Volcanoes can erupt invisibly. A sudden, swift explosion in an isolated location may be unobserved and still have worldwide impacts. The source of the large eruption of 1809 remains unknown. We still don’t know the culprits of the volcanic climate catastrophe of 536 and 540. The link between the year without summer of 1816 and the Tambora eruption was made only in the 20th century. The events of 43BC have recently been assigned to the Okmok volcano – who had expected that? The Hunga Tonga eruption could have been missed even just 50 years ago, leaving us to wonder where all that stratospheric water was coming from. Was Hunga Tonga the cause of the exceptional weather last year? We may never know – it is hard to know what part of the climate is volcanic and what part is our very own global heating. But it seems eminently plausible.
Even the largest eruptions may be invisible. We see the impacts but miss the source. The largest eruption of the past millennium was in 1257; it was a mystery until very recently when it was shown to have come from an unexpected volcano, Samalas in Indonesia.
The second largest (possibly) eruption of the millennium, in the 1450’s, is a particular enigma. It has turned into a major dispute: the cause, the effects and the date have all been questioned. Science does not always give us certainty – it can also question what we thought we knew. But volcanoes will not be denied. This is their story.
The calendar was uncommonly complicated. It was based on a 52-year cycle where each year was given both a number and a name. There were four names and the numbering went up to 13, so that (a bit like a pack of cards) the calendar could distinguish 52 years. The four names were Rabbit, Reed, Flint Knife, and House. The year ‘1 Rabbit’ would be followed by ‘2 Reed’ and so on up to 13 Rabbit. After these 13 years the numbering went back to the beginning, so that the following year would be 1 Reed. This allowed for 4 counting periods or 52 years, after which there would again be a year 1 Rabbit. It sounds remarkably similar to the calendric complexity in Jostein Gaarders’ book ‘The Solitaire Mystery’. The book delves every deeper into the cycles of time in order to reconnect a broken family – using a pack of cards and a joker. It is so easy to get lost in time.
This particular year ‘1 Rabbit’ ran (probably) from January 24, 1454 to January 23, 1455. It was a bad time. Over a period of several years, bad weather and frost had devastated the harvest; food supplies had dwindled and finally ran out. The worst crisis the empire had ever experienced was to become known as the famine of 1 Rabbit.
At the time, the empire ruled the Valley of Mexico. In spite of the name, the Valley of Mexico is a highland plateau, more than 2 km high, located in the centre of Mexico. It is surrounded on all sides by mountain ranges, part of the Trans-Mexican Volcanic Belt. There is no drainage out of the valley. Water running down from the high mountains collected in the Valley in several large lakes. The lowest lake, Texcoco, was the end point of the flow and it was salty. The other lakes provided fresh water , though it could also become saline at times. With a mild, equitable climate, a water supply and a fertile soil, the population exploded. At the time of this 1 Rabbit, a million people lived here. Much later, the lakes were drained and the entire Valley was metropolised. It is now the location of Mexico City, one of the largest cities in the world. I was there once, at a time when there was still extensive damage visible from a major earthquake. The earthquake waves had been amplified by the dry lake bed and this had increased the damage. Nature is not easily denied. The early-morning views of the mountains around Mexico City were magnificent. But soon the pollution would rise and the morning views disappear.
Famine in the Triple Empire
The Valley of Mexico was the heart of the fearful Aztec empire. The empire was build on an alliance of three of the peoples living in the region: Tenochtitlan (who eventually would become dominant), Tetzcoco, and Tlacopan, each of whom had settled around one of the various lakes. The alliance was not all powerful: there were nearby nations who managed to maintain their independence, leading to the almost ritualised warfare of the War of the Flowers. Why does paradise so often brings out the worst in people? Perhaps it takes Rainbow Fizz, the sensory drink of the Solitaire Mystery, to survive paradise.
The stories report that the problems started with early frosts, perhaps 3 years before the famine. In the Valley of Mexico, crops are planted in late spring, to make use of the summer rains. Harvest is in November or December. Frost can happen here on occasion: it is not common but temperatures can drop a few degrees below freezing during the winter months. An early frost could badly damage a late harvest. The famine began with such an early frost, so unusual that it killed not only the corn and the plants but also the trees. This happened two years in succession. The frost occurred everywhere above 2 km altitude which covers the entire Valley. Two years of summer drought worsened the situation. In the reports (compiled much later by the Spanish invaders), these two events follow one another, but it seems not unlikely that the drought and early frost were related and happened in overlapping years. The famine was caused by a few years of quite unusual weather.
Food became scarce. For a while people managed from stored food. Once this was gone, they scavenged for wild foods: roots, cactus leafs and the green of corn plants. But these carried little nutrition and people started dying. The three Aztec kings opened up the royal food storages but these too ran out. Now the kings gave permission for the people to leave. From the stories, it appears that the famine covered all lands above 1 km altitude: food was still available in the tierra caliente, the lower lying land at the coast. Many went to these low-lying region around the Gulf of Mexico where they sold their children in return for food. (They could retrieve the children later if they paid for the food those children had consumed.)
This famine was exceptional for the region. Famines are more common in regions with a shorter growing season, which can be badly affected by a period of poor weather. This is one reason they were regular in medieval Europe. With little buffer against bad times, hunger could set in quickly. The Valley’s longer season was more resilient against poor weather, and agriculture was sufficient reliable that a one-year store of food was achievable and provided a sufficient buffer in normal circumstances. But these were not normal times.
The precise sequence of events is open to discussion. The reports claim four years of famine, two with frost and two with drought. But the descriptions show some repetition which suggest there may have been some overlap. The climate upheaval may therefore have lasted 2, 3 or 4 years. The famine ended after the New Fire ceremony which was held at the beginning of the new fifty-two-year Calendar cycle. This was at the beginning of 2 Reed. That would have been in 1455.
The Aztec empire learned lessons from this disaster. They expanded their rule to cover Mexico all the way to the coast. This expansion brought the people who had left for the tierra caliente back into the fold. But it did not provide more food resources for the Valley: travel was too slow and hard (in the absence of either the wheel or horses) and the carriers would have needed more food for themselves than they could carry. The central cities therefore remained dependent on their own agriculture. The management of the lakes was improved to ensure adequate supply of fresh water. That was done by building dikes to stop back flow from the lowest, salty lake. That allowed for longer periods of irrigation.
The famine and population movement had an unintended consequence: there was now a shortage of victims for the required ritual human sacrifices. That was solved in an innovative way. There were three small, independent kingdoms near the Valley of Mexico: Tlaxcala, Huexotzinco, and Cholula, located in the Tlaxcala-Pueblan Valley. In spite of its overwhelming size, the Aztec empire never conquered these close neighbours. Even the Spanish were surprised by this. From the mid 1450’s the Aztec alliance battled against these independent kingdoms in the Wars of the Flowers. The battles were indecisive and it appears the Tlaxaca were often given advance warning of where the Aztec would approach. When the Spanish asked Muteczuma, why, having those enemies surrounded, they did not finish them off once and for all, the reply was: “We could easily do so; but then there would remain nowhere for the young men to train, except far from here; and, also, we wanted there to always be people to sacrifice to our gods.” The wars provided the human victims. The situation was complicated with the independent kingdoms often working with other enemies of the Aztecs, but there were also long periods of truce where the opposing kings would visit each other for feasts. This situation lasted for the entire period from the end of the famine of 1 Rabbit to the arrival of the Spanish. At that time, the Tlaxcala, clearly not happy with the situation, asked the Spanish the help them. They provided twenty thousand soldiers for the Spanish siege of Tenochtitlin which ended the Aztec empire.
The world beyond
The famine of 1 Rabbit did not stand on its own. Other places in the world also had problems with the climate. In China, the spring of 1453 was plagued by snow which came so late that it damaged the crops. The following winter was very cold with part of the Yellow Sea reportedly frozen.
Tree rings in the UK and in North America were narrow in 1453 and 1454. Frost damage was seen in Californian alpine bristlecone pine trees. These trees are particularly well suited to finding frost damage in annual rings during the growing season. Sometimes the event that caused it can even be dated precisely. For instance, frost damage in rings that formed in 1884, seen in numerous trees, happened on a very cold night 9-10 September. The summer of 1884 had been cold, delaying the growing season. In normal years the growth would already have been completed by the start of autumn, making the rings frost-proof. In cold years growth would be delayed into autumn, and this made the rings more susceptible to damage by an early frost, which happened in 1884. Significant frost damage was found in tree rings from 1453.
Climate on the rocks
Past climates are also recorded in the glaciers of Greenland and Antarctica. The snow incorporates pollutants in the air. One of these pollutants is sulphate, which is emitted by volcanoes. A large eruption can cause a notable increase in the sulphate concentrations. The ice forms from the seasonal snowfall, and this leaves recognizable annual layers in the ice. Count the layers, and it is possible to find out which year (and on occasion which month) the eruption occurred. The counting of ice layers has become more accurate over the years.
For the period around the famine of 1 Rabbit, there is indeed a significant sulphate layer in the ice, both in Antarctica and in Greenland. The layer counting has not always been easy, though. In Antarctica the peak was found in seven different ice cores, but not always counted to the same year. In some cores the year was counted around 1453-1455, others gave 1450 or even 1460. The problem is that every core can suffer from counting errors (a layer may be missed if little or no snow fell that year, or a large snow event may look like a double year). The counting may also have started from the wrong year. The layers only become recognizable after some decades when the snow becomes compressed by the overlying layers: this makes it hard to know when to start the count. This is solved by counting from a known event, often the 1258/9 sulphate peak, but not all cores include this. Sometimes the counting was started from the 1453 peak itself, assuming that was the right year. That is fine if it happens to be right, but gives false assurance if the wrong date was used! Greenland also had problems. There were several sulphate peaks in the ice, a few years apart: it wasn’t entirely clear which peak (if any) coincided with 1 Rabbit.
Plummer in 2012 (https://cp.copernicus.org/articles/8/1929/2012/cp-8-1929-2012.pdf) re-dated one of the Antarctic ice cores and found that the eruption occurred in 1458. This date agreed with some of the results from Greenland. The sulphate amount in their core was almost twice as large as that of Tambora! They found indications for two eruptions, with the main one around 1458 and a smaller one in 1453.
This did not fit well with the tree ring data which showed a strong effect only around 1453. Cole-Dai in 2013 (https://agupubs.onlinelibrary.wiley.com/doi/full/10.1002/jgrd.50587) looked in more detail, and found strong evidence for the two eruptions, where the 1453 one was only seen well in Greenland and was therefore a northern volcano, while the much larger 1458 eruption was seen in both hemispheres but stronger in Antarctica and therefore likely came from the southern tropics.
Sigl et al in 2015 (https://www.nature.com/articles/nature14565.pdf) clarified the discussions but ended up muddying the waters. They found that the older Greenland ice cores (older than 1257) had been misdated by up to 5 years. This shifted the Eldgja eruption from 934 to 939, the currently accepted year. But this change did not affect the younger dates where the two eruptions remained at 1453 and 1458. The former was seen mainly in Greenland and therefore came from a northern eruption while the 1458 eruption was tropical. They found that the 1458 eruption was the second largest in the past 2000 years, after Samalas (1257) but ahead of Tambora (1815), based on the sulphate deposition. The 1453 (northern) eruption was much smaller. They ranked it’s climate effect as the 4th worse among the known eruptions of the past 2500 years– but this climate impact was dated in their paper to 1453, not 1459 (the climate is worst a year after a major eruption). There seemed a major disconnect here between the eruption and its impact.
A later paper confirmed the importance of the 1458 eruption. Gautier et al in 2019 (https://www.nature.com/articles/s41467-019-08357-0) showed that sulphur isotopes can be used to distinguish stratospheric from tropospheric eruptions. The 1458 peak in Antarctica had the third highest stratospheric signature in the last 2000 years, twice as strong as Tambora. This really appeared to have been a very large explosive eruption.
The climate excursion was initially thought to have been caused by an eruption from Kelud in Indonesia. In the early 1990’s several papers pointed out that it could instead fit the eruption from Kuwae, in Vanuatu in the southern Pacific, which had been carbon dated to between 1420 and 1460. Kuwae’s location would fit quite nicely with the indications for an eruption in the southern tropics.
Kuwae is invisible. It is known only from oral histories in the region. It is the Atlantis of the South Pacific, a non-existing place, an island that had sunk amidst destruction. The local people still know where it was, in a gap between the islands of Epi and Tongoa. The oral stories contain details such as where each of the villages had been within this gap, and how they were destroyed when the land exploded.
In the 1890’s, Kuwae was recognized as a submarine caldera. The dark line on the map indicates the caldera rim. The coast line of each island facing the caldera consists of steep cliffs, part of this caldera rim. And if there was any doubt about the volcanic nature of the hole, there is still some activity (basaltic) within the 10-km wide caldera.
The local oral histories talk about the eruption The stories list 30 generations of chiefs before the eruption: the eruption which occurred in the middle of their history. A layer of burned trees and human-made artifacts underneath the ash shows that indeed, this was a green and populated land before the eruption.
The eruption split the island and forced the surviving people to flee to Efate where the refugees caused conflicts. The stories tell that the eruption came unexpectedly. There had been no history of volcanic eruptions at this location. (In fact, although the local islands here have a volcanic basement, much of the coast consists of raised coral reefs.) A large tsunami is mentioned, but the oral history puts the tsunamis and earthquakes before the main eruption. The eruption may initially have started on a submarine flank.
The bottom of the Kuwae caldera is around 400 meters below sea level. Originally it had been much deeper: there is as much as 380 meters of tuff at the bottom of the caldera. (A new volcano has also grown there). The coastal cliffs expose tephra which is 120 meters thick, consisting of several layers. The layering shows how the eruption proceeded. The first part was a longer-lasting maar-like eruption, where magma interacted with water. This was followed by the explosive ignimbrite event which led to the caldera collapse. As the eruption progressed the ejecta became dry: in the battle of magma against water, the magma won. Several vents were active, including some outside the caldera.
The layers indicate that it was a fast eruption: there is no evidence of any weathering in between layers. It may have lasted hours, days, or perhaps a week. During the eruption the magma evolved from basaltic andesite to dacite (the majority) . The current submarine vent in Kuwae is back to basaltic andesite.
Afterwards, Tongoa’s devastation was total. The island was covered under meters of ash. Over time vegetation and animals returned, but there were exceptions: unique in the archipelago, Tongoa was left without snakes! Grass dominated the vegetation for a while, until the trees re-established themselves. Recovery even from a major volcanic eruption can be fast in a wet, tropical climate. But it took more than 200 years before the new forest had stabilised.
A few human survivors returned to southern Tongoa only 6 years after the eruption. But northern Tongoa, closest to Kuwae, was resettled only generations later – so long after that these new settlers brought a different language with them. There are still two different languages spoken in north and south Tongoa!
This was a large eruption, as already shown by the fact that the caldera is 10 km in size. The oral histories state that there had been a cone before the eruption. The volume of the caldera, assuming a large cone, is estimated at 35 km3. That is sufficient for a VEI-7 if it was all ejected explosively.
Carbon dating put the eruption somewhere around 1420 to 1460. The Kuwae eruption has been blamed for a sharp decline in trade in the Pacific at that time, and for a major tsunami in New Zealand. We can now add to this the climate shock of 1453 with a large global temperature drop of around 1C. And of course there was the sulphate in the ice, used to be put a precise date on the eruption, on the assumption that Kuwae was indeed the source of this sulphate.
But there are problems with this interpretation. One may wonder why there were survivors at all. No other VEI-7 or large VEI-6 has left us with a local oral history, not Samalas nor Tambora. The reason is a lack of survivors: for Tambora, there were no survivors within 20 km and the local culture was wiped out. This lack of oral histories is true even for Eldgja. (We know for Laki how precarious local survival was, and how dependent on help from outside which would have been lacking at the time of Eldgja.) But for Kuwae, people managed to flee who had seen the eruption itself. The ash layers on the more distant islands are not as thick as might be expected. Resettlement within 6 years also suggests that the southern parts of Tongoa were not as badly hit, although the fact that the settlers stayed away from northern Tongoa for generations shows how bad the devastation was there.
It has been suggested that the caldera had formed in several smaller eruptions. But this is not what the oral tradition reports, nor is it consistent with the layering seen in the caldera walls. The presence of survivors could be due to the initial maar-like eruption, which could have given people cause to flee before the major explosion. Coastal eruptions can put much of the ash into the sea, if the wind is right. But the size of the Kuwae eruption and its association with the sulphate ice layer have been questioned. We don’t know.
We are left with problems. The major eruption is dated to 5 years after the associated climate havoc, and the identified culprit is in some doubt. We have an eruption in search of a volcano, and a climate excursion in search of an eruption!
Evidence from the ice
As we have seen, the ice indicates there were two eruptions. The first one was in 1453, and affected the northern hemisphere only. The second one was worldwide and therefore tropical, albeit with higher deposits in Antarctica which suggests it came from the southern tropics. This second eruption was a major one.
Hartman, in 2019, (https://www.nature.com/articles/s41598-019-50939-x) took the bait. They went through the Antarctic ice core with a microscope to find tephra: particles from the eruption which had managed to get this far, blown by the wind. It is called cryptotephra. It is only seen for very large or local eruptions. Hartman et al search a 10-cm thick section of the ice core. Five particles were found, roughly 10 micron in size.
The composition of these grains did not agree particularly well with Kuwae. The particles were rhyolitic with 75% SiO2 while Kuwae grains were typically in the 60-70% range. The sodium fraction was also a little higher than that of Kuwae. The composition is similar to that of Kaharoa (New Zealand) or Reclus (at the southern tip of Chile). Kaharoa can be ruled out as New Zealand did not have a major eruption around this time. They argue that the tephra has to come from a volcano near Antarctica because tephra does not travel far. The paper therefore argues for Reclus.
Reclus’ eruption history is not well known, but it has had several rhyolitic explosion. However, it is too far south for the Greenland ice signal. The paper therefore argues that there were two eruptions in the same year, one near Antarctic and one in the north. This disagrees though with the isotopic evidence that the sulphate came from a major stratospheric eruption.
The grains can not have come from Kuwae, and so there must have been another eruption. But whether these grains are from the same eruption as the sulphate remains an open question.
There is another way to date eruptions, using coral. Coral gets its carbonate from sea water. This includes the oxygen and carbon isotopes in the sea water and these isotopes can be affected by eruptions. Coral grows slowly, and a particular layer can be dated. Abram in 2020 (https://www.nature.com/articles/s41586-020-2084-4) analyzed corals from the Indian ocean south of Sumatra. A clear 13C spike was found in the coral. A similar spike occurred after Tambora, confirming a potential volcanic cause. The coral was dated using uranium and thorium decay to around 1452. They attributed this spike to Kuwae. The year disagrees with the ice core dates which put it at 1458. The later date might still be possible for the coral but at low probability.
The most reliable record of climate in prehistory comes from tree rings. Counting the rings gives the precise year, and measuring the width of the annual ring gives the growing quality of the year. A thin ring can come from a cold or dry year. Frost damage indicates that the growth in summer was so slow that it still continued into late autumn, and/or that frost came early. The best trees to use are from higher altitudes, where the growing season is shorter and which are most affected by any temperature variations: their rings show the clearest signal.
The evidence for poor tree growth in the 1450’s was already well established. The case was revisited in 2017 by Esper (https://link.springer.com/article/10.1007/s00445-017-1125-9), who collated a record from 18 different sites across the northern hemisphere. Not every location was equally affected. The strongest signals came from higher latitudes in Europe and Asia, and from the northwest of north America with the exception of one record from British Columbia which showed warming when all other records showed cooling. In the coldest year, summer temperatures were down by −0.4°C in the Swiss Alps to a staggering −6.9°C in the northern region of the Ural mountains.
There are two different ways of studying the tree rings. The one already mentioned is the so-called TRW method, where TRW stands for ‘tree ring width’ (science is very good at obscuring the obvious). The second method is called ‘MXD’: this stands for ‘maximum latewood density’ and refers to the thickest cell wall within a tree ring, measured using X-ray. It appears that the TRW is good at measuring effects over several years, since it depends in part on the health of a tree and this can be affected by previous years. The slow recovery of the tree from a volcanic eruption can be captured well. But for the sudden response to an eruption, the MXD method is found to be better. It very clearly shows the onset of the cool conditions.
Neither method is sensitive to winter temperatures: tree rings cannot show whether a winter is cold or mild. But in fact winter temperatures are less affected by volcanic eruptions. The sulphate clouds reduce the intensity of sunlight, and this has a larger impact in summer when there is more sunshine. There is little strength in the Sun in winter (especially at higher latitudes) and reducing this further does not make much difference. Europe (as an example) gets most of its winter warmth from the sea, not the Sun. The term ‘volcanic winter’ is misleading: it should be volcanic summer – meaning that there isn’t one. A volcanic winter means that the British ‘summer’ rules the world.
The plot above shows the MXD record of Esper et al. The temperatures themselves are obtained from the MXD data by measuring the correlation between MXD and temperature for a 20th century data set. The number shown in the plot is obtained by adding up all 20 records.
The year 1453 comes out as exceptionally cold: the average over each site is -2.5C. (The total of -50C looks a bit strange but comes from 20 (number of sites) times -2.5C (per site).) The temperature remains a bit low for more than a decade. In contrast, the year 1459 is not as exceptional. Based on this, one would have guessed that the major eruption was in 1453 (or 1452), not 1458.
This study shows the impacts across the northern hemisphere. We also have the coral record from the (Indonesian) tropics which shows an increase in 13C uptake, possibly because of cooler water. How about the southern hemisphere? Here, the climate record is much less well known. There are fewer trees available for detailed study, and the effects of eruptions is expected to be less because there is so much more sea in the south. Oceans have a tempering effect which keep the temperatures more constant. Land responds much quicker to changes in solar radiation. Even Tambora’s year without summer, which was a tropical eruption affecting both sides of the globe, is much less clear in the tree records of the southern hemisphere.
There is a recent study of tree rings in South America, by Morales in 2020 (https://www.pnas.org/doi/10.1073/pnas.2002411117). They were looking for evidence of past droughts, based on the expectation that drought will be the main cause of poor tree growth in this region. This expectation is based on the rainfall variability that comes from the El Nino/La Nina cycles. The study extends from the tropics to Patagonia.
The results are shown above, where the vertical axis shows the percentage of the studied area that has wet or dry conditions in a certain year. They do not directly consider temperature, but it is of course possible that some ‘dry’ periods were in fact ‘cold’. Red shows times and regions which are extremely dry. The red lines correlate well with some eruptions. Below is the same plot, but now with some eruptions indicated by the grey lines: 1452, 1458, 1640 (Parker?), 1815 (Tambora) and the unknown eruption of 1695 which we have discussed here in the past.
This plot is the best evidence for southern hemisphere climate effects from eruptions. In contrast to South America, the (sparse) New Zealand tree record does not show this effect. It is plausible that indeed the main effect on South America is from suppressed rainfall coming from the Pacific.
It is interesting that both the 1453 and 1458 eruptions appear to show up. Both volcanoes affected the southern hemisphere. The published data does not show where the drought regions were in those years, so we can’t tell whether the effects extended across South America or perhaps were mainly close to the tropics. The result is suggestive but not conclusive.
To get a reality check on how well ‘tree weather’ compared to real weather, I went through a compilation of weather records from Western Europe. The descriptions indicate cool summers in 1451 and 1452, before the eruption. In 1453 the spring is already cool and the trees blossom very late. June is very cold and August very wet, with a very late and poor wine harvest. In 1454 this scenario repeats, with a claim that Belgium had 7 weeks of rain every day in spring and summer. In 1455 there is frost in June, and the wheat harvest is poor: there is famine. Only in 1457 and 1458 do good harvests return. But extreme weather hits in 1460 with an exceptionally cold winter. The Baltic freezes over, and people travel across the ice from Poland to Sweden. The following summer, though, is normal.
This confirms the poor summers during the early and mid 1450’s, with a proviso that the two summers before the eruption were also not great. The grape (wine) harvest in northern France is late (late September) from 1453 to 1456. Although such a late harvest does happen more often (it also happened in 1451), a series of four such years in a row is very unusual. The harvest was late again in 1459, but now only for one year.
All this should be interpreted cautiously. Weather is not a perfect proxy for climate. In most years one can find examples of unusual weather. It gives context to the tree record but does not replace them.
Many will be familiar with the reports of strange weather in Constantinople in 1453 (just before the city fell) with descriptions of a poor harvest from the city gardens, a 3-hour darkness from a lunar eclipse and fog. They have been highlighted in many articles on the 1453 climate excursion. However, these reports are doubtful. The description of fog comes from a document written much later. Its claim that fog is unusual here is not correct: winter and spring fog is not unusual in Istanbul, caused by cold air from Russia coming in over the Black Sea. The lunar eclipse did not occur on the claimed day but a few days later and was only partial: the description is exaggerated. These reports cannot be used as evidence because of the lack of confirmation. There are also no similar reports from any other area in the region: if this had been a widespread volcanic effect, it would have been described in other places as well.
Back to the ice
We are left with a major problem. The ice cores show a minor, probably northern eruption in 1452 or 1453 and a large southern-tropical one (was it Kuwae?) in 1458. The climate shows evidence of a major eruption in 1453, with evidence from tree rings, coral and human history, while there is at most a minor effect after 1458. What have we overlooked?
The Greenland dates for the ice cores would appear to be well established. The Vatnajokull eruption in 1477 gives a clear marker in the ice seen both in sulphate and in tephra, and is not long after the two eruption signals of 1458 and in 1453. There is a proviso here. We cannot be fully certain that this eruption was indeed in 1477. There is not a single description of it in Icelandic reports, in spite of this being the second largest eruption in Iceland of the past 1000 years. Carbon dating gives a 1480+-11 date. The association with 1477 comes from a report of ashfall in northern Iceland from March 1477. The association of the sulphate spike with Vatnajokull is supported by the cryptotephra, but the evidence for the year is more circumstantial. It is likely, but not certain.
The plot from Abbott et al. 2021 (https://cp.copernicus.org/articles/17/565/2021/) illustrates the situation well. It shows the sulphate concentrations at three stations in Greenland, whilst the red line shows one Antarctic curve, WDC. The 1453 and 1458 peaks are seen in both hemispheres; all other peaks are absent from Antarctica. The 1458 event has a stronger signal in Antarctica and the 1452 is stronger in the north. This suggests eruptions in the corresponding hemispheres, likely in the southern (1458) and northern (1452) tropics as the sulphate did manage to spread across the equator. The cryptotephra in the South Pole core suggests a far southern eruption in 1458, possibly in Southern Chile, but this seems inconsistent with the Greenland signal. An eruption closer to the tropics would fit better.
Suspicions have been raised about the dating of some of the Antarctic ice cores. The date of 1458 comes from counting of annual layers in ice cores from the Law Dome in Antarctica. (The Law Dome is a small, isolated ice cap near the Antarctic coast opposite Australia.) Similar counting from the Siple Station ice core from West Antarctica ice core puts the date of the large spike at 1455; several other ice cores also give an earlier year (Gao et al 2007 https://agupubs.onlinelibrary.wiley.com/doi/epdf/10.1029/2005JD006710). Is it possible one of these is missing or is double counting some layers? Both teams appear convinced that their result is correct. The timelines have been checked with known other eruptions. The closest major southern eruption is Huaynaputina, dated to 1600. Both locations find this eruption at approximately the right time. The eruption in 1257 is also used as a check. There are discrepancies of 1-2 years but that is not unexpected. In conclusion, the Law Dome and WAIS cores gives an undisputed age of 1458 whilst the Siple core and the South Pole core give an equally certain date of 1453.
There is no clear solution to these discrepancies. The climate records show that the largest cooling was in 1453. There is some evidence for cooling in 1459 as well but not nearly as strong. The Law Dome dating indicates that the 1458 eruption was by far the largest, which implies that a smaller eruption in 1453 had a disproportionate impact on the climate. If the Siple and South Pole core dates are correct, then the 1453 eruption was much larger and came from the southern tropics, whilst the 1458 eruption seen only in Greenland must have come from a northern eruption at higher latitudes, for instance Alaska or the Aleutian arc. Another peak in the Antarctic ice cores is then dated to 1448 – a third eruption which came before the others.
Could the Law Dome have been miscounted? If so, this also happened in the separate WAIS core. Perhaps it is not impossible: there is an almost 150 year hiatus in the Antarctica cores, where no significant eruption was seen in the ice between 1460 and 1595. Perhaps something was missed, a few years of less snow making two years appear as one.
Or perhaps, the climate did not behave as expected and responded much stronger to a smaller eruption than to the large eruption 6 years later.
How large was this eruption, really? Was it indeed the second largest eruption in the millennium? It turns out, that is not easy to answer. The amount of sulphate that ends up in an ice core also depends on the vagaries of the local weather. Different ice cores can give different results, just because the wind took the aerosols to one place but not another. The plot below (made from data in Sigl et al. 2015) compares one Antarctic ice core (WDC) to three Greenland ones. The Antarctic core looks much cleaner: that is in part due to Iceland which erupts frequently and causes a lot of volcanic pollution in Greenland even from minor eruptions. NGRIP is particularly affected by Iceland. Laki, of course, completely dominates the Greenland record. The arrows indicate the three major eruptions, Tambora, Kuwae(?) and Samalas.
In the Greenland ice cores, the main eruption in the 1450’s was smaller than Samalas or Tambora. In the WDC Antarctic ice core, it was larger; this is also the case for the South Pole ice core. Combining Greenland and Antarctica makes it the second largest, ahead of Tambora. It is up in the air: different ice cores give different ratios, probably due to details of wind and precipitation.
The comparison indicates that most of the sulphate from the major eruption in the south. On the other hand, the fact that the climate took four years to recover indicates that there was a major injection into the stratosphere, which did reach the north over time and which took a long time to settle.
On balance, it seems more likely that the major eruption was in 1453 rather than 1458.
But was it Kuwae? The tephra in the Antarctic ice suggests it wasn’t: it indicates a more rhyolitic eruption which does not fit Kuwae. But they may have found grains from a separate eruption. The strong sulphate signal in Antarctica suggests a location in the southern tropics. Kuwae fits the location and the crater size, and it did have a large eruption at approximately this time.
What other options are there? South America is attractive, but there is no obvious 10-km wide, very young caldera – and this was only a few years before the Spanish extensively explored South America. How about New Guinea or the island volcanoes north of there? Or perhaps Indonesia, an island chain with a track record of VEI-7 eruptions?
The second eruption is either in the northern tropics or at higher latitudes, depending which year it was in. There is no obvious candidate in Central America, and Japan is too well documented to miss a large event at this time. Can we exclude Africa? There are more options further north. The jury is out.
The mystery remains. This may well have been the second largest eruption of the millennium – and we can’t even agree on when it happened!
To quote again from the Solitaire Mystery: “When you realize there is something you don’t understand, then you’re generally on the right path to understanding all kinds of things.” And it is important to understand what happened almost 600 years ago. For the past 200 years, we have not had a volcano with major worldwide climate effects. Krakatoa and Pinatubo reduced the global temperatures only a little. Have we become complacent? All the largest eruptions of the past two centuries were unexpected. Only one person saw Krakatoa as a danger before its climatic explosion: Verbeek. Pinatubo only gave a few months warning – just enough for evacuations, Hunga Tonga came out of the blue. Tambora was not even recognized as volcanic. And we cannot find the sources of some of the largest eruptions of the past 2000 years.
The Aztec used a cyclical calendar where eventually the same year would come again. And so it is with volcanoes. Eruptions have not ceased. The next major eruption will happen – even though it may come from a very unexpected volcano. 1 Rabbit will return. Are you ready?
Albert, May 2023
References (not a complete list!)
- The Aztec “Flowery War”: A Geopolitical Explanation
Barry L. Isaac, Journal of Anthropological Research 1983 39, 415-432
- The Famine of One Rabbit: Ecological Causes and Social Consequences of a Pre-Columbian Calamity
Ross Hassig, Journal of Anthropological Research 1981 37, 172-182
- Two likely stratospheric volcanic eruptions in the 1450s C.E. found in a bipolar, subannually dated 800 year ice core record
Jihing Cole-Dai et el., JGR Atmospheres 2013, 118, 7459-7466
- The 1452 or 1453 A.D.Kuwae eruption signal derived from multiple ice core records: Greatest volcanic sulfate event of the past 700 years
Gao et al. Journal of Geophysical Research: Atmospheres 2006, 111, 12107
- An independently dated 2000-yr volcanic record from Law Dome, East Antarctica, including a new perspective on the dating of the 1450s CE eruption of Kuwae, Vanuatu
C. Plummer et al. Clim. Past, 2012, 8, 1929–1940
- Timing and climate forcing of volcanic eruptions for the past 2,500 years
M. Sigl et al. Nature 2015, 543, 562
- Six hundred years of South American tree rings reveal an increase in severe hydroclimatic events since mid-20th century
M. Morales et al. PNAS 2020, 117, 16816-16823
- Coupling of Indo-Pacific climate variability over the last millennium
N. Abram et al. Nature 2020, 579, 385
- Northern Hemisphere temperature anomalies during the 1450s period of ambiguous volcanic forcing
J. Esper et al, Bulletin of Volcanology 2017, 79, 41
- 2600-years of stratospheric volcanism through sulfate isotopes
E. Gautier et al, Nature Communication 2019, 10, 466
- Cryptotephra from the Icelandic Veiðivötn 1477 CE eruption in a Greenland ice core: confirming the dating of volcanic events in the 1450s CE and assessing the eruption’s climatic impact
Abbott et al., Climate of the Past, 2021, 17, 565–585
- Annually resolved Southern Hemisphere volcanic history from two Antarctic ice cores
J. Cole-Dai, Journal of Geophysical Research: Atmospheres1997, 102, 16761-16777
- South Pole ice core record of explosive volcanic eruptions in the first and second millennia A.D. and evidence of a large eruption in the tropics around 535 A.D.
D. Ferris et al. Journal of Geophysical Research: Atmospheres, 2011, 116, 17308