the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
The Impact of Background ENSO and NAO Conditions and Anomalies on the Modeled Response to Pinatubo-Sized Volcanic Forcing
Helen Weierbach
Allegra N. LeGrande
Kostas Tsigaridis
Abstract. Strong, strato-volcanic eruptions are a substantial, intermittent source of natural climate variability. Background atmospheric and oceanic conditions such as El Niño Southern Oscillation (ENSO) and the North Atlantic Oscillation (NAO) also naturally impact climate on regular time scales. We examine how background conditions of ENSO and NAO impact the climate's response to a Pinatubo-type eruption using a large (81-member) ensemble of model simulations in GISS Model E2.1-G. Simulations are sampled from possible background conditions under the protocol of the coordinated CMIP6 Volcanic Model Intercomparison Project (VolMIP) – where aerosol forcing with time, latitude, and height. We analyze paired paired anomalous variation (perturbed - control) to understand changes in global and regional climate responses under positive, negative, and neutral ENSO and NAO conditions. In particular, we find that for paired anomalies there is a high probability of strong (~1.5 °C) post-eruptive winter warming for negative NAO ensembles with analysis coincident with decreased lower stratospheric temperature at the poles, decreased geopotential height, and strengthening of the stratospheric polar vortex. Historical anomalies (relative to climatology) show no mean warming and suggest that strength of this response is impacted by control conditions. Again using paired anomalies, we also observe that positive and negative ENSO ensembles relax the ENSO anomaly in the first post-eruptive Boreal Winter while neutral-phase ensembles are variable and show no clear response. In general, paired anomalies give insight into the evolution of the climate response to volcanic forcing, but are significantly impacted by background climate conditions present in control conditions.
- Preprint
(3054 KB) -
Supplement
(937 KB) - BibTeX
- EndNote
Helen Weierbach et al.
Status: final response (author comments only)
-
CC1: 'Review of acp-2023-54 by Alan Robock', Alan Robock, 31 Mar 2023
Review of Weierbach et al.
I recommend rejection or major revisions. The paper purports to be a set of calculations that follows the VolMIP protocol for volc-pinatubo-full simulations (Zanchettin et al., 2016). But unfortunately, they were not done correctly. The paper kept mentioning “background conditions” for the experiments, but did this mean the conditions at the time of the eruption? Both ENSO and NAO actually evolve in response to eruptions, so they are not background. I kept thinking, shouldn’t it be “initial conditions?” Then I figured out the problem. Zanchettin et al. (2016), of which I am a co-author, says:
“Initialization is based on equally distributed predefined states of ENSO (cold/neutral/warm states) and of the North Atlantic Oscillation (NAO, negative/neutral/positive states). … The recommended ENSO index is the NH winter (DJF, with January as reference for the year) Nino3.4 sea-surface temperature index, defined as the spatially averaged, winter-average sea-surface temperature over the region bounded by 120–170° W and 5° S–5° N. The recommended NAO index is calculated based on the latitude–longitude two-box method by Stephenson et al. (2006) applied on Z500 data, i.e., as the pressure difference between spatial averages over (20–55° N; 90° W–60° E) and (55–90° N; 90° W–60° E).”
However in this paper, the ENSO and NAO states were chosen for the year AFTER the eruptions, not for the initial conditions. This is a completely different experiment, and the authors do not explain why they did it that way.
It also seems that the GISS model does not allow radiative heating of the stratosphere, which would change stratospheric circulation and affect the AO and NAO. The model also cannot produce the observed El Niño after the 1991 Pinatubo eruption.
For these reasons, the results and the conclusions in this paper cannot be supported.
Any revision would have to address the concerns below and also each of the 48 comments in the attached annotated manuscript. (Comments were stopped after line 270 because of the erroneous results.
This paper left out at least five important references, which give conflicting results to the conclusions here.
Lines 63-66: You left out Zambri and Robock (2019), who showed (their Fig. 15) that no matter what the initial ENSO state, the WACCM model, in response to the 1783 Laki eruption, shows an increase SST in the Niño3.4 region of 0.5-1.0°C.
Lines 70-76: You left out Coupe et al. (2021), who showed that cooling of the Maritime Continent and tropical Africa produced an El Niño response when forced with soot aerosols in the stratosphere.
Line 88: You left out three important papers.
Zambri and Robock (2016) shows that if you look at just the first winter after large eruptions since 1850 in the Coupled Model Intercomparison Project 5 historical simulations, most models do produce a winter warming signal, with warmer temperatures over NH continents and a stronger polar vortex in the lower stratosphere. Zambri et al. (2017), which was written by one of the authors of the paper being reviewed here, showed the same thing for the last millennium.
Coupe and Robock (2021) showed that when there is an El Niño in the winter after a large volcanic eruption, as there was in observations after the 1982 El Chichón and 1991 Pinatubo eruptions, the NCAR CAM5 AMIP Large Ensemble shows winter warming for every ensemble member (their Fig. 1).
Lines 151-152: I don’t understand how specific years were sampled for ENSO and NAO conditions. Each of these has time scales that span different years and are usually stronger in NH winter. So how were the years identified with respect to the phase of each of these phenomena? Is there attention paid to the phase being strong at the time of the simulated eruption?
Lines 170-174: The technique of comparing simulations that start with identical initial conditions, but with and without volcanic eruptions, will not give results that identify the effects of volcanic eruptions unless multiple ensemble members are used for each experiment, because natural weather variability (chaos) will also be a large part of the differences. Yes, the weather will be the same for a few days, but will evolve differently, so how can you determine which is causing the differences in the pairs, forcing or internal variability?
I don’t understand the NAO results at all. Did your implementation of volcanic aerosols allow them to heat the stratosphere in the Tropics? If not, you did not force the climate system correctly. This tropical heating should produce a positive NAO, which should produce winter warming because of the increased polar vortex. See Coupe and Robock (2021).
References
Coupe, Joshua, and Alan Robock, 2021: The influence of stratospheric soot and sulfate aerosols on the Northern Hemisphere wintertime atmospheric circulation. J. Geophys. Res. Atmos., 126, e2020JD034513, doi:10.1029/2020JD034513.
Coupe, Joshua, Samantha Stevenson, Nicole S. Lovenduski, Tyler Rohr, Cheryl S. Harrison, Alan Robock, Holly Olivarez, Charles G. Bardeen, and Owen B. Toon, 2021: Nuclear Niño response observed in simulations of nuclear war scenarios. Communications Earth & Environment, 2, 18, doi:10.1038/s43247-020-00088-1.
Zambri, Brian, and Alan Robock, 2016: Winter warming and summer monsoon reduction after volcanic eruptions in Coupled Model Intercomparison Project 5 (CMIP5) simulations. Geophys. Res. Lett., 43, 10,920-10,928, doi:10.1002/2016GL070460.
Zambri, Brian, Allegra N. LeGrande, Alan Robock, and Joanna Slawinska, 2017: Northern Hemisphere winter warming and summer monsoon reduction after volcanic eruptions over the last millennium. J. Geophys. Res. Atmos., 122, 7971-7989, doi:10.1002/2017JD026728.
Zambri, Brian, Alan Robock, Michael J. Mills, and Anja Schmidt, 2019: Modeling the 1783–1784 Laki eruption in Iceland, Part II: Climate impacts. J. Geophys. Res. Atmos., 124, 6770-6790, doi:10.1029/2018JD029554.
Review by Alan Robock
-
RC1: 'Comment on acp-2023-54', Davide Zanchettin, 07 Apr 2023
I read with interest the manuscript by Weierbach et al. and I think it could be a valuable contribution to the VolMIP special issue, but pending revisions as detailed below. In my opinion, the revision should account for improvements in methodology as well as presentation and writing.
The main element of novelty of this study is the found “dampening” of NAO anomalies by the volcanic perturbation and its consequences for the post-eruption winter warming. However, this novel aspect is not fully investigated/understood, while there are results that seem ancillary. Since this single model study builds on the multi-model experiment volc-pinatubo-full, I think it should more strongly connect to the descriptive paper of the experiment (Zanchettin et al., 2022) for the general description to then focus on the novel aspects and possible model specificities that characterize GISS-E2.1-G. So, I recommend digging more into the main result and reduce the presentation of results that seem less insightful for the focus of the study (for instance paragraphs 3.1 and 3.2).
An obvious question that raises but remain unanswered is the cause of the different stratospheric response at high-latitudes in the different NAO sub-ensembles. There is no real investigation or discussion about the possible underlying mechanisms, but this seems to me central to establish a dependency on initial conditions. Also, this limits to put the GISS results in the context of future analyses with other models contributing to the same experiment. At least, I suggest checking the literature, for instance the Toohey et al. paper, for insights. Also, the mentioned strong correlation between ENSO and NAO in GISS could be relevant in shaping post-eruption NAO variability but is not considered. This could "bias" the results, or render them model specific, so should be at least discussed. Having a large ensemble, this could be checked by further stratifying responses around both, NAO and ENSO states. The “propagation” of anomalies from the stratosphere to the troposphere is not shown but should. This seems to me a necessary step to establish that it is the stratosphere that drives the tropospheric changes under all conditions. This could be done, for instance, with a figure of vertically resolved zonal average zonal wind anomalies at different time steps, for the different initial conditions. Another valuable figure could include maps of gridded 500 hPa geopotential height anomalies linked with the different initial conditions. Such maps would illustrate possible asymmetries and specificities in both, the meridional structure of atmospheric circulation and across the sub-ensembles with different initial conditions (as mentioned above, clarify potential remote sources of anomalies over the North Atlantic).
As a last note on the NAO response, in their admittedly simpler (full-ensemble) analysis, Zanchettin et al. (2022) reported a tendency “toward positive NAO anomalies in the first post-eruption winter in GISS-E2.1-G”. An explicit discussion here about this result seems appropriate.
Further regarding analyses, as a general comment, statistical support in the assessment of differences across ensembles (or sub-ensembles) is crucial. The authors mention ANOVA at some point, but do not show the associated p-value for the significance. Significance is then reported or only mentioned only occasionally. This must be amended. Also, all figures should report the ensemble envelope, not just the mean, as this alone can be deceptive. I recommend including a section 2.2 on “data analysis” where statistical methods are described, and terminology presented (see below).
Concerning the analysis of ENSO, the fact that the authors do not identify an El Nino-like response is very likely linked to the fact that the Nino3.4 index “as is” includes the volcanically induced cooling of the whole tropics, which must therefore be removed before investigating dynamical responses of ENSO. The most used approach is based on “relative SST” and is discussed in several papers, for instance Khodri et al. (2017) and Zanchettin et al. (2022). I strongly recommend the authors to revise the ENSO analysis to account for this. Note that using the relative SST method, Zanchettin et al. (2022) report the GISS-E2.1-G “showing a slight warm ENSO anomaly in 1992 in the ensemble-mean”, so contrasting the result reported here in this version of the manuscript.
Then, I read the comment by Alan Robock, and I agree that the phrasing may lead to some misunderstanding especially for those not familiar with the VolMIP protocol. This seems to me highlighted in this manuscript as the wording is often vague and does not stick to a well defined terminology.
The issue: For the volc-pinatubo experiments the selection of initial conditions was based on the idea to sample initial conditions that would lead, without the eruption, to a “controlled” diversity of ENSO and NAO conditions, ultimately to avoid sampling biases of internal variability. Therefore, the sampled ENSO and NAO refers, in time, to the first post-eruption winter as correctly described in this manuscript. What is sampled, therefore, are initial states that may capture preconditions to (or developing) different states of ENSO and NAO. This issue is presented in the volc-pinatubo-full multi-model ensemble by Zanchettin et al. (2022), where the choice, by some groups, to target the last per-eruption winter rather than the first post-eruption winter is also illustrated (see section 2.2.8 and Figure 2 of Zanchettin et al., 2022).
The same paper also discusses an adjustment of the VolMIP protocol for future experiments so that “the ENSO mean state and tendency on the period from the last pre-eruption winter to the onset of the eruption is considered instead of the state during the first post-eruption winter as in the original VolMIP protocol.” So, indeed this acknowledges a potential issue that concerns the VolMIP protocol, not this specific study.
The solution I suggest: To avoid misunderstanding I would suggest using a stricter wording with clear terminology and clear definitions. I still think it is viable to use a nomenclature for the ensembles as NAO positive/NAO negative/NAO neutral, especially if in reference to Zanchettin et al. (2022): indeed, in this paper the sub-ensembles with different initial conditions are labeled as, for instance, NAO+, NAO- and NAO0. Then, the text should be simplified as much as possible by referring generally to initial conditions, rather than specific examples that could be deceptive. For instance, at line 57, instead of “i.e. what state of NAO the climate system would normally be in” (normal refers to some average…) one should write something along these lines to be accurate: “i.e., what state of NAO the climate system would be in if the eruption did not occur” or, better, simply avoid the quoted sentence. Or, at line 126/127, one could rephrase: “background ENSO and NAO” simply with “initial conditions”.
Part of the problem above is that the manuscript does not seem as polished as it should. I spotted several typos, only some of which are reported below. So, please carefully check the manuscript in the revision.
Minor/specific comments:
Line 45-50: the cited paper mainly focused on decadal changes. Another relevant paper here, focusing on ENSO and interannual time scales, is Pausata et al. (2020), already cited in other parts of the manuscript.
Line 56: only atmospheric? This should concern all aspects of the coupled Earth system, not only the atmosphere.
Line 57: maybe it could be useful to briefly introduce this feature of post-eruption climate evolution
Line 66-69: this also concerns dynamics so I would put this in the next paragraph. I think this section is a bit a back and forth between characterization of post-eruption ENSO anomalies and mechanisms, which can be confusing for a reader. I suggest some reorganization.
Line 88: large numbers of ensembles à large ensembles
Paragraph 1.1.2: The paper by Toohey et al. (2014) could be cited here, as they question volcanic aerosol heating as a dominant mechanism for the post-eruption strengthening of the polar vortex.
Line 110: typo (withing)
Line 126-129: I must admit that I struggle to understand the exact meaning of these questions. I recommend some rephrasing. For instance, everyone would agree that different initial conditions would cause some ensemble spread (inter?) in the response even to a super eruption. So, what do you mean here? Similarly, what “small variations in the climate system” means is unclear, although this sounds like a repetition of the first question. Please also report always “post-eruption” of “volcanically forced” for the sake of clarity (for instance “post-eruption change in ENSO”)
Line 136-138: Isn’t GISS-Model E2.1 a coupled climate model? It seems you address this as only the atmospheric component. Please clarify.
Line 148-154: I don’t understand these sentences and the described method. For the VolMIP protocol, combinations of ENSO and NAO states should be sampled from a control run, then the associated states should simply be used to initialize simulations including volcanic forcing. The method that is described is confusing. Also, what is a background condition and co-condition?
Lines 161-162: If you are using only 40 simulations, why do you explain all this? It is confusing and unnecessary.
Line 163: what is NINT?
Line 173-174: unclear, the term historical with respect to control is used to describe transient simulations versus unperturbed simulations, so it highlights the type of forcing used (variable, constant). The point is that paired anomalies are calculated as “step-by-step” differences, so they do not include the effect of ongoing unperturbed (or otherwise forced) variability; anomalies from climatologies are instead deviations from a time average (in this case still from the control run), so include ongoing variability. You may also refer to Zanchettin et al. (2022).
Line 181: how GISS model compares with other models is also described in detail by Zanchettin et al. (2022), based on a subset of the simulations used here, see their section 4.2. There, GISS showed some distinct characteristics compared to other models. This should be reported here. Besides, as the author used paired anomalies (as also used in Zanchettin et al., 2022, but for other analyses, not for the radiative flux anomalies) it would be interesting to discuss how this affects the estimation of uncertainties in radiative imbalances. The ensemble spread is quite small in your calculations.
Section 3.1: please provide the direction of all changes (upward or downward). I guess Figure S2 right is for downward flux?
Line 202: normal conditions are not an anomaly of zero, but a range around zero.
Figure 1: please check label (°C)
Section 3.4.1: how is the NAO index defined? Is it the same as the VolMIP protocol? Anyway, this must be reported. Also, I recommend standardizing the index (for instance using mean and variance of the control run), so that the shown changes can be expressed in terms of standard deviations, so relative to the variability of the index.
Line 256: the neutral NAO ensemble is one
Figure 4 and associated text: MSU data are observational, as far as I know. Where are these shown in Figure 4? Why is an observational dataset brought into analysis at this point, whereas for all other analyses there is no such comparison? Volc-pinatubo are idealized experiments. Of course, model-data comparisons are possible, but should be presented and discussed properly.
Line 281: please report p value. The correlation in the right panel seems to be largely determined by a stratification across sub-ensembles based on initial conditions.
Line 511: the paper has been published: https://doi.org/10.5194/gmd-15-2265-2022
Line 283: ensembles à realizations
Line 297: which historical conditions? Please be detailed in the description of the data that are used (as I suggest above, please add a section on data processing and associated terminology). I would avoid “historical” in this context and use paired anomalies and deviations from climatology. For instance, I am certain most readers would misunderstand the statement at line 393 as well.
Figure 7: what are the percentiles shown in the box-whisker plots?
Line 305-306: this sentence has no meaning to me. I understand what you want to say, but it is not what is read.
Line 315-319: these sentences are also hard to read and understand… I recommend rewriting this part.
Toohey M, Krüger K, Bittner M, Timmreck C, Schmidt H. The impact of volcanic aerosol on the Northern Hemisphere stratospheric polar vortex: mechanisms and sensitivity to forcing structure. Atmos Chem Phys. 2014;14:13063–79. doi:10.5194/acp-14-13063-2014.
Citation: https://doi.org/10.5194/acp-2023-54-RC1 -
RC2: 'Comment on acp-2023-54', Anonymous Referee #2, 11 Apr 2023
This manuscript investigated the impacts of background ENSO and NAO conditions on the responses to Pinatubo-like forcing. Specifically, the authors focus on paired anomalies of different conditions and show that the winter warming can be found with the paired anomalies. The research topic is interesting and crucial, but the provided evidence is not precise enough and several critical scientific issues are not addressed. Therefore, I do not suggest this manuscript to be published in Atmospheric Chemistry and Physics before the authors revise the manuscript to clarify their points and to discuss the issues more comprehensively.
Major comments:
- The results in this manuscript highly depend on the paired anomalies, and their definition and interpretation are not well-discussed. The authors should provide more guidance on how to interpret the paired anomalies and why the results are different from the anomalies calculated with the control.
- For the discussion related to the polar vortex and winter warming, the authors should check the histogram of the polar vortex strength since the perturbed +NAO and -NAO ensembles have a mean value close to climatology. This may indicate that the NAO state does not have a significant impact when imposing volcanic forcing. The authors should compare the histogram of the control and perturbed 81 ensembles. If the histogram looks similar, it should be considered that there is no significant impact from the chosen NAO states for this model. This is also an issue for the winter warming part.
- The discussions of control and perturbed NAO and ENSO ensembles are, in general confusing. Since the authors use “positive NAO ensemble” but do not say whether this is the volcanic forced positive NAO ensemble or the control positive NAO ensemble. The authors should make the description more intuitive and easier to follow in order to precisely deliver their arguments.
Detail comments:
- Line 8, “pair anomalies” needs to be explained.
- Line 9, “winter warming” of what?
- Lines 12-13, what does it mean for ‘relax ENSO anomaly’?
- Lines 22-24, any reference for this?
- Lines 54-55, there are papers using large ensemble to study volcanic impact, such as Zanchettin et al., (2022). Please include them and discuss the significance of this manuscript.
Zanchettin, D., Timmreck, C., Khodri, M., Schmidt, A., Toohey, M., Abe, M., ... & Weierbach, H. (2022). Effects of forcing differences and initial conditions on inter-model agreement in the VolMIP volc-pinatubo-full experiment. Geoscientific Model Development, 15(5), 2265-2292. - Lines 77-78, there are possibilities of not having El Niño response in different models, such as the aerosol distribution (Ward et al., 2021). Please includes more details of the possibilities.
Ward, B., Pausata, F. S., & Maher, N. (2021). The sensitivity of the ENSO to volcanic aerosol spatial distribution in the MPI large ensemble. In open review for ESD. Earth System Dynamics, 12, 975-996. - Lines 97-98, there should be more recent modeling studies for reference.
- Lines 180 and 183, please use the same format for the unit.
- Line 217, Khodri et al. (2017) uses relative Niño3.4 to indicate the El Niño signal, as so does some others. Please also discuss whether the El Niño signature also does not exist when considering the relative Niño3.4.
- Lines 222-223, How do the authors define “greatly” even though clear differences are found between ENSO states?
- Line 225, where is the evidence/reference for ‘not at all on background NAO phase’?
- Line 230, the anomalies in Figure 3 are confusing. Is it a pair-wise anomaly? If yes, please state it; if not, I think the anomaly is not necessary.
- Lines 230-237, the description is hard to follow, especially for the part for +NAO and -NAO. Does it simply mean that the precondition of NAO does not hold anymore after the volcanic forcing? That is, the precondition of NAO does not impact the volcanic responses in this model.
- Line 238, I cannot infer this argument from Figure 3. If, in total, the histogram really has a reduction of the strong cases, then this argument is valid, but with only Figure 3, this is not the case.
- Line 253, ‘E).averaging’?
- Section 3.4.2, are there corresponding evidence (figures) for the arguments/results?
- Lines 255-256, which pattern? And why it can lead to “is driven primarily in pressure changes over the polar region”?
- Line 262, (Miller et al.)?
- Line 262, Figure 4 should also show the control simulations of +NAO and -NAO.
- Line 265 “north of 60◦”? north of 60°N?
- Lines 269-272, I cannot follow whether the authors are discussing the control or the perturbed ensembles. This happens for the entire manuscript.
- Line 285, same as previously. The authors need to check whether the anomaly is representable. If the control +NAO has a strong signature, but the perturbed +NAO is close to climatology. Then Figure 6 may be showing the -1*control +NAO signature, meaning that the perturbed +NAO follows the Gaussian distribution and the pre-condition does not change the response of the volcanic eruption.
Citation: https://doi.org/10.5194/acp-2023-54-RC2 - The results in this manuscript highly depend on the paired anomalies, and their definition and interpretation are not well-discussed. The authors should provide more guidance on how to interpret the paired anomalies and why the results are different from the anomalies calculated with the control.
Helen Weierbach et al.
Helen Weierbach et al.
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 268 | 93 | 13 | 374 | 32 | 4 | 4 |
- HTML: 268
- PDF: 93
- XML: 13
- Total: 374
- Supplement: 32
- BibTeX: 4
- EndNote: 4
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1