the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 01 Sep 2026
More than Laki. Compound Climate Shocks and Icelandic Volcanism in the Nordic World of the 1780s
Abstract. The article investigates the 1783–1784 Laki flood lava eruption as a case of regionally differentiated volcanic climate impacts and compound climate–society shocks in the Nordic countries. Drawing on recent advances in paleoclimate reanalysis (ModE-RA), early instrumental weather observations, dendroclimatology, and local environmental histories, it reconstructs seasonal and spatial patterns of climate anomalies and their societal repercussions from Iceland to Finland. A model–data comparison reveals marked regionality: while ModE-RA indicates pronounced post-eruption cooling from 1784 to 1786, instrumental and proxy evidence show strong sub-regional contrasts, including a preceding unrelated severe spring–summer drought in southern Scandinavia and notable discrepancies between ensemble simulations and observed autumn 1783 and winter 1783/84 temperature patterns. The study conceptualizes the Laki crisis as a compound event in the Nordic countries, in which volcanic cooling interacted with antecedent warmth and drought, atmospheric pollution, societal tensions, and institutional vulnerabilities. Pre-eruption drought and grain harvest failures in Denmark and southern and central Sweden, eruption-related toxic haze and vegetation damage in Iceland, and subsequent cold seasons combined to produce cascading impacts on harvests, food prices, mortality, and social protest, with peak demographic and political effects lagging 2–3 years after the eruption and food crises. By systematically integrating “archives of nature” and “archives of society” across multiple Nordic polities, this study illustrates how an interdisciplinary climate history framework can capture the regional complexity of volcanic forcing and climatic anomalies, clarify tensions between models and observations, and explain divergent societal trajectories under compound climatic stress.
Competing interests: At least one of the (co-)authors is a member of the editorial board of Climate of the Past.
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 paper. While Copernicus Publications makes every effort to include appropriate place names, the final responsibility lies with the authors. Views expressed in the text are those of the authors and do not necessarily reflect the views of the publisher.- Preprint
(3335 KB) - Metadata XML
- BibTeX
- EndNote
Status: open (until 19 Nov 2026)
- RC1: 'Comment on egusphere-2026-5137', Anonymous Referee #1, 03 Sep 2026 reply
-
CC1: 'Comment on egusphere-2026-5137', Claudia Wieners, 06 Oct 2026
reply
This is an interesting article. But I think the events in Iceland are not always represented as clearly as they could, and the section on dempgraphic impacts could be more sysematic, e.g. providing baselines, not just isolated numbers.
Events in Iceland (and the Faroe islands)?
Are the events in Iceland itself part of your original research, or just a comparison case based on literature research?
Within the Nordic countries, Iceland experienced by far the worst impacts from the eruption, yet it receives the least discussion. Iceland is, for example, not shown on the maps in fig 3-4, and it seems that you do not analyse primary data (e.g. parish data, weather annals…) from that country.
Also, nothing is said about the Faroe islands.
There is nothing wrong with focusing on mainland Scandinavia (which I loosely define here as Norway, Sweden, Finland, Denmark), especially since there is already quite some literature on the effects of the eruption in Iceland, and the authors’ affiliation list suggests the team’s expertise may lie more in mainland Scandinavia than in Iceland. But then it would make sense to state explicitly which regions are your main focus of original research, or a comparison case based on literature research, or excluded.
Examples of aspects that are discussed for mainland Scandinavia, but not for Iceland, are listed below. Again, you can choose not to pursue these topics, if your focus is clearly on mainland Scandinavia.
Interaction between (prior) weather and the eruption.
For mainland Scandinavia, it is discussed in considerable detail how prior weather affected harvest shortfall, in particular the 1783 spring drought, but also weather in 1782.
In Iceland, the eruption likewise interacted with prior weather; in particular, bad weather in Northeast Iceland led to poor hay harvest and loss of livestock already before the eruption (Wieners and Hálfdanarson, 2024).
Possibly your re-analysis does not yield reliable weather information for Iceland (tree ring data might be scarce there), but a compilation of weather-related information from annals is here:
https://www.vedur.is/media/vedurstofan/utgafa/hlidarefni/annalar_sig-th.pdf and the letters from Icelandic officials in the source Gunnlaugson et al 1984, which you cite, also show this.
Food aid.
Two very good articles (in Icelandic) about food aid by the Danish government to Iceland during the famine are in the book “Skaftáreldar 1783-1784”, of which you cited other chapters. In Wieners 2020, this material is reviewed in English. None of these however discusses food aid, or the lack thereof, to Iceland with harvest failures in Denmark, so it might be interesting to follow that link.
End of Trade monopoly.
Not only the Norwegian grain monopoly, but also the Iceland trade monopoly was lifted after the eruption. The latter may have partly been due to the fact that the Iceland trade company (which at that time was run by the crown) had become bankrupt. An excellent source on this is Gísli Gunnarsson’s book, https://babel.hathitrust.org/cgi/pt?id=uc1.b3524053&seq=15.
Social unrest.
You make an interesting case that the food scarcity and unsatisfactory trade arrangements contributed to the Lofthus uprising in Norway. Interestingly, nothing of that sort happened in Iceland, despite much worse famine, and an even more restrictive trade monopoly. Why? Because Icelanders were more religious and many saw the famine as divine punishment rather than something that good governments should prevent? Because Denmark was so far away and rising up against them was impractical? Because population was so thinly distributed that it was harder to gather and start a revolution? Because people feared they depended too much on Denmark for vital imports? Because the Icelandic “elite” was quite “Danicised” and happy with the status quo, whereas most ordinary farmers had other problems in their mind? …
Minor Comments
There are some, mostly minor, inconsistencies and imprecise statements about events in Iceland (I am not an expert in the effects elsewhere, so I cannot judge). In some cases, statistics are presented in a not very systematic way, for example lacking a baseline value for clearer interpretation. I list these instances below.
Line 49: Skaftáreldar translates as Skaftá Fires, not Fire (eldar is plural).
Line 92-93: Thordarson and self 2003 say in their abstract that 95Mt(SO2) reach the “lower stratosphere and upper troposphere”, i.e. considerably less may have ended up in the stratosphere. Consider checking.
Line 127: “poisoning livestock and people”: This is quite imprecise. Livestock certainly received lethal amounts of poison, namely from fluorine in ash. Humans experienced health issues attributable to air pollution (respiratory irritation etc), but these were likely not lethal, or at least not enough to cause significant excess mortality. There is no strong evidence that humans experienced fluorine poisoning. So unless you count non-lethal health issues as “poisoning”, humans were not poisoned, even though livestock were poisoned at a large scale. See Wieners and Hálfdanarson, 2024.
Line 130 ff: While it is true that there were respiratory diseases in mainland Europe, some relatively recent papers suggest it was not as deadly as some of the older works, e.g. Grattan’s articles, suggest. For example Heaviside et al., 2021 re-analysed the simulations by Smith et al., 2011 but found much lower mortality because of lower dose-response curves (https://doi.org/10.1016/j.scitotenv.2021.145549) and Balkanski et al., 2018 found that modelled aerosol and SO2 concentrations in France were too low to cause large excess mortality (https://doi.org/10.1038/s41598-018-34228-7).
Line 132: Casey et al seems to lack in the reference list. If you refer to this article: https://escholarship.org/uc/item/5d64k9sr I wonder how reliable it is, given that it suggests unrealistically low childhood mortality in Sweden.
Line 229: Magnús Stephensen was not the governor (stiftamtmaður) of Iceland. His father had been deputy governor (amtmaður) until summer 1783. At the time of the eruption, he was a student at Copenhagen (later in life, he held high offices in Iceland, but not that of governor). He was sent to Iceland, together with Hans von Levetzow of the Rentekammer (finance department of the government) to investigate the effects of the eruption. In 1785, he published a lengthy report, though he probably got a lot of information from the local pastor, Jón Steingrímsson. So if you want to list local observers, Jón is a much better example.
Line 440: “warm summer cut short by frosts in June” -> For England, there was a discussion (probably in one of Grattan’s articles) on whether there was actually frost in summer, or whether the effect of the sulphurous haze made plants look as if frost had occurred (either because material accretion on leaves, or because of leaves withering). Could it be that the same happened in Norway?
Line 450: amount of hay and grain is abundant, but of poor quality, since it was harvested while green…. “it” probably refers to grain, but not to hay (which is typically harvested as green grass). onsidr reformulating.
Line 465ff: This is a bit imprecise. Prevailing practices suggested to slaughter animals after the round-up, which in turn was after the hay harvest. In other words, normally farmers waited how much hay they got before deciding how many animals to keep during winter. There is no evidence to my knowledge that people postponed slaughter until after regular slaughter time.
In the vicinity of the eruption, fluorine poisoning happened so fast that animals quickly lost weight and died. In this region, it was regretted (according to Jón Steingrímsson) not to have slaughtered the animals unusually early, namely as soon as symptoms occurred, “while they still had meat on their bones”. Here, a hope for a speedy end of the eruption and opportunity for the animals to recover may have played a role.
In regions further away, the disease only manifested itself gradually over summer and autumn. Probably there the slaughter happened at the usual time. It may have happened that in the course of winter the hay became scarce, both because the winter was long and cold, and because the hay appeared less nutritious than normally. In that case, farmers may have regretted not to have killed more animals during the (usual) slaughtering moment, but this would then not be related to hopes for a speedy end of the eruption.
As for animals becoming inedible: Jón Steingrímsson described foul-tasting meat, but the animals were eaten nonetheless… though, as mentioned, they had less meat than they would have otherwise.
Line 486: “preceded the eruption’s impacts”: Rather strong wording if you are still disentangling whether and which such impacts happened. In particular, if there were (probably) good conditions in 1784, one can at least doubt whether harvest reductions in 1785 and 1786 were due to the eruption. The direct effect like acid rain should have stopped by then, and whether there still was such a strong climate effect even after the sulphate aerosol was likely mostly washed fro the stratosphere, might at least need some discussion.
Similar line 607 “volcanic shocks”, line 797 “a single eruption generated a spectrum of climatic anomalies…”
Line 518: Iceland and Norway, which exported timber, minerals and fish in return: Bit misleading, which country does the “which” refer to? Iceland certainly did not export timber (it imported Norwegian timber through the Danish monopoly traders), and no or hardly any minerals except maybe some sulphur, though it did export dried fish.
Line 530: By 1785 -> by the end of 1785 (the two deadliest years were 1784 and 1785).
Also, the excess mortality was 1/6 rather than 1/5.
Fig 8: It would be helpful to add a map of average crude death rates (for example, average 1770-1800, if enough data is available, otherwise, some reference period that gives a good compromise between availability and usefulness), and maybe also a map of the standard deviation. This helps to estimate whether the presented crude death rates are actually “really bad” or “quite normal”.
You could also aggregate over bigger regions in a table, e.g. Southern, Central and Northern Norway, Souther, Central and Northern Sweden, etc, and maybe also Iceland, for better comparison. (You can get the numbers for Iceland in Wieners and Hálfdanarson 2024.)
Line 541: “impacts stopped short of demographic catastrophe”, what does that mean? Is there a threshold above which something counts as demographic catastrophe, and this threshold was just about nor reached? Or was a demographic catastrophe just narrowly prevented by strong measures taken just-in-time?
Line 548: smallpox: These happened every 5 years, quite regularly. So it is not a priory clear whether the bad harvests and cold winter had anything to do with it, or whether the smallpox just happened because 5 years were over.
Line 553: Denmark experienced increased mortality… crude deathrates of 31/1000: To judge whether this was bad or not, it would help to have a reference value. (In Iceland, this death rate would have been pretty normal, but Iceland had very high infant mortality.) Same remark holds for line 556 (how bad was Norway’s deathrate).
More broadly speaking, this discussion is a bit fragmented, for example, it is mentioned that “death continued to exceed births in 1787”, but not by how much or from when on. If this information can be found elsewhere, it is OK to only highlight specific findings, but then I’d like a reference to a more complete description. Or if this is unpublished statistics, then some more systematic information, for example a timeline plot of death and birth rates over the years 1781-1787 plus reference values, would be helpful.
Line 583: “reduced to 4 skillings”: from which starting value? “reduced from 8 to 4 skillings” says much more.
Line 608: “foreign supplies increased fivefold”: But were these significant? For example, if foreign exports increased from 0.5% to 2.5% of total exports, that would still be very minor.
Line 720: One could argue about whether the “famine of the mist” is a good example of a compound event. Unrelated prior bad weather played some role in depleting food reserves in the northeast, as mentioned above. Other than that, the drivers (as opposed underlying to long-term vulnerability), i.e. direct eruption damage from lava flows swallowing farms etc, vegetation damage (acid rain), livestock loss from poisoning (from fluorine) or starvation (from lack of hay), and at least to some extent subsequent weather anomalies, can all be related to one cause, the eruption. The agricultural and trade arrangements were not so much drivers (a more or less acute event) but underlying factors increasing vulnerability to drivers.
So in my view, it is much easier to argue that mainland Scandinavia had a compound event, multiple similarly strong drivers combining in one crisis. In Iceland, the eruption overwhelmed everything.
Line 732: Population displacement: it seems to me that this is hardly discussed for mainland Scandinavia in the current paper, except for mentioning beggar children roaming the country side in southern Sweden. (There was considerable displacement in Iceland though, mostly towards fishing regions in the West.)
Line 735: “reduce but not eliminate famine”: Was there a famine in mainland Scandinavia? According to modern definitions, death rates were probably too low to qualify as “famine”, so in that sense, famine was averted there.
Citation: https://doi.org/10.5194/egusphere-2026-5137-CC1
Viewed
| HTML | XML | Total | BibTeX | EndNote | |
|---|---|---|---|---|---|
| 295 | 289 | 51 | 635 | 57 | 50 |
- HTML: 295
- PDF: 289
- XML: 51
- Total: 635
- BibTeX: 57
- EndNote: 50
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
This study combines the ModE-RA palaeoclimate reanalysis, early instrumental observations, tree-ring data, historical documents, grain-trade records and demographic data to reconstruct the climatic and societal impacts of the 1783-1784 Laki eruption across the Nordic region. Its most valuable move is to argue against Laki as a spatially uniform "European disaster" with immediate effects, and to show how pre-eruption warmth and drought, post-eruption cooling, market structures and institutional vulnerabilities interacted. The source base is unusually broad, and the regional framing deserves to be developed. The core inferential chain, however, is not yet adequately supported. Observations already assimilated into ModE-RA are reused as if they provided independent validation; an ensemble mean is compared directly with a single historical realization; the compound/cascading framework does not consistently distinguish hazards, preconditioned vulnerabilities and downstream impacts; and temporal sequence is used to infer a climate–society causal cascade. Consequently, harvest failures, warmth, drought and political-economic tensions already present in 1781–1783 are too readily folded into a “Laki-triggered” narrative. As it stands, the evidence supports an interpretive, hypothesis-generating interdisciplinary synthesis, but not the stronger volcanic-attribution and societal-causation claims made in the current version. I therefore recommend major revision.
Main comments
1. Compound hazards, vulnerabilities and cascading impacts are not kept apart
Lines 712-734 and Figure 10 place climatic anomalies, pollution, institutional dependence, slow information flow, market shocks, disease and political outcomes in one compound/cascading frame, without saying which are hazards, which vulnerabilities and which downstream impacts. Compound-event typologies also separate preconditioned, multivariate, temporally compounding and spatially compounding cases; the manuscript gives one broad definition instead of assigning a type to each region. A workable split would be: hazards (antecedent warmth and drought, volcanic pollution, the following cold seasons, early frosts), preconditions (earlier harvest failures, import dependence, the grain monopolies, poverty), cascading impacts (crop loss, prices, nutrition, disease, mortality, protest) and response feedback (imports, relief, migration, reform).
2. The societal analysis rests on temporal proximity and has no common comparative design
Section 3 says only that demographic, administrative and narrative records were combined and cross-checked. It gives no search scope, inclusion criteria, assessment of source independence, translation practice, unit of comparison or alternative explanations. The results and discussion nevertheless use causal verbs: "driven by", "helped seed", "averted", "resulted in". Please apply one comparative design at different levels (e.g., polity, region, locality) with the same indicators at each: climatic exposure, harvest, grain prices, imports, relief, disease, mortality, protest. If quantitative identification is not possible, state that the method is comparative-historical process tracing and label each conclusion as demonstrated, a plausible contributing factor, associated but not identified, or speculative.
3. ModE-RA is not independent of the instrumental, tree-ring and documentary records
The manuscript states that many of the series used here were assimilated into ModE-RA. It then reports that ModE-RA "very closely reproduces the observations of instrumental and proxy data across all four seasons", and the conclusion presents reanalysis, proxies and instruments as three mutually supporting lines of evidence. Agreement with assimilated data is an assimilation fit, not a test. ModE-RAclim does not escape this: it replaces the transient prior with randomly drawn years of ModE-Sim, removing the forced signal from the prior but keeping ModE-Sim covariances and the same observations, so it is not an observation-only product. Please add a record-level table of which stations, chronologies and documentary series were assimilated and which were not, describe comparisons against assimilated data as assimilation-fit or consistency checks rather than independent validation, and test the reconstructed spatial anomalies against fully withheld observations.
4. An ensemble mean is not a historical realization, and there is no no-Laki counterfactual
Section 4.1.3 and Figure 6 set the ModE-Sim ensemble mean against ModE-RA and the instrumental records season by season and read the differences as model-data disagreement. An ensemble mean suppresses internal variability and keeps mainly the common forced response, whereas the observed year is one forced response plus one realization of internal variability. The question is not whether the observations equal the mean but whether they fall outside the ensemble distribution. ModE-Sim is also an atmosphere-only ECHAM6 ensemble with prescribed SST and sea-ice boundaries, so its mean smooths the Laki response as well. It is not a no-Laki experiment, and it cannot generate the sea-ice and ocean heat transport feedback. Please show the full ensemble, or at least the 5th-95th range, the rank of the observations within it with exceedance probabilities, and the posterior spread, SD ratio and assimilation increments. Paired Laki / no-Laki simulations or a formal attribution analysis would be needed to provide a more appropriate basis for attribution. Without them, "the radiative response ... led to" and "link the external forcing" should be reduced to consistency with the expected response to volcanic forcing.
Specific comments
1. Lines 266-288 call instrumental observations "the most reliable source of information for assessing past climates" and then work with mixed baselines, 1961-1990 at l. 281 and 1761-1790. Please give, for the series used, instrument exposure and siting, unit conversions, gaps, breakpoints and homogenization uncertainty.
2. Please define the starting point of each lag separately eruption, harvest failure, price peak, epidemic onset and report them one by one.
3. Section 5.2 and the conclusion trace the food crisis through the abolition of the grain monopoly to the declaration of independence in 1814. As currently written, the chain is teleological: the 1814 outcome organizes the interpretation of the earlier events.
4. The 1783 density signal is doing two jobs at once. Lines 244-246 and 677-678 attribute low northern MXD in 1783 to aerosol acting on plant physiology; the same low density also carries the reconstructed northern cooling, in a region the manuscript says has no local instrumental coverage. If those chronologies entered the assimilation under the usual MXD-temperature relationship, direct aerosol damage would be read as cold and spread spatially through the model covariances.
5. Was Laki the trigger of the whole crisis, or a further shock entering a sequence already under way?
6. The three regional groups come from post hoc visual sorting and do not establish latitudinal control. Please use predefined geographical groups, or a reproducible clustering or hierarchical model, controlling for proxy type, target season, elevation and autocorrelation, and report sample depth, EPS/Rbar, calibration-verification statistics and reconstruction uncertainty.
7. Table A1 gives the longitude of Arjeplog as 182 deg E; Table A2 gives 18.20 deg E, which is correct. Cross-referencing is inconsistent throughout: the appendix labels two consecutive tables "Table S2" and "Table A2"; the main text cites Tables S1-S3 and Figures S1, S2 and S4 for material that appears as Tables A1-A2 and Figures A1-A3; and Figure A3 labels its third panel "Group C" where the text says group 3.
8. Figures 3 and 4 use a six-year baseline, 1777-1782; Figure 6 uses 1751-1780; the tree-ring work uses 1733-1782, 1734-1782 or 1773-1782. Line 390 notes that for temperature the two reanalysis baselines agree to within 0.2 deg C in every season, which is useful, but that check is stated only for the area-mean temperature. It does not cover the precipitation field in Figure 4, the spatial patterns, or the three tree-ring baselines. The short baseline also contains the three cold winters of 1780-1782. Please adopt one primary baseline and show how the maps and the tree-ring thresholds respond to the alternatives.
9. Figures 3, 4 and 6 show central estimates only, with no ensemble spread, no posterior uncertainty and no indication of how strongly each region is constrained. Please mask sparsely constrained areas or show the SD ratio alongside.
10. Figure 4 is labelled "large-scale precipitation" while the text compares it with rain-gauge totals. Please state whether the model variable excludes convective precipitation; if the definitions differ, the two are not directly comparable.