
This study investigates changes in the color of the sky at twilight caused by a volcanic eruption injecting sulfur in the stratosphere in the context of ongoing perturbation of a geoengineering stratospheric aerosol injection (SAI) experiment. The authors combine CESM2 (WACCM6)-MAM4 model simulations, radiative transfer calculations using SCIATRAN, and CIE/sRGB colorimetry. Three scenarios are examined, involving volcanic injections of 10 or 50 Tg of SO₂, for an observer located at 15° N during the first and fifth months following the eruption.
This work represents an interesting extension of research into volcanic sunsets, using an original multi-model approach. I found the discussion of Mie scattering, anomalous scattering and the formation of greenish hues particularly interesting, going beyond a simple visual analysis of images. The figures effectively illustrate the spatial structures and color changes depending on the viewing angle (which is not obvious to do), although, in my opinion, one should bear in mind the difference in how they appear on screen compared with on paper.
Â
However, I do have a few reservations before the manuscript is published in ACP. The manuscript should give a more detailed description of the MAM4 aerosol model and does not discuss the aerosol optical/microphysical properties (e.g. shape of size distribution) and their impacts on the resulting sky colors. We have no information about how sulfate is treated in MAM4 in terms of size range and microphysical evolution (growth, sedimentation). The colors obtained depend heavily on the particle size distribution, the refractive index and the wavelength. The authors report median radii of 500 to 700 nm for the coarse mode, but the full parameters of the log-normal distributions used in SCIATRAN are not specified in the description of the method. The choice of two log-normal distribution modes is not justified (see ‘Specific comments’ below). The geometric standard deviations, concentrations, refractive indices and any spectral dependence thereof should be specified, and the sensitivity of the images obtained to variations (or uncertainties) in these parameters should be discussed.
The color modelling methodology (chapter 2.3) is described a bit too briefly, and the reader is presented with equations that, whilst understandable, are not accompanied by a sufficiently detailed discussion of the choices made.
I am not totally convinced that the (x,y) coordinates are sufficient to describe all aspects of visual appearance, notably perceived saturation and contrast effects with neighbouring regions. They constitute a useful colorimetric indicator, but how close do they come to a complete simulation of human perception?
Individual image normalization is justified by the authors for visualizing scenes with strong different luminance levels, but to me it may limit the comparison of their appearance and brightness across the aerosol injection scenarios proposed in the manuscript. As a result, if one is seeking to identify unusual visual phenomena caused by injections of stratospheric volcanic aerosols, it could be useful to present both the sRGB images (which are useful for illustrating the phenomena), the CIE (x,y) coordinates (in order to compare the chromaticities) and non-normalized spectral luminances or radiances (to determine the physical significance of the signals). Such information could be given at least in a supplementary material.
Â
Â
Specific comments:
L60-63: why this choice of 2020-2039 to maintain the global mean temperature at 1.5°C while the volcanic eruption is chosen on January 2055? Does the temperature value used here (1.5°C) correspond to a specific CMIP scenario? Please explain. Why a simulation experiment over such a long period (35 years) initialized in 2035? The reader is confused by the various time values that do not match clearly and that suffer from lack of explanations.
L67: the authors indicate several volcanic eruptions whereas only one us specified. Do you mean several scenarios? This needs some clarification.
L69: the authors specify that the calculations were done at the 15°N latitude where the greatest effects are expected. However, looking at Reff contours in Fig. A1, the aerosol signal is stronger at -20 – 0 ° in latitude. Is there an explanation for this discrepancy?
L84-89: the choice of a bimodal distribution is not clear to me. Most of the recent (except Hunga at an early stage of the plume propagation) eruptions show monomodal log-normal distributions. Also, MAM4 is a modal model with fixed modes. Without any discussion about the aerosol sizes covered by the modes, how the microphysical model distributes the sulfate aerosol among these modes, the choice of a bimodal distribution cannot be clearly justified. Furthermore, how the authors derive the bimodal parameters (rm, sigma, N) for each mode? This is a high number of degrees of freedom.
Â
Â
Â
Citation: https://doi.org/10.5194/egusphere-2026-3347-RC2
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 437 | 239 | 56 | 732 | 55 | 54 |
- HTML: 437
- PDF: 239
- XML: 56
- Total: 732
- BibTeX: 55
- EndNote: 54
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
With respect to the 1883 Krakatau eruption, I want to make two small comments:
1. The correct spelling is Krakatau, and not Krakatoa. Even though the British spelled it wrong, it should be spelled the way the Indonesians spell the name of their own volcano.
2. I recommend you consider this paper, which is relevant to your paper. One idea, which I support, is that the sky in The Scream was based on what Munch saw after the Krakatau eruption, and the paper goes into detail about volcanic sunsets.
Prata, Fred, Alan Robock, and Richard Hamblyn, 2018: The sky in Edvard Munch’s The Scream. Bull. Amer. Meteorol. Soc., 99, 1377-1390, doi:10.1175/BAMS-D-17-0144.1.