Identifying climate variables that interchange with volcanic eruptions as cooling forces during the Common Era&rsquo;s ice ages

Seip, Knut Lehre; Grøn, Øyvind; Wang, Hui

doi:https://doi.org/10.5194/egusphere-2024-1874

Preprints

https://doi.org/10.5194/egusphere-2024-1874

Preprints

20 Aug 2024

| 20 Aug 2024

Identifying climate variables that interchange with volcanic eruptions as cooling forces during the Common Era’s ice ages

Knut Lehre Seip, Øyvind Grøn, and Hui Wang

Abstract. Volcanism is known to be an instigating factor for the Late Antique Little Ice Age (LALIA, 536–660) and the Little Ice Age (LIA, 1250–1850), but little is known about when the effect of volcanism ends, and which other mechanisms prolong a cold period that includes the ice-ages’ cold periods, but also continued periods with persistent cooling. Here we show, with a high-resolution lead-lag method, where the stratospheric aerosol optical depth (SAOD) generated by volcanic emissions ceases to precede the Northern Hemisphere summer temperature (NHST). We find that five climate mechanisms cool the Northern Hemisphere (percentage time in parentheses): SAOD (51 %), total solar irradiance (TSI, 2 %), the North Atlantic oscillation (NAO, 11 %), the interdecadal Pacific oscillation (IPO, 28 %) and CO₂ (16 %). The last four variables overlap, and altogether the five climate variables cover 89 % of the cold period that includes LALIA and LIA. In contrast, we find an increase in atmospheric CO₂ over a brief period just after large volcanic eruptions. During the cold period, the five variables lead NHST, are in a cooling mode, and have sufficient strength to cool the Northern Hemisphere.

Received: 21 Jun 2024 – Discussion started: 20 Aug 2024

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Preprint (PDF, 1209 KB)

Supplement (678 KB)

Download & links

Knut Lehre Seip, Øyvind Grøn, and Hui Wang

Status: closed

RC1:
'Comment on egusphere-2024-1874', Anonymous Referee #1, 29 Aug 2024

This is the third time I have reviewed this paper, earlier versions of which were submitted to Bulletin of Volcanology and Science of the Total Environment. In both cases I recommended rejection, and my recommendation is the same here. Resubmitting papers without fixing them wastes editors’ and reviewers’ time. This may not even be only the third time, as I may have not been sent other versions to review.
The fundamental problem is that a purely statistical approach is not correct for this problem. Such a statistical approach ignores the physics of the system and is confusing. Correlation is not causation. The forcing from volcanic eruptions is very short, but the response has a much larger time scale, and is non-linear. In the case of the Little Ice Age, a resetting of the state of the Arctic climate prolonged the response for decades and centuries. This is already well known. The authors claim “to our knowledge, few studies examine the timings when the effect of volcanism ceases, and other mechanisms continue the cooling effect.” But to my knowledge there are many papers that address this. For example, look at papers referenced in this submission such as Miller et al. (2012), the title of which is “Abrupt onset of the Little Ice Age triggered by volcanism and sustained by sea-ice/ocean feedbacks.” By ignoring the physical system and its response, it is not possible to understand how climate changes. By treating SAOD, TSI, NAO, IPO, and CO₂ as normalized time series, and not with units and actual radiative forcing, the results are meaningless, particularly as NAO and IPO are internal variability and not forcing. The conclusion that CO₂ changes are as important as volcanic eruptions is ridiculous, if you account for the magnitude of their forcings, except for the period since 1850 when CO₂ and other greenhouse gases dominate. The claim on lines 529-530 that “Volcanic eruptions instigated directly or indirectly increases in atmospheric CO₂ concentrations for 136 ±42 years” is completely wrong, and based on statistical artifacts of the analysis that have no basis in physics. CO₂ emissions from volcanic eruptions for the past 2000 years have been very small. They have been important for massive flood basalt eruptions, but the contribution from eruptions during the period studied is trivial.
The authors mix up external forcing and the internal climate system. Oceanic and atmospheric variations (such as the NAO) are internal responses to external forcings, such as volcanic eruptions and solar variations.
Some of the references are messed up with parts of them on different lines, such as Miller et al. (2012).

Citation: https://doi.org/10.5194/egusphere-2024-1874-RC1
- AC1: 'Response to EGUSPHERE-2024-1874', Knut Seip, 02 Sep 2024
  
  R … Resubmitting papers without fixing them wastes editors’ and reviewers’ time.
  Response. I find the introduction to the review personal, and to me this contrast with the reviewer choosing to be anonymous. Second, we have “fixed” the paper, partly based on the present reviewer’s recommendations. However, there are issues the reviewer pointed to that we do not agree with. We appreciate the opportunity to respond to the reviewer’s – sometimes harsh (the term ridiculous was used!) –comments. And for the record, there have been other reviewers that have been positive to the study.
  R The fundamental problem is that a purely statistical approach is not correct for this problem. Such a statistical approach ignores the physics of the system and is confusing.
  Response. The paper is basically statistical, but we invoke time series, e.g., the North Atlantic Oscillation (NAO) and other ocean variability series that have been discussed extensively in physical terms. The ocean variability series has been shown to impact global temperature. Wu et al. (2019) is cited in the paper line 537, but a new article in Nature corroborate this result, Samset et al. (2024, p.3):” What was special about 2023 as rather that multiple ocean basins had warm anomalies at the same time”. Apart from this, to address issues around a phenomenon from a statistical perspective, is in our opinion correct and may give valuable information on the theme studied. We will add the reference to Samset et al. (2024).
  R. Correlation is not causation.
  Response. We are not sure if the referee assumes that we believe correlation is causation. We do not. We outline our method in Section 3 line 145 and below, we list three additional techniques, in addition to ordinary linear regression (OLR), that may strengthen a causal interpretation. Furthermore, we discuss causality interferences when there are known and unknown potential causes, line 235.
  R. In the case of the Little Ice Age, a resetting of the state of the Arctic climate prolonged the response for decades and centuries. This is already well known.
  Response. We are not sure what the reviewer means by “resetting”. There is a large literature on ocean teleconnections across the globe, and changes in the Arctic climate is one element. Other variabilities have also been given prominent roles, like El Niño and the Pacific decadal oscillation.
  R The authors claim, “to our knowledge, few studies examine the timings when the effect of volcanism ceases, and other mechanisms continue the cooling effect.” But to my knowledge there are many papers that address this. …” Miller et al. (2012) is given as an example.
  Response. The study by Miller et al. (2012) identify when summer cold begins (1275), but not when it ends. There are several studies that suggest mechanisms that would continue the cooling effect, but they are generic. The Miller et al. (2012, p. 5) study says: “Climate modeling reveals one such possible feedback mechanism.” Modeling studies have their advantage in being mechanistically driven, but also their drawbacks. Examples on this were given on Line 60. In contrast, the time series we use are observed series, although often based on proxies.
  R By ignoring the physical system and its response, it is not possible to understand how climate changes.
  Response. “To understand” is a complex and interesting expression. For example, you understand a phenomenon if you can refer to something well known (freezing water increase in volume, IPO gets warmer) and accepted a causal link between the familiar and what you want to understand (the water bottle breaks, GTA increases). The water bottle example is from Campbell (1952).
  R. By treating SAOD, TSI, NAO, IPO, and CO₂as normalized time series, and not with units and actual radiative forcing, the results are meaningless.
  Response. We treat the series in a normalized form to identify correlations and lead-lag relations. We also distinguish components of the series. To convert the series to a common unit is a challenge, but we refer to studies that suggest that e.g. NAO and IPO would have the strength to affect the global temperature, line 535. Presently we could have added a study by Samset et al. (2024) who focus on ocean basins having positive anomalies at the same time and on upwelling of previously stored deeper water with high temperatures. Text will be changed, a reference to Samset et al. (2024) will be added.
  R particularly as NAO and IPO are internal variability and not forcing.
  Response. NAO (indirectly) and IPO show characteristics of surface water temperatures in the Atlantic and the Pacific respectively, but they are also part of a teleconnection system and therefore exert force on other part of the global surface temperature system. (NAO is measured as sea surface pressure difference, but it is associated with sea surface temperatures in the literature). If GTA is restricted to the surface system, the oceanic deep waters will be a force outside this system. However, the referee is right, presently all the ocean waters are treated as one system. Text will be changed.
  R. The conclusion that CO₂changes are as important as volcanic eruptions is ridiculous, if your account for the magnitude of their forcings, except for the period since 1850 when CO₂and other greenhouse gases dominate.
  Response.The role of CO₂ in the atmosphere before, say 1850, has been revised in recent literature. It is probably not as small as previously believed. For example, Hansen et al. (2023) states that “We infer from paleoclimate data that aerosol cooling offset GHG warming for several millennia as civilization developed” The authors suggest that the effects of aerosols and GHG were comparable. However, their suggestions are discussed further in subsequent literature.
  R. The claim on lines 529-530 that “Volcanic eruptions instigated directly or indirectly increases in atmospheric CO₂concentrations for 136 ±42 years” is completely wrong and based on statistical artifacts of the analysis that have no basis in physics.
  Response. We give the source for the time series for CO₂ around line 140. To our knowledge, it is developed without references to volcanic emissions. Figure 3 shows that following three (pre-1850) volcanic eruptions, atmospheric CO₂ concentrations increase giving the numbers 136± 42 years. The CO₂ curve in the Figure is a slightly LOESS (0.1,2) smoothed version of the raw curve (Line 228, to avoid possible high frequency noise) and the droplines shows volcanic eruptions. KLS remade the calculations the result is correct, and the calculations are available from him.) We do not know what artefacts the referee refers to.
  R CO₂emissions from volcanic eruptions for the past 2000 years have been very small. They have been important for massive flood basalt eruptions, but the contribution from eruptions during the period studied is trivial.
  Response. We quote Werner et al. (2019) and Buono et al. (2023). We should not have restricted the Werner reference to one figure. Did we misunderstand the information in the Werner article? Text will be changed if the sources do not support our claim.
  R The authors mix up external forcing and the internal climate system. Oceanic and atmospheric variations (such as the NAO) are internal responses to external forcings, such as volcanic eruptions and solar variations.
  Response.. The GTA is a global surface phenomenon with measurements at the surface. If “the system” is the surface system, then the deep ocean waters are external to this system. This was the case for many early publications. It is also rational; the surface waters have been monitored for many years ≈ 1850, whereas deep waters have only recently, ≈ 1990, been monitored and understood, e.g., (Cheng et al., 2024). We believe the referee’s second statement is a conjecture. The causes of variations in NAO are not known, although several mechanisms have been proposed. (Meehl et al., 2013; Wu et al., 2019).
  R. Some of the references are messed up with parts of them on different lines, such as Miller et al. (2012).
  Response. We will take care of this.
  Literature
  Buono, G., Caliro, S., Paonita, A., Pappalardo, L., & Chiodini, G. (2023). Discriminating carbon dioxide sources during volcanic unrest: The case of Campi Flegrei caldera (Italy). Geology, 51(4), 397-401. https://doi.org/10.1130/G50624.1
  Campbell, N. R. (1952). What is science? Dover Publications.
  Cheng, L. J., Abraham, J., Trenberth, K. E., Boyer, T., Mann, M. E., Zhu, J., Wang, F., Yu, F. J., Locarnini, R., Fasullo, J., Zheng, F., Li, Y. L., Zhang, B., Wan, L. Y., Chen, X. R., Wang, D. K., Feng, L. C., Song, X. Z., Liu, Y. L., . . . Lu, Y. Y. (2024). New Record Ocean Temperatures and Related Climate Indicators in 2023. Advances in Atmospheric Sciences, 41(6), 1068-1082. https://doi.org/10.1007/s00376-024-3378-5
  Hansen, J. E., Sato, M., Simons, L., Nazarenko, L. S., von Schuckmann, K., Loeb, N. G., Osman, M. B., Pushker Kharecha, Qinjian Jin, George Tselioudis, Andrew Lacis, Reto Ruedy, 9, Gary Russell, Junji Cao, & Li11, J. (2023). Global warming in the pipeline. Oxford Open Climate Change, 3.
  Meehl, G. A., Hu, A. X., Arblaster, J. M., Fasullo, J., & Trenberth, K. E. (2013). Externally Forced and Internally Generated Decadal Climate Variability Associated with the Interdecadal Pacific Oscillation. Journal of Climate, 26(18), 7298-7310. https://doi.org/10.1175/Jcli-D-12-00548.1
  Miller, G. H., Geirsdóttir, A., Zhong, Y. F., Larsen, D. J., Otto-Bliesner, B. L., Holland, M. M., Bailey, D. A., Refsnider, K. A., Lehman, S. J., Southon, J. R., Anderson, C., Björnsson, H., & Thordarson, T. (2012). Abrupt onset of the Little Ice Age triggered by volcanism and sustained by sea-ice/ocean feedbacks. Geophysical Research Letters, 39. https://doi.org/Artn L02708 10.1029/2011gl050168
  Samset, B. H., Lund, M. T., Fuglestvedt, J. S., & Wilcox, L. J. (2024). 2023 temperatures reflect steady global warming and internal sea surface temperature variability. Communications Earth & Environment, 5(1). https://doi.org/ARTN 460 10.1038/s43247-024-01637-8
  Werner, C., Fischer, T., Aiuppa, A., Edmonds, M., Cardellini, C., Carn, S., & Allard, P. (2019). Carbon Dioxide Emissions from Subaerial Volcanic Regions: Two Decades in Review. In B. Orcutt, I. Daniel, & R. Dasgupta (Eds.), Deep Carbon: Past to Present (pp. 188-236). Cambridge University Press.
  Wu, T. W., Hu, A. X., Gao, F., Zhang, J., & Meehl, G. A. (2019). New insights into natural variability and anthropogenic forcing of global/regional climate evolution. Npj Climate and Atmospheric Science, 2. https://doi.org/UNSP 18 10.1038/s41612-019-0075-7
  
  Citation: https://doi.org/10.5194/egusphere-2024-1874-AC1
- AC2: 'Reply on RC1', Knut Seip, 14 Oct 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1874/egusphere-2024-1874-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2024-1874-AC2
RC2:
'Comment on egusphere-2024-1874', Anonymous Referee #2, 10 Oct 2024

The authors have made a commendable effort in this manuscript, presenting original approaches to determining the onset and duration of cooling episodes caused by large volcanic eruptions during the Common Era. The study is interesting and has potential offering contributions to our understanding of volcanic impacts on climate.
However, the clarity of the manuscript should be improved. At times, it is difficult to follow due to long and complex sentences. For instance, the explanation of the LL hypothesis—"The LL hypothesis comes with a caveat; if two variables impact the temperature, the second variable may distort the temperature series so that the first series appears to lag the temperature series even if it is leading"—is quite dense and could benefit from simplification.
The first line of the discussion is actually helpful, but should be moved up to the introduction.
Overall, I recommend revising sections with similarly complicated phrasing to improve readability. A clearer structure would enhance the impact of this important work and make it more accessible to the wider audience.
Despite the effort and interesting approaches presented in the manuscript, my primary concern lies with the tree-ring width (TRW) reconstruction, which forms the foundation of the study. The authors apply a loess filter to the data (what happens if you use the raw data?), but they fail to mention that the reconstruction is based on tree-ring widths, which are known to exhibit biological memory effects (see Esper et al., 2015, among many other references). This omission is critical, as memory effects in TRW can extend the cooling signal by up to 10 years, whereas the actual volcanic forcing and cooling feedbacks may be much shorter in duration.
Additionally, the issue of volcanic dating raises further concerns. While the dataset of Sigl et al. (2015) is referenced, the authors do not consider the potential impact of the double-event effect (see Tejedor et al., 2021). This is a complex matter that could potentially undermine the validity of the results.
Therefore the paper could benefit from using Northern Hemisphere reconstructions based on maximum latewood density (MXD), such as those by Schneider et al. (2017) or Büntgen et al. (2024), as MXD reconstructions are known to mitigate the memory effects associated with TRW. This shift in proxy data would likely result in more accurate reconstructions of volcanic cooling episodes.
In Figure 3, if you include the titles (which i dont know if its allowed), you should also point out which are your NHST raw temperature and which are the PCAs..

Esper et al., 2015 (Dendrochronologia). https://www.sciencedirect.com/science/article/abs/pii/S112578651500048X
Tejedor et al., 2021 (P&P). https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2020PA004128
Schneider et al., 2017 (ERL).https://iopscience.iop.org/article/10.1088/1748-9326/aa7a1b
Büntgen et al. 2024 (GPC). https://www.sciencedirect.com/science/article/pii/S0921818123003107?via%3Dihub

Citation: https://doi.org/10.5194/egusphere-2024-1874-RC2
- AC3: 'Reply on RC2', Knut Seip, 18 Oct 2024
  
  Please see attachment
  
  Citation: https://doi.org/10.5194/egusphere-2024-1874-AC3
RC3:
'Comment on egusphere-2024-1874', Anonymous Referee #3, 07 Jan 2025

The manuscript presents an attempt to disentangle cooling effects after volcanic eruptions during the common era. Technically, the authors use a quite complicated approach to reveal some already well known causes of cooling (sulfate aerosols) with some minor contributions from other factors (solar irradiance and ocean feedback). The conclusions are rather superficial and some are even wrong, because they are based on a false interpretation of results and statistical artifacts (i.e. the effect of CO₂caused by volcanic eruptions is unrealistic, because emissions are very low and could not offset the cooling caused by aerosols, at least during the period of the Common Era)
Moreover, the overall style of the manuscript is also very hard to follow using a combination of bullet points, a loose mixture and listing of ideas and concepts in the context of volcanic eruptions, combined with a method that is in my opinion not really needed and too sophisticated to investigate the questions posed.
Also the section called „1.1 Hypotheses“ is not really a clear scientific setup of rigorously testing hypotheses using a Null hypothesis and an alternative hypothesis. Phrases like „support hypothesis“ should are misleading. The hypothesis can be falsified or not on a certain level of confidence. In this context also most of the statistical significance tests and according numbers provided are somewhat useless, as the very large numbers of degrees of freedom (n> 1000= renders even very small values to be stat. significant and therefore the stat. test looses its power.
The largest weakness of the manuscript is related to the application of a pure statistical analysis technique without any inference on potential physical mechanisms involved. For instance, authors use the time series of NAO and IPO, two internally generated modes of variability that of course also have a spatial manifestation, eg. with a clear temperature seesaw dipole structure in the case of NAO and tripole in case of IPO. This is not mentioned at all but is important to infer robust conclusions based on physical considerations. Also studies addressing the physical feedback of large eruptions on the state of the NAO and other modes of variability (e.g. monsoon, ENSO) are not explored and discussed (cf. early studies of Kirchner et al. 1993, and more recent by Zambri et al., 2017 and Tejedor, 2024).
I suggest rejection of the manuscript because there is no added value and most results are already presented in former studies (also cited by the authors). Second, authors only apply an overly complicated statistical concept without any discussion on physical mechanisms giving rise to cooling effects in the aftermath of volcanic eruptions during anomalously cold periods of the Common Era.

Citation: https://doi.org/10.5194/egusphere-2024-1874-RC3
- AC4: 'Reply on RC3', Knut Seip, 10 Jan 2025
  
  Response to Reviewer # 3 for “Identifying climate variables that interchange with volcanic eruptions as cooling forces during the Common Era’s ice ages.”
  Referee
  The manuscript presents an attempt to disentangle cooling effects after volcanic eruptions during the common era.
  Response
  Thank you for the short summary. Yes, we try to disentangle the causes for low temperatures during the Late Antique Little Ice Age (LALIA, 536-660) and the Little Ice Age (LIA, 1250 – 1850).
  Referee
  Technically, the authors use a quite complicated approach…
  Response
  We are sorry that the referee finds the method we use complicated. We have tried to explain it in various ways to make the explanation as easy to understand as possible. A method description was accepted in our article on paleoclimate: Seip et al. (2018), but we think the present description is easier to understand. (When I was a graduate student, I learned about the method and remember the “the right hand” rule, see https://en.wikipedia.org/wiki/Magnetic_field, second set of figures).
  The method is essential for the study: Following the cooling caused by volcanism (sulfate aerosols), other factors (solar irradiance and ocean variabilities) have to continue the cooling to cause an ice age. However, one has to show that the cool phase for these variabilities comes after the (relatively short) periods of cold caused by volcanism. An alternative method would have been cross- correlation methods, e.g.. Kestin et al. (1998), but our method is a high-resolution lead-lag (HRLL) method and better suited for the current issues.
  Referee
  ..to reveal some already well known causes of cooling (sulfate aerosols) with some minor contributions from other factors (solar irradiance and ocean feedback).’
  Response
  Maybe we misunderstand this statement, but it cannot be minor contributions from other factors to extend the cooling caused by volcanism (lasting for, say, a maximum of two years) and ice ages, lasting for decennials.
  Referee
  The conclusions are rather superficial and some are even wrong, because they are based on a false interpretation of results and statistical artifacts (i.e., the effect of CO₂caused by volcanic eruptions is unrealistic, because emissions are very low and could not offset the cooling caused by aerosols, at least during the period of the Common Era).
  Response
  We are not sure what the referee means by “statistical artifacts”, but suppose the concern refers to Figures 3b that show time series for CO₂, the Northern hemisphere surface temperature, NHST, and dates (droplines) for large volcanic emissions. The CO₂ values and the droplines for large volcanic emissions are from independent sources, our interpretation of the figure is based on visual observations, and not any formal statistics. See below on statistics for very long series n> 1000, and with cycle characteristics. “Figure 3b shows that atmospheric CO₂ increases after volcanic eruptions…”. (There are four / five incidents of large volcanic eruptions, so a formal statistical analysis would be difficult to defend.) However, compared to the general CO₂ levels, the four/five increases in CO₂are not negligible and could potentially increase NHST, but “on the balance, the cooling effects from oceans seem to outweigh the warming effect from atmospheric CO₂.” (Lines 561-2, our manuscript).
  Referee
  Moreover, the overall style of the manuscript is also very hard to follow using a combination of bullet points, a loose mixture and listing of ideas and concepts in the context of volcanic eruptions, combined with a method that is in my opinion not really needed and too sophisticated to investigate the questions posed.
  Response
  First, the referee may indicate that the method is hard to follow, the text itself is hard to follow, or both. As mentioned above, the method should be conceptually easy to understand, but the equation may be difficult to follow. (It is used in electrical engineering and in geophysics.) The text is written the usual way scientific texts are written, and we went through the text and found no bullet points. Second, the method is needed, you must be sure that cooling effect of the candidate cooling variables continue the low temperatures during the ice ages.
  Referee
  Also the section called „1.1 Hypotheses“ is not really a clear scientific setup of rigorously testing hypotheses using a Null hypothesis and an alternative hypothesis.
  Response
  We think the hypotheses are clear and testable. We could have added criteria for when the hypotheses would be supported (or fail), but from the context this would add text that we think is superfluous. The referee also comments that with series > 1000 samples long, regressions can be difficult to interpret. A second complication is that the series we discuss has cyclic characteristics, see comments below. However, the common Era gives time series are 2000 samples long, and we see no other options for its study.
  Referee
  Phrases like „support hypothesis“ should are misleading. The hypothesis can be falsified or not on a certain level of confidence. In this context also most of the statistical significance tests and according numbers provided are somewhat useless, as the very large numbers of degrees of freedom (n> 1000= renders even very small values to be stat. significant and therefore the stat. test looses its power.
  Response
  Thank you, this is a very valid comment. We assume that most readers will be aware of the issues with very large number of samples. Furthermore, applying regressions to cyclic, or pseudo cyclic series has additional issues. For example, the regression with SAOD and NHST has 2022 samples (years), both normality test and constant variance tests fail because of the inherent cyclicities in the variables. The power with alpha equal to 0.05 is 1.00. The regression parameters will reflect both the phase shift between the cycle periods in the series and the (normal) distribution of the samples in the series once periodicity in the series has been (hypothetically) removed.[1] However, reporting regression characteristics to the series summarizes the visual relations between the series. We should have added, and will add, that the regression characteristics are guiding characteristics that summarizes the visual impressions in the relevant figures. There are other ways to support hypotheses that deal with very long time series, e.g., in analytic and experimental chemistry, but we have not worked with such methods for a very long time and do not think that they are readily applicable to climate time series.
  Referee
  The largest weakness of the manuscript is related to the application of a pure statistical analysis technique without any inference on potential physical mechanisms involved. For instance, authors use the time series of NAO and IPO, two internally generated modes of variability that of course also have a spatial manifestation, e.g., with a clear temperature seesaw dipole structure in the case of NAO and tripole in case of IPO. This is not mentioned at all but is important to infer robust conclusions based on physical considerations.
  Response
  About NAO, we write: “If SLP is high on the Iberian Peninsula and low in Iceland, NAO is designated as NAO+. The NAO+ tends to cause warmer and wetter climate over the Northern Europe, so we will use NAO- in the calculations “line 123-24. A detailed description of the spatial structure of temperatures instigated by NAO would be beyond the scope of this article. Hernández et al. (2020) uses the term “tends to”, and we believe this is as close as you can come to characterize temperatures in the North Atlantic over a 2000 year period.
  Referee
  Also, studies addressing the physical feedback of large eruptions on the state of the NAO and other modes of variability (e.g., monsoon, ENSO) are not explored and discussed (cf. early studies of Kirchner et al. 1993, and more recent by Zambri et al., 2017 and Tejedor, 2024).
  Response
  The immediate impact of volcanic eruptions would be about two years, Tejedor et al. (2024 p. 5654). Although the authors study effects of low-latitude volcanic eruptions, Tejedor et al. (2024) and Tejedor et al. (2021), they find no firm evidence for effects on NAO and ENSO. We include the two variability series (NAO, IPO) that we believe has the most pronounced effects on NHST and that have been extended 2000 years back in time. NAO is not a temperature series but has been used in many studies to characterize temperatures in North Atlantic waters. (Ma et al., 2022).
  Referee
  I suggest rejection of the manuscript because there is no added value and most results are already presented in former studies (also cited by the authors).
  Response
  We believe that our study gives added value because it restricts the contributions from climate variability series to those portions of the series that i) comes after volcanic eruptions and ii) has a cooling effect. To our knowledge, no other studies have used these additional criteria. Typical formulations in earlier literature would be “…and ocean–sea-ice feedbacks have been connected to this cooling (Lehner et al., 2013; Moreno-Chamarro et al., 2017). “van Dijk et al. (2022 p. 1602). Tejedor et al. (2021 p. 2) show that “... characterization of enhanced probabilities of El Niño. following large volcanic eruptions.” and “These studies have generally produced contractionary results...”.
  Referee
  Second, authors only apply an overly complicated statistical concept without any discussion on physical mechanisms giving rise to cooling effects in the aftermath of volcanic eruptions during anomalously cold periods of the Common Era.
  Response
  We believe the statistics is conceptually easy to understand, and not more difficult than that a climate researcher would recognize it. Our cooling mechanisms are climate variability series that have been shown to play a prominent role in cooling the NHST before about 1850, e.g. Wu et al. (2019).
  Literature
  Kestin, T. S., Karoly, D. J., Yang, J. I., & Rayner, N. A. (1998). Time-frequency variability of ENSO and stochastic simulations. Journal of Climate, 11(9), 2258-2272. https://doi.org/Doi 10.1175/1520-0442(1998)011<2258:Tfvoea>2.0.Co;2
  Ma, X., Wang, L., Smith, D., Hermanson, L., Eade, R., Dunstone, N., Hardiman, S., & Zhang, J. K. (2022). ENSO and QBO modulation of the relationship between Arctic sea ice loss and Eurasian winter climate. Environmental Research Letters, 17(12). https://doi.org/ARTN 124016 10.1088/1748-9326/aca4e9
  Seip, K. L., Gron, O., & Wang, H. (2018). Carbon dioxide precedes temperature change during short-term pauses in multi-millennial palaeoclimate records. Palaeogeography Palaeoclimatology Palaeoecology, 506, 101-111. https://doi.org/10.1016/j.palaeo.2018.06.021
  Tejedor, E., Polvani, L. M., Steiger, N. J., Vuille, M., & Smerdon, J. E. (2024). No Evidence of Winter Warming in Eurasia Following Large, Low-Latitude Volcanic Eruptions during the Last Millennium. Journal of Climate, 37(21), 5653-5673. https://doi.org/10.1175/Jcli-D-23-0625.1
  Tejedor, E., Steiger, N., Smerdon, J. E., Serrano-Notivoli, R., & Vuille, M. (2021). Global Temperature Responses to Large Tropical Volcanic Eruptions in Paleo Data Assimilation Products and Climate Model Simulations Over the Last Millennium. Paleoceanography and Paleoclimatology, 36(4). https://doi.org/ARTN e2020PA004128 0.1029/2020PA004128
  van Dijk, E., Jungclaus, J., Lorenz, S., Timmreck, C., & Kruger, K. (2022). Was there a volcanic-induced long-lasting cooling over the Northern Hemisphere in the mid-6th-7th century? Climate of the Past, 18(7), 1601-1623. https://doi.org/10.5194/cp-18-1601-2022
  Wu, T. W., Hu, A. X., Gao, F., Zhang, J., & Meehl, G. A. (2019). New insights into natural variability and anthropogenic forcing of global/regional climate evolution. Npj Climate and Atmospheric Science, 2. https://doi.org/UNSP 18
  10.1038/s41612-019-0075-7
  
  [1] Calculating new β – coefficients as suggested by Wu et al (2019), we get for the equation:” Sin (0.1t) = 0,41 * Sin (0.1t-1.57) + 1,10 * sin(0.1t+0.39)) +0.00, R = 1.00, p < 0.001, n = 101” the coefficients: β (λ/4) = 0.329794 ≈ π/8 and β (π/8) = 0.729927 ≈ π/4.
  
  Citation: https://doi.org/10.5194/egusphere-2024-1874-AC4

Status: closed

RC1:
'Comment on egusphere-2024-1874', Anonymous Referee #1, 29 Aug 2024

This is the third time I have reviewed this paper, earlier versions of which were submitted to Bulletin of Volcanology and Science of the Total Environment. In both cases I recommended rejection, and my recommendation is the same here. Resubmitting papers without fixing them wastes editors’ and reviewers’ time. This may not even be only the third time, as I may have not been sent other versions to review.
The fundamental problem is that a purely statistical approach is not correct for this problem. Such a statistical approach ignores the physics of the system and is confusing. Correlation is not causation. The forcing from volcanic eruptions is very short, but the response has a much larger time scale, and is non-linear. In the case of the Little Ice Age, a resetting of the state of the Arctic climate prolonged the response for decades and centuries. This is already well known. The authors claim “to our knowledge, few studies examine the timings when the effect of volcanism ceases, and other mechanisms continue the cooling effect.” But to my knowledge there are many papers that address this. For example, look at papers referenced in this submission such as Miller et al. (2012), the title of which is “Abrupt onset of the Little Ice Age triggered by volcanism and sustained by sea-ice/ocean feedbacks.” By ignoring the physical system and its response, it is not possible to understand how climate changes. By treating SAOD, TSI, NAO, IPO, and CO₂ as normalized time series, and not with units and actual radiative forcing, the results are meaningless, particularly as NAO and IPO are internal variability and not forcing. The conclusion that CO₂ changes are as important as volcanic eruptions is ridiculous, if you account for the magnitude of their forcings, except for the period since 1850 when CO₂ and other greenhouse gases dominate. The claim on lines 529-530 that “Volcanic eruptions instigated directly or indirectly increases in atmospheric CO₂ concentrations for 136 ±42 years” is completely wrong, and based on statistical artifacts of the analysis that have no basis in physics. CO₂ emissions from volcanic eruptions for the past 2000 years have been very small. They have been important for massive flood basalt eruptions, but the contribution from eruptions during the period studied is trivial.
The authors mix up external forcing and the internal climate system. Oceanic and atmospheric variations (such as the NAO) are internal responses to external forcings, such as volcanic eruptions and solar variations.
Some of the references are messed up with parts of them on different lines, such as Miller et al. (2012).

Citation: https://doi.org/10.5194/egusphere-2024-1874-RC1
- AC1: 'Response to EGUSPHERE-2024-1874', Knut Seip, 02 Sep 2024
  
  R … Resubmitting papers without fixing them wastes editors’ and reviewers’ time.
  Response. I find the introduction to the review personal, and to me this contrast with the reviewer choosing to be anonymous. Second, we have “fixed” the paper, partly based on the present reviewer’s recommendations. However, there are issues the reviewer pointed to that we do not agree with. We appreciate the opportunity to respond to the reviewer’s – sometimes harsh (the term ridiculous was used!) –comments. And for the record, there have been other reviewers that have been positive to the study.
  R The fundamental problem is that a purely statistical approach is not correct for this problem. Such a statistical approach ignores the physics of the system and is confusing.
  Response. The paper is basically statistical, but we invoke time series, e.g., the North Atlantic Oscillation (NAO) and other ocean variability series that have been discussed extensively in physical terms. The ocean variability series has been shown to impact global temperature. Wu et al. (2019) is cited in the paper line 537, but a new article in Nature corroborate this result, Samset et al. (2024, p.3):” What was special about 2023 as rather that multiple ocean basins had warm anomalies at the same time”. Apart from this, to address issues around a phenomenon from a statistical perspective, is in our opinion correct and may give valuable information on the theme studied. We will add the reference to Samset et al. (2024).
  R. Correlation is not causation.
  Response. We are not sure if the referee assumes that we believe correlation is causation. We do not. We outline our method in Section 3 line 145 and below, we list three additional techniques, in addition to ordinary linear regression (OLR), that may strengthen a causal interpretation. Furthermore, we discuss causality interferences when there are known and unknown potential causes, line 235.
  R. In the case of the Little Ice Age, a resetting of the state of the Arctic climate prolonged the response for decades and centuries. This is already well known.
  Response. We are not sure what the reviewer means by “resetting”. There is a large literature on ocean teleconnections across the globe, and changes in the Arctic climate is one element. Other variabilities have also been given prominent roles, like El Niño and the Pacific decadal oscillation.
  R The authors claim, “to our knowledge, few studies examine the timings when the effect of volcanism ceases, and other mechanisms continue the cooling effect.” But to my knowledge there are many papers that address this. …” Miller et al. (2012) is given as an example.
  Response. The study by Miller et al. (2012) identify when summer cold begins (1275), but not when it ends. There are several studies that suggest mechanisms that would continue the cooling effect, but they are generic. The Miller et al. (2012, p. 5) study says: “Climate modeling reveals one such possible feedback mechanism.” Modeling studies have their advantage in being mechanistically driven, but also their drawbacks. Examples on this were given on Line 60. In contrast, the time series we use are observed series, although often based on proxies.
  R By ignoring the physical system and its response, it is not possible to understand how climate changes.
  Response. “To understand” is a complex and interesting expression. For example, you understand a phenomenon if you can refer to something well known (freezing water increase in volume, IPO gets warmer) and accepted a causal link between the familiar and what you want to understand (the water bottle breaks, GTA increases). The water bottle example is from Campbell (1952).
  R. By treating SAOD, TSI, NAO, IPO, and CO₂as normalized time series, and not with units and actual radiative forcing, the results are meaningless.
  Response. We treat the series in a normalized form to identify correlations and lead-lag relations. We also distinguish components of the series. To convert the series to a common unit is a challenge, but we refer to studies that suggest that e.g. NAO and IPO would have the strength to affect the global temperature, line 535. Presently we could have added a study by Samset et al. (2024) who focus on ocean basins having positive anomalies at the same time and on upwelling of previously stored deeper water with high temperatures. Text will be changed, a reference to Samset et al. (2024) will be added.
  R particularly as NAO and IPO are internal variability and not forcing.
  Response. NAO (indirectly) and IPO show characteristics of surface water temperatures in the Atlantic and the Pacific respectively, but they are also part of a teleconnection system and therefore exert force on other part of the global surface temperature system. (NAO is measured as sea surface pressure difference, but it is associated with sea surface temperatures in the literature). If GTA is restricted to the surface system, the oceanic deep waters will be a force outside this system. However, the referee is right, presently all the ocean waters are treated as one system. Text will be changed.
  R. The conclusion that CO₂changes are as important as volcanic eruptions is ridiculous, if your account for the magnitude of their forcings, except for the period since 1850 when CO₂and other greenhouse gases dominate.
  Response.The role of CO₂ in the atmosphere before, say 1850, has been revised in recent literature. It is probably not as small as previously believed. For example, Hansen et al. (2023) states that “We infer from paleoclimate data that aerosol cooling offset GHG warming for several millennia as civilization developed” The authors suggest that the effects of aerosols and GHG were comparable. However, their suggestions are discussed further in subsequent literature.
  R. The claim on lines 529-530 that “Volcanic eruptions instigated directly or indirectly increases in atmospheric CO₂concentrations for 136 ±42 years” is completely wrong and based on statistical artifacts of the analysis that have no basis in physics.
  Response. We give the source for the time series for CO₂ around line 140. To our knowledge, it is developed without references to volcanic emissions. Figure 3 shows that following three (pre-1850) volcanic eruptions, atmospheric CO₂ concentrations increase giving the numbers 136± 42 years. The CO₂ curve in the Figure is a slightly LOESS (0.1,2) smoothed version of the raw curve (Line 228, to avoid possible high frequency noise) and the droplines shows volcanic eruptions. KLS remade the calculations the result is correct, and the calculations are available from him.) We do not know what artefacts the referee refers to.
  R CO₂emissions from volcanic eruptions for the past 2000 years have been very small. They have been important for massive flood basalt eruptions, but the contribution from eruptions during the period studied is trivial.
  Response. We quote Werner et al. (2019) and Buono et al. (2023). We should not have restricted the Werner reference to one figure. Did we misunderstand the information in the Werner article? Text will be changed if the sources do not support our claim.
  R The authors mix up external forcing and the internal climate system. Oceanic and atmospheric variations (such as the NAO) are internal responses to external forcings, such as volcanic eruptions and solar variations.
  Response.. The GTA is a global surface phenomenon with measurements at the surface. If “the system” is the surface system, then the deep ocean waters are external to this system. This was the case for many early publications. It is also rational; the surface waters have been monitored for many years ≈ 1850, whereas deep waters have only recently, ≈ 1990, been monitored and understood, e.g., (Cheng et al., 2024). We believe the referee’s second statement is a conjecture. The causes of variations in NAO are not known, although several mechanisms have been proposed. (Meehl et al., 2013; Wu et al., 2019).
  R. Some of the references are messed up with parts of them on different lines, such as Miller et al. (2012).
  Response. We will take care of this.
  Literature
  Buono, G., Caliro, S., Paonita, A., Pappalardo, L., & Chiodini, G. (2023). Discriminating carbon dioxide sources during volcanic unrest: The case of Campi Flegrei caldera (Italy). Geology, 51(4), 397-401. https://doi.org/10.1130/G50624.1
  Campbell, N. R. (1952). What is science? Dover Publications.
  Cheng, L. J., Abraham, J., Trenberth, K. E., Boyer, T., Mann, M. E., Zhu, J., Wang, F., Yu, F. J., Locarnini, R., Fasullo, J., Zheng, F., Li, Y. L., Zhang, B., Wan, L. Y., Chen, X. R., Wang, D. K., Feng, L. C., Song, X. Z., Liu, Y. L., . . . Lu, Y. Y. (2024). New Record Ocean Temperatures and Related Climate Indicators in 2023. Advances in Atmospheric Sciences, 41(6), 1068-1082. https://doi.org/10.1007/s00376-024-3378-5
  Hansen, J. E., Sato, M., Simons, L., Nazarenko, L. S., von Schuckmann, K., Loeb, N. G., Osman, M. B., Pushker Kharecha, Qinjian Jin, George Tselioudis, Andrew Lacis, Reto Ruedy, 9, Gary Russell, Junji Cao, & Li11, J. (2023). Global warming in the pipeline. Oxford Open Climate Change, 3.
  Meehl, G. A., Hu, A. X., Arblaster, J. M., Fasullo, J., & Trenberth, K. E. (2013). Externally Forced and Internally Generated Decadal Climate Variability Associated with the Interdecadal Pacific Oscillation. Journal of Climate, 26(18), 7298-7310. https://doi.org/10.1175/Jcli-D-12-00548.1
  Miller, G. H., Geirsdóttir, A., Zhong, Y. F., Larsen, D. J., Otto-Bliesner, B. L., Holland, M. M., Bailey, D. A., Refsnider, K. A., Lehman, S. J., Southon, J. R., Anderson, C., Björnsson, H., & Thordarson, T. (2012). Abrupt onset of the Little Ice Age triggered by volcanism and sustained by sea-ice/ocean feedbacks. Geophysical Research Letters, 39. https://doi.org/Artn L02708 10.1029/2011gl050168
  Samset, B. H., Lund, M. T., Fuglestvedt, J. S., & Wilcox, L. J. (2024). 2023 temperatures reflect steady global warming and internal sea surface temperature variability. Communications Earth & Environment, 5(1). https://doi.org/ARTN 460 10.1038/s43247-024-01637-8
  Werner, C., Fischer, T., Aiuppa, A., Edmonds, M., Cardellini, C., Carn, S., & Allard, P. (2019). Carbon Dioxide Emissions from Subaerial Volcanic Regions: Two Decades in Review. In B. Orcutt, I. Daniel, & R. Dasgupta (Eds.), Deep Carbon: Past to Present (pp. 188-236). Cambridge University Press.
  Wu, T. W., Hu, A. X., Gao, F., Zhang, J., & Meehl, G. A. (2019). New insights into natural variability and anthropogenic forcing of global/regional climate evolution. Npj Climate and Atmospheric Science, 2. https://doi.org/UNSP 18 10.1038/s41612-019-0075-7
  
  Citation: https://doi.org/10.5194/egusphere-2024-1874-AC1
- AC2: 'Reply on RC1', Knut Seip, 14 Oct 2024
  
  The comment was uploaded in the form of a supplement: https://egusphere.copernicus.org/preprints/2024/egusphere-2024-1874/egusphere-2024-1874-AC2-supplement.pdf
  
  Citation: https://doi.org/10.5194/egusphere-2024-1874-AC2
RC2:
'Comment on egusphere-2024-1874', Anonymous Referee #2, 10 Oct 2024

The authors have made a commendable effort in this manuscript, presenting original approaches to determining the onset and duration of cooling episodes caused by large volcanic eruptions during the Common Era. The study is interesting and has potential offering contributions to our understanding of volcanic impacts on climate.
However, the clarity of the manuscript should be improved. At times, it is difficult to follow due to long and complex sentences. For instance, the explanation of the LL hypothesis—"The LL hypothesis comes with a caveat; if two variables impact the temperature, the second variable may distort the temperature series so that the first series appears to lag the temperature series even if it is leading"—is quite dense and could benefit from simplification.
The first line of the discussion is actually helpful, but should be moved up to the introduction.
Overall, I recommend revising sections with similarly complicated phrasing to improve readability. A clearer structure would enhance the impact of this important work and make it more accessible to the wider audience.
Despite the effort and interesting approaches presented in the manuscript, my primary concern lies with the tree-ring width (TRW) reconstruction, which forms the foundation of the study. The authors apply a loess filter to the data (what happens if you use the raw data?), but they fail to mention that the reconstruction is based on tree-ring widths, which are known to exhibit biological memory effects (see Esper et al., 2015, among many other references). This omission is critical, as memory effects in TRW can extend the cooling signal by up to 10 years, whereas the actual volcanic forcing and cooling feedbacks may be much shorter in duration.
Additionally, the issue of volcanic dating raises further concerns. While the dataset of Sigl et al. (2015) is referenced, the authors do not consider the potential impact of the double-event effect (see Tejedor et al., 2021). This is a complex matter that could potentially undermine the validity of the results.
Therefore the paper could benefit from using Northern Hemisphere reconstructions based on maximum latewood density (MXD), such as those by Schneider et al. (2017) or Büntgen et al. (2024), as MXD reconstructions are known to mitigate the memory effects associated with TRW. This shift in proxy data would likely result in more accurate reconstructions of volcanic cooling episodes.
In Figure 3, if you include the titles (which i dont know if its allowed), you should also point out which are your NHST raw temperature and which are the PCAs..

Esper et al., 2015 (Dendrochronologia). https://www.sciencedirect.com/science/article/abs/pii/S112578651500048X
Tejedor et al., 2021 (P&P). https://agupubs.onlinelibrary.wiley.com/doi/full/10.1029/2020PA004128
Schneider et al., 2017 (ERL).https://iopscience.iop.org/article/10.1088/1748-9326/aa7a1b
Büntgen et al. 2024 (GPC). https://www.sciencedirect.com/science/article/pii/S0921818123003107?via%3Dihub

Citation: https://doi.org/10.5194/egusphere-2024-1874-RC2
- AC3: 'Reply on RC2', Knut Seip, 18 Oct 2024
  
  Please see attachment
  
  Citation: https://doi.org/10.5194/egusphere-2024-1874-AC3
RC3:
'Comment on egusphere-2024-1874', Anonymous Referee #3, 07 Jan 2025

The manuscript presents an attempt to disentangle cooling effects after volcanic eruptions during the common era. Technically, the authors use a quite complicated approach to reveal some already well known causes of cooling (sulfate aerosols) with some minor contributions from other factors (solar irradiance and ocean feedback). The conclusions are rather superficial and some are even wrong, because they are based on a false interpretation of results and statistical artifacts (i.e. the effect of CO₂caused by volcanic eruptions is unrealistic, because emissions are very low and could not offset the cooling caused by aerosols, at least during the period of the Common Era)
Moreover, the overall style of the manuscript is also very hard to follow using a combination of bullet points, a loose mixture and listing of ideas and concepts in the context of volcanic eruptions, combined with a method that is in my opinion not really needed and too sophisticated to investigate the questions posed.
Also the section called „1.1 Hypotheses“ is not really a clear scientific setup of rigorously testing hypotheses using a Null hypothesis and an alternative hypothesis. Phrases like „support hypothesis“ should are misleading. The hypothesis can be falsified or not on a certain level of confidence. In this context also most of the statistical significance tests and according numbers provided are somewhat useless, as the very large numbers of degrees of freedom (n> 1000= renders even very small values to be stat. significant and therefore the stat. test looses its power.
The largest weakness of the manuscript is related to the application of a pure statistical analysis technique without any inference on potential physical mechanisms involved. For instance, authors use the time series of NAO and IPO, two internally generated modes of variability that of course also have a spatial manifestation, eg. with a clear temperature seesaw dipole structure in the case of NAO and tripole in case of IPO. This is not mentioned at all but is important to infer robust conclusions based on physical considerations. Also studies addressing the physical feedback of large eruptions on the state of the NAO and other modes of variability (e.g. monsoon, ENSO) are not explored and discussed (cf. early studies of Kirchner et al. 1993, and more recent by Zambri et al., 2017 and Tejedor, 2024).
I suggest rejection of the manuscript because there is no added value and most results are already presented in former studies (also cited by the authors). Second, authors only apply an overly complicated statistical concept without any discussion on physical mechanisms giving rise to cooling effects in the aftermath of volcanic eruptions during anomalously cold periods of the Common Era.

Citation: https://doi.org/10.5194/egusphere-2024-1874-RC3
- AC4: 'Reply on RC3', Knut Seip, 10 Jan 2025
  
  Response to Reviewer # 3 for “Identifying climate variables that interchange with volcanic eruptions as cooling forces during the Common Era’s ice ages.”
  Referee
  The manuscript presents an attempt to disentangle cooling effects after volcanic eruptions during the common era.
  Response
  Thank you for the short summary. Yes, we try to disentangle the causes for low temperatures during the Late Antique Little Ice Age (LALIA, 536-660) and the Little Ice Age (LIA, 1250 – 1850).
  Referee
  Technically, the authors use a quite complicated approach…
  Response
  We are sorry that the referee finds the method we use complicated. We have tried to explain it in various ways to make the explanation as easy to understand as possible. A method description was accepted in our article on paleoclimate: Seip et al. (2018), but we think the present description is easier to understand. (When I was a graduate student, I learned about the method and remember the “the right hand” rule, see https://en.wikipedia.org/wiki/Magnetic_field, second set of figures).
  The method is essential for the study: Following the cooling caused by volcanism (sulfate aerosols), other factors (solar irradiance and ocean variabilities) have to continue the cooling to cause an ice age. However, one has to show that the cool phase for these variabilities comes after the (relatively short) periods of cold caused by volcanism. An alternative method would have been cross- correlation methods, e.g.. Kestin et al. (1998), but our method is a high-resolution lead-lag (HRLL) method and better suited for the current issues.
  Referee
  ..to reveal some already well known causes of cooling (sulfate aerosols) with some minor contributions from other factors (solar irradiance and ocean feedback).’
  Response
  Maybe we misunderstand this statement, but it cannot be minor contributions from other factors to extend the cooling caused by volcanism (lasting for, say, a maximum of two years) and ice ages, lasting for decennials.
  Referee
  The conclusions are rather superficial and some are even wrong, because they are based on a false interpretation of results and statistical artifacts (i.e., the effect of CO₂caused by volcanic eruptions is unrealistic, because emissions are very low and could not offset the cooling caused by aerosols, at least during the period of the Common Era).
  Response
  We are not sure what the referee means by “statistical artifacts”, but suppose the concern refers to Figures 3b that show time series for CO₂, the Northern hemisphere surface temperature, NHST, and dates (droplines) for large volcanic emissions. The CO₂ values and the droplines for large volcanic emissions are from independent sources, our interpretation of the figure is based on visual observations, and not any formal statistics. See below on statistics for very long series n> 1000, and with cycle characteristics. “Figure 3b shows that atmospheric CO₂ increases after volcanic eruptions…”. (There are four / five incidents of large volcanic eruptions, so a formal statistical analysis would be difficult to defend.) However, compared to the general CO₂ levels, the four/five increases in CO₂are not negligible and could potentially increase NHST, but “on the balance, the cooling effects from oceans seem to outweigh the warming effect from atmospheric CO₂.” (Lines 561-2, our manuscript).
  Referee
  Moreover, the overall style of the manuscript is also very hard to follow using a combination of bullet points, a loose mixture and listing of ideas and concepts in the context of volcanic eruptions, combined with a method that is in my opinion not really needed and too sophisticated to investigate the questions posed.
  Response
  First, the referee may indicate that the method is hard to follow, the text itself is hard to follow, or both. As mentioned above, the method should be conceptually easy to understand, but the equation may be difficult to follow. (It is used in electrical engineering and in geophysics.) The text is written the usual way scientific texts are written, and we went through the text and found no bullet points. Second, the method is needed, you must be sure that cooling effect of the candidate cooling variables continue the low temperatures during the ice ages.
  Referee
  Also the section called „1.1 Hypotheses“ is not really a clear scientific setup of rigorously testing hypotheses using a Null hypothesis and an alternative hypothesis.
  Response
  We think the hypotheses are clear and testable. We could have added criteria for when the hypotheses would be supported (or fail), but from the context this would add text that we think is superfluous. The referee also comments that with series > 1000 samples long, regressions can be difficult to interpret. A second complication is that the series we discuss has cyclic characteristics, see comments below. However, the common Era gives time series are 2000 samples long, and we see no other options for its study.
  Referee
  Phrases like „support hypothesis“ should are misleading. The hypothesis can be falsified or not on a certain level of confidence. In this context also most of the statistical significance tests and according numbers provided are somewhat useless, as the very large numbers of degrees of freedom (n> 1000= renders even very small values to be stat. significant and therefore the stat. test looses its power.
  Response
  Thank you, this is a very valid comment. We assume that most readers will be aware of the issues with very large number of samples. Furthermore, applying regressions to cyclic, or pseudo cyclic series has additional issues. For example, the regression with SAOD and NHST has 2022 samples (years), both normality test and constant variance tests fail because of the inherent cyclicities in the variables. The power with alpha equal to 0.05 is 1.00. The regression parameters will reflect both the phase shift between the cycle periods in the series and the (normal) distribution of the samples in the series once periodicity in the series has been (hypothetically) removed.[1] However, reporting regression characteristics to the series summarizes the visual relations between the series. We should have added, and will add, that the regression characteristics are guiding characteristics that summarizes the visual impressions in the relevant figures. There are other ways to support hypotheses that deal with very long time series, e.g., in analytic and experimental chemistry, but we have not worked with such methods for a very long time and do not think that they are readily applicable to climate time series.
  Referee
  The largest weakness of the manuscript is related to the application of a pure statistical analysis technique without any inference on potential physical mechanisms involved. For instance, authors use the time series of NAO and IPO, two internally generated modes of variability that of course also have a spatial manifestation, e.g., with a clear temperature seesaw dipole structure in the case of NAO and tripole in case of IPO. This is not mentioned at all but is important to infer robust conclusions based on physical considerations.
  Response
  About NAO, we write: “If SLP is high on the Iberian Peninsula and low in Iceland, NAO is designated as NAO+. The NAO+ tends to cause warmer and wetter climate over the Northern Europe, so we will use NAO- in the calculations “line 123-24. A detailed description of the spatial structure of temperatures instigated by NAO would be beyond the scope of this article. Hernández et al. (2020) uses the term “tends to”, and we believe this is as close as you can come to characterize temperatures in the North Atlantic over a 2000 year period.
  Referee
  Also, studies addressing the physical feedback of large eruptions on the state of the NAO and other modes of variability (e.g., monsoon, ENSO) are not explored and discussed (cf. early studies of Kirchner et al. 1993, and more recent by Zambri et al., 2017 and Tejedor, 2024).
  Response
  The immediate impact of volcanic eruptions would be about two years, Tejedor et al. (2024 p. 5654). Although the authors study effects of low-latitude volcanic eruptions, Tejedor et al. (2024) and Tejedor et al. (2021), they find no firm evidence for effects on NAO and ENSO. We include the two variability series (NAO, IPO) that we believe has the most pronounced effects on NHST and that have been extended 2000 years back in time. NAO is not a temperature series but has been used in many studies to characterize temperatures in North Atlantic waters. (Ma et al., 2022).
  Referee
  I suggest rejection of the manuscript because there is no added value and most results are already presented in former studies (also cited by the authors).
  Response
  We believe that our study gives added value because it restricts the contributions from climate variability series to those portions of the series that i) comes after volcanic eruptions and ii) has a cooling effect. To our knowledge, no other studies have used these additional criteria. Typical formulations in earlier literature would be “…and ocean–sea-ice feedbacks have been connected to this cooling (Lehner et al., 2013; Moreno-Chamarro et al., 2017). “van Dijk et al. (2022 p. 1602). Tejedor et al. (2021 p. 2) show that “... characterization of enhanced probabilities of El Niño. following large volcanic eruptions.” and “These studies have generally produced contractionary results...”.
  Referee
  Second, authors only apply an overly complicated statistical concept without any discussion on physical mechanisms giving rise to cooling effects in the aftermath of volcanic eruptions during anomalously cold periods of the Common Era.
  Response
  We believe the statistics is conceptually easy to understand, and not more difficult than that a climate researcher would recognize it. Our cooling mechanisms are climate variability series that have been shown to play a prominent role in cooling the NHST before about 1850, e.g. Wu et al. (2019).
  Literature
  Kestin, T. S., Karoly, D. J., Yang, J. I., & Rayner, N. A. (1998). Time-frequency variability of ENSO and stochastic simulations. Journal of Climate, 11(9), 2258-2272. https://doi.org/Doi 10.1175/1520-0442(1998)011<2258:Tfvoea>2.0.Co;2
  Ma, X., Wang, L., Smith, D., Hermanson, L., Eade, R., Dunstone, N., Hardiman, S., & Zhang, J. K. (2022). ENSO and QBO modulation of the relationship between Arctic sea ice loss and Eurasian winter climate. Environmental Research Letters, 17(12). https://doi.org/ARTN 124016 10.1088/1748-9326/aca4e9
  Seip, K. L., Gron, O., & Wang, H. (2018). Carbon dioxide precedes temperature change during short-term pauses in multi-millennial palaeoclimate records. Palaeogeography Palaeoclimatology Palaeoecology, 506, 101-111. https://doi.org/10.1016/j.palaeo.2018.06.021
  Tejedor, E., Polvani, L. M., Steiger, N. J., Vuille, M., & Smerdon, J. E. (2024). No Evidence of Winter Warming in Eurasia Following Large, Low-Latitude Volcanic Eruptions during the Last Millennium. Journal of Climate, 37(21), 5653-5673. https://doi.org/10.1175/Jcli-D-23-0625.1
  Tejedor, E., Steiger, N., Smerdon, J. E., Serrano-Notivoli, R., & Vuille, M. (2021). Global Temperature Responses to Large Tropical Volcanic Eruptions in Paleo Data Assimilation Products and Climate Model Simulations Over the Last Millennium. Paleoceanography and Paleoclimatology, 36(4). https://doi.org/ARTN e2020PA004128 0.1029/2020PA004128
  van Dijk, E., Jungclaus, J., Lorenz, S., Timmreck, C., & Kruger, K. (2022). Was there a volcanic-induced long-lasting cooling over the Northern Hemisphere in the mid-6th-7th century? Climate of the Past, 18(7), 1601-1623. https://doi.org/10.5194/cp-18-1601-2022
  Wu, T. W., Hu, A. X., Gao, F., Zhang, J., & Meehl, G. A. (2019). New insights into natural variability and anthropogenic forcing of global/regional climate evolution. Npj Climate and Atmospheric Science, 2. https://doi.org/UNSP 18
  10.1038/s41612-019-0075-7
  
  [1] Calculating new β – coefficients as suggested by Wu et al (2019), we get for the equation:” Sin (0.1t) = 0,41 * Sin (0.1t-1.57) + 1,10 * sin(0.1t+0.39)) +0.00, R = 1.00, p < 0.001, n = 101” the coefficients: β (λ/4) = 0.329794 ≈ π/8 and β (π/8) = 0.729927 ≈ π/4.
  
  Citation: https://doi.org/10.5194/egusphere-2024-1874-AC4

Knut Lehre Seip, Øyvind Grøn, and Hui Wang

Supplement

https://doi.org/10.5194/egusphere-2024-1874-supplement

Knut Lehre Seip, Øyvind Grøn, and Hui Wang

Viewed

Total article views: 1,337 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	Supplement	BibTeX	EndNote
394	781	162	1,337	27	7	7

HTML: 394
PDF: 781
XML: 162
Total: 1,337
Supplement: 27
BibTeX: 7
EndNote: 7

Views and downloads (calculated since 20 Aug 2024)

Month	HTML	PDF	XML	Total
Aug 2024	81	33	3	117
Sep 2024	73	15	6	94
Oct 2024	58	29	5	92
Nov 2024	58	303	11	372
Dec 2024	98	374	83	555
Jan 2025	26	27	54	107

Cumulative views and downloads (calculated since 20 Aug 2024)

Month	HTML	PDF	XML	Total
Aug 2024	81	33	3	117
Sep 2024	73	15	6	94
Oct 2024	58	29	5	92
Nov 2024	58	303	11	372
Dec 2024	98	374	83	555
Jan 2025	26	27	54	107

Viewed (geographical distribution)

Total article views: 1,372 (including HTML, PDF, and XML) Thereof 1,372 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 21 Jan 2025

Short summary

The Little Ice age in the Northern Hemisphere is suggested to be instigated by volcanic eruptions, but cooling of the earth that is due to volcanism last for a few years whereas the little ice age lasted for 600 years. So, even if there are several volcanic eruptions, what continue the cool temperature during the ice age? Here, we use a new technique that let us know when volcanic cooling cease, and which mechanisms lead the temperature and are strong enough to cool the Northern Hemisphere.


Total:	0
HTML:	0
PDF:	0
XML:	0