Direct Observation of Spallation Bombs During The 2021 La Palma Eruption, Canary Islands, Spain


 Expulsion of incandescent pyroclasts of >64 mm from active volcanoes are known as ‘volcanic bombs’1-3, which can pose significant hazards to life and manmade structures. Most volcanic bombs are considered to fall ballistically, enabling the prediction of a volcanic hazard radius4,5. Here we describe a bomb type that was ejected ballistically, but subsequently travelled downslope a steep volcanic flank, posing an important but hitherto unrecognized impact and fire hazard. The volcanic bombs were observed to fall from the summit of the newly formed La Palma cone in late October 2021 and fell on a soft tephra substrate, travelling downhill for over a kilometer in some instances. The bombs spalled incandescent lava fragments en route, creating a significant fire hazard far beyond the range of ballistic transport. Spallation bomb hazards ought to be considered in volcanic risk assessment, necessitating an increase of hazard radius exceeding a factor of two on steep-sided volcanoes with ballistic bomb activity.


Introduction
Observations made by volcanologists over the last century have led to the description of distinct types of 'bombs' that form during volcanic eruptions. Volcanic bombs are de ned as pyroclasts, or fragments of incandescent lava, larger than 64 mm in diameter [1][2][3] . Due to their size, masses, temperatures, and ballistic trajectories, volcanic bombs represent signi cant hazards to life and infrastructure both during and after impact and need to be carefully considered during volcanic hazard assessment [4][5][6] . Indeed, volcanic bombs are a major cause of fatalities in global volcanic eruptions 7 and are a major hazard in volcano 'tourism' management 8 . During basaltic volcanic eruptions, common types of bombs have included ballistically ejected varieties, including 'breadcrust' bombs, as well as rounded features that form with ballistic ejection, such as accretionary balls (also known as lava balls), that form through addition of mass to their outer portions during rolling within lava ows. Ballistic volcanic bombs have been of particular interest for both understanding volcanic processes to examine their in-ight dynamics 4,5,9 , the mechanisms of magma fragmentation 10 , as well as the dangers that they present for the destruction of manmade structures, and the preservation of life [11][12][13] , and they are a key aspect when considering the growing consensus that lower magnitude eruptions, such as that at La Palma, can cause signi cant local, regional and global damage and potential disasters 14 .
The 2021 eruption on La Palma began at ~14:12 UTC on the 19th September and ceased major activity on 14th December 2021. It is the largest, longest, and most destructive eruption on the island in recorded history, following previous historical eruptions in AD 1585AD , 1646AD , 1677AD -1678AD , 1712AD , 1949AD , and 1971 emanating from the prominent Cumbre Vieja volcanic ridge in the southern part of the island of La Palma 15 . The eruption has led to signi cant destruction, with the loss of ~2800 man-made structures, >80 km of roads, and ~12 km 2 of crops (>1200 Hectares) (e.g., Refs 16,17). Here, we document a phenomenon that occurred at the eruption site and vent in October of the 2021 La Palma eruption, where incandescent lava bombs were observed to travel by gravity along the anks of the newly formed cone, offering insight into formation of a speci c form of previously unrecognized volcanic bomb and associated volcanic hazards.

Observations And Results
Observations were made in the o cial safety exclusion zone to the northeast of the village of Las Manchas on the western ank of the Cumbre Vieja on La Palma (28°36'32.8"N, 17°52'27.6"W), approximately 900 m southwest of the new basaltic cinder cone (28°36'54"N, 17°52'7"W). Visits were made before (26th October), during (27th October) and after (28th October) the fall of the volcanic bombs at the same site. The observation of spallation bombs occurred during the eruption phase of the new cinder cone volcano between 14:30 to 15:00 local time on 27th of October, 2021, and took place during a period when a lava was emanating from the northwestern ank of the cinder cone and owed through the northern outskirts of the village of Las Manchas. On the 26th October, part of the author team (JMDD, HG, FJPT, MA, and GG) had visited the same area and found it to be subject to intense (>4 cm/day) basaltic lapilli fall. During the afternoon of the 27th October, smaller ash and sand sized basaltic particles were falling in the area ( Figure S1).
At approximately 14:30, 27th October, another part of the team (JMDD, HG, VRT, and JCC) approached the ank of the volcano to collect xeno-pumice fragments seen to have fallen and deposit amongst the tephra during the 26th October lapilli fall. The team approached the location following the route of the LP-212 road where it was noted that ~20 roughly spherical 30-70 cm diameter basaltic bombs that were still warm to the touch occurred in the vicinity that had not been observed on the previous day. These lava bombs were well-consolidated basaltic rocks and were distributed across a region where the steep sided cinder cone intersected the lower angled sides of the Cumbre Vieja. The eld party noted the fall of large blocks and bombs by ejection from the central plume for about 200-500 m from the site of ejection, onto the anks of the steepening cinder cone ( Figure 1). During this time-period, elliptical slope-parallel indents or hollows were noted in the basaltic lapilli and sand sized ash substrate. The presence of these indents on the up slope-side of some of the bombs showed that they had been formed by travel in a rolling motion along the ground of the volcanic bombs. Some bombs showed no obvious hollows behind them suggesting that they had fallen signi cantly prior to the team's arrival in the area and the hollowed areas had been lled with ash and lapilli. The observed indents showed that some of the volcanic bombs had travelled by bouncing substantially more than 10 meters per bounce in the area of observation. The exact bombs or balls responsible for some of the tracks could not be identi ed in all cases, suggesting that they had come to rest signi cantly down slope (>0.5 km) of the observation site meaning a travelled distance of more than 1.5 km (Supplementary Video S1).
At 14:34, a large bomb was noted to be ejected and to tumble down the southwestern ank of the cinder cone (Supplementary Video S2). Ejection occurred from the central ash-laden vent and the bomb most likely ew ballistically from the vent, prior to tumbling and bouncing down the side of the cone. The bomb was easily observed to fall by the plume of dust left behind during the tumbling and bouncing phases.
After settling, the bomb was approached and found to be ~60 cm in diameter with a partially vitrophyric and dark scoriaceous crust, with a lighter grey outer basaltic shell and incandescent basaltic material in the center (Figure 2a). There was no indication for any signi cant accretion to the bomb, and no coating of ash or lapilli that might be expected from accretion. The bomb was imaged and carefully hammered to remove a sample (LP2106JD; Figure 2b). The bomb was noted to have formed an increasingly closely spaced 'rolling' track as it slowed down and eventually halted near the eld party's position. While noting aspects of the bomb's morphology, another bomb was observed to fall at 14:42 local time During observation of the bomb that fell at 14:34, a bomb was recorded to fall at 14:54 local time (Supplementary Video S3). This bomb also formed a dust trail during its transit down the slope of the cone and was seen to bounce and to 'spall' incandescent material. Based on approximation, the bomb travelled initially at speeds more than 60 km/h prior to losing energy along the shallowing slope and on impact with the soft lapilli/sand sized particle substrate. This bomb was halted within a stand of dead trees and combustion of organic material took place (Supplementary Video S3). Two minutes later (14:56 local time; Supplementary Video S4) a new bomb was observed to follow a similar track and similar speed down the side of the cone; like the previous bomb, the object spalled or fragmented incandescent material and was slowed by the substrate as well as by collision with a preexisting lava ow (Figure 3ad). On halting, the bomb was approached and found to be more ovoid in shape than the rst observed falling bomb, at around 70 cm in length and 40 cm in width. One side of the bomb, which had been observed to spall on impact, was incandescent, exposing an interior with apparent 'concentric' features ( Figure 2c).
Returning to the site at 14:00 on the 28th October 2021 revealed that minor ashfall had occurred during the preceding ~23 hour period. No more than about half a dozen new spallation bombs were noted in the area suggesting that the main phase of activity to produce these speci c spallation bombs was between the evening of the 26th October 2021 and ~15:00 on the 27th October 2021, a period of <24 hours. The ashfall was su cient to partly to totally ll some of the hollows observed to have formed during the previous day.
To con rm a heritage with the recent eruption, the bomb sample (LP2106JD) was analyzed for majorand trace-element compositions along with i) the recent lava ow to the north on to which some of the bombs spalled; ii) the lava that was erupting coincident with the formation of the spallation bombs; iii) lavas erupted earlier, i.e., in September and earlier in October 2021, and iv) undisturbed tephra deposits immediately underlying the spallation bombs. The spallation bomb has 7 wt.% MgO, with a major element composition like the underlying tephras and associated lavas (6.4-7.8 wt.% MgO) and similar trace element abundances including similar to marginally higher Ni (130 ppm) and Cr (348 ppm) contents compared with the associated lavas and tephras (Ni = 98-130 ppm; Cr = 242-328 ppm; Table S1). These values are higher than those measured in the lavas erupted from the volcano in September (MgO= ~6.4 wt.%; Ni = ~60 ppm; Cr = ~105 ppm) and con rm the heritage of the spallation bombs from the ongoing eruptions on the 27th October 2021 ( Figure S2) rather than being reworked volcanic blocks.

Discussion
The basaltic bombs or balls that fell on the 27th October 2021 at La Palma were observed to be ejected and then to y ballistically, bounce and roll as solid incandescent materials ( Figure 4). They were recorded to roll and bounce down the mountain side, cushioned in part by the relatively soft lapilli and sand-sized substrate. Since the materials were ejected, we elect to call them bombs rather than balls (the term 'egg' has been used in the popular press, but does not re ect the diverse shape of objects observed).
The materials were observed to bounce and roll, and to lose, rather than gain material. For this reason, these bombs are not accretionary lava balls such as the famed 'Teide's Eggs' (Refs, 18, 19), but rather experienced erosive loss, by spallation of material, sometimes in large chunks, from their exteriors ( Figure  3). For these reasons, we term the phenomena volcanic 'spallation bombs'. This term is consistent with the requirement for viscous lava to provide shape and with the observation of fragmentation or spalling during their fall (Supplementary Videos S2-S4). Remarkably, spallation bombs during the 27th October 2021 were well-preserved due to their fall through a soft substrate of ash and lapilli. The observation of the characteristics of these bombs suggests that, on impact with hard substrate, such as a solidi ed lava ow, they will likely fragment and their preservation may then be considerably limited.
Another notable aspect of the spallation bombs is the presence of a 'track' formed during tumbling and Indian Ocean 23 , Stromboli, Italy 11 , and Bogoslof Volcano, Alaska 24 . These reports bear passing similarity to those now observed and recorded in detail on La Palma in 2021 and imply that the phenomenon of spallation bomb activity is in fact far more widespread, but that gravitational fall after ballistic fall has not been well constrained. This is particularly problematic when considering distance of fall from vents of preserved projectiles from older eruptions as well as from a hazard perspective (e.g., Refs. 4,5,8,10,25) ( Figure 4). Without recognition of gravitational tumbling and fall, ballistic distances and, hence, eruption column energy, may be overestimated. A notable feature of the falling spallation bombs on La Palma is that they came to rest proximal to the edge of a preexisting lava ow. Had the fall of these spallation bombs not been viewed directly, then their concentric inner portions and proximity to the lava ow might well have led to their misidenti cation as accretionary bombs or 'lava balls' associated with the lava ow itself.
An important feature of the observations of spallation bombs at La Palma was the controlled environment in which they were observed. All of these phenomena fell within the 'exclusion zone' determined by authorities, where only permitted scientists were allowed to make measurements and observations. The large exclusion zone incorporated the possibility of lavas owing to the sea, and a notable feature of the 27th October 2021 spallation bombs was their coincident formation at the time of lava emanation from the 2021 La Palma volcano. This lava, which owed over the preexisting 19th to 30th September lava ow, led to renewed destruction in the village of Las Manchas. Prior to the 26th October 2021, lava emanation from the volcano had been restricted for several days, and there was continuous explosive eruption of lapilli and bomb-sized material from between three to four vents along a NW-SE oriented ssure at the top of the volcanic summit.
Hazard assessment in connection to spallation bombs, include, in the rst instance, that their ballistic ejection was followed by gravitational tumbling on the soft tephra substrate. This meant they had the potential to travel far greater distances than ballistically ejected bombs alone, especially on the steep anked cone of the 2021 eruption that itself took place on the ank of the Cumbre Vieja ridge. As such, they represent a signi cant impact hazard well beyond the area of ballistic transport and in some senses represent similarities to a limited particle pyroclastic density current common to many more explosive volcanoes 1,2 . Hazard assessment on volcanoes must therefore take into consideration this additional gravitational impetus for volcanic bomb hazard even in a basaltic and hence generally low explosivity environment. In the case of La Palma, the excluded zone was su cient to accommodate the spallation bomb hazard and offers an excellent case study in appropriate volcanic hazard management. In the case of the observations made here, team members were aware of the hazard and able to spot the spallation bombs. Elsewhere globally where this might not be the case, a sudden apparent appearance of spallation bomb hazard during an eruption is conceivable. For instance, when visibility is reduced due to weather or additional eruptive phenomena, or if rolling bombs suddenly change direction, as the videos in the supplementary material demonstrate that their exact travel trajectory is not always predictable. This previously unaccounted impact hazards to affected areas may necessitate the widening of volcanic exclusion zones beyond the areas affected by ballistic (strombolian) eruption effects only. The fragmentation and spallation of the bombs against harder objects can also lead to the potential for multiple spalled hot fragments and shrapnel outside of the immediate (ballistic) danger zone.
Another critical aspect of hazard is the hot, incandescent nature of the bombs. One bomb was witnessed to land in dead tree stands, leading to smoke from combustion of organic materials (Supplementary Video S3). The potential to ignite forests, timber structures or other ammable constructions would be severe from spallation bombs, since such objects travel further than other bomb types and due to spallation and fragmentation, which can lead to incandescent material spalling and igniting res at several sites simultaneously. This issue is exempli ed well beyond La Palma. For example, on Mt. Ontake, Japan, casualties have been recorded from ballistic ejecta, and consequently, improved designs have been proposed for wooden structures to protect property and life 11 . Volcanic ballistic projectiles have also been considered a hazard on Mt. Chinshin, Taiwan 7 , but we stress that spallation bombs will greatly increase the extent of pyroclastic hazard on volcanoes well beyond the range of ballistic transport only. Speci cally, the volcanic spallation bombs of the 2021 La Palma eruption suggest that the danger from volcanic pyroclasts is at least double that normally assumed from ballistic bombs alone and should be considered in future hazard risk assessment at active volcanoes of this type, as well as in the study of historical volcanic bomb deposits. This hazard may be particularly important where volcanoes and society (including 'volcano' tourism) closely intersect. At La Palma, the authorities took a cautious but sensible approach to volcanic hazard, enabling observations of spallation bomb phenomena in a restricted exclusion zone, thus keeping populations safe throughout the entire history of the 2021 eruption. The observation of this hitherto unrecognized volcanic hazard adds to the growing consensus that lower magnitude eruptions, such as that at La Palma, can cause signi cant local and regional catastrophe.

Observations and Video Preparation
Observations were made on the 26th and 27th of October and recorded as photographs and videos. All media was geotagged and edited to include observation date and time and converted to common digital formats to ensure playback on most devices. Videos are provided in the supplementary information.
Bulk rock major-and trace-element abundance analyses Hand samples were sawn, surfaces cleaned and then approximately 100g of material was disaggregated using an alumina crusher. The crushed material was then coned-and-quartered and ~30-40 g of material was turned to a ne (sub-10 micrometer) rock our using a SPEX alumina shatterbox. Analytical procedures were undertaken at the Scripps Isotope Geochemistry Laboratory (SIGL). For trace-element abundances, ~100 mg of sample powder was digested in Te on-distilled concentrated HF (4 mL) and HNO 3 (1 mL) for >72 hrs on a hotplate at 150°C, along with total procedural blanks and terrestrial basalt and andesite standards (BHVO-2, BCR-2, BIR-1a, AGV-2). Samples were sequentially dried and taken up in concentrated HNO 3 to destroy uorides, followed by doping with indium to monitor instrumental drift during analysis, and then diluted to a factor of 5,000. Major-and trace-element abundances were determined using a ThermoScienti c iCAP Qc quadrupole inductively coupled plasma mass spectrometer (ICP-MS) in low resolution mode and all data are blank-corrected. Reference materials were analyzed as unknowns to assess external reproducibility and accuracy (see also Ref. 26). Reproducibility was generally better than 5% relative standard deviation (Table S1). For major elements, it should be noted that silica content was attained by difference. This means that, for some samples with signi cant water content (e.g., ashes within tephras) the SiO 2 content may be overestimated. The other major element oxides are not subject to these uncertainties. The limits of quanti cation using these techniques were signi cantly below the measured values reported.  The eruption column and associated debris fall during the 27 th October 2021 showing spallation bombs in the foreground. Unlike conditions on the previous day, where ejected material occurred all along the NW-SE trending ssure, material was dominantly ejected from the center and NW end of the ssure system on that day (also see Supplementary Video Content).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.