In August 2014, melt intruded 48 km from Bárðarbunga along a lateral dike, first propagating 5 km toward the southeast before turning north-eastward, and eventually erupting at Holuhraun (Figure 1). During the 6 month long eruption, 1.5 km 3 of magma was erupted (Pedersen et al., 2017), and Bárðarbunga caldera collapsed as melt flowed out from beneath it (Gudmundsson et al., 2016). Intense seismicity accompanying the collapse was recorded by a dense local seismic network (Ágústsdóttir et al., 2019), including over 75 M w > 5 earthquakes (Gudmundsson et al., 2016).Caldera collapses have rarely been studied in such detail as in the 2014-2015 Bárðarbunga-Holuhraun volcanic rifting episode. Gudmundsson et al. (2016) reported 65 m of incremental, highly asymmetric subsidence at Bárðarbunga during the eruption from GPS measurements and radar profiling. Two synthetic aperture radar (InSAR) images, spaced 1 day apart, captured a M w ∼ 5 earthquake, identifying the trace of an inner caldera ring fault near its northern rim (Figure 1). Further work by Ágústsdóttir et al. (2016, 2019) and Woods et al. (2019) provided detailed analysis of the seismicity along the dike path and within the caldera throughout the eruption. In

We use a dense seismic network on the Reykjanes Peninsula, Iceland, to image a group of earthquakes at 10–12 km depth, 2 km north-east of 2021 Fagradalsfjall eruption site. These deep earthquakes have a lower frequency content compared to earthquakes located in the upper, brittle crust and are similar to deep long period (DLP) seismicity observed at other volcanoes in Iceland and around the world. We observed several swarms of DLP earthquakes between the start of the study period (June 2020) and the initiation of the 3-week-long dyke intrusion that preceded the eruption in March 2021. During the eruption, DLP earthquake swarms returned 1 km SW of their original location during periods when the discharge rate or fountaining style of the eruption changed. The DLP seismicity is therefore likely to be linked to the magma plumbing system beneath Fagradalsfjall. However, the DLP seismicity occurred ~ 5 km shallower than where petrological modelling places the near-Moho magma storage region in which the Fagradalsfjall lava was stored. We suggest that the DLP seismicity was triggered by the exsolution of CO2-rich fluids or the movement of magma at a barrier to the transport of melt in the lower crust. Increased flux through the magma plumbing system during the eruption likely adds to the complexity of the melt migration process, thus causing further DLP seismicity, despite a contemporaneous magma channel to the surface.

<div> <p><span>Using a dense network of seismometers located on the Reykjanes Peninsula of Iceland we image a cluster of earthquakes located </span><span>at a depth of 10-15 km, </span><span>beneath the brittle-ductile transition</span><span> and active before and during the Fa</span><span>gradalsfjall</span><span> eruption.</span> <span>The </span><span>deep </span><span>seismicity has markedly different properties to those earthquakes located in the upper, brittle crust with a lower frequency content and a high b-value suggesting that fluids and/or high temperature gradients could be involved in their initiation. Detailed relocation of the deep seismicity reveals that the locus of the activity shifts southwest after the onset of the eruption, suggesting that although the location of the deep seismicity is unlikely to be the source for the magma which erupted, nevertheless the eruption and the deep earthquakes are linked. We interpret the deep earthquakes</span> <span>to be induced by the intrusion of magma into the lower crust. In such an interpretation, the intruded region could be offset from the conduit that transports the magma from the source region near the base of the crust to the surface. </span><span> </span></p> </div>

<p>The 2021 Fagradalsfjall eruption on Iceland’s Reykjanes Peninsula was preceded by more than 12 months of elevated seismic and inflationary activity, beginning around December 2019. On 24<sup>th</sup> February 2021, an exceptionally intense episode of seismicity covering the length of the Peninsula marked the initiation of a dyke intrusion, which continued to develop until the 19<sup>th</sup> of March 2021, when melt first erupted at the surface. During the intrusion, more than 80,000 microearthquakes marked the propagation of melt, first northeast towards Mt Keilir, then to the southwest, eventually forming a 10 km-long dyke. These events were recorded by a dense local seismic network and detected and located using QuakeMigrate[1].</p> <p>We present relative relocations of the seismicity, and tightly constrained focal mechanisms for earthquakes from the dyke intrusion period. The high precision of the relative relocations reveals fine scale structure in the region, which is studied in relation to the orientation of fault planes rupturing in individual earthquakes, thus providing insight into the mechanism of dyke propagation and the controls on faulting in the region. We find that the strikes of the fault planes of individual earthquakes differ from the overall trend of dyke propagation across several propagating seismic swarms.</p> <p>We compare our findings for the Fagradalsfjall seismicity to the 2014-2015 Bárðarbunga-Holuhraun intrusion and eruption seismicity [2], in the context of the contrasting tectonic settings, and markedly different precursory activity.</p> <p>1: Tom Winder, Conor Bacon, Jonathan D. Smith, Thomas S. Hudson, Julian Drew, & Robert S. White. (2021). QuakeMigrate v1.0.0 (v1.0.0). Zenodo. https://doi.org/10.5281/zenodo.4442749</p> <p>2: Woods, J., Winder, T., White, R. S., and Brandsdóttir, B., 2019. Evolution of a lateral dike intrusion revealed by relatively-relocated dike-induced earthquakes: The 2014–15 Bárðarbunga–Holuhraun rifting event, Iceland. https://doi.org/10.1016/j.epsl.2018.10.032</p>

<p>The 6-month long fissure eruption that started in Geldingadalir valley within Mt. Fagradalsfjall, Reykjanes Peninsula, SW Iceland, on 19<sup> </sup>March 2021 was preceded by three weeks of intense seismic activity associated with a ~10 km long NE-SW oriented dyke intrusion, along the Fagradalsfjall volcanic system. This was the first eruption in over 800 years on the Peninsula. A multi-institutional seismic network, installed prior to the dyke intrusion, comprises 27, 3-component instruments (25 broadband and 2 short-period instruments) covering the whole Reykjanes Peninsula. Here we focus on the Fagradalsfjall area (~12x10 km) with 4 instruments located within a 2.5 km radius of the observed dyke seismicity. Accurate automatic earthquake locations using a new detection and location algorithm QuakeMigrate[1] obtain an order of magnitude higher number of earthquakes than conventional location methods. For high precision locations, events are cross-correlated and then relatively relocated using GrowClust[2]. Here we present detailed earthquake location results from 18 September 2021 to 30 September 2022. This period comprises i) the 2021 post-eruptive seismicity along the 10 km long 2021 dyke path; ii) an earthquake swarm about 5 km NE of the eruption site at 5-7 km depth in October; iii) a 5 day-long dyke intrusion in December 2021 that failed to breach the surface; iv) a 5-day-long dyke intrusion that breached the surface on 3 August 2022, and led to a 6 week-long fissure eruption in Meradalir, located about 0.5 km NE of the 2021 eruption site.</p> <p>We find that the failed dyke in December 2021 and the 2022 dyke that successfully breached the surface share many of the same features. They both propagated at similar depths of 3-6 km, in the pathway of the initial 2021 dyke and both show some sparser seismicity closer to the surface. The time span of their propagation is almost identical; both are propagating for around 5 days, with similar lengths of about 6 km, which is considerably shorter than the 10 km long 3-week 2021 dyke propagation. They differ, however, in their location with respect to the 2021 eruption site. The failed 2021 dyke intrusion propagated mainly SW of the 2021 eruption site, whereas the successful 2022 dyke propagated NE of it. Interestingly, our results suggest that during the initial phases of the 2022 dyke intrusion, two dykelets propagate in opposite directions simultaneously.</p>