the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
Volcanically Forced Cooling over the last 600 years in Northern Patagonia
Abstract. We present the first tree-ring (TR) based summer surface temperature reconstruction (1382–2017) for the Southern Hemisphere that expresses strong volcanically forced cooling. The Northern Patagonia (NPAT) reconstruction is based on ring-width (RW) and Blue Intensity (BI) parameters measured from Araucaria araucana trees sampled from 6 locations across the middle to southern end of the species' range in Argentina. This multi-TR-parameter reconstruction explains 53 % of the summer surface temperature variance (1903–2017), which is on par with the fidelity of density or mixed TR parameter-based reconstructions from the Northern Hemisphere. The reconstruction coheres strongly with surface mean air temperatures for a large region in South America, including sea surface temperatures well into the southeastern Pacific. The warmest 10-year period is 2008–2017, corresponding to the last ten years of the reconstruction, while the coldest period is 1455–1464. The coldest reconstructed austral summer is 1971, followed by 1460. For both the NPAT reconstruction and a range of model simulations, superposed epoch analysis, using significant tropical eruptions since the 1400s, indicates a significant post-eruption T+1 mean surface cooling of ca. 0.4–1.0 °C, depending on which volcanic events are used. The degree of relative cooling is on par, or even stronger, with the cooling represented by individual TR records used in the Northern Hemisphere N-TREND reconstruction suggesting that the volcanic response in northern Patagonia over the last 6 centuries is equivalent, or even more extreme, to what is observed in many Northern Hemisphere locations. Our results indicate that the use of ring-density parameters is of paramount importance for assessing past volcanically forced cooling in the Southern Hemisphere, but the dating and seasonality of the eruptions as well as the continentality and mid-latitude location of the woodland sites may also be important factors for capturing the signal of volcanic cooling.
- Preprint
(3925 KB) - Metadata XML
-
Supplement
(234 KB) - BibTeX
- EndNote
Status: final response (author comments only)
-
RC1: 'Comment on egusphere-2026-4785', Anonymous Referee #1, 14 Oct 2026
-
RC2: 'Reply on RC1', Anonymous Referee #2, 12 Nov 2026
Summary: The authors present a tree-ring reconstruction of summer temperatures from 1382–2017 using ring-width and density measurements from trees in northern Patagonia. They demonstrate that the recon captures a strong volcanic cooling signal, which they claim is a significant finding for the SH. The reconstruction reveals significant post-eruption cooling of approximately 0.4-1.0°C, comparable to responses documented at many NH sites. The findings highlight the importance of ring-density parameters for detecting past volcanic cooling in the SH, while also identifying eruption timing, seasonality, and site characteristics as important influences on the strength of the reconstructed signal.
General Remarks: This is an important contribution to the literature and a valuable collection of high-resolution temperature proxies, specifically density measurements in tree-rings, that will be of great use to the community. The analyses that the authors provide on volcanic responses in South America also prompt new questions about how the climate system responds to volcanism, and is therefore a relevant and thought-provoking finding. The paper should be published after addressing the minor and specific points that I outline below.
Specific Comments
Line 50: I find both the definitiveness and the citations for the statement that the "importance of common era hemispheric scale paleoclimate reconstructions for the attribution of drivers of past and present temperature change is unquestioned" to be a little much. I would temper the statement a bit (unquestionable is hard to defend) and consider using some more recent D&A literature, including the most recent IPCC assessment as support.
Ln 50-53: The authors are mixing a few things here. The first sentence in the paragraph talks about CE recons generally, and the second sentence shifts specifically to tree-ring based recons. The citations in the second sentence are also specific to dendro records. The uncertainties in the large-scale recons are legion and this paragraph should not present arguments necessarily specific to tree rings. The final sentence also switches back to more general NH recon work, but is still constrained to growing season interpretations specific to tree rings. There is a lot of NH and global work with multi-proxy recons that should be considered here, some of which has also looked in detail at volcanic responses. My point is that the dendro-specific framing is a little off and also overlooks a lot of multi-proxy work, that is certainly very dependent on tree-ring records, but broader than what the authors awkwardly discuss here.
Ln 66: The authors are again mixing tree-ring only recon results with multi-proxy efforts like the Neukom et al. and Tejedor et al. papers. It is true that the multi-proxy efforts strongly depend on dendro records, and ring-width specifically, but they incorporate other records and some include MXD proxies. I would also note that papers like Tejedor et al. (2021) specifically find that in the NH, the volcanic signal does not appear to be attenuated like other ring-width recons, making the argument the the authors are making about the SH harder to defend, i.e. that ring width is the source of the attenuated signal in the SH. He et al. (https://doi.org/10.1175/JCLI-D-25-0179.1) also provides a recent assessment across a host of global field reconstruction products, all of which give evidence of significant SH cooling after large tropical eruptions.
Ln 78: There are a lot of "howevers" happening in this paragraph, which makes it a little choppy to follow. More importantly, however, the description here could be clearer. The point I believe the authors are trying to make is that if an eruption happens before the austral summer, the signal may be detected in the SH prior to the expression of the cooling in boreal summer. It is merely a timing point (the reverse could also happen) and it is not clear why the authors choose to make this a relatively complicated point.
Ln 215: There are a lot of decision iterations in the approach to the reconstruction and the authors could state more clearly the recon recipe that they ultimately converged upon. In other words, what decisions were ultimately used to derive the final recon, which they use subsequently in their analyses? That said, this line gives the impression that despite various validation exercises, the final recon is simply calibrated on all of the instrumental data. I may be mistaken, but if that is the case it is unwarranted. Whatever final recon is used, should be accompanied by out of sample validation statistics. Such statistics for differently calibrated recons are innapplicable to a final calibration that uses the entire record. If I am interpreting this decision correctly, the authors need to choose a different calibration that allows them to provide applicable validation statistics that allow us to evaluate the skill of whatever calibrated version they use in the rest of their paper.
Ln 244: Clearly identify the years over which these events were identified.
Ln 255: I am confused by what the authors did to treat age uncertainty. My impression is that they choose another year within the possible age uncertainties and test their results with that shift. Why not evaluate a +/- range, given that symmetric uncertainties are typically how they are reported for the eruption years. In other words, if the eruption year is estimated to be 1808, why not evaluate both 1807 and 1809?
Figure 3: I find this figure layout confusing. The inset plots with the panel lettering makes them awkwardly placed within the larger plot. The layout would make more sense with the table and eigenvalue spectrum outside of the PC loadings plot.
Figure 4: The labeling scheme across the top of the plot includes pMonths, which are not defined. I assume this means months in the previous year, but this should be explicitly stated in the figure caption.
Ln 388: This phrasing about explaining 35% of the temperature variance in the 1450s is confoundingly misleading. The authors mean that in the nest that extends back to the 1450s, 35% of the variance in the *calibration or validation interval* (which is it?) was explained. There is no way of knowing quantitatively that the 1450s variance was explained to that degree and the language should be more precise here.
Ln 442: What grey boxes are we talking about? The ones in the maps on the lower right? The description says they denote a "period of increased uncertainty," which makes them sound like they should be in a time series not a spatial plot. This needs to be clarified.
Ln 463-464: The authors need to qualify what they mean by "the first time that such significant cooling" has been identified in the SH. There is a wide range of work that has identified cooling in response to volcanism in the SH. A recent example is He et al. (https://doi.org/10.1175/JCLI-D-25-0179.1), which does a comprehensive evaluation of field reconstructions globally, and characterizes hemispheric responses for all of the recons. Many recons show significant cooling in the SH. How do the authors see their result as different from these findings? To be clear, the authors' work is useful regardless of whether it is the first to show significant cooling in the SH, it is just that their claims are bold and poorly qualified given other work that has been published.
Ln 629: What do the authors mean by a "dynamical situation." Moreover, the argument presented here needs to be better qualified. At first blush, it is also not clear why the NH and SH should be different, given that the important dynamics are in the stratosphere and the direct cooling response should be driven principally by the change in radiation balance driven by the stratospheric aerosols (cooling differentials are driven more by heat capacity arguments than "dynamical situations." The authors simply need to make a clearer argument about what they are explicitly suggesting here.
Ln 637: EVA is never defined in the text and the EVA-ensemble is discussed here as if it has been well described in the text. It has not.
Citation: https://doi.org/10.5194/egusphere-2026-4785-RC2
-
RC2: 'Reply on RC1', Anonymous Referee #2, 12 Nov 2026
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 408 | 231 | 64 | 703 | 88 | 68 | 67 |
- HTML: 408
- PDF: 231
- XML: 64
- Total: 703
- Supplement: 88
- BibTeX: 68
- EndNote: 67
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This paper presents a new tree-ring based reconstruction for an area in Northern Patagonia. The paper detects significant cooling following volcanic eruptions which given the previously noted lack of response in the Southern Hemisphere is a very interesting finding. The paper is very well written and the analysis seems rigorous and clearly described. I especially like the modelling based results in the second half of the paper which really add to the context for the significance of the conclusions reached here. I have a few minor comments but think that this paper is almost ready for publication without any major changes.
My one more substantial comment is around low frequency variability. The detrending of the data is described at length in section 2.1. Given that there will be interest in multi-decadal and longer timeseries (some of which is included in this paper) it would be good if you could discuss a little more whether you think these have been preserved in this reconstruction. You seem to suggest that they are by a comparison over the instrumental era (line 393). However, I note that you have said that your approach “greatly reduces the removal of any potential warming signal that could be expressed in these data” so I was not sure whether this period could be used to validate the full record (if it has been given special treatment). I also note that your new record has notably less decadal variability than others in figure 7 and think this should be commented on.
The other question I have regarding the reconstruction itself is whether we could expect some autocorrelation given the results in figure 4 which show some dependence on the previous year in some but not all variables – or does the method account for this somehow?
Minor points:
Line 27 – coheres seems like a strange use of word to me, would correlates be better.
Line 54-55 – I would consider adding a citation to Ballinger et al 2026 (DOI 10.1088/1748-9326/ae1bbc) here – which shows substantial cooling in an instrumental dataset
Line 140 – “initially” – this implies you also used other protocols – did you?
Section 2.3 – I think it would be good to highlight the difference in how the eruption is prescribed in UKESM experiments compared to the PMIP experiments. I would also recommend adding a brief description of the EVA-ensemble here for completeness.
Caption figure 6 -? HadCRUT5
Figure 9 – I note a slight (but non-significant) cooling in year 0 – do you think this is worth discussing? Could it be due to a dating error for one of the eruptions?
You have concentrated your model analysis for good reason on tropical eruptions. Given the recent work indicating the importance of NH eruptions on NH climate (Toohey et al 2019: https://doi.org/10.1038/s41561-018-0286-2) do you expect a similar effect here with SH eruptions?
In the discussion/conclusion you discuss possible areas of interest to detect the rsespose to volcanic eruption in the SH. It looks from figure 13 that South Africa has a strong response – would a proxy reconstruction be possible here? Also would you expect to see a response in other types of proxy archives other than tree-rings?