Introduction

Basaltic volcanoes are generally characterized by the effusive nature of their eruptions, associated with mild explosive events, such as lava fountaining and/or Strombolian explosions. The occurrence and intensity of these events are influenced by factors such as the degree of gas-melt decoupling, the magma ascent rate, and the extent of bubble coalescence within the magma1,2,3. If magma ascent rate is high, bubble coalescence is minimized, leading to a high population of millimeter to centimeter size bubbles in the shallow conduit, ultimately sustaining a fountaining eruption3,4,5. In contrast, with a slow rising magma, gas slugs can develop and migrate upward within the ascending melt to produce Strombolian explosions3,4,5. These manifestations are thought to be induced by a low magma viscosity that generally inhibits fragmentation. Sub-Plinian to Plinian explosions are rare at basaltic volcanoes, but yet their occurrence is now documented at several volcanoes (e.g., ref. 6), including the 8.6 ka eruption of La Vache and Lassolas in the Chaine des Puys7, the 2.2 ka eruption of Mt. Fuji (Japan), the 1125–1050 CE eruption of Sunset Crater (Arizona, USA)8, the 1999 CE eruption of Shishaldin (Alaska, USA)9,10,11, the 2.1 ka Masaya Triple Layer eruptions (Nicaragua)12,13,14,15,16, the 122 BCE eruption of Etna (Italy)17 and the 1886 CE eruption of Tarawera (New Zealand)18,19. These past basaltic Plinian events have been studied mainly by investigating their tephra deposits and modeling. The gas emissions associated with these past events miss precise estimation. Therefore, the 2017–2018 eruption of Ambae, in the Vanuatu archipelago, stands out as a rare example of witnessed sub-Plinian or Plinian basaltic event. Here we provide a comprehensively characterization of this eruption, by presenting its gas emissions and surface thermal heat flux. This is combined with the chemistry of bulk rocks and melt inclusions from our previous study20, now supplemented with bulk rock chemistry of phase 4, which was not included in our 2019 work. By comprehensive examination of these datasets, we attempt at shedding light into the potential processes that culminated in the observed sub-Plinian eruption, with an emphasis on the role of volatile content and surface degassing.

The 2017–2018 eruption of Ambae

The short chronology of Ambae’s volcanic activity is summarized in ref. 21 and amended in Table 1. The volcanic activity on Ambae resumed in the early 1990s after more than 300 years of dormancy. Gradually, the Crater Lake Voui became acidic and the gaseous emanation through the lake destroyed the lush forest, first on the lake’s islets and then around the crater22,23. The first eruptive manifestation was a phreatic event (Table 1) which peaked in March 199522 when meter-size bubbles at the surface of the 40×106 m3 of Voui crater lake23 were observed and a gray ashy-column was seen rising above the summit of Ambae volcano22. Ten years later, in 2005, a sustained surtseyan eruption was observed21,24 that lasted a month and a half, from end-November 2005 to mid-January 200621,25 The area around the Voui crater lake was extensively covered by ash but no impact was reported along the coast where the majority of the island’s population live. Five months later, in June 2006, the color of Voui Lake water entirely changed from blue to red in response to ion oxidation21.

Table 1 Chronology of Ambae’s volcanic activity from ref. 21 with details on the 2017–2018 eruption (this work)
Full size table

The 2017–2018 eruption, the subject of this work, is the third eruptive event at Ambae since activity resumption. This major event lasted a whole year, from September 2017 to August 2018. The activity went through 4 distinct intense eruptive phases, separated by episodes of relative calm phase as summarized in Table 1. The first intense manifestation (phase 1) was the onset of the eruption, from September 5–10, 2017. Large explosions on September 6 released about 38 kilotons (kt) of SO2, measured by the (Ozone Monitoring Instrument (OMI) on NASA’s Aura strellite (https://so2.gsfc.nasa.gov). The activity then became relatively quiet between September 10 and 21. During a visit to the summit on October 17, 2017, a large block exceeding 3 m3 was found at about one kilometer away from the active vent with a large impact crater on the ground. No other rock of this size was found beyond the rim during the later phases of the eruption. The second strong eruption phase (phase 2) commenced around September 22, 2017, when the Moderate Resolution Imaging Spectroradiometer (MODIS) thermal alerts detected the first and very high thermal anomaly induced by a lava outburst in Voui crater (http://modis.higp.hawaii.edu/; www.mirovaweb.it). Local people who ventured to the summit during this period reported the lava fountaining activity. Aerial observations made on October 3 and October 6 confirmed the presence of lava flowing out at the base of a new cone, while at least three vents maintained strombolian activity at the top. The high thermal energy associated with phase 2 persisted until November 9, 2017, whilst SO2 emissions were extremely high from October 14 to November 23, 2017, when eruption phase 2 ended. The strongest daily SO2 emission reached 39 kt (October 22), while the highest Volcanic Radiative Power (VRP) of 2270 MW was recorded on November 5. From end-November 2017 to early-January 2018, there was a lull in eruptive activity. No thermal anomalies were observed and SO2 emissions were relatively modest compared to the prior strong outgassing phases. The strong eruption phase 3 started around January 4, 2018 with an intense lava fountaining, but no lava flow, observed during a field survey. In contrast to phase 2, only intermittent thermal anomalies were observed during this phase. The highest thermal energy (912 MW) was observed on March 5, 2018, about 14 days before a period of intense eruptive degassing that lasted from March 19 to April 14. The maximum daily SO2 release from this phase was 101 kt on April 6. Phase 3 lasted approximately 3 months and ended on April 14. It was followed by another lull in activity from mid-April to mid-June 2018. The intense eruption phase 4 commenced on June 21 with the occurrence of strong thermal energy at the surface. In contrast to early phases, the thermal energy was relatively small. The highest thermal energy release during this phase was recorded on July 7 (353 MW) just 10 days before the extreme eruptive degassing that lasted between July 17 and August 3, 2018. During this period, the maximum daily SO2 output peaked at ~390 kt, the highest of the entire eruption (Fig. 1; Supplementary clip 1).

Fig. 1: Ambae Eruption: Over 390 kt of SO2 dispersed across 4000 km in the South Pacific.
figure 1

More than 390 kt of SO₂ were measured during the most intense eruptive phase of Ambae, dispersing across numerous South Pacific countries over a distance of 4,000 kilometers from the source. The thumbnail shows the eruptive column of July 27, 2018—photo courtesy, Veronica Legler / Peace Corps Vanuatu.

Full size image

This succession of intense eruptive periods and lulls has strongly challenged the volcanic risk management during the crises. Two distinct evacuations of the 11,000 inhabitants of Ambae were carried out. About 2/3 of the island’s vegetation was damaged by acid rain and heavy ash fall. Ash deposits about one meter thickness were observed in the caldera at the top of the volcano, while along the coast, ash deposits of 1 to 2 centimeters were reported. Several buildings collapsed due to the weight of the ash deposits. Numerous lahars have been reported during and after the crisis and the transported materials have transformed the rocky coasts into wide black sand beaches26. The most destructive lahar occurred during the night of 30–31 March 2018 at 1–2 am. It swept away an entire village called Waluembue. By chance, the event occurred in two phases. The first, relatively small flow forced the inhabitants to flee their houses, just before the 2nd flow, which occurred half an hour later that swept away all the houses in its path26. Despite the intensity and long duration of the eruptive event, only two direct victims were reported, and both were killed by lahars.

Results

Very high levels of SO2 emitted during the Ambae eruption in 2017–2018

Residents of Ambae witnessed the first explosions on September 5, and by September 6, 2017, OMI satellite data indicated the presence of ~38 kt SO2 over Ambae (Table 2). This substantial sulfurous cloud over Ambae was a surprise because such a large amount had never been documented on this particular volcano. SO2 emissions then gradually decreased to background levels over the next 3 days. The total mass of SO2 discharged over this period of extreme degassing (hereafter referred to as Extreme Degassing Period 1, EDP1) is 76 kt (Fig. 2, Table 2). Two series of airborne DOAS (Differential Optical Absorption Spectroscopy) traverses and scanning measurements carried out after this first extreme degassing (Supplementary table 1) indicate a mean SO2 flux fluctuating between 350 ± 120 t/d on the October 3 and >5500 ± 1700 t/d on October 6, 2017. The former value is comparable to the satellite estimate (292 t, Fig. 2) but likely underestimates the SO2 emission rate since the scanning profiles are only partially completed. A month later, after this first intense SO2 emission, satellite observations revealed another period of extreme SO2 release (EDP2) which commenced on October 14 and persisted until November 23, 2017. This time satellites measured a gradual increase from >11 kt on October 14 to a maximum of >39 kt on October 22, 2017. For nearly a month and a half, more than 516 kt of SO2 were injected into the atmosphere at an average daily emission rate of 16 kt/d (Table 2). This 2nd extreme degassing phase was followed by 3.5 months of relative calm with persistent strong degassing, characterized by an average daily SO2 output of >4 kt. On March 19, 2018, a third period of extreme SO2 degassing (EDP3) began with ~8 kt measured above the volcano. Over the following weeks, emissions steadily increased, peaking at over 101 kt on April 6, 2018. Emissions then dropped to background levels by April 14, 2018. This third extreme degassing phase lasted nearly 4 weeks and resulted in a total release of 945 kt of SO2, with a daily average of >36 kt. For three months after this third extreme degassing, Ambae experienced a period of relative calm during which the average daily SO2 output dropped to a mean value of ~2 kt. This relative calm ended on July 17 when the fourth and most intense eruption phase (EDP4) started. EDP4 released a total of 1074 kt of SO2 into the atmosphere with a daily average of 72 kt. The maximum daily SO2 mass was 390 kt, measured by Tropospheric Monitoring Instrument (TROPOMI) on July 28, 2018 (Fig. 1). This was the largest single emission of the entire 2017–2018 eruption, and included large stratospheric injection of SO227. EDP4 lasted only 17 days, ending on August 3, 2018 and marking the end of a year-long period of eruptive activity. The cumulative mass of SO2 released during the four phases of extreme degassing totals up to 2.6 ± 0.4 Tg. When the emissions from relative calm phases are also taken into account, the total SO2 released into the atmosphere during the Ambae 2017–2018 eruption increases to 3.3 ± 0.2 Tg of SO2 (Fig. 2). This figure exceeds the SO2 released by several Volcanic Explosivity Index (VEI) 4 eruptions, such as the 1982–1983 Galunggung eruption (Java, Indonesia) or the Kasatochi eruption (Alaska, USA) in 200828 (Fig. 2). If emissions from the 4 most intense eruptive phases are considered separately, one can observe- that EDP3 and EDP4 released more SO2 than the VEI 3 eruptions of Nevado del Ruiz (0.66 Tg, 1985) and Redoubt (0.61 Tg, 2009), whilst the SO2 discharge by EDP2 is higher than that from the VEI 4 eruption of Merapi (0.44 Tg, 2010) (Fig. 2). Even EDP1 released more SO2 than the VEI 3 eruption of Anathan (0.07 Tg, 2005). Therefore, Ambae stands out as one of the largest contributor of eruptive SO2 to the atmosphere, and may rank 6th according to the list compiled for eruptive contributions in the period 1978–201428.

Table 2 Total of recorded SO2 mass break down into eruptive phases
Full size table
Fig. 2: SO2 emissions from Ambae eruption: daily discharge trends and comparative analysis.
figure 2

A Daily SO2 mass recorded over Ambae during the whole eruptive period. The 4 phases of strong SO2 discharge are highlighted and identified as EDP1, EDP2, EDP3 and EDP4. The maximum daily value for each phase is indicated (in red). The curve corresponds to the cumulative SO2 values over the entire eruption period and highlights the gradual increase in SO2 discharge from EDP1 to EDP4, EDP1 being the onset eruptive discharge. The two yellow points correspond to the only DOAS SO2 flux results. B The largest SO2 loading per eruption in the period of 1978–201428. The contributions from the four eruptive phases and the total SO2 released by the Ambae 2017–2018 eruption are shown in red.

Full size image

A CO2-rich magmatic system

The gas composition of Ambae had never been measured before this work, partly because most of the volcanic gas dissolved into the 40×106 m3 of water of the crater lake of Voui23, before the 2017–2018 eruption. The process has maintained a very high acidity level of the Voui Lake21. However, during the first two phases of the eruption, successive lava flows and tephra deposits led to the formation of a new cone of about 1 km in diameter and 150 m high. When reaching the summit on May 2018 (this work), the new cone had a crater of ~300 m in diameter, through which magmatic gas escaped directly into the atmosphere. Only a limited period of time was spent on the rim of this cone, due to the high level of activity, to record the gas composition of this volcano for the first time ever. Our Multi-GAS derived in-plume concentration time series, presented in Fig. 3, exhibit a good correlation between the SO2 and other measured gas species, suggesting a common magmatic source. The derived gas to SO2 ratios are 60–260, 0.10–0.12, 1.6–5.3 and 0.01–0.02 for H2O/SO2, H2S/SO2, CO2/SO2 and H2/SO2 respectively (Table 3). The range of CO2/SO2 ratios observed likely reflect distinct compositions for the several vents active inside the crater at the time of our survey. Assuming that the averaged ratios are representative of the system, then the magmatic gas composition of Ambae is inferred at 97 mol.% H2O, 2 mol. % CO2, 0.7 mol. % SO2, 0.07 mol. % H2S and 0.01 mol. % H2 (Table 3). Scaling these gas ratios to the amount of SO2 obtained from satellite data (3.3 Mt) yields the equivalent of 134 ± 4 Mt of H2O, 7.7 ± 0.2 Mt of CO2, 0.2 ± 6E-3 Mt of H2S, and 0.001 ± 4E-5 Mt of H2 (Table 3) for the entire eruptive period, which is a considerable contribution to the atmosphere. The total CO2 released is substantial for a single volcano, comparable for instance to the annual CO2 budget passively released each year from all South America volcanoes (6.4 Mt)29, and represents ~20% of the global CO2 from active volcanoes through passive degassing (39 Mt/yr)30. Hence, Ambae can be considered as a notable CO2 source. The CO2/ST ratios (ST being the sum of SO2 and H2S) of Ambae gas fluctuates between 1.4 and 4.7 (Table 3) and plot at the upper range of the arc volcanism representative field in the H2O-CO2-ST ternary diagram (Fig. 3). This CO2-rich character of the Ambae gas is in contrast with observations made at other volcanoes of the Vanuatu arc (Mount Garet, Ambrym, Yasur) that display low-carbon gas31,32,33 (Fig. 3). However, this Ambae gas composition overlaps with that of Merapi, and is close to that of Bromo, two Indonesian volcanoes fueled by volatile-rich sources derived from dehydration of a subducting slab containing carbon-enriched subducted sediments34,35. Therefore, our results suggest that combination of (1) the magma source that fueled the Ambae eruption being enriched in CO2, and (2) the magmatic gas that sustained the eruption being separated at depth in the plumbing system (hence resulting CO2-enriched), as highlighted by the CO2-SO2-H2S ternary plot (Fig. 3). A deep and CO2-enriched source is also supported by melt inclusions (Fig. 3, Supplementary Table 2, and see complete dataset in ref. 20), that show elevated CO2 contents (up to >3800 ppm), relative to volatile contents in basalts from other subduction zones (from ref. 36). The CO2 versus H2O diagram for Ambae inclusions shows that the most elevated CO2 levels were observed in EDP1 samples (>3800 ppm). CO2 contents then decreased to ~2000 ppm in EDP2, and to <1000 ppm in EDP3 and possibly EDP4 (Fig. 3), indicating eruption of more shallowly stored magma. The absence of melt inclusion analysis for Phase 4 necessitates the assumption that its volatile contents are comparable to those of Phase 3. This assumption is supported by several factors, including the comparable eruptive intensity, the duration of eruptive discharges, the quantity of SO2 emitted, and the bulk compositions of tephra deposits, which collectively suggest analogous magmatic conditions. In any case, the above results support the idea that Ambae volcanic activity is in part sustained by a carbon-rich source, having a CO2/ST ratio greater than 4, following the classification in ref. 37 Two mechanisms may possibly contribute to this high CO2 source, including the subduction of ~650 m sediments with enriched carbonate content38, carried by the d’Entrecasteaux Ridge39, in front of Ambae volcano; and the westward and upward invasion of Indian Ocean-type MORB mantle (IOTMM) in the central part of Vanuatu and more particularly beneath Ambae, in response to the opening of the north Fiji basin40 and the effect of buoyant ridge subduction40. The IOTMM is an enriched mantle source40,41 that cannot be interpreted as induced by the DER input into the mantle wedge since the isotopic signatures point to a long term (>500 Ma) enrichment processes41. The proportion of CO2 contributions from these two sources remains to be determined. Furthermore, this result should be used with caution, because no gas composition measurements have been conducted outside of the eruption period.

Fig. 3: Gas compositions and melt inclusions reveal a transition from deep to shallow magmatic carbon-rich sources beneath Ambae during the eruption.
figure 3

A Fluctuation of gas concentrations recorded on the rim of the new cone developed during the Ambae eruption. The correlations of the measured gases to SO2 are provided. B Ternary diagrams showing a deep to shallow magmatic carbon-rich source beneath Ambae in comparison to other volcanoes of the Vanuatu arc. The CO₂-SO₂-H₂S diagram is adapted from ref. 83,84, while the CO₂-H₂O-ST diagram is adapted ref. 33. The data sources include ref. 33 for Gaua, ref. 30 for Ambrym, ref. 31 for Yasur, ref. 34,35 for Merapi and Bromo, and the present study for Ambae. C Comparing the maximum CO2 versus H2O content from Ambae inclusions with inclusions in basalt from various subduction zones. The diagram is adapted from ref. 35, with sourced from ref. 36,85 and other references therein. The pink shaded zone highlights the decreasing trend in CO2 levels throughout the eruptive event, indicating a progressive reduction in CO2 contributions from a relatively deep source (>16 km) during EDP1, transitioning to ~10 km in EDP2, and ultimately approaching the surface in EDP3. This gradual decline occurred alongside a relatively modest but consistent H2O content, which remained stable until the shallower depths of EDP3.

Full size image
Table 3 Gas composition and total degassing budget obtained by combining MultiGAS and satellite data
Full size table

Lava discharge volume decreased over the eruption period

MODIS thermal alerts (MODVOLC; http://modis.higp.hawaii.edu/) and MIROVA (Middle Infrared Observations of Volcanic Activity—https://www.mirovaweb.it/) are widely recognized tools for volcano monitoring, particularly during volcanic crises42,43. In the case of the 2017–2018 Ambae eruption, they provided a valuable means to monitor the activity. The first thermal alert of the 2017–2018 eruption was issued on September 22, 2017 with a VRP of 457 MW, and during the whole eruption period MIROVA recorded 102 thermal alerts distributed into three distinct phases. The first phase, which ran from September 22 to November 9 (hereafter referred to as Thermal Radiation Phase 1 or TRP1), was the most energetic, emphasized by three significant VRP pulses, including 1424 MW on October 1, 883 MW on October 19 and 2270 MW on November 5, 2017. The second phase of thermal activity (TRP2) was observed between January 4 and April 11, 2018. This is the longest period of thermal manifestations at the surface, but in contrast to TRP1, was characterized by more intermittent thermal release behavior. The maximum VRP value of 912 MW was measured on March 5, 2018. The third and final period of thermal manifestation (TRP3) was observed between June 21 and July 26, 2018. This is the shortest period of thermal activity with a maximum VRP value of only 353 MW on July 7, 2018. The cumulated lava volume erupted at the surface, obtained following the approach of ref. 44 highlights a high mean extrusion rate of 4.3 × 105 m3/d (~5 m3/s) in TPR1, which then fell to 8.8 × 104 m3/d (~1 m3/s) in TRP2, and then to 1.2 × 105 m3/d (~1 m3/s) in TRP3. The strongest mean discharge rate during TRP1 corresponds to the period of lava fountaining, lava flow and strombolian activity. TRP2 and TRP3 periods were sustained by lava fountaining, strombolian activity and explosive discharges. The cumulative discharged volume for the whole eruption period amounts to 0.034 km3 with a lower and upper bounds of 0.016 and 0.050 km3 respectively (Supplementary Fig. 1).

Transition to a more primitive basaltic source

The bulk compositions of the tephra samples collected during the four strong eruption phases are presented in Supplementary Table 2. The SiO2 vs. total alkali plot (Fig. 4) shows a transition from a basaltic andesite composition in EDP1 to a composition intermediate between basalt and trachy-basalt composition in EDP2 and EDP3, and finally to a more primitive basaltic composition during EDP4. This large range of SiO2 content (47.8-54.8 wt. %) in the 2017–2018 eruptive products contrasts with past eruptions that show bulk rock compositions restricted to basalts (Fig. 4). The highest SiO2 content of 2017–2018 eruption corresponds to a large block that fell about a kilometer from the active vent during the onset phase. In terms of alkali content, samples from 2017 to 2018 event are relatively more alkali-rich than previous eruptions, with the exception of the samples from phase 1 that plot well below the alkaline/sub-alkaline line. The Harker diagrams, drawn relative to MgO content (Fig. 4), point to high Na2O (2.48–3.51 wt. %), K2O (1.56–1.90 wt. %) and low CaO (9.22–10.21 wt. %) compared to data from past works. The Fe2O3 content from the 2017–2018 eruption (9.70–10.34 wt. %) is in the lower range of past eruptions while the Al2O3 content varies in a narrow range (14.21–16.78%). The MgO content in our samples has a relatively narrow range (5.17–10.23 wt. %), compare to past data, but display a relative increase from EDP1 to EDP4, paralleled by a decreasing trend of Na2O, Al2O3, SiO2, TiO2 and K2O (Fig. 4). Such a trend may indicate that deeper magma source gradually reached the surface over the course of the eruption, as evidenced by melt inclusion results (Fig. 3). Trace elements (Supplementary Table 2) show a typical arc magma signature, with enrichment in highly incompatible elements (e.g., Ba contents range from 400 to 600 ppm), and negative Nb anomalies (Fig. 4). The flat MREE to HREE (Sm/Yb between 1.84 to 2.91) patterns suggest that residual garnet was not present in the mantle source. All the trace element patterns are similar. In terms of ratios, Ba/Th, Ba/Nb and Zr/Nb correspond respectively to 197–223, 313–335, 40–42 for EDP1, to 236–238, 201–325, 26–40 for EDP2, to 230–574, 215–291, 27–39 for EDP3 and to 259–301, 191–301, 28–39 for EDP4. These ratios are in the range of subduction zones (0.1–100) for Zr/Nb, whilst relatively higher for Ba/Th and Ba/Nb which may suggest fluid contributions from the subducted sediments and subducted slab (e.g. ref. 45). The volatiles from melt inclusions results (Supplementary Table 2; ref. 20) indicate strong variations in CO2, S, Cl and F, as highlighted in ref. 20, with, respectively, 24–3899 ppm, 440–980 ppm, 1878–7243 ppm, 558–717 ppm in EDP1, similarly 12–2353 ppm, 208–1541 ppm, 2176–5348 ppm and 492–711 ppm in EDP2, and 8–946 ppm, 152–1183 ppm, 1435–3957 ppm, and 393–622 ppm for EDP3. Many processes can contribute to this wide range of volatile concentrations, including degassing and gas loss, crystallization, or post entrapment processes46. However, as noted in ref. 20, crystallization is unlikely to be the primary factor driving these fluctuations. Assuming that the highest volatile concentrations in each eruption phase are representative of the volatile content in the melt, one can observe that the chlorine and carbon dioxide content in the melt decreased over the eruption period, while the fluoride content remained relatively stable. Such fluctuation may reflect the progressive volatile exsolution as the magma ascended to the surface. These volatile concentrations are within the range of magmatic volatiles observed in the context of subduction47, with the exception of Cl, which appears relatively enriched, yet supporting the idea of fluid contributions from the subducted slab and associated sediments. The water content (H2O) varies from 1.01–1.60 wt. % in EDP1, to 1.15–2.13 wt. % in EDP2 and finally 1.19–2.23 wt. % in EDP3. Although a slight increase can be observed over the eruption period, in consistent with a magma ascent, these concentrations are rather low for the subduction zone (Fig. 3), where the global average figure is around 3.9 wt. %48. Unfortunately no melt inclusions were analyzed for EDP4 and we assumed that water content was in the range of Phase 3.

Fig. 4: Evolution of magma composition throughout the eruption compared to past eruptions.
figure 4

A Lava and tephra bulk composition obtained in this work (Supplementary Table 2) are plotted (in red) for comparison with the past results of Warden (1970)86, Gordon (1974; 1977)87,88, Eggins (1993)89 and Sorbadere et al. 90. Melt inclusion results (from ref. 20) show a wide range of composition from basaltic trachy-andesite to basalt. The bulk rock and tephra composition from the 2017–2018 show a transition from basaltic andesite and trachy-basalt to basalt compositions. EDP1 corresponds to the onset eruption phase the produces material with relatively high silica content, whilst phase 4 corresponds to the final eruption phase. Phase 2 and phase 3 correspond to intermediate eruption phases. B The Harker diagrams of Ambae samples showing a strong olivine control on the past data87 contrasting with a relatively restricted MgO content in this work. The CaO content is lower than the past data, while the K2O and Na2O are higher, suggesting a relatively evolved source compared to results from past studies86,88,89,90. Of the two series defined from the TiO2 content in the past works88, only the low-TiO2 series is obtained in this work. The trace element concentrations normalized to N-MORB91 indicate a typical subduction arc source. Results are from bulk rock analyses.

Full size image

Discussion

Eruption triggered by new magma injection

In contrast to the existing data in the literature, the bulk compositions of the samples collected during the eruption onset (EDP1) notably, the large ejected block, display a relatively high silica content (Fig. 4, Supplementary Table 2). The block was stained with sulphur, suggesting regular exposure to a magmatic gas flow. It is thus likely that this block was part of the conduit, which was cleared at eruption onset. The SiO2 content of this rock was probably increased by the deposition of amorphous or microcrystalline hydrothermal silica in voids and veins49. Such void sealing could reduce rock porosity, which in turn promotes pressure build-up during EDP1. The latter unleashed 76 kt of SO2, an amount surpassing the output of many VEI 3 eruptions such as Shishaldin (63 kt, 1999) or Makian (50 kt, 1988)28, and suggesting the involvement of magmatic processes. The transition from basaltic andesite to a more alkaline basaltic source (Fig. 4, Supplementary Table 2) suggests the arrival of a fresh magma into a relatively evolved melt in the reservoir. Such phenomenon is known to lead to over-pressure (e.g. ref. 50), as new magma disrupts the equilibrium in the reservoir, leading to the exsolution of volatiles, changes in magma composition and viscosity, promoting bubble nucleation, bubble growth, pressure buildup, magma ascent, and a subsequent eruption.

On Ambae, the process was likely accelerated by the rapid heating of the hydrothermal system as magma ascended towards the surface. This led to a swift conversion of subsurface water into steam, triggering volume expansion, which in turn intensified pressure buildup, ultimately resulting in a phreatomagmatic eruption. Similar patterns have been observed during eruptions at other volcanoes51,52,53.

A basaltic sub-Plinian event

The first papers on the 2017–2018 Ambae eruption did not agree on the strength of the event. They assigned a VEI 354, which means an eruptive column height between 3 and 15 km and a tephra volume of less than 108 m355, to a Plinian eruption20 which suggests a VEI of at least 4 with an eruptive column height that can exceed 25 km and an ejected volume ranging from 0.1 to 10 km356. The volume of the ejected materials is yet to be properly assessed. Alternatively, when looking at the plume heights derived either from the Wellington Volcanic Ash Advisory Centre (VAAC) or from satellite observations57 (Fig. 5), it appears that not only did the maximum height of the plume emitted from Ambae vary between 2 and 20 km, but also that it overall increased over the course of the eruptive activity. The maximum height of 20 km was reported on July 28, 2018. This height evolution is consistent with the high SO2 emissions (Fig. 2; Fig. 5) and emphasizes the increase in eruption intensity from EDP1 to EDP4. Assuming that the plume heights remained at the top of the buoyant column then based on the table compiled in ref. 55, phase 1 (EDP1) was a VEI 2, phase 2 (EDP2) a VEI 3, phase 3 (EDP3) a VEI 4 and phase 4 (EDP4) a VEI 4, with maximum assumed plume heights of 1.5 km, 6.1 km, 18 km, and 20 km respectively (Fig. 5).

Fig. 5: VRP and SO2 mass reveal a distinct pattern repeated across the four intense eruption phases.
figure 5

The juxtaposition of the cumulative values of volcanic radiative power (VRP) and sulfur dioxide (SO2) mass reveals a discernible pattern that can be divided into 5 distinct episodes that characterize each of the three eruption phases. It also underlines the time lag between the appearance of hot magmatic features at the surface and the SO2 surges. Lava fountain episodes are characterized by a rapid rise in VRP, which typically precedes the intense explosive activity. Plume heights derived from data provided by the Volcanic Ash Advisory Center (VAAC), satellite imagery, and on-site observations, collectively indicate an increase of the eruption duration, corresponding to an escalation in eruptive intensity.

Full size image

Relating the maximum daily values of SO2 in each eruption phase to the volcanic explosivity index (VEI) according to the following formulation28,58:

$${{{\rm{VEI}}}}=\left[{\log }_{10}\left({{{{\rm{SO}}}}}_{2},{{{\rm{Tg}}}}\right)+3.15\right]/0.71$$
(1)

We obtain a VEI 2 for EDP1 and EDP2 with 38 kt and 39 kt respectively (Table 2), while EDP3 with 101 kt may be classified as VEI 3 and EDP4 with 390 kt would be a VEI 4. These two approaches show comparable results and emphasize that the VEI of the eruption can be different depending on the eruption phase. This could explain the divergence of classification in the early works. Overall, based on the above results, Phase 1 may be classified as VEI 2, Phase 2 as VEI 2-3, Phase 3 as VEI 3-4 and Phase 4 as VEI 4. These results confirm what was experienced on the ground, e.g., an eruption of very high intensity, never before experienced or described in the short history of Ambae volcano (Table 1) which increased during the eruption period. The total amount of SO2 released by the eruption is among the top 5 largest SO2 contributors through eruption on a global scale28. These observations suggest that Ambae volcano, despite its basaltic magma composition (Fig. 4), is capable of producing at least sub-Plinian VEI 4 events.

High magma supply rate and volatile-rich source

The amount of magma needed to degas the volume of gas released into the atmosphere during an eruption can be estimated from the ratio of degassed sulfur and the difference between the sulfur in the melt inclusions and in the matrix glass36:

$${{{{\rm{M}}}}}_{{{{\rm{magma}}}}}=\left({{{{\rm{M}}}}}_{{{{\rm{Sulfur}}}}}\_{{{\rm{gas}}}}\times {10}^{6}\right)/\left({{{{\rm{M}}}}}_{{{{\rm{Sulfur}}}}}\_{{{\rm{inclusion}}}}-{{{{\rm{M}}}}}_{{{{\rm{sulfur}}}}}\_{{{\rm{matrix}}}}\right)$$
(2)

where Mmagma is the mass of the magma (in kg) that sustained the eruption, Msulfur_gas is the mass of sulfur (in kg) released into the atmosphere, Msulfur_inclusion is the amount of sulfur (in ppm) in the melt inclusion and Msulfur_matrix is the mass of sulfur (in ppm) in the matrix glass. Assuming that the highest sulfur content in the olivine-pyroxene melt inclusion reflects the undegassed magma source (Supplementary Table 2), and using the above formula, along with the basaltic densities derived from bulk compositions59 (Table 4), it is possible to estimate the magma volume required for the eruption. For EDP1, EDP2, EDP3 and EDP4, the required magma volumes are estimated to be 0.014 km3, 0.059 km3, 0.142 km3 and 0.161 km3 respectively (Table 4), totaling 0.376 ± 0.025 km3. Relating these volumes to the duration of each phase yields magma supply rates of 27 m3/s, 17 m3/s, 61 m3/s, and 104 m3/s, respectively. These figures are well above typical rates for basaltic sources, such as those observed at Kilauea (0.6-5.7 m3/s)60 or Etna (~0.3 m3/s)61, the latter remained at constant rate for decades. Ambae’s mean magmatic supply rate of 52 m3/s doubles that of nearby Ambrym volcano, where a rate of 25 m3/s was necessary to renew the entire 0.5 km3 magma reservoir within 240 days31. This high supply rate likely accounts for Ambrym’s status as a top global magmatic degassing source, characterized by continuous intense degassing until 201828,62,63. As evidenced in instances like La Soufrière64, these finding suggest a robust link between magma supply rate and volcanic degassing, implying a direct influence on eruption intensity. The relationship is starkly illustrated during VEI 4 phases (EDP3 and EDP4), sustained by exceptionally high magma supply rate of 61 m3/s and 104 m3/s respectively, coinciding with extreme degassing, each phase releasing 0.9 Mt and 1.1 Mt of SO2. If we plot for a series of key eruptions36, the associated released SO2 mass vs. the volume of degassing magma (Eq. 2), the Ambae 2017–2018 stands out for its exceptional gas productivity relative to other basaltic eruptions. This elevated gas/magma proportion (Fig. 6) at Ambae is comparable to that of Ambrym volcano with its 0.5 km3 and a continuous strong degassing28,31,63,65 prior to the 2018 eruption. Ambrym has experienced very strong eruptions in the past, including a VEI 4 in 1950 and a probable VEI 6 in 50 A.D66. Ambrym and Ambae share the same geodynamic context and are both influenced by the slightly oblique collision of the buoyant D’Entrecasteaux Ridge (DER)67, aided by the influx of a more fertile mantle material with isotopically distinct Indian Ocean-type MORB character40,41,68. The location of Ambrym Volcano well to the south of the southern edge of the D’Entrecasteaux Ridge (DER) may explain the difference in CO2 levels compared to Ambae volcano. Ambae’s location above the carbonate-rich sediments ~650 m-thick transported by the subducted DER may explain its CO2-enriched magmatic source39. The interaction between DER and the Vanuatu arc began 2–3 million years ago, approximately 100 km to the south of its current position, and migrated northward due to arc curvature and oblique subduction69. It is believed that the collision extensively fractured the DER, resulting in deep faults that allow fluids to penetrate and hydrate the oceanic lithospheric mantle70. The relatively high thermal parameters of the Australian plate at depth, enables hydrated minerals to reach deeply into the mantle before dehydrating70. These processes may play a crucial role in the formation of volatile-rich, deep magma sources, that may subsequently trigger large-scale basaltic explosive eruptions, as demonstrated by the activity on Ambae. Recognizing the fundamental role of volatiles in regulating magma ascent from depth, it is reasonable to assume that they contributed to the rapid magma ascent observed on Ambae20. These results therefore suggest that rapid magma ascent towards the surface required to produce basaltic Plinian eruptions71 reflected a magmatic sources richly saturated in volatiles. With a high volatile content, the magmatic mixture becomes more buoyant due to an increased volume fraction of exsolved gas, thereby expediting magma ascent.

Table 4 The volume of magma that sustained 2017–2018 eruption of Ambae
Full size table
Fig. 6: Relationship between SO2 mass and magma volume across different magma types, emphasizing basaltic gas-rich events.
figure 6

A Relationship between SO2 mass and the corresponding magma volume sustaining past eruptive events. It distinguishes the magnitude of gas output during eruptive events as a function of magma type. The blue, black and red lines are regression lines between SO2 releases during eruptions and corresponding magma source volumes for basalt, andesite and dacite respectively. The dashed lines are included to clearly distinguish the individual eruptions for better visualization. Bezymianny (1956) and Nevado del Ruiz (1985) eruptions were gas-rich events, whilst Soufrière Hills eruption (1995–1998) was a gas-poor event for andesitic magma. Agung (1963) and Arenal (1985–1996) eruptions are gas-rich events contrasting with Galunggung (1982-1983) for basaltic-andesite magma. Similarly, in the case of basaltic magma, Ambae 2017–2018 eruption stands out as a gas-rich event. The relatively calm phases contributed significantly to the overall release of SO₂. Ambrym’s situation in 2005–2015 shares this same analysis with its annual 2.5 Tg of SO262 while the 2018 eruption was gas-poor event suggesting a degassed source. Dacite magmas are poor-gas source. Data are recalculated from ref. 36.

Full size image

The role of degassing in the transition from effusive to explosive activity

The analysis of the interplay between VRP variations and SO2 emissions throughout the eruptive period unfolds a consistent pattern, delineated into five distinct episodes (Fig. 5). Indeed, excluding the phreatomagmatic onset phase (Phase 1), the subsequent three eruptive phases can be divided into the following 5 episodes (Fig. 5): a period of relative calm following a vigorous eruptive phase (here after as EPS1 or Eruption Phase Sequence 1), followed by a phase characterized by magma emergence at the surface (EPS2), featuring lava fountains, lava flows, and strombolian activity (lava flows being restricted to eruption phase 2). The 3rd episode is characterized by sustained lava fountaining (EPS3), leading to a sudden gas surge that triggered explosive manifestations (EPS4), and finally a phase marked by persistent of gas surges whilst the top of magma column receded to depth (EPS5). This pattern illustrates the transition from effusive to explosive character, underscoring that the primary controlling factor is gas. Each eruptive phase commenced with an effusive type manifestation that progressed into explosive as gas increased. A general trend can also be discerned for the eruption as a whole, in which the amount of gas released during the explosive phases increased from EDP2 (0.5 Tg SO2), being more effusive nature, to EDP3 (0.9 Tg SO2) and then EDP4 (1.1 Tg SO2). This increase further emphasizes that the intensity of eruptive activity is driven by the amount of available volatiles, which facilitate a high rate of magma supply and an extremely fast ascent speed (15–270 km/h)20. This rapid ascent led to substantial undercooling and subsequent rapid crystallization72, a process known to increase magma viscosity, which is a critical precursor to fragmentation. The rapid magma ascent from the shallow reservoir (0.5–3 km)20, posed a challenge for effectively decoupling of volatiles from the melt. This is another crucial process for highly explosive eruptions in basaltic volcanoes6,71, as seen in phase 3 and phase 4. The likelihood of producing basaltic explosive events increases with a lower H2O content, as observed on Ambae, with 1–2 wt. % of H2O (Fig. 3; Supplementary Table 2), compared to the 2–6 wt. % typically observed in basaltic arc magmas47. Recent studies indicate that low H2O content increases magma viscosity in the conduit and that even small amounts of external water can induce a shift from explosive to lava fountain eruption styles71.

While many studies have implicated high levels of CO2 in magma as responsible for more explosive eruptions8,73,74,75, this topic remains controversial. At Ambae, the decrease in CO2 concentrations in melt inclusions, dropping from 3900 ppm to 900 ppm, amid an escalation eruption intensity (Supplementary Table 2) does not support this idea. However, this decrease can be attributed to the exsolution of CO2, a process that not only enhances magma buoyancy, facilitating its ascent, but it may also increases the amount of volatiles present adding more pressure in the reservoir and conduit.

Conclusion

The 2017–2018 Ambae eruption stands out as a unique case in which the total gas emissions of a basaltic sub-Plinian event have been quantified in parallel with thermal heat fluxes. The data show very high levels of degassing with a total eruptive release of 130 Tg H2O, 6.8 Tg CO2, 3.3 Tg SO2, 0.2 Tg H2S and 0.001 Tg H2. The total SO2 output ranks Ambae at 6th position on the list of eruptions with quantified SO2 output. The total CO2 release was substantial for a single volcano, equivalent to the annual CO2 output from of all Latin American volcanoes. Surprisingly, the magma volume sustaining this enormous gas output is modest compared to other basaltic volcanoes, implying a volatile-rich magma source beneath Ambae. We propose the volatile-rich nature of Ambae magma is probably linked to the regional geodynamic environment. Indeed, the subduction of the extensively fractured D’Entrecasteaux Ridge, with its carbonate-rich sediments, associated to the high thermal gradient and the westward uprising fertile mantle likely favor the formation of a volatile-rich magma. A high content of dissolved volatiles can lead to extensive exsolution, which increases magma buoyancy and may result in a higher magma supply and faster ascent. The rapid ascent of the magma creates challenges in separating volatiles from the melt, which favor magma into a highly explosive basaltic eruptions. The results also highlight the role of gas surges in the transition from effusive to explosive volcanic activity. Without a substantial influx of volatiles, the 2017–2018 eruption at Ambae would likely have been predominantly effusive.

Methods

SO2 mass from Satellite sensors

The quantity of SO2 released during the Ambae 2017–2018 eruption was estimated using daily SO2 mass captured above the Ambae volcano by the Ozone Mapping and Profiler Suite (OMPS) and the Ozone Monitoring Instrument (OMI). The data series, available online at https://so2.gsfc.nasa.gov/, were compiled and analyzed for the entire eruptive period, from September 5, 2017, to August 3, 2018. The most intense SO2 output of July 28, 2018 was followed thanks to TROPOMI measurements. The nearby Ambrym volcano also contributed to the daily SO2 mass, but its contribution was estimated and then excluded from the total SO2 mass, especially during relative calm phases, based on the area covered and the corresponding DU according to the method in ref. 51 During the four extreme degassing phases, Ambrym’s output was negligible and therefore ignored in the daily values. Yasur volcano, located in the south of Vanuatu, had a negligible impact throughout the eruption period. In contrast there was a notable SO2 output from the Tinakula eruption on October 21, 2017, but that contribution was also excluded in the compilation.

Airborne and scanning DOAS measurements

In the early eruption phase, two DOAS campaigns were carried out, on October 3 and October 6, 2017, using airborne traverses and stationary scanning measurements. Airborne traverses were performed following the approach outlined in ref. 21 using a Brittan-Norman Islander flying beneath the plume at 8–12 km from the source whilst scanning DOAS were carried out beneath the plume on the western part of Ambae at the about 14 km from the source using the same approach outlined in ref. 76 Due to the increasing activity and the effective cost of the campaigns, no further measurements were carried out.

Volcanic Radiant Power (VRP) and Time Average Discharge Rate (TADR)

The variation of thermal radiation emitted during the eruption period by hot surfaces at the summit of Ambae volcano was recorded by the Moderate Resolution Imaging Spectroradiometer (MODIS). These data were made available to the scientific community and observatories through MODVOLC (MODIS Volcanic Activity Monitoring), MIROVA (Middle InfraRed Observation of Volcanic Activity) and Mounts (Mounts Project). In this work, the thermal dynamics of the eruptive activity are based primarily on data from MIROVA who provided VRPs in near real time77. Specifically, the VRP is calculated as follows:

$${{{\rm{VRP}}}}=18.9\,{{{{\rm{A}}}}}_{{{{\rm{pixel}}}}}{\sum}_{{{{\rm{i}}}}}^{{{{\rm{npix}}}}}{\left({{{{\rm{L}}}}}_{{{{\rm{MIR}}}},{{{\rm{alert}}}}}-{{{{\rm{L}}}}}_{{{{\rm{MIR}}}},{{{\rm{bk}}}}}\right)}_{{{{\rm{i}}}}}$$
(3)

where npix is the number of alerted pixels,LMIR,alert is the pixel integrated MIR radiance of the ith alerted pixel, LMIR,bk is the MIR radiance of the background (average radiance of pixels surrounding the anomaly), Apixel is the pixel size (1 km2 for the resampled MODIS pixels), and 18.9 is a constant of proportionality77

The VRP (in W) is then convert into TADR (m3/s) according to the following equation44:

$${{{\rm{TADR}}}}={{{\rm{VRP}}}}/{{{{\rm{C}}}}}_{{{{\rm{rad}}}}}$$
(4)

where Crad is the radiant density expressed in J/m3 and can be predicted on the basis of the silica content of the erupted products:

$${{{\rm{Crad}}}}=6.45\times {10}^{25}\times {\left({{{{\rm{X}}}}}_{{{{\rm{SiO}}}}2}\right)}^{-10.4}$$
(5)

where XSiO2 is the silica content (wt. %) of the erupted magma. The cumulated TADR over the eruption period has allowed broad estimation of extruded magma volume.

Gas composition

The gas composition of Ambae was measured using a Multicomponent Gas Analyser System (Multi-GAS)78,79. This portable instrument measures the concentrations of CO2, SO2, H2S, H2, as well as ambient atmospheric pressure (P), temperature (T), and relative humidity (RH). This latter is converted into plume H2O concentration following the approach outlined in ref. 80:

$${{{{\rm{H}}}}}_{2}{{{\rm{O}}}}={6.1121} * {\left(1.0007+3.46{{{{\rm{P}}}}}^{-6}\right)} * \exp {\left(\left({17.502}^{* }{{{\rm{T}}}}\right)/\left(240.97+{{{\rm{T}}}}\right)\right)} * \left({{{{\rm{Rh}}}}/100}^{* }{10}^{6}\right)$$
(6)

where H2O is the water vapor content in ppmv, P is the pressure in mbar, T is the air temperature in °C, and Rh is the relative humidity in %. The CO2 is measured with a commercial infrared spectrometer (range 0–3000 ppm) whereas SO2, H2S, H2 gases were quantified using specific electrochemical sensors (typical range 0–200 ppm). On May 22, 2018, the MultiGAS was first positioned on Voui crater rim, but given the distance to the active vent and the dynamic of the degassing, the plume never reached the device. The decision was then taken to move the system to the rim of the new cone in the Voui’s crater. The system was placed in the plume, however given the relatively dangerous exposure to the activity, only about a quarter of an hour has been successfully devoted to this gas recording.

Field sampling and Laboratory analyses

Three field campaigns at the summit of the Ambae volcano were carried out on October16, 2017, January 6, 2018 and May 21-22, 2018. During these expeditions ash, scoria and rock samples were collected for laboratory analyses. Complementary sampling were carried out on the western part of Ambae island by local inhabitants, immediately after ash falls. Samples collected were analyzed for major elements using ICP-AES, after dissolution of 100 mg of finely crushed sample by alkaline fusion. Trace-element concentrations were measured by ICP-MS after the dissolution of 100 mg of each sample, following the approach outlined in ref. 81 Comparison with repeated analysis of international standards (AVG-2, BIR-1, and BEN) was used to validate major and trace element data. For the major elements, the precision is better than 5%, whilst the precision is less than 10% for all the trace elements. Analyses were carried out in Laboratoire Magma et Volcans (Clermont-Ferrand, France) and Laboratoire Géosciences Océans (Brest, France).

Volatile (H2O, CO2, Cl, F, S) content in melt inclusions and matrix glasses is sourced from ref. 20 which provides a detailed description of the methodology, including post entrapment crystallization and modeling.