Imagine an Atlantic island affected by a deep and complex rift, with half the country pulled east, towards Europe, and half pulled west, leaning towards America, but the northern part actually feeling closer to Scandinavia whilst the southern half doesn’t know where it is going. The rifting causes frequent eruptions with significant financial consequences. Its very parliament got started in a rift. Fleeing the local divisions, its people long ago settled in newly-discovered America.
But let’s not discuss the UK any further. Iceland is more interesting and easier to understand. Its faults are also complex, but geological rather than political, unaffected by political campaigns.
Rifting in the MAR
Iceland is famous as a place where the mid-oceanic spreading ridge breaches the surface of the sea. This happens rarely. It makes it possible to study the breaking point of continental drift without getting your scientific feet -and instruments- wet. We know rather little about how volcanic eruptions in the spreading ridges work. GPS monitors do not work immersed in salt water where satellite signals cannot penetrate. Iceland really is the place to go.
The Earth has 60,000 kilometer of spreading centres. An amazing number! If you would consider this as one volcanic dyke (and why not?), the length would reach 1/6th of the way to the Moon. The rifts form not because of rising magma: this is a common misconception. They also do not form because of hot spots, even if sometimes the two are found in close proximity. Hot spots, as the name implies, are spotty, not rifty. The two most famous hot spots, Hawaii and Yellowstone (they are the most famous because they are in the US and therefore erupt in English) have not caused spreading rifts.
Instead, rifts form because in an ocean far, far away, the edge of a plate started to sink. This sinking is a common malady of oceanic plates. Over time they cool, and as they cool they get denser, and sit a bit deeper on the mantle. After 100 million years they are down to 6 kilometer (the standard depth of an old ocean). If they cool any further – as they must, without new heat – they begin to sink fast. This sinking pulls in both the plate itself and the neighbouring one, to cover the hole caused by the sinking. Now the sinking plate finds itself below the neighbour, which is why it is called subduction.
The pulling in of the neighbouring plate causes long-distance stress. Two ends of this plate may find themselves being pulled in opposite directions. Somewhere in the middle, wherever is the weakest link, it will tear. The tear allows the material from the mantle to come up, and this upwelling brings heat -and magma- to the surface. But where exactly the tear happens can be a matter of chance, and it may even happen in more than one place. Once a small tear happens, it can spread, ripping through the plate like tearing paper, albeit at geological speed. If it started in several places, these rips may not meet. You can get a patchwork of linear tears, like a teenager’s pair of jeans; the tears can be hundreds of kilometers apart. The links between them are called transform faults. And a break may heal, for instance when a new break which developed somewhere else takes the stress away. Now you have an extinct spreading centre, an inactive scar at the bottom of the sea.
Once the upwelling begins, new magma is inserted in the crack at a rate which -not accidentally- exactly fills the growing gap. The oceanic crust begins to move, with the youngest material closest to the break. The rift becomes a broad, high ridge, typically 2 to 3 kilometer above the sea floor and 50 kilometer wide, with a central depression which may be a few kilometer wide which is the actual spreading centre. The crust is thicker underneath the ridge.
Only 2% of the spreading centres are on land. But almost everything we know about how they work comes from those 2%, and in particular from Iceland. The rifts in Iceland are not exactly like the ones underneath the sea. They are above water for a reason, a source of heat which raises the crust, and which is absent in other places. This may change how some things works.
In Iceland, spreading is intermittent, with episodes 100-1000 years apart. Rifts may widen by several meters in one year, and remain stationary for many centuries afterwards. Further from the spreading centre, 50-100 kilometer away, the movement becomes smooth, not intermittent. Magma is injected at depth at some location. It finds its way into the rift by forming dykes in the upper crust, vertical sheets which may be 100 km long. These dyke-forming episodes is what gives the sudden spreading, at least at that place. Obviously the average rate over several episodes cannot be larger than the real spreading rate. Krafla caused rifting of 4-8 meters. This is equivalent to 200-400 years of average movement (which is 1.9cm per year at this location). So there shouldn’t be a repeat at this location any time soon.
Iceland is transected by the Mid-Atlantic Rift (the MAR), which here forms several several branches of varying activity levels. Coming from the south, at sea the MAR seems to head for western Iceland, but close to Iceland it begins to bend to the east, and comes on land at the Reykjanes peninsula where it heads almost due east: this is the Reykjanes volcanic zone. It is highly volcanically active, but episodic, with many eruptions between 940 and 1250 AD but nothing since. At Hengill the rift turns to the northeast, closer to the original direction of the MAR, and becomes the WVZ, the western volcanic zone. It runs to midway in Iceland, turns further to the north – and disappears.
Parallel to the WVZ, about 100 kilometer to the east, runs the EVZ, the eastern volcanic zone. It more or less connects Katla to Askja, and it has a similar southwest-northeast orientation as the WVZ. Volcanically, the EVZ is much more active than the WVZ. There is a vague extension into the southern ocean at Vestmannaeyjar but otherwise it ends near Katla and does not extend further south. Connecting the WVZ and the EVZ is a transform fault, the SISZ (Southern Iceland Seismic Zone) which doesn’t do volcanoes (much) but specialises in earthquakes. Th SISZ is about 70 kilometers long and 10 kilometers wide.
Around Askja the EVZ turns north and becomes the (hold it) northern volcanic zone or NVZ. It runs into the sea and ends there. Along the north coast runs the Tjörnes Fracture Zone (TFZ), which acts as a transform region. It contains two distinct faults, the Húsavík–Flatey fault (HFF) and the Grímsey Oblique Rift (GOR). (Almost every rift map of Iceland which google shows you also has a transform fault in the centre of Iceland, called the MIB, but this one is totally inactive and doesn’t seem to exist).
Finally, north of Iceland where the TFZ ends, the MAR continues as the Kolbeinsey Ridge. It runs north to northeast, slightly tilted away from the direction if the southern MAR. It would have connected well to the WVZ had it not been for the gap in activity in north Iceland.
There is some volcanic activity in Iceland away from the main rift zone, on the western and south eastern edges of Iceland. The Öræfajökull Volcanic Zone is an example of this. It may even sit on its own rift, in which case the EVZ would need a new name as the Öræfajökull rift would be further east. We don’t really know: the glacier covering the area is hiding what is going on, at least for now. Elsewhere, it is notable how quiet the region between the WVZ and EVZ is. This small region is an island of stability amidst an ever-changing environment.
Looking at the different zones in detail shows that they tend to contain a number of parallel rifts. That is true even on the Reykjanes peninsula where each volcano has a fissure swarm which extends southwest-northeast, at 30 degrees from the rift. This shows that when the MAR turned east, it did so reluctantly. The main stress direction still follows that of the MAR itself. At the northern tip of the fissure swarms, they tend to turn more to the east, towards the WVZ.
How do all the rifts and rift-lets move? That has become a lot clearer in recent years. There are nowadays many GPS stations across the country, and these continuously measure their position, as a geological satnav system. The MAR spreads at a few centimeters per year, and you would expect Iceland to show the same. Over a few years, this is easily measurable with a stable (i.e. not located on top of a glacier) GPS. It shows a fascinating pattern.
There are three different types of rifts in Iceland. The NVZ, EVZ, and WVZ are pure spreading rifts. The movement is perpendicular to the rift, so the area shows extension – these rifts are highly volcanic but relatively weak in earthquakes. The SISZ and HFF are pure transform faults, with strike-slip motion along the rift. There is very little volcanic activity here but plenty of earthquakes. Finally, the Reykjanes rift and the GOR show a combination, with both strike-slip motion and extension. The Reykjanes volcanic zone is reported to show two different modes, depending on the cycle of volcanic activity. When the volcanoes are ‘off’, the movement is mainly strike-slip, and when they are ‘on’ it is mainly through magma injection into the fissure swarms, with extension.
The image above shows the NVZ area, and is from a paper by Metzger and Jónsson (2014). In this area, there are a number of parallel rifts and fissures on the land, running into the sea where they fade away. The arrows show the motion: the NVZ is rifting at an average rate of 1.9 centimeter per year. When plotting the motion, you have to choose a frame of reference, which is normally either the European plate or the North America one. Occasionally the difference is split and the map shows north sides moving away at equal speed. You may wonder why not always choose the centre of the rift as a velocity of zero? There are two reasons. First, you don’t quite know where the centre is. Second, the two sides may not be moving at the same speed. Spreading centres are caused (mostly) by the pull of a distant subduction zone. The pull may be stronger on one side than on the other. In the map shown here the arrows are plotted relative to the North America plate. You can see that the rift zone is actually quite wide, perhaps 40 kilometers. Within this zone, the velocities change smoothly: you can see the arrows get longer further to the east.
It is not easy to say where the real boundary between the plates is. It is not a narrow fault but a diffuse area. This is different from the MAR, which is typically a depression a few kilometer wide, in the centre of a wide elevated area. Perhaps Iceland isn’t all that representative: the structure here is quite different.
The movements in the EVZ and WVZ is more complex. Each takes up part of the spread. The total spreading rate is the sum of the EVZ and WVZ. (The two zones are not exactly parallel which adds to the complexity.) This gives about the same total movement or a little higher, at 2.2 cm per year. The EVZ has a spreading rate around 1.9 cm per year in the north, decreasing to 1.1 cm per year at the southern end. The WVZ shows a decreasing extension rate in the opposite direction, from 0.7 cm per year in the south to less than 0.3cm per year in the north. At Hengill, its southern end, a third of the spreading comes from the WVZ and two third from the EVZ. The WVZ fraction decreases to the north. It seems the WVZ is slowly dying, north to south. However, the WVZ hasn’t died yet, and none of it has become entirely inactive. The EVZ takes up most of the extension, and it may be in the process of extending southward, but its extension becomes very minor south of the SISZ. There is little or no rifting in the Katla area.
Because of the difference in extension rates along the WVZ, and the counter change in the EVZ, the microplate in-between the EVZ and WVZ is rotating a bit, anti-clockwise. People have claimed to see this effect in the GPS motions, although it is marginal. Perhaps this motion explains Hekla, situated at its southeast corner, where the rotation may be creating a widening of the SISZ. However, this is speculative and the rotation of the microplate, or even its very existence, are not undisputed.
The spreading rate of 1.9 to 2.2cm of rifting per year is a bit less than the commonly quoted value of around 2.5cm/year. But the lower value fits the MAR better. Spreading rates in the MAR are highest in the south Atlantic, at around 3cm/year, a decline going north. The lowest spreading rates are found near Iceland, at indeed a rate of 1.9cm/year. It is an interesting observation that the most active part of the MAR (by far) has the slowest spreading. However, it makes sense when realizing that continental drift is caused by pull from the sinking plate in the subduction zone, and not by the activity at the spreading ridge.
The two transform faults are expected to show transverse motion. The sliding motion causes earthquakes, and indeed both these regions are regular shakers. In the north, in the TFZ, the HFF is currently locked and has not had a major failure since 1872. The failure prior to that was in 1755, meaning a major earthquake in this region may be slightly overdue, and the village of Husavik is in the firing line. The accumulated stress has been estimated as equivalent to an event of just below M7. Luckily, the Krafla rifting episode, which began in 1975, released a bit of the stress here and is it possible that this rifting is the reason the area of Husavik has so few earthquakes. But this reprieve will be temporary.
The SISZ has significant earthquakes about once every 50 years. It last had a major event in 2000, with two M6.6 earthquakes three days apart. Earthquakes up to M7 are possible here. It is not entirely clear why the SISZ is a pure transform fault, as it doesn’t run exactly in the direction of continental drift. The fault formed 2-3 million years ago, but slowly changed into a transform fault as the EVZ extended south. The SISZ is crossed by many short NS faults, perpendicular to the main transform fault. The earthquakes tend to occur on these faults. Some of these faults seems to have formed in the WVZ, and drifted east in the general continental drift to their present location.
The best-known rift in Iceland is associated with its ancient parliament. From 930 to the 18th century, this is where the Althing met. Thingvellir is located along the WVZ, where there isn’t that much rifting. One side is America, but the other side isn’t Europe: it is the in-between micro plate. The European plate only begins on the other side of the EVZ, 100 kilometer away.
Thingvellir is very unlike other rift valleys in Iceland. It is a 5-kilometer wide graben with steep boundary faults, located within the WVZ. The whole area is subsiding, as a sinking block, now partly filled by a lake. The subsidence is still continuing, at 1.5 mm per years, for a total of 40 meter during the holocene. At this location the WVZ is spreading at 0.7 cm per year, but this rate is distributed over the 50-kilometer wide WVZ and only a part is spreading of the graben itself. The structure seems the most mid-ocean-like rift of Iceland’s.
Iceland’s rifts do not normally become grabens: the numerous eruptions fill them in too quickly with fresh lava. The suggestion has been made that the fact this has not happened here shows that there isn’t as much magma available in the area: the rift here is magma-starved. There have been eruptions in the area, and eventually the rift may fill in again, even though at the moment it is dropping. But this is not the place where the main spreading centre is. Most of the spreading is in the EVZ.
GPS monitors show not only the horizontal motion of the drift. They also measure vertical displacement. Growing magma chambers can push up the ground above them, and we are familiar with inflation as a sign for an impending eruption. But the GPS pattern of inflation shows something funny in Iceland.
GPS station have been around in Iceland for decades. Plotting all the data shows that there has been a general pattern of inflation. This has been noticed in several papers, most recently by Kathleen Compton. The figure shows on the left, the rate of inflation, in mm per year, as of 2014. Known, local events were removed from the trend, so you won’t find anything from the Bardarbunga eruption. The significant deflation seen at sites where geothermal energy is harvested is also removed. The remaining pattern shows general inflation in the centre and the south, and some deflation at the northern edge and Reykjanes. No magma chamber can be that big. And in fact the inflation is accelerating. The right side shows the acceleration at each location, from 1995 to 2015. What is going on? Is Iceland heading for a super-eruption and should we start evacuating the Atlantic Ocean? In fact, it is not volcanic at all. Surprisingly, Iceland’s rising is caused by global warming.
The reason for the general inflation is that Iceland’s big glaciers are melting and thinning. The mass of ice pushed down the land, over an area that is quite a bit larger than the glacier itself. Now that the glaciers are declining in weight, the land is beginning to rise up. And the fact that the rise is accelerating shows that the ice melt is getting faster. Some of the measured rifting rates in Iceland, which were around 2.5 cm per year, in hindsight were affected by this inflation. GPS measures line-of-sight changes to the satellites, and this includes a vertical component. by ignoring the unrecognized inflation, the spreading rates were overestimated.
Iceland has two centuries of ice melt to look forward to. At the end, parts of the country may have risen up by several meters.
All surface of the Earth has come from rifting. It has brought ocean floor up from the mantle, and the lighter material that has formed the continents. But there are few places we can see the process in action. Iceland shows how the world was build. In all its complexity.
Plate boundaries, rifts and transforms in Iceland
JÖKULL 58, pp. 35-58 (2008)
The present kinematics of the Tjörnes Fracture Zone, North Iceland, from campaign and continuous GPS measurements
S. Metzger, S. Jónsson, G.J.K. Danielsen, S. Hreinsdottir, F. Jouanne, T. Villemin
Geophys. J. Int., 192, pp. 441–455 (2013)
Glacial rebound and plate spreading: Results from the first countrywide GPS observations in Iceland
T. Árnadóttir B. Lund W. Jiang H. Geirsson H. Björnsson P. Einarsson T. Sigurdsson
Geophys. J. Int. 177, pp 691-716 (2009) https://academic.oup.com/gji/article/177/2/691/2023257/Glacial-rebound-and-plate-spreading-Results-from (open access)
Climate-driven vertical acceleration of Icelandic crust measured by continuous GPS geodesy
Kathleen Compton, Richard A. Bennett, Sigrún Hreinsdóttir
Geophysical Research Letters, 42, pp. 743–750 (2015)