The eruption at Fagradalsfjall emitted a total amount of SO2 on the order of 0.3–0.9 Mt (Pfeffer et al., in preparation) and a total bulk volume of lava of 150 ± 3 × 106 m3 (Pedersen et al., 2021). Although the 150 million m3 bulk volume makes this eruption of average size for Iceland (Gudmundsson et al., 2008), it was small compared with the most recent basaltic eruption in Iceland, within the Bárðarbunga volcanic system in 2014–2015, when 9.2 Mt of SO2 was released and 1.4 km3 of lava was extruded (Pedersen et al., 2017; Pfeffer et al., 2018) during a similarly 6-month long eruptive period. However, the Fagradalsfjall eruption was distinctively hazardous, given its vicinity to inhabited areas and critical infrastructure, as well as the high number of visitors to the site, representing new challenges for both scientists and civil protection representatives. Its location between the International Airport of Keflavík and the Reykjavík Capital area, facilitated the access of numerous visitors to the eruption site. Holuhraun, in contrast, was an extremely remote location and access to the site was restricted during the 2014–2015 Bárðarbunga eruption. The Fagradalsfjall eruption therefore provides an excellent case study of a popular “touristic” eruption, which both individuals and tour companies came to visit. The former for experiencing the forces of nature, the latter for obvious economic interests (Donovan, 2018).
Based on the data from the Icelandic Tourist Board (2022) about 356 thousand people went to visit the eruption site during the period 24 March 2021–7 January 2022. During the first month of the eruption the number of visitors per day reached 6000 and was comprised of mainly local tourists. Since June, in correspondence to the border opening (following the COVID restriction release by the government), the tourism became more international. Despite the high number of visitors to the area, rough hiking trails and elevated SO2 levels there were no fatalities.
Touristic eruptions require adequate monitoring and response, designed to support mitigation actions and decision-making procedures. Throughout the eruption, the goal was to provide a safe access route to the site, via regular and frequent (initially daily) meetings between the IMO’s surveillance room and its managers, selected scientists, the Civil Protection representatives, the police and rangers from the environmental institute that patrol the area on a daily basis. The latest observations, forecasts, and hazard assessment were communicated and discussed. Those attending the meeting had the chance to look at the scientific data and to consider the implications for their actions in the field, like rerouting the walking paths, diverting people to more secure areas, and closing the area when needed because of very bad weather or expected very high concentration of gas at ground level.
Throughout the duration of the eruption, the types of associated hazards were changing along with variations in the eruptive style, reflecting a dynamic eruption which required a dynamic response. The hazards included: air pollution (both SO2 and sulphates), acid precipitation, water and ground pollution, electrification of air, and, on a more local scale, ground fracturing, opening of new eruptive fissures, lava flows and lava spill over, tephra fallout and moss fires. Throughout the evolution of the eruption, the hazard assessment was also changing and needed to be adaptable to account for additional volcanic phenomena. However, different volcanic hazards had different temporal and spatial scales and for this reason different approaches were used for dealing with a local assessment and a proximal zone (within 50 km). No direct hazards were identified that posed a significant threat to the far-field, even though an occasional increase in SO2 concentration was detected as far as Akureyri on 25 June (~ 280 km NE of the eruption site). In addition to the Reykjavík Capital area, where about 237,000 people live (Statistics in Iceland, 2022), there are several inhabited centers within a distance of 40 km from Fagradalsfjall: Keflavík and Njarðvík (19 600), Grindavík (3 500), Sandgerði (1 900), Garður (1 700), Vogar (1 300) and Hafnir (100). Resulting in a total over 264 000 persons, corresponding to about 72% of the entire population of Iceland, living next to the eruption and exposed directly to primary volcanic hazards.
As soon as the eruption started a forecasting system was implemented at the IMO with the purpose of informing people living on the Peninsula about the anticipated danger due to volcanic pollution. The CALPUFF dispersal code was reinstated (used during the Holuhraun eruption) to reproduce the SO2 emission and transport from Fagradalsfjall (Barsotti et al., 2020; Scire et al., 2000). The two-days forecast ground concentration of SO2 was available on IMO’s web. The hourly forecast was shown using the color system used by the EAI (Directorate of Health, 2021) which follows the reference SO2 thresholds. Two additional maps were provided that showed the most likely impacted areas within temporal windows of 6- and 24-hours. These maps had no specific reference to the intensity of the polluting event, but rather indicated where and when the volcanic cloud was expected to affect areas at the ground. These additional maps were introduced to deal with the expectation of a very detailed and precise forecast, that was actually very hard to obtain. Indeed, the closeness of the target locations to the eruption, the highly temporal variable plume height and the relatively low frequency of SO2 flux estimates, reduced the capability of providing a reliable quantitative forecast (Barsotti, 2020). People were asked to check on the EAI website for the real-time measurement at the closest municipal station whenever the plume was forecasted to be near to them.
On a more local scale, the hazard assessment focused on dynamic phenomena that would have the potential to threaten people visiting the eruption site. The possibility of sudden fissure openings, high levels of volcanic gas, advancing lava fronts and lava outbreaks toward hiking paths, and occasional tephra fallout, all needed to be considered to ensure safe access. When the first “dangerous area” map was designed, the aim was to identify the area where additional openings were more likely to occur, based on the proximity to the modeled dyke intrusion at depth. This area was considered a high-danger region and people advised not to enter it. Subsequent hazard maps accounted for the extension of the lava field (as reconstructed mainly via aerial photogrammetry) by adding an extra buffer zone of about 100 m and was intended to prevent people from being subject to sudden lava outbreaks and exposure to fluid lava and high temperatures. The lava flow model results were not included into such maps as their temporal uncertainty was too large and the maps would have been prone to public misunderstanding. A very basic numerical tool was used to draw the areas anticipated to be affected by ejected lava spatters. The on-line ballistic trajectory tool developed by Mastin (2001) was adopted to identify preliminary contours of the hazard area effected by volcanic clasts fallout. Both calm and high-wind conditions were considered, as the wind in the Fagradalsfjall area could reach very high speeds (through the eruption duration the strongest measured wind was 34 m/s on the 12 May, and with wind gusts up to 48 m/s). The density of the volcanic pyroclasts was so low (roughly 500 g/m3) that they were transported much further in the presence than absence of wind. Two circles of 500 m and 650 m radii were eventually added to the map, identifying the area where people were advised that in given conditions such fallout might occur, see Fig. 10. This area was primarily used by the air-traffic controller for helicopters to advise on where they might land when bringing people to the eruption. The hazard map (Fig. 10) was formally published on-line by the Icelandic Meteorological Office on 4 May 2021 and endorsed by the Civil Protection which passed it through key dissemination channels like Safe Travel (safetravel.is).
As the forecast of gas pollution over such a small area would have required a very precise numerical model and high-resolution meteorological data that were not available, the decision was taken to not attempt to model it, but to inform people that pollution due to volcanic gas was expected anywhere surrounding the eruption site and they were asked to check the daily written forecast prepared by the weather forecaster on-duty, available on IMO’s website, as well as directing them to the real-time data streaming from the stations installed at the eruption site which was available on EAI’s website. In the first weeks of the eruption, when the rescue team volunteers detected elevated gas concentrations with their hand-held sensors, they would instruct people to leave the area and move up wind and to elevated places to avoid direct exposure. Advice from the CP was given to not bring young babies or pets to the eruption site. The level of pollution of surficial water was regularly reported to civil protection and recommendations were given to discourage to bring dogs or horses who might drink polluted water puddles.
Mitigation actions and coordinated response, as those listed above, guaranteed a protected visit to the eruption site for most people. However, it is also important to mention those occasions when support and intervention was required by local rescue teams. This included search and rescue for people lost during bad weather (mainly in the beginning of the eruption). Three people needed to be rescued by helicopter and several have been collected after being trapped by lava flows; more than 30 people asked for help because of exhaustion and/or hypothermia, and possibly gas poisoning (up to 30 people visited the Poison center at the Hospital for checks); 25 people suffered fractures and more than 30 needed assistance for other types of injuries (Landsbjörg, personal communication).
A completely new scenario was anticipated in the event of the lava flow advancing down Nátthagi valley and towards the sea on the south coast as no other natural obstacles would prevent the lava reaching the coastline within the remaining 1.5 km long distance. Eventually, the lava did not fill Nátthagi, but preparations for this possibility started as soon as the lava progressed further into the valley. The lava flow model, initialized for a short-term forecasting scenario, identified the most likely entry point into the sea (Fig. 11).
In Iceland, lava flow last entered the sea during the Heimaey eruption in 1973. This would trigger additional hazards to take into consideration. Amongst others the most relevant hazards considered were:
1) the production and release into the atmosphere of abundant HCl;
2) the potential for phreato-magmatic activity and the consequent production of ash in the atmosphere.
The recent experiences and lessons learnt during eruptions in Hawai’i from 2018 onwards, were initially considered to start designing the necessary instrumentation setup to monitor the air-quality aspects (Kullman et al., 1994; Mason et al., 2021). A preliminary plan discussed and agreed with representatives at the EAI included:
Installation of HCl sensors for real time monitoring nearby the entry point as well as in the closest inhabited town of Grindavík;
Regular filter-pack acquisitions to measure Cl, Fl and S compounds and heavy metals;
Installation of Optical Particle Counters nearby the entry point and in Grindavík to assess the intensity of polluting events.
The mobile-lidar was moved to the Hópsnes Peninsula by Grindavík (Fig. 1) and the co-located calibrated camera was set pointing toward the likely entry point. The initial plan was that the camera would be used in assessing the plume height and enable a rapid mass eruption rate estimate and the initialization of ash dispersal numerical models. Indeed, the vicinity of the airport required a quick response in the event of the production of ash and tephra. The plan, agreed together with the air navigation service provider ISAVIA, was to immediately change the aviation color code to red if the lava began flowing into the sea, along with an initial SIGMET (SIGnificant METeorological information) (Barsotti et al., 2020) to be issued identifying the most likely area affected by ash contamination. A re-assessment would be done as soon as additional data became available to understand the severity and the intensity of the explosive component.