The Southern Peninsula (Suðurnes) is the western part of the Reykjanes Peninsula; the border is drawn to exclude Reykjavik and its suburbs. Suðurnes is the first point of call for most visitors to Iceland. The main international airport, Keflavik, is located here: the road to Reykjavik runs along the entire northern coast of Suðurnes. It is a windy, rugged, barren place. From the airport (if it is not raining), you see fields of ancient lava and a few distant cones left by past eruptions, the most obvious being Keilir. There are few towns of any significance. Most people live along the northern side. The area around Keflavik accounts for some 20,000 people. Vogur, on the north coast, has some 1500 people. The only major settlement on the southern coast is Grindavik, with a population of 3000.

There is a reason for this. The Reykjanes rift zone runs in-land along the southern coast, dividing the European and North American plates. It is a zone of regular volcanism, which has been especially active in the past 5 years. The north coast is much further from the rift and is much less affected by the lava flows. (Keflavik is entirely out of lava range.) It s therefore older and more eroded. The coast gives easy access to the sea, and this sea is a bay, partly protected from the Atlantic waves. For a culture long based on fishing, this is important. The south coast is exposed to the wild Atlantic and has few bays which give a protected harbour. Grindavik is just about the only one. The volcanism has created a hillier, rougher landscape near the coast. Grindavik’s harbour is created by a promontory, from a lava flow dating to only 2500 years ago. But what volcanoes give, they can also take away. To use a Viking analogy, Danegeld can buy a few years of volcanic peace, but not forever. Eruptions don’t repeat but do return, and building behind a lava flow means a risk of being in the path of the next one.

Faults and rifts

As a reminder, Iceland is a deviation of the Mid-Atlantic Ridge. From the south, the MAR bends a bit eastward, with a larger bend when reaching the Reykjanes Peninsula. In the central part of Iceland, the rift separates into several distinct branches, which towards the north bend back to the west, eventually to become again an extension of the MAR. The Reykjanes Peninsula is a bit unusual, in the sense that the rift here has both a spreading and a transform component. It is the only place on-land in Iceland where this happens.

Volcanic systems

But Iceland is more than a collection of faults. Dotted along the various faults are volcanoes, some large, some small, some with, some without calderas and each with different frequency of eruptions. Some produce enormous lava flows, others produce explosions and some do both. In Iceland, individual volcanoes are typically 20-30 km apart. But what counts as a separate volcano?

Because one volcano may erupt from different locations, they are called volcanic systems. To be classed as such, the system should have some (or all) of a number of aspects: a central volcano (or a centre of production), a distinct chemical composition, a fissure swarm, a caldera and a geothermal system. Torfajokull, for instance has four of these. It only lacks a fissure swarm – although the caldera is disected by different swarms, those belong to other volcanoes.

At Suðurnes, there are no central volcanoes or calderas, but there are plenty of fissure swarms and geothermal systems, and a range of different chemistries. Four volcanic systems have been identified: Reykjanes, Svartsengi, Krysuvik and (just across the Suðurnes border) Brennisteinsfjöll. Further east lies Hengill, which does have a central system. Since 2021, Fagradalsfjall can be considered as a further separate system or as part of Svartsengi.

Taken together, these are substantial systems! They are responsible for 5% of Iceland’s lava output, an impressive amount given the absence of actual volcanoes. But no eruptions had been recorded here on-land for many centuries. This region seems to erupt episodically, with a series of events across the entire peninsula over a few centuries (involving all systems apart from Hengill), followed by some 8 centuries of quiescence. The last such active episode was between 950 and 1250. And the quiet phase that followed has just come to an end.

Fagradalsfjall

Looking back through VC’s archives, the earliest mention of Fagradalsfjall (where the 2021 eruption occurred) was in July 2017, when a tectonic earthquake was followed by a magma intrusion: https://www.volcanocafe.org/unrest-at-fagradalsfjall/ It seemed an unlikely place for an eruption. Fagradalsfjall was not one of the recognized volcanic centres and had not erupted for the past several periods. And indeed, when things became serious in January 2020, it was not here but to the west at Thorbjorn, in the Svartsengi system – which was a recognized system. It started with another seismic swarm on 21 January 2020, followed by a sill intrusion. Two more such episodes ocurred at this location in 2020. The Reykjanes and Krysuvik systems also showed uplift at this time, and Fagradalsfjall showed earthquake activity but no inflation. Then, on 20 October, after a period of inflation a M5.6 earthquake ocurred west of Krysuvik. And magma was on the move.

In this region, the eruptions do not occur on the rift itself. The rift mainly does transform movement which does not generate easy openings for magma – a lazy substance, which, like trickle, always goes for the easiest path. Instead the extension (‘rifting’) is done on a series of faults that run north-south within a few kilometer of the rift zone, and turn NE-SW further away. It became a race between these faults, to see which one would give way first: during 2020, activity alternated between different systems on Suðurnes. There has been a question whether on the peninsula each system triggers the next one, or that the trigger is from deeper magma. Events since 2020 suggest it is the latter: magma is pushing up everywhere, and over several centuries one fault after another fails. Once one fault is filled with magma (solidifying over time, after an eruption), the next one comes under assault.

It was only in February 2021 that Fagradalsfjall became the full focus of activity, with swarming that included 8 magnitude-5 earthquakes. But why here? It turned out, during the summer of 2020 an old fault here had failed. This fault had already shown itself during the 2017 swarm, as a line of increased shear. Analysis of INSAR data over 2020 showed movement of around 1 cm across this fault. It appears to be an ancient fault, as it runs on a somewhat different angle to others in the region. This fault was reactivated by the seismic activity. While other faults failed to fail, this one provided an opening for the magma. When the eruption started in March 2021, it was near this line and later in the year, when the eruption became focussed on a single vent, that vent was exactly on this fault line. Never trust a fault.

And so a dike formed along the fault and an eruption started in March – the most touristic eruption ever in Iceland, being close to the capital, close to the airport, easily reachable yet remote, with coffee available at the main parking area and some nice bars in Grindavik, while the lava catchment area was large enough to contain the flow with little damage. You may remember Iceland’s first lava management experiments: an earth wall which stopped lava from flowing out on the lowlands, and finally police putting up yellow tape at the exit towards the coast with signs ‘Lava – stop’ which immediately caused the flow to cease. Three more dike formations and two more eruptions followed, but after October 2023 activity ceased at Fagradalsfjall. The fault was exhausted.

Sundhnúkur

Instead, the magma went back to its first plan of escape: Svartsengi. And where Fagradalsfjall became a touristic highlight, this location was more serious. The geothermal plant, the hot lakes and the town of Grindavik were now in danger. Here at VC, commenters expressed serious doubt as to whether Grindavik could survive. But survive it did, thanks to a combination of Iceland’s experience with preventative lava management and a considerable amount of luck. The crisis came very suddenly, on 20 November 2023. It was a close call.

As at Fagradalsfjall, the Grindavik crisis had precursors. The INSAR mapping over 2020 and 2021 found movement on more than 1200 (mainly short) faults on Suðurnes, of which more than 1000 had not been known before. Around 10 of those unknowns ran through Grindavik itself. These faults had shown movement over the two years, but had not generated earthquakes: the movement was aseismic. The cause of the movement was the growing sill to the north: around the edges of the sill, the ground stretches and this stretching reactivated the old unknown and unnoticed faults. Old faults can lie hidden in the landscape, but in this case they were just covered up and made invisible by buildings and roads.

An eruption in the Grindavik area was both not suprising and not expected. The Sundhnúkur crater row, which lies closest to Grindavik, had not erupted in the most recent cycle. The lava fields which surround it are dated to 2300 years ago, placing in two episodes ago. The Svartsengi system did erupt in the most recent episode but not in this location. Thorbjorn itself is even older: it dates to the ice age. The hill is dissected by an undated graben which lies along the direction of the crater row. It indicates that considerable rifting happened at one time, but this rifting did not cause a local eruption. There is an older series of craters just to the east, dated to 8000 years ago. The region is a persistent zone weakness, albeit one that does not erupt every cycle.

Diking crisis

After several phases of inflation under Svartsengi, activity resumed in late October 2023. Earthquakes began on 25 October and inflation was seen two days later. The new sill was located at a depth of 7 km, and grew by 9 million m³ over the next two weeks. This was an amount similar to that of the previous inflation periods. But this was the drop that overflowed the bucket. At 7am (UTC) on 10 Nov, the magma began to break out of the sill, moving 3.5 km northward along the Sundhnúkur crater row. The amounts were still small, and no inflation was seen at the surface. At 15:23 there was an M4 earthquake and suddenly the magma was moving southward, forming a dike underneath the crater row.

Now things moved fast. Earthquakes came fast and strong, showing where the tip of the dike had reached. But 6pm, it passed Hagafell with an M5.2 event. Thirty minutes later the dike reached Grindavik and an hour later, it had moved underneath the sea. Rapid deflation commenced by 4:30pm. Deflation and extension both reached a massive one meter by 8pm. After that, it slowed down but parts of Grindavik would sink by 1.3 meter. This activated faults all over town. Iceland knows how to handle emergencies, and the town had been evacuated on time.

The diking had been extreme. At 2 to 4 km depth, the dike was 8 meters wide. The volume of the dike was 130 million m³. At its peak, the inflow into the dike was 7400 m³/s! This is the highest rate every recorded, with the ground moving sideways by 25cm per hour. It is 100 times larger than at the Fagradalsfjall dike, and 30 times higher than at Bardarbunga in 2014. Krafla came closest, at 2000-3000 m³/s during the early phase. But where was all this magma coming from? The dike volume was three times more than the sills that had formed over the previous years. The magma was tapping into reservoirs ( a ‘magma domain’) below the recent sills.

And why did it move so fast? The area of the dike must have been under very low stress: the region was ready for a bit (more than a bit) of extension. It was primed, perhaps after 2300 years of preparation.

Survival

The big question is, though, how Grindavik survived the events of 10 November. Why did such a massive, instant magma flow not cause an eruption? Possibly, the reason lies in the deeper reservoirs. The magma there is buoyant, with slightly lower density than the surrounding crust. But the buoyancy was not quite sufficient. It reached neutral buoyancy (equal density between lava and rock) at 3 km depth. The push to the surface wasn’t sufficient. There was not enough gas pressure to push. But perhaps the main reason was the lack of stress: there was no stress on the dike which would have helped to push it up.

And so it ended. Grindavik survived. Roads, buildings and even the golf course were badly damaged, but not destroyed. The harbour is deeper than it used to be which perhaps will become useful. The golf course may need some reshaping.

But this was not the end, of course. Inflation resumed, and on 18 December, the feared eruption did happen. Another one followed in January, and on 16 January 2024, lava entered the town. But the eruptions were now focussing much further north on the fissure swarm: Grindavik was at the end of events, not the centre. Three houses were lost, but the town survived, perhaps mainly thanks to the earthen berms that had been built.

Iceland’s government offered to buy the houses in Grindavik, so people could move if wanted. And if conditions improved, they could also buy back the houses. It was an eminently sensible scheme. But so far, only those three houses have burned, although other houses may have earthquake and subsidence damage.

It is not clear whether the Sundhnúkur eruptions will continue. Magma inflow has slowed and the rift space created by the extension has been filled – if only in part. The major part of the eruption seems to be over, but smaller events may still occur. But any action will be far from the coast and Grindavik is unlikely to be endangered again.

The next Reykjanes eruption may even move on entirely. Before Fagradalsfjall, all volcanic systems on Suðurnes showed signs of inflation. Krysuvik seems to be calm at the moment, but the Reykjanes system still shows seismic swarms, a little off-shore. Following the idea that activity at the Reykjanes Peninsula tends to migrate westward, perhaps we should look here for the next phase – whether this year, this decade, or later.

Ever since 2020, commenters at VC have been concerned about Grindavik. Events came close, but this is the town that would not die. Iceland threw its magma and lava against it, but resilience, berms and a fair amount of luck came to the rescue. Today, Grindavik lives. Long may it continue.

Albert, May 2026

References

Fracturing and tectonic stress drive ultrarapid magma flow into dikes.
Freysteinn Sigmundsson et al, 2024 Science, 383, 1228–1235. https://www.science.org/doi/10.1126/science.adn2838

Halldór Geirsson: “Meter-Scale Deformation on the Reykjanes Peninsula 2020-2024: a Volcano – Tectonic Rifting Episode”. https://www.nordicgeodeticcommission.com/wp-content/uploads/2024/05/2024_03_12_HG_Reykjanes_s.pdf