Growth and evolution of the lava dome and coulée during the 2020–21 eruption of La Soufrière, St Vincent Open Access

The extrusion of lava at the surface of a volcano in the form of a dome is a common feature of many volcanic eruptions in convergent settings e.g. Soufrière Hills, Montserrat (Watts et al. 2002), Mount St Helens (Vallance et al. 2008), Merapi (Surono et al. 2012), and Redoubt, USA (Diefenbach et al. 2013). Approximately 6% of all volcanic eruptions worldwide involve the growth of some form of lava dome (Calder et al. 2015).

The La Soufrière volcano on St Vincent (13.33° N; 61.18° W) is located in the northern part of the island (Fig. 1). Historical eruptions have consisted of both explosive (1718, 1812, 1902–03, 1979) and effusive (1784, 1971–72) activity, with some eruptions consisting of both explosive and effusive phases (e.g. 1979, Aspinall et al. 1973; Shepherd et al. 1979; Cole et al. 2019). The eruptive products have consisted largely of basaltic andesite magmas (Cole et al. 2019; Fedele et al. 2021).

The summit (at 1204 m a.s.l) is dominated by a nearly circular, 1.5 km wide crater, with nearly vertical walls that are 100–300 m high. Prior to the 2020–21 eruption, the Summit Crater contained a large 120 m high × 800 m wide lava dome that was erupted during the most recent eruption in 1979 (Fig. 1) (Shepherd et al. 1979). The dome was surrounded by a dry ‘moat’, except in the northern sector of the crater where the 1979 dome was in contact with the base of the inner crater walls. Weak fumarole activity was present in two areas, on the eastern and southern sectors of the 1979 dome, and there was a small lake in the SE sector of the Summit Crater.

The 2020–21 eruption consisted of two distinct phases. The initial, effusive phase was characterized by the extrusion of a small lava dome inside the Summit Crater, adjacent to the 1979 lava dome. This occurred between 27 December 2020 and 9 April 2021. It was followed by an explosive phase between 9 and 25 April 2021, which consisted of more than 30 discrete sub-Plinian and Vulcanian explosions (Cole et al. 2023; Sparks et al. 2023). This paper describes the growth and evolution of the lava dome and coulée (Table 1). A companion paper uses the volume and growth rate data to investigate the rheological properties of the magma in both the effusive and explosive phases of the 2020–21 eruption (Stinton et al. 2023), while papers by Weber et al. (2023) and Morrison-Evans et al. (2023) discuss additional aspects of the lava and plumbing system.

Growth of the lava dome was monitored primarily through photogrammetry (e.g. Diefenbach et al. 2013), using images acquired from the rim of the Summit Crater or aerial images from fixed-wing aircraft, helicopters and consumer-grade unmanned aerial vehicles (UAVs). The image processing program ImageJ (Abràmoff et al. 2004) was used to estimate size, volume and growth rate for oblique images acquired during observation flights or from the crater rim. For images acquired by helicopter or UAV aerial photograph surveys, the photogrammetry software Agisoft Metashape Professional was used to generate 3D models from which the volume and growth rate could be calculated.

There is some uncertainty on the volume and growth rate data calculated from the photogrammetry models, derived from the positional errors of the helicopter or UAV and also from the lack of a high-quality pre-eruption digital elevation model (DEM). Positional errors for the four UAV surveys are in the range ±0.98 to 3.34 m, while for the helicopter-based survey the error is estimated to be of the order of ±10 m. Using a conservative estimate of 5 ± 2.5 m for uncertainties in length and height yields a minimal uncertainty of approximately 3% for the 19 March 2021 photogrammetry survey. In addition to this, the lack of high-quality pre-eruption DEMs mean assumptions had to be made concerning the general shape of the new lava dome. During the latter stages of the dome growth, the dome was assumed to have a purely flat base and exhibited a pure half-ellipsoidal shape, except for the final four photogrammetry surveys when the surface volume tool in ArcGIS Pro was used. In reality, where it reached the 1979 lava dome and the inner slopes of the Summit Crater wall, the new lava dome had a slightly trapezoidal cross-section. Consequently, the volume data presented in Table 2 are overestimated by as much as 25–30%.

The photogrammetry surveys were conducted at intervals of up to several weeks constrained by access and safety concerns (the UAV-based surveys were conducted from locations along the rim of the Summit Crater). Between surveys, radar and multispectral imagery from the European Space Agency's Sentinel-1 and -2 satellite constellations and from Planet.com were used to track the extent of the footprint of the new lava dome (Fig. 2).

Limited precursory activity was detected prior to the emergence of lava on 27 December 2020. Beginning in November 2020, the Seismic Research Centre (SRC) observed a small increase in seismicity while slight increases in fumarole activity were observed on the 1979 lava dome during a visit to the Summit Crater on 16 November 2020 by staff from the Soufrière Monitoring Unit (SMU) of the National Emergency Management Organisation (NEMO). Seismicity averaged 2 events per day until 23 December, after which it decreased to very low levels of less than 0.6 events per day (Joseph et al. 2022; Thompson et al. 2022).

Although not initially recognized (see Joseph et al. 2022), pre-eruption deformation in Global Positioning System (GPS) data was identified beginning in July 2020 (Camejo-Harry et al. 2023). This inflation signal, which continued throughout the effusive phase, was best modelled as a volume increase at a depth of approximately 18 km (Camejo-Harry et al. 2023). In addition, a much shallower signal, modelled as a north–south-orientated dyke down to 0.7 km depth, was identified in Interferometric Synthetic Aperture Radar (InSAR) imagery acquired 19–31 December.

Post-eruption analysis of imagery acquired by the Advanced Baseline Imager on the GOES-16 geostationary satellite revealed the presence of some thermal anomalies over the Summit Crater before the emergence of lava at the surface (Thompson et al. 2022). The first significant anomaly (>0.9 K above the background) occurred on 13 December and was followed by a second anomaly on 22 December. These anomalies corroborate the visual observations of SMU staff on 16 November of minor increases in fumarole activity on the 1979 lava dome.

The growth and evolution of the new lava dome and its evolution into a coulée at La Soufrière can be divided into six stages (Table 1).

Following the limited precursory activity, the first indication of significant surface activity was a thermal anomaly detected by the NASA Fire Information Resource Management System (NASA-FIRMS, Schroeder et al. 2014). Imagery collected by the Visible Infrared Imaging Radiometer Suite (VIIRS) instrument at 17:30 UTC on 27 December 2020 contained a single anomaly centred in the SW sector of the Summit Crater, between the 1979 lava dome and inner crater wall. It was not until 29 December, however, that visual observations confirmed that this thermal anomaly was due to the presence of lava at the surface.

Given the little to no seismicity recorded at the time, the exact time when lava first emerged is not known. However, following a call to the public, the SRC received a series of photographs taken by tourists from the crater rim on 27 December at approximately 14:00 UTC. The photographs clearly showed a small, roughly circular uplifted area measuring approximately 40 m in diameter (Fig. 3a), located about 400 m SW of the presumed conduit that had fed the 1979 lava dome. Based on the timing of the photographs, the emergence of lava likely began several hours prior to the detection of the thermal anomaly. This is corroborated by a Dove Classic and a SuperDove image (from Planet.com) acquired at 14:13 and 14:39 UTC, respectively, on 27 December that appear to show a dark area on the floor of the Summit Crater with a small white plume, neither of which were present in similar imagery from 16 December.

The second stage occurred between 28 December 2020 and approximately 3 January 2021. Visual observations from the crater rim by two members of the SMU on 29 December confirmed the presence of a small, very dark-coloured lava dome forming in the SW sector of the Summit Crater (Fig. 3b), between the 1979 lava dome and the inner wall of the Summit Crater. The small dome had a remarkably near-perfect hemispherical shape, with an upper surface that consisted of a ropey-like texture covering the majority of the surface and a lower region that comprised a rubbly apron of debris that formed a ‘skirt’ around the base. The rubbly debris is presumed to result from the fragmentation of lava as the dome expanded over the surface of the crater floor. A small depression was present in the summit of the dome, through which gas and steam vented. This feature, which persisted throughout the growth of the lava dome, was presumed to be roughly located above the vent in the floor of the Summit Crater.

Based on the photographs acquired on 29 December from the crater rim, the dome had a radius of c. 50 m, giving it a volume of c. 0.26 × 10⁶ m³. Additional photographs acquired on 30 and 31 December from aircraft showed continued growth of the hemispherical dome, such that by 31 December, the lava dome had a radius of approximately 70 m and a volume of 0.72 × 10⁶ m³. The time-averaged extrusion rate during this stage was low, varying from 1.5 to 2.6 m³ s⁻¹ (Table 2; Fig. 4). The little to no seismicity that had occurred since 20 December continued throughout this phase, indicating that emergence of the lava and then growth of the hemispherical dome occurred essentially aseismically.

Sometime around 3 January 2021, the eastern margin of the new dome began encroaching on the 1979 lava dome. This resulted in a change in shape from hemispherical to more elliptical and a concentration of growth towards the NNW, west and SSE.

On 5 January, photographs acquired from the rim of the Summit Crater revealed that the dome had increased significantly in size to measure 200 × 160 m across, with a maximum thickness of 70 m, giving a volume of approximately 1.4 × 10⁶ m³. A further 0.12 × 10⁶ m³ was added by 6 January, bringing the total volume to 1.53 × 10⁶ m³ and the dome dimensions to 260 × 160 × 70 m. The time-averaged extrusion rate during this stage varied little between 1 and 2 m³ s⁻¹ (Table 2; Fig. 4).

Observations made from the Summit Crater rim on 5 January 2021 showed that there was very little in the way of rockfall activity, a trait that continued throughout the dome's existence. The elliptical dome continued to show the ropey-like carapace and rubbly skirt that was present on the hemispherical dome. In addition to venting through the small depression in the summit of the lava dome, steam and gas were also venting from along the contact between the 1979 dome and the new dome.

Growth of the elliptical dome continued until approximately 13 January 2021 when it filled the width of the moat between the 1979 lava dome and inner wall of the Summit Crater.

The first (helicopter-based) aerial photogrammetry survey over the growing dome was carried out on 18 January. This survey showed that between 6 and 18 January, the lava dome reached a total volume of 3.25 × 10⁶ m³ and measured 364 m long by 214 m wide and was 80 m high. Thus, c. 1.77 × 10⁶ m³ had been added at an average rate of c. 1.67 m³ s⁻¹. Staff from the SMU carried out a further four aerial photogrammetry surveys between 24 January and 19 March by small UAVs from the Summit Crater rim. These surveys revealed a more than fourfold increase in total volume, reaching an estimated 13 × 10⁶ m³ (Table 2). Due to safety concerns, from 19 March until the destruction of the lava dome, its size was tracked using satellite imagery rather than by conducting UAV surveys that required people to be close to the growing dome. By 4 April, the lava dome measured c. 1 km long, 250 m wide and 105 m high and occupied c. 25% of the moat surrounding the 1979 lava dome (Figs 2 & 4; Table 2).

Observations from helicopter flights over the lava dome between 14 and 18 January revealed some changes to morphology. As a result of the new dome becoming constrained by the 1979 lava dome and the inner wall of the Summit Crater, growth of the new dome occurred primarily in north-northwestern and south–southeasterly directions away from the original vent (Figs 2 & 3c). This confinement and flow away from the original vent meant the new dome acquired the characteristics of a lava coulée (Fink and Anderson 2000; Calder et al. 2015). Tracking of the growth of the coulée in satellite imagery showed that the two lobes appeared to grow independently of each other (Fig. 3). The ropey-like surface of the dome continued, but aerial photographs and observations from the helicopter flights suggested that the morphology varied slightly between proximal (to the vent) and distal parts of the lobes/coulée. In the proximal area, the ropey-like ridges formed a spiral-like pattern around the depression in the central, vent area, while in the more distal parts, the spiral pattern morphed into curvilinear features that were aligned perpendicular to the direction of growth of the lobes. Small levee-like features along the edges of the dome where it was in contact with the inner crater wall and the 1979 lava dome also developed. In addition, towards the terminus of the northern lobe, the surface of the lobe developed a blocky texture (Fig. 5b), first seen on 3 March 2021 and several small spine-like features (c. 10 m high) appeared on 19 March 2021 (Fig. 5c). Identified in images acquired by a camera that had been installed on the crater rim, the spines appeared to move southeastwards away from the central vent region of the coulée and did not change in height after they were first observed. The spine–like features persisted until the destruction of the dome and no other similar features appeared.

Thermal imagery of the new lava dome was collected between 14 and 18 January 2021 using a FLIR T650sc handheld camera during helicopter flights and from the ground during a sampling mission on 16 January 2021. The thermal images acquired from the helicopter revealed two main areas with elevated temperatures on the surface of the lava dome. There was a broad area surrounding the depression in the summit of the dome where temperatures, as measured from more than 500 m away, were in excess of 300°C. A second main area of elevated temperatures correlated with the skirt from which very occasional rockfalls occurred. These rockfalls revealed warmer parts of the shallow interior of the lava dome where temperatures of up to 590°C were recorded in thermal images acquired at distances of approximately 50 m from the front of the lava coulée.

On 16 January 2021, samples of the lava were collected from a small rockfall deposit at the front of the northern lobe of the lava dome. Visual observations of the samples on collection showed that the lava varied from black to dark grey and reddish-brown in coloration, and was also scoriaceous in nature. Geochemical analysis of the samples revealed a basaltic andesite bulk composition and a phenocryst assemblage consisting of plagioclase, clinopyroxene and Fe–Ti oxides with rare olivine and abundant gabbroic clots (Joseph et al. 2022). This is broadly similar to previous eruptions (Fedele et al. 2021).

Seismicity associated with the dome growth underwent a significant change on 17 January, when average event rates suddenly jumped from less than 2 events per day to more than 60 events per day. Low frequency events (in the range 0.5–5 Hz, Lahr et al. 1994; Green and Neuberg 2006) were recorded only by the closest seismic stations, indicating a shallow source. The events were interpreted as being related to the emplacement of the lava dome, either as it moved across the surface or from the magma moving up through the new conduit feeding the dome (Joseph et al. 2022).

Beginning sometime on 6 or 7 April 2021, the style and rate of growth suddenly changed. Images from a camera installed on the rim of the Summit Crater captured rapid inflation of the central (vent) region of the lava coulée, such that it developed a prominent bulge with clear incandescence observed at night over the crater rim from the Belmont Observatory and the jetty in the nearby town of Chateaubelair. This rapid inflation was associated with cyclical episodes of vigorous venting from fractures that radiated out from the depression in the central part of the coulée. The venting, which followed a volcano tectonic swarm on 5–6 April and may also have coincided with the occurrence of banded tremor on 8 April (Joseph et al. 2022), included steam, gas and minor amounts of ash, and generated plumes that extended downwind of the Summit Crater.

Although it was not possible from photogrammetry to obtain volume and growth estimates between 19 March and the coulée's destruction, analysis of several sources of satellite radar imagery suggested that the growth rate increased by an order of magnitude to 17.5 m³ s⁻¹. This rapid spike in growth rate resulted in a final dome volume estimated at 19.4 × 10⁶ m³ (Dualeh et al. 2023).

The rapid inflation and venting continued until the coulée was destroyed in a series of powerful sub-plinian explosions beginning at 12:41 UTC on 9 April 2021. On the basis of component analysis of the ash fall sequence from the explosions and satellite radar imagery collected on 10 April, the new lava coulée was destroyed by the initial explosions that occurred on 9 and 10 April (Cole et al. 2023). In its place, an 800 m wide, 300 m deep crater was formed as a result of the explosive sequence, which also removed parts of the 1979 lava dome and exposed its feeding conduit. No evidence of the conduit that fed the 2020–21 lava dome could be identified in the new crater.

The historical eruptive history of La Soufriére, St Vincent consists of a mix of explosive and effusive eruptions (Cole et al. 2019). The last two eruptions, in 1971–72 and 1979, both involved the extrusion of significant volumes of lava either with (1979) or without (1971–72) an explosive component and similarities between the 2020–21 and these two historical eruptions can be made. A brief summary of both eruptions follows. More information about the 1971–72 eruption can be found in Aspinall et al. (1973) and Shepherd and Aspinall (1982), while further detail on the 1979 lava dome is presented in Shepherd et al. (1979), Graham and Thirlwall (1981) and Huppert et al. (1982).

Due to the absence of a near-vent instrument, no precursory seismicity was recorded prior to the onset of the 1971–72 eruption (Shepherd and Aspinall 1982). The first indication that an eruption was underway came in the form of abnormal activity (steaming, discoloration, and a sulfurous smell) in the crater lake, which is thought to have started sometime in late September 1971 (Aspinall et al. 1973). It was only after seismometers were installed on the crater rim that small events identified as originating from the Summit Crater were recorded (Shepherd and Aspinall 1982). By 3 November, the lake surface temperature measured 81.5°C and the level had risen 26 m above normal. Lava broke the surface of the lake on 20 November, by which time it was estimated that c. 43 × 10⁶ m³ of lava had been extruded at c. 10⁶ m³ per day or 1 m³ s⁻¹ (Aspinall et al. 1973). Following appearance of the lava island, it continued to rise at a rate of between 0.5 and 1 m per day. By the time growth stopped in March 1972, the island had reached a thickness of approximately 295 m and an estimated 80 × 10⁶ m³ of lava had been erupted (Aspinall et al. 1973). The lava that formed the island initially emerged as a series of long, narrow ridges radiating from a central depression that coalesced to form a roughly circular, relatively flat-topped and steep-sided island 600 m in diameter. The ridges consisted of lenticular slabs or sheets 0.5 m thick and tens of metres in width/length that may have represented units separated by cooling joints. Joint surfaces exposed lava that was brownish, oxidized and vesicular. No evidence of shatter zones or true slickensides in joint surfaces was identified, although columnar jointing perpendicular to slabs was seen, as were pillow-like structures in the northern and western parts of the island (Aspinall et al. 1973). Chemical analysis of lava samples showed it to be a dark-grey basaltic andesite with phenocrysts of plagioclase, olivine and pyroxene, with a silica content of 54.1% and chemically comparable to magmas associated with previous eruptions (Aspinall et al. 1973).

The 1979 eruption was preceded by more than 12 months of precursory activity that, in addition to minor changes in the lake behaviour, also included significant low frequency seismicity. This period of unrest was followed by a series of large phreatomagmatic explosions over a period of 13 days (Shepherd et al. 1979; Shepherd and Sigurdsson 1982). The explosions removed c. 25% of the 1971–72 lava dome, consumed the crater lake and raised the crater floor by about 28 m above the pre-existing lake level. Shortly after the explosive sequence ended, lava was observed extruding onto the crater floor and this continued though to October 1979. On 7 May 1979, the lava dome had reached c. 30 m high and 60 m in diameter. The lava dome consisted of five lobes radiating from a central vent, giving the dome a roughly circular footprint. Growth of the lava dome lobes was discontinuous with periods of little to no lateral spread while the dome still increased slightly in height. The dome spread laterally in two separate ways: through the accretion of loose, newly-cooled blocks tumbling down the flanks; and also, partly by the horizontal movement of large blocks or sections of the lobe fronts that caused crater floor sediments to pile up like a bulldozer pushing earth ahead of it. By 2 October, when growth ceased, the dome measured 133 m high and 868 m in diameter. The final volume was estimated at 47.1 × 10⁶ m³, while the extrusion rate peaked at c. 10 m³ s⁻¹ early in the dome's growth, but averaged c. 3 m³ s⁻¹ over the whole period of dome growth. As with the previous dome-building eruption in 1971–72, samples of the lava showed it to be basaltic andesite with c. 45% phenocrysts or plagioclase, clinopyroxene, orthopyroxene, titanomagnetite and minor olivine.

Comparing the effusive phase of the 2020–21 eruption with both the 1971–72 and 1979 eruptions highlights a number of significant differences with one or both of the previous dome-building episodes, and also one similarity (Table 3).

Beginning with precursory activity, as summarized above and described elsewhere (Joseph et al. 2022; Thompson et al. 2022; Camejo-Harry et al. 2023), there was only limited precursory activity prior to the onset of lava extrusion in December 2020, some of which was only identified in post-eruption analysis of satellite imagery (Thompson et al. 2022). This limited precursory activity is similar to the apparent lack of observed activity prior to the 1971–72 eruption (though this is complicated by the lack of near-vent seismometers and the presence of a large lake in the Summit Crater), but is in total contrast to the 12 months of strong precursory activity prior to the onset of explosive activity in 1979. The limited precursory seismicity prior to the emergence of the 2020–21 lava dome (Joseph et al. 2022; Latchman and Aspinall 2023) suggests that the source magma found an easy route to the surface that likely involved pre-existing pathways that had not been completely sealed off by solidified magma. Given the relative short period of time since the previous eruption (in 1979), Stinton et al. (2023) suggest that this may not have been enough time for magma to have cooled and fully crystallized. In contrast, prior to the 1995–2010 eruption of Soufrière Hills, Montserrat, there was over 3 years of precursory seismicity that had also been preceded by at least two other seismic crises at roughly 30-year intervals, suggesting the fresh magma needed to create new pathways to the surface following more than 300 years of inactivity (Ambeh and Lynch 1996; Aspinall et al. 1998). This is supported by the deformation signal identified in InSAR imagery as a shallow dyke (Joseph et al. 2022; Camejo-Harry et al. 2023) that occurred during a period of limited seismicity (Joseph et al. 2022; Thompson et al. 2022). In contrast, the increase in seismicity observed prior to the onset of the explosive phase, clearly indicates that new magma rising from the deep was having to create new pathways to reach the surface. Both of these factors suggest that the lava in the new dome was remobilized from remnant magma residing in shallow levels of the plumbing system. Stinton et al. (2023) and Weber et al. (2023) discuss the implications of this further.

The estimated 16–19 × 10⁶ m³ extruded during 2020–21 is less than half of that erupted in 1979 (c. 47 Mm³) and roughly a fifth of that erupted during the 1971–72 eruption (c. 80 Mm³). The timing of explosive activity is another obvious difference between the three eruptions. In 2020–21, the eruption transitioned from an effusive phase to a highly explosive phase, completely destroying the 2020–21 lava coulée as well as most of the 1979 lava dome. In contrast, the 1979 eruption began with a very violent phase of explosive activity that was quickly followed by a month's-long effusive phase. These key differences between the 2020–21 and 1979 eruptions are linked and can be attributed to the source for the magma for the 2020–21 eruption. The explosive phase at the start of 1979 effectively cleared the path for fresh magma to reach the surface and produce the large lava dome. In contrast, the magma that was sourced for the lava coulée in 2020–21 is thought to be a degassed and partially crystallized magma left over from previous eruptions. This remnant magma was squeezed out by fresh gas-rich magma that once close enough to the surface resulted in the violent explosive phase in April 2021. Stinton et al. (2023) expand on this as part of a conceptual model for the 2020–21 eruption.

The morphology of the domes in 1971–72, 1979 and 2020–21 was also quite different. The 2020–21 dome started out as a near-perfect hemisphere (Fig. 3b) as the lava initially pushed up through the floor of the Summit Crater and then spread slowly over the crater floor. This morphology is similar to the near-circular, low profile axisymmetric dome produced in analogue experiments by Fink and Griffiths (1998), although the extrusion rate for the dome in 2020–21 is quite low, while those of the experiments were among the highest used. As extrusion continued, the shape of the 2020–21 dome became elliptical (Stage 3) and then evolved into an elongated dome with two lobes that exhibited coulée-like features (Stage 4, Fig. 3c). This overall change in shape was caused by confinement between the 1979 lava dome and the inner walls of the Summit Crater. The transition from unconfined (in Stage 2) to confined flow (Stages 3 and 4) restricted the ability of the lava to spread evenly across the moat surrounding the 1979 dome, resulting in the lava coulée that was eventually destroyed. While not a true coulée, the application of the term here is appropriate because of the transitional nature between a true lava dome and a true lava flow, as per the definitions of Fink and Anderson (2000) and Calder et al. (2015). Growth of the 2020–21 coulée is in strong contrast to the lava extrusions in 1971–72 and 1979, both of which were largely unconfined. The growth of the 1971–72 lava dome provides the starkest contrast given that the lava extrusion occurred when the Summit Crater was occupied by a large lake. The lava island that formed in 1971–72 consisted of a series of long narrow ridges radiating from a central vent that eventually coalesced to form a relatively flat-topped construction 600 m in diameter and nearly 300 m thick (similar to the lobate domes produced by Fink and Griffiths (1998)). In certain areas, columnar jointing and pillow structures were observed (Aspinall et al. 1973) neither of which were observed at any time during in 2020–21 (due to the complete absence of a lake or other surface water). A crater lake was still present at the start of the 1979 eruption, but this was consumed by the powerful phreatomagmatic explosions, fallout from which raised the crater floor by more than 25 m. The new dome eventually consisted of five large lobes that coalesced to give the dome an overall circular footprint 868 m in diameter and 133 m high (Huppert et al. 1982).

The surface texture exhibited by the lava can be loosely described as ropey, with small, decimetre-scale ridges radiating out in a spiral from a depression in the summit believed to be over the original vent (Fig. 5a), in a similar manner to the curvilinear features observed by Fink and Griffiths (1998) on their axisymmetric dome. A ‘skirt’ surrounded the base of the lava dome, which appeared to expose lava that was more massive in appearance. Observations made during the sampling mission on 16 January revealed that the lava was also variably scoriaceous, based on samples collected from what appeared to be one of the rare rockfalls that occurred. The ropey surface texture and the skirt remained throughout the life of the dome and its evolution into a coulée. Textural changes that did occur involved (1) the development of a blockier texture in the distal region of the northern lobe, and (2) the appearance of several small spines on the surface of the dome in mid-March.

The blocky texture appeared late in the evolution of the coulée, forming only in the distal region of the northern lobe (Fig. 5b). The blocks present on the surface were not carried from the vent. The distribution of the blocks was not random, but developed in curvilinear features. The blocky texture enhanced disintegration of the lobe in the skirt, such that a small apron of debris accumulated at the foot of the lobe. In contrast to this, the blocks on the 2020–21 lava coulée appear to be a result of the break-up of the surface of the more distal part of the lobe, most likely related to cooling of the lobe and shrinkage of the surface due to enhanced spreading of the lobe as it entered a section of the moat that appeared to be slightly wider. This method of block formation due to enhanced spreading and low extrusion rate leading to break up of a surface crust is in contrast to block formation identified by Watts et al. (2002) on the andesitic lava dome at Soufrière Hills, Montserrat. Here block formation occurred primarily over the vent and was associated with high extrusion rates. The resultant blocks were carried away from the vent as extrusion continued and also broke apart with time and distance from the vent, leading to a decrease in block size with distance from the vent. Block formation at the vent has also been observed on lava domes at other volcanoes including Mount St Helens and Mount Unzen (Fink and Anderson 2000).

Several small spine-like features appeared on the surface of the lava coulée on 19 March 2021 (Fig. 5c). The spines were c. 10 m high and were quickly carried away from the central region of the coulée to the SE. The occurrence of these spines is curious. Watts et al. (2002) showed that spines formed on the lava dome at Soufrière Hills, Montserrat during periods of very low extrusion rates (<0.5 m³ s⁻¹). While a photogrammetry survey was carried out on the day the spines were identified, the period covered by this survey is 35 days, and the average extrusion rate was estimated at 2.11 m³ s⁻¹, one of the highest averages for the 2020–21 coulée to that point. This implies that the spines formed at the end of a period of relatively high growth, although it is entirely possible for the growth rate to have slowed significantly in the days prior to the survey on 19 March 2021. Tracking of the spines via the remote camera proved difficult as the camera, which was used to obtain views over as much of the coulée as possible from a fixed location, could be rotated and zoomed. Therefore, it was not possible to ensure a consistent field of view, and no more UAV surveys were carried out due to safety concerns.

Measurements made to monitor the growth of the 1979 dome estimated the slopes at the margins to be c. 35°, in complete contrast to the near-vertical margins of the 2020–21 one where not confined by the Summit Crater slopes and the 1979 lava dome. This difference in margin slopes, combined with the loosely compacted nature of the recently emplaced tephra deposits associated with the explosive phase of the 1979 eruption, can explain why some of the lobes in 1979 acted like a bulldozer deforming the tephra sediments as the lobes advanced across the crater floor. There was no evidence of this sort of behaviour during the growth of the 2020–21 coulée, with observations suggesting the lava moved entirely across the surface of the crater floor without deforming the sediments and burying the few obvious rockfall deposits that were seen.

Using the growth data in Fig. 4 and Table 2, Stinton et al. (2023) applied the model of Huppert (1982) and Huppert et al. (1982) to estimate viscosity at two different times in the growth of the lava coulée. Assuming axisymmetric flow for the hemispherical dome (Stage 2), an estimate of 4.6 × 10¹⁰ Pa s was obtained, while during the coulée growth stage (Stage 4), a viscosity of 1.4 × 10¹¹ Pa s was estimated assuming the growth of the coulée could be approximated as 2D spreading due to the lateral confinement forcing growth in just two directions. Despite the large (up to 85%) uncertainties on these estimates (see Stinton et al. 2023, for details), they are broadly similar to the viscosity estimate of 2 × 10¹¹ Pa s obtained by Huppert et al. (1982) for the 1979 lava dome. The slightly higher viscosity estimated for the 2020–21 lava coulée during the latter stages of growth could be attributed to increased friction on the lava resulting from its confinement between the inner walls of the Summit Crater and the 1979 lava dome (see Stinton et al. 2023, for more detail).

Between emergence of the dome on 27 December 2020 and the last survey on 19 March 2021 (Stages 1–4), the extrusion rate averaged 1.85 m³ s⁻¹. This is broadly similar to the estimated long-term averages for both the 1971–72 and 1979 eruptions (c. 1 and 3 m³ s⁻¹, respectively), although the 1979 lava dome is estimated to have grown at rates of over 10 m³ s⁻¹ during the early stages. However, during Stage 5 when rapid inflation of the central region of the coulée occurred, extrusion rates in excess of 17 m³ s⁻¹ were estimated by Dualeh et al. (2023) in post-eruption analysis of satellite radar backscatter imagery. Variations in extrusion rate can be related to several factors, including changes in conduit diameter or changes in the magma properties. Based on estimates of magma rheology, Stinton et al. (2023) infer that the conduit that fed the 2020–21 eruption may have measured between 30 and 80 m in diameter. This compares favourably with the diameter of the conduit that fed the 1979 lava dome (which was exhumed during the explosive phase of the 2020–21 eruption), which measures c. 70 m in diameter, and also with the size of the uplifted area photographed on 27 December 2020 at the very beginning of the effusive phase. Stinton et al. (2023) explore the implications of the conduit size further in their conceptual model of the eruption. The rapid increase in extrusion rate for the last 3–4 days of the 2020–21 effusive phase can be attributed to an increase in driving pressure resulting from the ascent of fresh, gas-rich magma. This hypothesis is supported by the occurrence of increased degassing, presence of ash venting and incandescence as well as changes in seismicity (Joseph et al. 2022). This idea that it was not until the explosive phase of the 2020–21 eruption occurred that fresh magma reached the surface (Cole et al. 2023; Stinton et al. 2023) further supports the hypothesis that the lava extruded during the effusive phase was a degassed remnant left over from previous historical eruptions. Stinton et al. (2023) describe this hypothesis in more detail.

Analysis of samples collected from the front of the lava coulée and from pyroclastic density current deposits associated with the explosive phase showed the magma erupted in 2020–21 was a basaltic andesite containing an assemblage of plagioclase, clinopyroxene and Fe–Ti oxides, along with sparse olivine (Joseph et al. 2022). This is very similar to the magmas erupted in 1971–72 and 1979. This similarity is unsurprising given that the lava in the 2020–21 coulée is hypothesized to be remnant degassed magma from these previous eruptions (Stinton et al. 2023). The make-up of the 2020–21 magma is also very similar to magma erupted from older historical eruptions (e.g. 1902–03), indicating that the magmatic system at St Vincent has undergone only minor evolution over the last 1000 years (Cole et al. 2019; Fedele et al. 2021).

After 41 years of quiescence, the La Soufrière volcano in the northern part of the island of St Vincent began erupting in December 2020 after limited precursory activity. The eruption consisted of two phases: an effusive phase that saw the extrusion of lava on to the floor of the Summit Crater, and a highly explosive phase that consisted of over 30 discrete sub-Plinian and Vulcanian explosions (Cole et al. 2023).

Growth of the dome, which began sometime on 27 December 2020 and continued through to the destruction of the lava coulée on 9 April 2021, was monitored using satellite imagery, photogrammetry and visual observations. An estimated 16–19 × 10⁶ m³ of lava was extruded at a long-term average of 1.85 m³ s⁻¹. The extrusion began as a hemispherically-shaped dome on the floor of the Summit Crater, but eventually evolved into a lava coulée that was confined on its eastern and western margins by the 1979 lava dome and the inner walls of the Summit Crater. By the time it was destroyed on 9 April 2021, it measured c. 1 km long, had a maximum width of c. 240 m and was up to 110 m thick.

The morphology and shape of the 2020–21 coulée are a clear reflection of the transition from unconfined to confined flow in Stage 3 of the dome's evolution, when the shape changed from hemispherical to elliptical to eventually forming the two lobes of the lava coulée. Although classified as a coulée, the presence of two lobes that appeared to move independently of each other suggests that the confinement of the 2020–21 coulée and the low extrusion rate had a significant control on the growth directions, given that the floor of the Summit Crater over which the lobes moved was comprised of tephra from the 1979 eruption with little to no observable slope. Changes in the width of the moat in which the coulée was confined may also have resulted in the development of blocky and spine-like features more commonly associated with rates of growth not observed during the 2020–21 St Vincent eruption.

Differences in the level of precursory activity associated with the 2020–21 eruption and earlier eruptions on St Vincent in 1971–72 and 1979, as well as prior to the onset of the 1995–2010 eruption of Soufrière Hills, Montserrat, suggest that lava extruded in 2020–21 on St Vincent exploited pre-existing pathways from previous eruptions, which had not been completely sealed by cooled magma.

Petrologically speaking, the lava erupted in 2020–21 was very similar to that from the 1971–72 and 1979 eruptions and is hypothesized to be a remnant plug of strongly degassed and partially crystalline lava left over from previous eruptions. The rapid inflation, surge in extrusion rate, occurrence of ash venting, increased degassing and the increase in seismicity support the hypothesis of Stinton et al. (2023), that the remnant lava was pushed out by a rapidly rising batch of fresh, gas-rich magma that fuelled the resulting explosive phase (Cole et al. 2023).

Due to safety concerns arising from the nature of the rapidly evolving eruption, it was not always possible to gather sufficient data on various aspects of the growth and evolution of the 2020–21 lava coulée. Consequently, some aspects of the evolution of the lava coulée remain unanswered, and warrant further research. This includes topics such as the emplacement of domes comprising lava left over from previous historical eruptions, and the impacts of confinement on dome growth.

The author is indebted to the following people and organizations for assistance in the collection of imagery that were used in the photogrammetry and interpretation of the lava dome: Dominic Noon, Greg Scot and Timothy Francis of CalvinAir Helicopters; Prof. Richard Robertson, UWI Seismic Research Centre; and Kemron Alexander, Dewayne Price and Rommel Defreitas of SMU-NEMO. The author would like to thank Jon Fink and an anonymous reviewer for comments that helped refine the manuscript.

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

AJS: writing – original draft (lead).

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

The datasets generated during and/or analysed during the current study are not publicly available as the data are covered by the Montserrat Volcano Observatory Research and Data Policy. A copy of the data policy is available on request.