Space is a precious resource. We hoard it and guard it. Together with air, water, warmth and tomato ketchup, it is one of the essential ingredients for life. We are happy to share empathy, food, and money, but letting someone else invade our personal space is a big step well beyond that. Social distancing is a fundamental part of our lives. The walls around us are invisible, but strongly defended. It is why the internet has become such a big part of our lives: distance makes the heart grow fonder.
It is not just us. The crows feeding in the field keep their equal distance to each other. This is abandoned only in danger. When birds flock and animals herd, it is a sign of predation. There is safety in numbers, but there is happiness in separation. And the plants do it too. Immovable trees keep their distance from each other. In a semi-desert, you’ll find small bushes which occupy their own personal space. Other bushes stay well away. This is not government-imposed, but is because of water constraints. Each bush needs a certain amount of water and it takes a certain area to collect that. Any other bush encroaching on the water space will not get enough and will quickly shrivel and fail. Each bush needs space to grow. This is the basis of ecology.
It is not just the living world. Volcanoes too need space to grow, and they too have ecology. Each needs a certain amount of magma, and if another volcano gets too close, a battle for this precious resource will ensue. The real battle happens out of sight, underground where magma chambers collect their precious resource. The competition is not so much between volcanoes but between their magma chambers. But we judge what we see. So to us, it is a competition between volcanoes. It is a geological ecology.
Can we see this? Just like the crows, the bushes and the people, do volcanoes have their ideal separation? There is not much published on what constitutes adequate personal space for a volcano. How can we find out? The volca-biologist would plant a series of volcanoes and see which ones grow. The volca-physicist would measure the growth rate of volcanoes, and plot this against the number of volcanoes per area. The volca-psychologist would talk to volcanoes and find out how stressed they feel: are they about to blow their top? The volca-geologist would apply for a big grant and travel the world. We can’t do any of these. So let’s look at Iceland instead. All volcanic roads go to Iceland, after all.
Between Icelandic volcanoes
To find out whether volcanoes have their ideal distance, I compiled a list of active or at least non-extinct volcanoes in Iceland (to be precise, I just copied it from wikipedia and removed a few entries). It is a bit of a funny list which mixes central volcanoes and rifts, but I kept them as long as there was a single centre for the eruption (Laki mountain was kept, but the Laki and Eldgja rift eruptions were removed). The list is complete for now, but every time an old system re-awakens, it may need adding to the list. The way things are going, the next one to join may be Thorbjörn. But it is important not to include extinct volcanoes. In the competition for resources, the dead don’t count. They have no ecological needs apart from respect. And the dead outnumber the living: all of Iceland is an volcanic graveyard. Let’s stick to the living.
With this list, I calculated the distance between every conceivable pair of volcanoes. This was done using the coordinates given by wikipedia (accurate to 1 km or so) and applying the haversine equation to get the proper distance on the curved surface of the earth.
It gave me an old-fashioned looking distance table, but that is fine by me. Each distance is listed twice on the table, depending where you start from, and the empty diagonal are distances whether the starting point and end points are the same location. The only aspects that differ from those kind of tables that used to be attached to road atlasses (remember those?) is that (1) these are distances between very unsafe places, and (2) the distances are as the (socially distant) crow flies, if a crow could be convinced to fly from one volcano to the next. A sample of the table is shown below. The full table is available at the VC shop. The distances are given in kilometers, by the way, as volcanoes have long gone metric. (Only Yellowstone still erupts virtual cubic miles.) The accuracy is limited to around 1 km: don’t try to land your plane using this table because you will miss. My apologies to the crow.
For information on each of the volcanoes, the best source is Carl’s mammoth guide.
If you look through the list used here, you’ll find some volcanoes that are not on the main land, and one is even stuck below sea. But wikipedia included them, so I did too.
Sometimes it is obvious which volcano is the nearest neighbour to another volcano. Sometimes it isn’t, and by eye there seems to be two volcanoes at about the same distance, just in a different direction. The real one can be found by looking for the smallest number in each row or column of the table (the full table, not the extract above). The result is shown in the table below. Each row in this table shows the name of the volcano, its height in meters, the year of its last eruption, the distance to the nearest neighbour in km, the name of that nearest neighbour and its height and last time of eruption.
|Volcano||height||last eruption||distance (km)||neighbour||height||last eruption|
|Hengill||803||90 AD||15||Hrómundartindur||540||10,000 BC|
|Hverfjall||420||500 BC||10||Fremrinámur||939||800 BC|
|Kolbeinsey Ridge||5||1755||122||Theistareykjarbunga||564||750 BC|
|Prestahnúkur||1386||7550 BC||36||Hveravellir||1360||950 AD|
(If you like, the table can be downloaded from the VC check-out as a word document.
Interestingly, a nearest neighbour can be a one-sided affair. Sometimes it is a monogamous relation involving pair-bonding, much like albatrosses. For instance, the relation between Katla and Krakatindur is mutual. But sometimes it is not: the nearest volcano to Askja is Holuhraun, but the nearest volcano to Holuhraun is not Askja. This is more of an open relation. A particular dominant volcano may have to keep several partners happy.
A quick glance through this table suggests that a typical distance is some 20 km, but with many variations. A calculation shows that the average separation is 21 km, and the median is 20 km. Does that make sense? Could we have predicted this?
Let’s assume that volcanoes are distributed randomly over all of Iceland. Our slightly edited list has 46 volcanoes. For a surface area of Iceland of 103,000 km2, that gives us on average 1 volcano per 2200 km2. A quick calculation finds that typically, the nearest volcano to you (assuming you are in Iceland) should be about 27 km away. That also applies to each volcano: if you are an average Icelandic volcano (we’d love to hear from you), the nearest volcanic neighbour should be that far away. In reality, only 8 of the 46 volcanoes have a nearest neighbour 26 km or more away (and one of those is on the Kolbeinsey Ridge and perhaps should not be in this study). They are a bit too close together, and therefore Iceland’s volcanoes are not randomly placed across the country.
Of course you already knew this. They are located in several volcanic zones which together cover some 30% of the country. If all volcanoes are in these zones, the typical distance to the nearest neighbour would be less than 20 km. That is more like what we see.
But is the distribution random within these zones? Did Thor drop volcanoes at random within the predesignated areas, or did he try to space them out?
The plot shows the answer. It takes the distance between each pair of nearest volcanoes, and plots them as a histogram. There is a broad peak between 10 and 25 km, with a sharp drop-off for larger separations. The dashed line shows the expected curve if the volcanoes are distributed randomly within the area of volcanic activity in Iceland. It also shows a broad peak, at similar separations, but the random distribution doesn’t have the steep fall-off.
So it looks like close to other volcanoes, Iceland’s boom-boxes are quite independent, but the volcanoes don’t like being too far apart. It is less like the endless field of well-separated crows, and more like animals on the African plains staying within walking distance of the water holes. In science terms, the volcanoes ‘cluster’. Or to a volca-ecologist, they ‘flock’, ‘herd’, or ‘school’.
The volcanic zones in Iceland are about 50 kilometers wide. Are we perhaps seeing the effect of this limited width, beyond which the volcanic fields lie fallow? If you draw a circle which is larger than 50 km diameter around a volcano, part of the circle falls in the area where there are no active (or potentially active) volcanoes, and which lacks suitable partners.
In fact, the typical distance to a nearest neighbour is something like half the width of the volcanic zone. This suggests there typically are two volcanoes across the zone, and these act as nearest neighbours. We can explain why this might be the case. An Icelandic volcanic zone is centred on a spreading rift. A spreading rift spreads – that is how it makes it living. Now assume that a volcano develops on or near the rift. The spreading will slowly take this volcano off to one side. While this happens, another volcano may form and start to move away in the other direction. The volcano dies when it gets to some 20-30 km from the rift, but by that time other volcanoes are already taking its place.
What the plot may tell you is that you indeed tend to have two volcanoes at any one time, on either side of the rift. These volcanoes will not have the same age but they will have the same fate.
The reality is a bit more complex, as it always is. Each of the volcanoes itself develops a rift, which can take up some of the spreading. A series of parallel rifts develops, rather than just one main one. But the basics of the model works well.
But is there a minimum distance? This is not obvious from the plot, but there are indications in the table that there are few pairs closer than about 10 km. And of the 8 such pairs in the table, only two involve pairs where both have recently been active. If I look at the 26 volcanoes that have erupted in the last 2020 years, only two (8%) have a nearest neighbour less than 10 km away. In contrast, of the 19 volcanoes for which the most recent activity is older, five (26%) have a closer neighbour than this.
So it seems that older, sedate volcanoes can snuggle together, but more dynamic ones need their space. To test this, I removed all volcanoes listed as ‘dormant’ from the list, and reran the nearest-neighbour algorithm. I could have been harsher and also excluded the ones listed as ‘unknown’ or have an eruption date more than a few thousand years in the past, but it becomes a bit arbitrary. I went for the minimum exclusion. A few volcanoes now had to find new partners, as their previous nearest neighbour was no longer. The new partnerships are listed below, with the separation in km and in brackets the separation to the previous partner. The dormant volcanoes are themselves also removed from the above table: they had nearest neighbours at 6, 8, 11, 20, 25, 25, and 26 km.
Now I have remade the distance plot for nearest neighbours, with rather large bins (10 km) because we are looking at small differences.
With these rather large bins, the difference is notable for the first bin only: the number of pairs with separation less than 10 km has almost halved. So indeed, close pairs are more likely to involve one (or two) dormant volcanoes. It is hard to prosper in the shadow of an active neighbour.
So in conclusion, there is a minimum distance of 10 km below which Icelandic volcanoes become uncomfortable. But this separation applies mainly to more frequently active volcanoes. Close neighbours are tolerated if they are dormant.
It takes a village
The well-known expression is that it takes a village to raise a child. (This may explain a lot about New York City.) It is not just the nearest neighbour: we should look at the full set of distances between volcanoes. This is done below.
The plot shows a rapid increase, up to about 20 km: this is the typical separation between closest volcanoes. It is followed by a gradual increase up to 100 km, with a fairly flat peak at larger distances, after which it declines again. The gradual increase shows the effect of clumping together of volcanoes: the increase stops when it reached the typical size of a clump. Thus, a clump of volcanoes is typically some 100 km across. If the volcanoes had been distributed randomly over all of Iceland, the plot would show an increasing trend until it ran out of Iceland. Once the distance becomes more than half the diameter of Iceland, there will be fewer pairs: volcanoes near the centre of Iceland run out of volcanoes to pair up with. This is seen in the plot from about 250 km onward. But between 100 and 250 km, the plot is fairly constant. This means that the distribution is not uniform but condensed: most volcanoes are close together but others are sparsely distributed well outside the clump.
I have mentioned that the nearest neighbour plot shows the behaviour of the spreading rift. In contrast, the pairing plot (also called the two-point correlation) shows the effect of the hot spot. The size of a volcanic area is something like 100 km. This is the area over which the hot spot can melt through the crust. Beyond this, the hot spot runs out of heat.
The closest pairing is that of Eldfell and Helgafell, on the island of Heimaey. The two are less than 1 kilometer apart! One is a recent eruption, and the other one erupted 6000 years ago. It appears that the underlying magma found the same weakness in the crust at both times, but during the recent eruption the old (cold) conduit had become unusable. A new one was therefore formed. In a larger volcano, this might have counted as two eruptions from the same volcano, but here they are listed as separate events.
We have already mentioned that the closest pairs tend to consist of less frequently active volcanoes. If a volcano has not erupted in a very long time, and perhaps is listed as dormant or unknown, there is a good chance that its nearest neighbour is also quite inactive. There are exceptions, but typically, quiet volcanoes have quiet neighbours. And the opposite is trued as well: if a volcano erupts frequently (i.e. has a fairly recent date of last eruption), the neighbour has a good chance of doing the same.
It is obvious from the table that nearest neighbours tend to have similar heights, perhaps not surprisingly. Adjacent volcanoes often have access to similar amounts of magma to build up their edifices.
And one final comment: do you know which is the best connected volcano in Iceland? There are various ways to define this. I used as definition the number of volcanoes between 20 and 40 km distance. This excludes most of the closest neighbours, and instead gives an indication of how dense the volcanic field is far enough away to not be affected by the particular volcano, but close enough to be inside the ‘clump’ (if one exists at this location). Curious? Have a guess. I’ll reveal the answer in a few days.
What does this tell us? It is all about the magma. This is the precious resource for which there is strong competition. Just like African animals congregate around the water holes, so Iceland’s volcanoes cluster around the magma hole. And these are greedy creatures. An Icelandic volcano is the elephant in the pool, quite capable of finishing the whole supply itself, leaving little for the other animals. Dormant volcanoes don’t drink much, and they are happy to be denied access while lurking nearby. But a typically active Icelandic volcano needs a monopoly on a drinking area of some 10 km diameter. No other dynamic volcano can be tolerated within this area.
Not every area in Iceland is equally productive. The eastern volcanic zone from Katla to Askja is best at producing magma. Most of Iceland’s lava erupts here. So you might expect that volcanoes elsewhere in Iceland, who have to live on drier ground (as far as magma is concerned) would be further apart, just like those semi-desert bushes grow further apart as conditions become more desert-like. But the table does not show this. The typical distance between volcanoes is the same across Iceland. It seems that if the magma production is less, volcanoes grow less active but not more distant.
Volcanoes are warts growing on magma chambers. The real activity is in those chambers. Sometimes the chambers fill up and leak out a bit: the volcano erupts. At other times the chamber is less full and the wart on top is sitting idle. When we talk about the ecology of volcanoes, it is really those magma chambers that do the competing. In our African analogy, we are counting the oxpeckers rather than the buffaloes they feed on. The magma chambers are locked in a struggle for life. Their food is magma and only the fittest will get it. Where do magma chambers get their magma from?
The melt production happens deep down, near the bottom of the crust. This is the underground aquifer, sitting around and below the spreading rift. The magma chambers have their roots into this aquifer and draw up the magma, each to their own chamber. This is why adjacent volcanoes often have similarities: their magma chambers feed on the same aquifer. And a magma chamber will not allow a nearby chamber to grow. Only some 10 kilometers away is magma independence achieved. The real personal space of volcanoes is underground.
As an aside, you may guess that the depth of the magma chambers is comparable to half the distance between adjacent volcanoes since otherwise a volcano could too easily access the wrong chamber, in an act of volcanic piracy or parasitism.
I mentioned the migrating volcanoes. Now we can understand why a new volcano does not start growing as soon as the previous one has left the rift. It can’t do anything until that volcano is some 10 km away. It also explains why the new volcano will probably start on the other side of the rift, as this is further from the old one. But why do we not get a long line of volcanoes? Wy do they die? This is because the aquifer is mainly located around the rift. The magma chamber has its roots into this area. As the volcano migrates, like wildebeest on the savannah, the roots (ok – not quite like wildebeest and perhaps more like ents) become angled towards the rift. But the magma needs a steep enough angle to move up. As the volcano becomes more distant, the angle shallows; the magma flow is slowly impeded and the volcano dries up and dies. It will survive only if it is lucky enough to find another aquifer, and begins to fight another volcano for this resource.
Schooling in isolation
People need both company and space. So do animals. And so do volcanoes. For they too interact, compete, and predate each other. It is a volcanic ecology, perhaps missing only the packs of roaming wild volcanoes attacking the herds and stealing their magma. In Iceland, volcanoes need some 10 km of social distancing for maximum comfort. Perhaps in this, Iceland’s volcanoes resemble its people. They need their space to grow.
Albert, June 2020
And remember the challenge question: which do you think is the best connected volcano in Iceland?
Answer Defining the connection as the number of other volcanoes between 20 and 40 km distance, there are two winners. The top prize (shared) goes to Torfajökull and Eyjafjallajökull, with 6. Third prize (shared) goes with Þórólfsfell, Trölladyngja and the angry one, Grímsvötn, each with 5.