Introduction

Calbuco volcano is an active, hazardous volcano located in the Southern Andes of Chile (Stern et al. 2007) that lies 30-32 km E of the cities of Puerto Montt and Puerto Varas (Fig. 1). During the afternoon of 22 April 2015, Calbuco volcano started a new eruptive cycle following only minor precursory activity, which comprised mainly a seismic swarm 3 hours before the start of the eruption (Valderrama et al. 2015); the eruption then generated two successive sub-Plinian phases in 12 hours. These phases produced >15 km high eruption columns with their associated fall deposits to the NE, and pyroclastic density currents (PDCs) that travelled up to 6 km from the vent over most of the flanks of the volcano (Fig. 1). Primary and secondary lahars were also generated on the S, N and NE flanks during and following the eruptions. Airborne tephra caused disruption of flights in the Chilean and Argentinean airspace. In proximal areas tephra fall most strongly affected the people of Ensenada (Fig. 1), a small village located 10 km NE of the volcano summit. After the two main sub-Plinian events, weak activity continued for a couple of weeks until the first half of May, before returning to baseline levels.

Fig. 1
figure 1

Location map of Calbuco volcano and distribution deposits from fallout, PDCs and lahars of the 22-23 April 2015 eruption. Red dots with numbers indicate the location of the stratigraphic sections

Full size image

In this article we report results of a comprehensive study focusing on the products emitted during the two sub-Plinian events. We focus on the stratigraphy and regional distribution of the main fallout and PDC units, characterization of eruptive dynamics in terms of volume and intensity of eruptive phases, and the main petrologic features of eruptive products. In a recent paper, Van Eaton et al. (2016) presented satellite observations of the eruption that agree with our interpretation of the timing of the deposits. In another recent work, Romero et al. (2016) presented the first analysis and results on the tephra fall deposit of the eruption. Although their volume estimates coincide with those of this study we interpret the timing and nature of the activity differently. Furthermore we present the first stratigraphic studies on the PDC deposits and characterize the juvenile material of both fallout and PDC deposits (mineral chemistry, whole-rock major and trace elements, SEM analyses). These results provide insights into the eruption style and its possible triggering mechanisms.

Geological setting

Calbuco volcano (41°S, 2,003 masl) is located in the Southern Volcanic Zone (SVZ, Fig. 1) of the Andes and has been one of the most active and hazardous Chilean volcanoes during historical times (since the XVII century). Eruptive activity started at ~300 ka (Sellés and Moreno 2011) and was mainly characterized by effusion of andesitic lavas and domes and their associated PDCs, mainly block-and-ash flows, as well as “cold” and “hot” lahars (López et al. 1992; Castruccio et al. 2010; Sellés and Moreno 2011; Castruccio and Clavero 2015). Holocene products are better preserved and also include sub-Plinian to Plinian fall deposits and a basaltic-andesite ignimbrite emplaced at about 6.5 ka (Sellés and Moreno 2011). A distinctive hummocky terrain on the lower northern flank is the result of two sector collapses that occurred in postglacial times (Clavero et al. 2008). Historical activity (Petit-Breuilh 1999) includes sub-Plinian eruptions (1893-1895; 1929 and 1961) and other less powerful, Vulcanian eruptions (1906-1907, 1917, 1932, 1945 and 1972).

Chronology of the 22-23 April 2015 eruption

Calbuco volcano started a new eruptive cycle on 22 April 2015 at 18:05 local time. It is interesting to note that on 27 May 2014, a VT event of local magnitude Ml=3.0 was recorded at 16 km depth beneath the volcano (SERNAGEOMIN 2014a). The seismic network of Calbuco consisted of two seismometers located 5-6 km away from the summit and one infrared camera. Regional seismic reports (Valderrama et al. 2015) indicated a rise in the number of low local magnitude (Ml<0.9) volcano-tectonic (VT) events beneath the volcano during the 3 months before the eruption. In December 2014 seismic stations recorded 5 events (4 VT and 1 Long-period –LP) and only 4 in January. During February there was an increase in the number of events (40) and during March 57 events were recorded (SERNAGEOMIN 2014b, 2015a, b, c). Then, three hours before the beginning of the eruption there was a notable increase in shallow seismicity below the volcano, with the occurrence of more than 200 VT and LP earthquakes (SERNAGEOMIN 2015d). Seismic activity and underground noises were felt and reported by inhabitants near the volcano (5-10 km away from the volcano summit) for a few hours before the eruption.

The eruption started suddenly with the emission of an eruptive column from the summit crater of the volcano, which reached a height of approximately 15 km within a few minutes (Fig. 2a, b). This first eruptive phase lasted ~1.5 hours, with plume dispersion to the NE (SERNAGEOMIN 2015e, f). Occasional, minor column collapses (Fig. 2a) were reported on all flanks of the volcano (Mella et al. 2015), together with the occurrence of primary lahars on the S flank. After the end of this first phase (~19:35 local time, reported visually and seismically), continuous harmonic tremor activity was reported starting at 21:55 local time (SERNAGEOMIN 2015g). On 23 April at 01:00 local time, a second eruptive phase started with the generation of an eruptive column that reached a maximum altitude of 17 km (estimated with satellite images, SERNAGEOMIN 2015h, i, Bertin et al. 2015, Van Eaton et al. 2016). This phase lasted about 6 hours with similar plume dispersion to the NE as the first phase. As this second phase occurred during the nighttime few visual observations were possible, although the lower part of the eruptive column was partially observable due to incandescence at the vent and very intense lightning inside the column (Fig. 2c). We could observe directly that the intensity of both incandescence and lightning varied during this second phase, with a notable increase after 3:30 local time, which coincides with observations made by Van Eaton et al. (2016). These authors attribute this increase to a major occurrence of PDCs. Towards the end of this second phase (~7:00 local time, noticed seismically and by the lack of incandescence and lighting), there was a notable increase in the occurrence of VT events, with a maximum magnitude of 3.2 (Valderrama et al. 2015).

Fig. 2
figure 2

a Start of the first phase at 18:05 of 22 April 2015. Notice the pyroclastic flows at the base of the developing eruptive column. Picture taken from the N of the volcano (photo by Lester Acuña). b 15-16 km high eruptive column of the first phase. Osorno volcano to the left. Plume dispersion is to the NE. Picture taken from Puerto Varas, 30 km W of the volcano. c Second phase on 23 April. Notice the incandescence and lightning inside the column. Picture taken from Frutillar, 40 km to the NW of the volcano at 2:15 AM

Full size image

During the two eruptive phases, PDCs reached 6 km from the vent on the NE and SW flanks, and subsequent lahars reached Chapo Lake on the S flank (Fig. 1) where a number of houses and fish farming buildings were destroyed (SERNAGEOMIN 2015i). As a consequence of these two eruptions, air traffic was disturbed in southern Chile and Argentina. The town of Pucón located 230 km to the N (Fig. 1) was completely darkened during the morning of 23 April (6:00-12:00 local time, reported by local inhabitants). Fall deposits blocked the main roads up to 25 km NE of the volcano and many minor buildings collapsed due to the load of tephra (15-40 cm thick) accumulated on their roofs.

In the next days activity decreased notably, with sporadic eruptive events that generated weak plumes (< 2 km high above the vent) and minor column collapses. On 30 April, at 13:08 local time, a third phase occurred, but it was of much lower intensity and magnitude than the previous two. It lasted only a couple of hours, with a 3-5 km high eruption column above vent and a SE dispersion axis (SERNAGEOMIN 2015j). In the following weeks the eruptive and seismic activity decreased gradually, and on 28 May the alert level from SERNAGEOMIN was lowered to baseline, pre-eruption levels.

Methods

Fall deposits at proximal to medial distances (up to 50 km from the vent) were analyzed in the field on 23-25 April and 19-24 May 2015. We did not notice significant compaction of the deposits between these two periods. We sampled 57 sites in proximal and medial areas (5 – 100 km from vent) along the main roads to the NE of the volcano in Chile. No sample site was closer than 5 km from the vent because of safety reasons and difficulty of access.

In order to reconstruct the isopachs and isopleths, we measured the thicknesses of the whole fall deposit and report the maximum size of clasts in the coarsest layers in the fall deposit by averaging the lengths of the 3 orthogonal axes of the 5 largest clasts for each measurement sample location.

We analyzed 10 tephra fall samples from 3 outcrops along the downwind axis at 6, 14 and 27 km from the vent for componentry and grain-size analysis. For each sample we determined the mean size and sorting parameters (Inman 1952). Additionally, we analyzed the componentry in the 0.5-1 and 2-4 mm size range, separating the different types of clasts with the aid of a binocular magnifying lens and weighing each class. We also made density measurements on selected clasts by determining weights in air and water (Archimedes principle) using clasts sealed in wax.

We conducted a series of petrographic and chemical analyses of the juvenile material. Major elements of mineral and matrix glasses of 7 selected samples from the fallout and PDC deposits were analyzed at Laboratoire Magmas et Volcans (Université Blaise Pascal, Clermont-Ferrand, France) using a CAMECA SX-100 microprobe. The operating conditions were 15 kV accelerating voltage, 10–12 nA beam current, and 10 s counting time for minerals; and 15 kV accelerating voltage, 8 nA beam current, 5-10 μm beam size and 10 s counting time for matrix glass measurements. Major and trace element concentrations of 18 samples from juvenile blocks and bombs of the PDC and the tephra fall deposits were analyzed at the Laboratoire Domaines Océaniques (Université de Bretagne Occidentale, Brest, France). Chemical analyses were obtained by ICP-AES (Inductively Coupled Plasma-Atomic Emission Spectroscopy).

Products of the April 22-23, 2015 eruptions on the NE flank

General stratigraphy

The 22-23 April eruptions generated a series of fallout, PDC and lahar deposits distributed on various flanks of Calbuco (Fig. 1). The tephra fall deposits were distributed mainly to the NE of the volcano, with a total thickness of 52 cm at 6 km from the vent (Fig. 3a) that thinned to 5 cm at 27 km along the downwind axis. We identified four subunits in the fall deposit: layers 0, 1, 2 and 3 from base to top (Fig. 3a, b, Table 1). The deposits contain mainly juvenile basaltic andesite clasts (54.5 – 55.4 wt.% SiO2) with minor amounts (< 10 vol.%) of lithics. The two lower layers (0 and 1) are composed mainly by brown, vesicular scoria (Fig. 4a). The upper layer (3) is dominated by dark grey, dense scoria (Fig. 4b), while layer 2 is a mix of both types. A third type of juvenile clast is a more vesicular white pumice that makes up < 1 vol.% of the deposit (Fig. 4c, Table 1).

Fig. 3
figure 3

a 4 layers of the tephra fall deposit, 52 cm thick 6 km NE of the crater. b Tephra fall deposit, 23 cm thick 16 km from the vent. c Río Blanco-Frío valley, one month after the eruption; note strong smoking from the PDC deposits, which had been eroded by secondary lahars. Locations of stratigraphic sections indicated d PDC lobe (unit I) from the second phase of the eruption. Notice the concentration of scoria bombs at the front. Layer 3 of the fall deposit partially covers the lobe. e PDC deposit layer C. Notice the undular contact with deposit D (above) and the abundance of whitish, highly crystalline blocks (WB) and cauliflower scoria bombs (CB). Shovels for scale. f detail of sharp contact (dashed line) between 2 pyroclastic density current beds; the upper contains brown scoria fragments incorporated from the lower one (BS). A “cauliflower” scoriaceous bomb (CB) is also shown

Full size image
Table 1 Field description of the stratigraphic units of the 22-23 April 2015 Calbuco eruption
Full size table
Fig. 4
figure 4

a Brown Scoria from fall deposit layer 1. b Dark-grey, dense juvenile from fall deposit layer 3. c White, highly vesicular pumice from fall deposit layer 3

Full size image

PDCs were generated in some of the main valleys and gullies around the volcano, reaching up to 6 km from the vent (Fig. 3c, d, e, f). We investigated in detail the PDC deposits located in the Frio-Blanco-Hueñu Hueñu river system on the N and NE flanks of the volcano. At least 8 successive pyroclastic density currents inundated parts of the Río Frio and Río Blanco valleys (Units A to I from base to top, Fig. 5, Table 1). Individual deposits have thicknesses of 2-4 m, while the whole sequence has a thickness of 20-30 m. These PDC deposits were emplaced above the two lower tephra fall layers (0 and 1) and below the upper fall deposit layer (layer 3, Fig. 5). The PDCs are partially eroded (and dissected) by a series of secondary lahar deposits, forming a pyroclastic terrace in the valley, situated 20-30 m above river level.

Fig. 5
figure 5

Stratigraphic scheme (not to scale) of 3 sections in the Río Blanco – Frío valley, showing the relationships between the tephra fallout and PDC deposits. Map insert indicates the location of each section

Full size image

Most of the PDC deposits observed on the NE flank are massive and rich in dense scoria bombs (Fig. 6a), although some units present diverse sedimentary structures (Table 1). Individual deposit units have a levée-channel morphology, with lateral margins and frontal lobes enriched in bombs up to 1 m in diameter (Fig. 3d). This is typical of pyroclastic flows in which grain size segregation causes concentration of coarse clasts at the top and margins, the latter forming static levées that delimit a central channel from which most material is drained and finally accumulates at the front to form the frontal lobe (e.g. Jessop et al. 2012).

Fig. 6
figure 6

a Dark-grey dense bomb from the upper unit I of the PDCs. b Whitish, highly crystalline bomb from the C unit of the PDC deposits. c Bomb from unit I of the PDC deposit. Notice layering with darker zones, with texture similar to scoria bombs. d Scoria bomb from unit T, with whitish, highly crystalline enclave

Full size image

Deposit C (Fig. 3e) contains a predominance of whitish, highly crystalline dense clasts (Fig. 6b) that were not recognized in any other deposit and have a different composition (andesitic, 58 wt.% SiO2) compared with the rest of the eruption products. There are also occasional clasts with enclaves and mixing textures in all units (Fig. 6c, d).

Distribution, volume and eruptive parameters

The tephra fall isopachs maps (Fig. 7a, b) show distribution mainly to the NE, with a main axis at ~N35°E, which is consistent with satellite observations of the plume dispersion during the eruption (Fig. 7e). Calculation of the deposit volume for each layer was not straightforward, mainly because it was not possible to identify individual layers in distal areas. We estimated the total volume of the two eruptive phases using the isopachs of the whole deposit (Fig. 7a, b), considering the exponential (Pyle 1989; Fierstein and Nathenson 1992), power-law (Bonadonna and Houghton 2005) and Weibull (Bonadonna and Costa 2012) methods (Fig. 7f). The exponential method with two segments gives a total bulk volume of 0.29 km3; the power law method, 0.28 km3 (using a distal integration limit of 400 km); and the Weibull fit, 0.38 km3 (see discussion for these differences and comparison with Romero et al. 2016).

Fig. 7
figure 7

a Proximal isopachs (in cms) of the 22-23 April eruptions (total deposit). b Distal isopach (0.1 cm) of the total deposit. c Isopleth (in cm) of layer 1. d Isopleth (in cm) of layer 3. e Satellite image (from NASA) showing the plume dispersion of the second phase. f Thickness versus square root of area of the total (phase 1 + 2) fall deposit from the eruptions with Weibull, power law and exponential fits

Full size image

According to the higher intensity of the second phase (Van Eaton et al. 2016), the relative thickness of each fall deposit layer, the timing of the PDCs and the correlation between fallout and PDC deposits, we assign layer 0 to the first phase and layers 1, 2 and 3 to the second phase. Romero et al. (2016) assigned layers 0 and 1 (units A and B in their work) to the first phase, but they did not consider the stratigraphic relationships between the PDCs and fall deposits. Furthermore, their eruption rate results indicate a nearly constant intensity through both main phases, whereas Van Eaton et al. (2016) clearly indicate a higher intensity for the second phase. We used proximal thicknesses to estimate the respective volumes of the first (layer 0) and second (layers 1, 2 and 3) eruptive phases at approximately 15% and 85% of the total erupted volume.

Figs. 7c and 7d show the reconstructed isopleths for layers 1 and 3. For layers 0 and 2 it was not possible to reconstruct isopleths as these units were not easily recognizable in medial and distal parts. Using the method of Carey and Sparks (1986), the estimated maximum column heights corresponding to these layers were 16.5 and 17 km above the vent, respectively, in agreement with satellite estimations of the second phase (Van Eaton et al. 2016). The maximum mass eruption rates estimated from these column height values according to Wilson and Walker (1987) are 2.4 x 107 kg/s and 2.7x107 kg/s.

The area covered by the PDC deposits on the NE flank is approximately 650,000 m2. Assuming a mean thickness of 15 m for the whole sequence, a volume of ~107 m3 is obtained in this sector and ~3x107 m3 (i.e. 0.03 km3) for the whole deposit, equaling less than 10 vol.% of the fall deposit volume. Taking into account the individual thicknesses measured, it is likely that each individual flow deposit volume is less than 106 m3. Satellite images (Google Earth) show that pyroclastic flow deposits on the W and SSW flank had slightly shorter runouts and probably slightly smaller volumes compared to those of the SE flank.

Componentry and grain size analysis

The componentry of all fall deposit samples is dominated by juvenile material. In the 0.5-1 and 2-4 mm ranges, lithic fragments represent less than 12 wt.% in all layers (Fig. 8a). Layers 0 and 1 are dominated by brown scoria (78 and 90 wt.% at 5 km from the vent and 95 and 96 wt.% at 14 km, respectively in the 2-4 mm range) with low contents of dense, dark grey scoria (10 and 5 wt.% at 5 km and 2 and 1 wt.% at 14 km). Layer 2 has greater abundances of dense dark-grey scoria (25 and 15 wt.% at 5 and 14 km respectively). Layer 3 contains the greatest abundance of dark grey dense scoria (78 – 90 wt.% versus 3 – 20% of brown scoria in the 2-4 mm range).

Fig. 8
figure 8

a Componentry in the 4 and 1 mm range of selected samples of the tephra fall deposit. b Grain size distribution of the tephra fall deposit samples at 6, 14 and 24 km from the vent

Full size image

All fall deposit samples plot in the tephra-fall field defined by Walker (1971), with calculated modes between -3Φ and -2Φ. The sorting of the samples ranges between 1.05Φ and 1.4Φ, indicating well-sorted deposits (Fig. 9a). All samples have very low amounts (< 5 %) of fine material (< 63 μm), except layer 3 which has 6 wt.% of fine material at 6 km from vent (Fig. 8a). All samples are unimodal, with a better sorting and finer grain-size with distance (Fig. 8b).

Fig. 9
figure 9

a Median and standard deviation of fall deposit and PDC samples from the eruption. Fields of fallout, pyroclastic density current and “surge” (dilute PDC) from Walker (1971) are also shown b Componentry in the 1 mm range of selected samples of PDC deposits. c Grain size distribution of PDC deposits from section 1. d Grain size distribution of PDC deposits from section 2

Full size image

Grain-size distributions of the PDC deposits are shown in Fig. 9c and d. Most samples are within the dilute and dense PDC fields of Walker (1971), although some units are well-sorted, similar to the fall deposits (Fig. 9a). Most of the samples show modes in the range between -1Φ and 1Φ and standard deviations between 2.5Φ and 4Φ. PDC deposits that are massive (layers A, B, C, F) show two modes, typically at -4Φ and 3Φ (Fig. 9a, b). Laminated PDC deposits are better sorted with only one mode at -1Φ, -3Φ or 5Φ. The componentry of selected samples in the 1 mm range is shown in Fig. 9b. Most of the samples show a similar proportion of brown vesicular and dark grey dense juvenile fragments. The PDC samples show lithic contents in the range of 7 to 43 wt.%, generally much higher than the lithic content of the fall deposit samples (<10%).

Petrological data

There are 4 types of juvenile clasts in the eruption deposits (Table 2). The two most abundant types are brown vesicular scoria and dark grey, dense scoria. SEM images reveal that brown scoria are more vesicular (30 – 60 vol.% versus 15 – 20 vol.%) than the dense clasts and have fewer microlites (Fig. 10a, b, c, d). Phenocryst content varies between 20 – 25 vol.%, except for whitish clasts from PDC unit C (90 vol.%). Vesicularity varies from <5 vol.% in dense whitish clasts to 90 vol.% in white pumices. The mineralogy of all types of clasts includes plagioclase, ortho- and clino-pyroxene, minor amounts of olivine and Fe-Ti oxides (Fig. 11a, b, c, d, Table 3). Plagioclase is normally zoned with tiny (10-20 μm) overgrowth rims rich in Na. There are also occasional reverse zoned plagioclase crystals (Fig. 11a). White pumice also contain amphibole (Fig. 11c).

Table 2 Description of the 4 types of juvenile fragments from the eruption
Full size table
Fig. 10
figure 10

SEM images of the brown scoria (a and c) and Dense dark-grey juvenile (b and d) of the tephra fall deposit. Notice the larger size of vesicles in a. Dark-grey samples have a higher density of vesicles but of smaller size. Notice in d the high content of microlites

Full size image
Fig. 11
figure 11

Microprobe images of selected minerals from fragments in tephra fallout and PDC deposits. a Plagioclase with oscillatory zonation in rims b Ortho and clino-pyroxene from brown scoria. c Amphibole from white pumice d olivine from sample of C unit of the PDC deposit

Full size image
Table 3 Main features of the mineral phases of the products of the Calbuco eruption
Full size table

All scoriaceous and dense juvenile are basaltic andesite (54.5-56.5 wt.% SiO2, normalized on an anhydrous basis, Table 4) with similar major and trace element compositions (Fig. 12). Pumiceous samples are slightly more silicic (56.3-56.5 wt.% SiO2), whereas the whitish, highly crystalline clasts from PDC unit C are andesitic (58.1 wt.% SiO2). These samples plot inside the field of Calbuco volcano compositions (López-Escobar et al. 1992; Sellés and Moreno 2011), which define a low-to-medium-K calc-alkaline magmatic series characterized by basaltic andesites and andesites with scarce basalts and dacites (Fig. 12a). The composition of 2015 Calbuco samples is very similar to the 1961 lava flow.

Table 4 Major and minor element composition for samples from the fallout and PDC deposits from the 22 and 23 April 2015 eruptions from Calbuco volcano
Full size table
Fig. 12
figure 12

a K2O v/s SiO2 b MgO v/s SiO2 c Rb v/s SiO2 and d La v/s SiO2 for selected samples of the 22-23 April eruptions. The only sample that has a more evolved composition is the whitish highly crystalline block from layer C of the PDC deposits. Also shown (star) 1961 lava flow composition

Full size image

Discussion

Dynamics of the April 2015 eruption and eruptive parameters

The eruption of 22-23 April 2015 consisted of two phases that lasted 1.5 and 6 hours, respectively, with eruptive columns with a maximum height of ~17 km above the crater, and generated PDCs that travelled up to 6 km from the vent, as well as the occurrence of lahars immediately after and during the weeks following the eruption. Visual reports and fall deposit characteristics indicate that the eruptive column during the first phase reached its maximum height very shortly after the start of the eruption and was relatively steady thereafter with little fluctuations. The second phase was more pulsatory and produced larger PDCs, which reached longer distances than during the first phase. We correlate layers 1, 2 and 3 with the second phase. We were unable to estimate column height from layer 2, but its smaller maximum clast size (Fig. 3a, b) suggests a lower column height during the deposition of this layer. We correlated this layer with the major occurrence of PDCs, because layer 0 and 1 are below the PDC units and the entire layer 3 is above. These observations agree with those of Van Eaton et al. (2016), who noted a stopping of the cloud expansion between 2:38 and 3:38 local time, which could indicate a decrease in the eruption rate. They also suggest onset of the major occurrence of PDCs at ~3:00 local time. Our observations indicate that eruption rate increased again at the end of the eruption, because layer 3 is reversely graded and lacks normal grading towards the top.

Due to the correlation of layer 2 of the tephra fall deposit with the PDC deposits, we suggest that the major occurrence of PDCs during the second phase was associated with eruption of a large proportion of denser clasts (indicated by the PDC componentry, Fig. 9) and partial column collapse, instead of an increase of the eruption rate. Pyroclastic flow deposits represent approximately 10 vol.% of the total eruption volume, indicating that they were a minor product of the eruptive columns compared with fallout.

Bertin et al. (2015) and Van Eaton et al. (2016) estimated a column height of ~ 15 km for the first phase based on the height of the umbrella cloud in satellite images. Our column height estimate for the second phase (~17 km) is similar; although they reported overshooting tops of 23 km. Romero et al. (2016) calculated lower heights for all the tephra fall layers (12 – 15 km). This could be due to uncertainties of clast density and/or size measurements, although Romero et al. (2016) did not recognize the reverse grading observed in layer 3, which could explain their lower value of column height.

The volume of the fall deposits obtained with the Weibull method is considerably larger (~30%, 0.38 km3) than those obtained with the exponential or power-law methods. This could be due to the lack of very proximal data (< 5 km from the vent), which does not permit us to identify an additional proximal segment in the exponential method. Uncertainty in the total extent of the deposit in distal areas is problematic for the power-law method. The distal fallout in Argentina left a very thin deposit (< 0.5 cm) that was quickly eroded by subsequent rain and wind. According to Bonadonna and Costa (2012) the Weibull function can overcome to some extent problems related to proximal and distal deposits, as it does not depend on integration limits and has fewer fitting parameters.

According to our volume estimations (Weibull method), the first phase emitted ~8 x 1010 kg in 1.5 hours, which implies a mean mass eruption rate of 1.4x107 kg/s. According to Wilson and Walker (1987) this value implies a mean column height of 14.7 km, close to the value reported by Bertin et al. (2015) and Van Eaton et al. (2016). The second phase emitted ~3.2 x 1011 kg in 6 hours, which implies a mean eruption rate of 1.5 x 107 kg/s, almost equal to that of the first phase. However, the calculated eruption rate for layers 1 and 3 is ~2.7x107 kg/s based on the column height calculated from isopleths. Thus, the eruption rate during sedimentation of layer 2 must have been lower, confirming our observations of the tephra fall deposit or alternatively we underestimated the volume of the second phase. Thus, we suggest that a total bulk volume of 0.38 km3 (0.15 km3 DRE assuming a deposit density of 1,000 kg/m3) is probably the best estimate and it is close to the estimate made by Van Eaton et al. (2016) of 0.58 km3. The eruptive volume classifies the eruption as a VEI 4 (Newhall and Self 1982) with an intensity of 10 and magnitude of 4 (Pyle 2000); both phases of the eruption are classified as sub-Plinian in agreement with Romero et al. (2016).

Layer 3 is noteworthy for the dominance of dense, dark grey juvenile clasts, as opposed to the more-vesicular and brown clasts that are dominant in the rest of the fall deposit (except at the top of layer 0 which is also enriched in dense juvenile clasts and indicates the end of phase 1). That layer 3 also has slightly higher content of fine-grained tephra might suggest a higher degree of fragmentation due to a phreatomagmatic component. However, that ash could include fine ash suspended in the atmosphere from previous PDCs. Additionally, the low content of lithic clasts together with the morphology and vesicularity of the majority of juvenile clasts indicate that the eruption was essentially magmatic, with little or no involvement of external water.

As stated previously both brown scoria and dark grey juvenile clasts from the fall deposit have a similar whole rock and trace-element compositions. A similar pattern of fall deposits with two distinct juvenile populations of the same composition, with denser clasts in the upper part of the deposit, was described for the 1992 sub-Plinian eruptions of crater Peak, Mount Spurr, in Alaska (Gardner et al. 1998) and also for the larger AD 79 Vesuvius and 1980 Mount St. Helens eruptions (Carey and Sigurdsson 1987; Klug and Cashman 1984). According to Gardner et al. (1998) and Scandone and Giacomelli (2001), the difference in vesicularity between the juvenile clasts was not due to heterogeneities in the magma reservoir but rather to variable ascent rates, decompression and degassing of the magma chamber during the eruptive process.

Comparison with previous Calbuco eruptions

Major eruptive events during historical times at Calbuco volcano occurred in 1792, 1893-1895, 1917, 1929 and 1961. These eruptions were mainly sub-Plinian events with eruptive columns ~12-15 km high and durations of a few hours for the paroxysmal phase. Between these events other minor eruptions also occurred in 1845, 1906-07, 1911-12, 1932, 1945 and 1972 (Petit-Breuilh 1999). Comparison with previous major eruptions (Petit-Breuilh 1999; Selles and Moreno 2011) suggests that the magnitude and intensity of the 2015 eruption were similar to those of earlier historical events.

There are three main differences that distinguish the 2015 eruption. First, there was a very short period of time (a few hours) between the first unequivocal signals of renewed activity and beginning of the eruptive activity. There was no observed gas release or small explosions, and although low-magnitude seismicity was registered in the previous 3 months, no fumaroles, noises or earthquakes were reported by inhabitants. Previous documented major historical eruptions of Calbuco volcano (1893-1895, 1929, 1961) were characterized by more-prominent precursory signals, such as seismic swarms felt by the inhabitants in surrounding areas and development of fumaroles weeks or months before the main explosive events (Petit-Breuilh 1999). Second, the interval between the two main phases was only a couple of hours, in contrast with typically weeks or months in previous explosive events. Third, the 2015 eruption was characterized only by explosive activity, without emission of lava flows or domes, which had been formed in all previous documented eruptions (Petit-Breuilh 1999).

We suggest that closely spaced eruptive pulses, fast ascent of magma, short duration of the eruption and the occurrence of only explosive activity during the 2015 eruption are consistent with fast ascent of magma from the crustal storage region to the vent. Woods and Koyagushi (1994), Eichelberger (1995) and Slezin (2003) have argued that high ascent velocities of magma prevent the escape of volatiles upwards through the melt or through porous host rock, hence favoring explosive eruptions. Scandone et al. (2007) also argued that pre-eruption VT seismicity is associated to rock fracturing and the developing and interconnection of the conduit network. For the Calbuco eruption, the brief 3-hour swarm of seismic events and the very low magnitude and number of VT events during months before the eruption indicate that the transport system opened quickly and magma ascended at high rates from the moment an overpressure threshold was reached in the magma chamber.

Triggering mechanisms

The sudden start and minimal precursory signals of the 2015 Calbuco eruption raise the question of what were the eruption's triggering mechanisms. Passarelli and Brodsky (2012) reviewed the potential causes for the duration of precursory activity (i.e. the unrest period) and found a positive correlation with repose time and silica content. However, the repose time before this eruption and composition of the 2015 magma are similar to those of previous eruptions, so these factors cannot explain the shorter unrest period of the 2015 eruption compared with previous ones.

For the 2015 Calbuco eruption we discard earthquake triggering because no major earthquake has been reported in the region for the past five years (Chilean National Seismic Center, www.sismologia.cl). We also discard magma-water interactions as a triggering mechanism; we show in previous sections that the tephra fall deposits have a low abundance of fine-grained tephra and scarce lithic fragments, typical of products from magmatic eruptions. Our data suggests that large scale magma mixing is also very unlikely because the composition of the emitted products is very homogeneous (with the exception of the highly crystalline blocks from unit C of the PDCs). There is some macroscopic textural heterogeneity (banding, enclaves; Fig. 6c, d) but with same (± 1 wt.% SiO2) chemical composition. The abundance of mineral disequilibrium textures and reverse zoning is also very low. In addition, according to White and McCausland (2016), the low seismic energy released by the precursory activity supports the hypothesis of little or no magma injection.

Our data suggest that a likely mechanism for triggering of the 2015 Calbuco eruption is volatile oversaturation, which built-up overpressure inside the magma chamber and caused the eruption. Preliminary microprobe results on amphiboles indicate a high content of water (4-5%) in the magma, and pressures of 180-240 MPa (Ridolfi and Renzulli 2012, see Supplementary Material 1), together implying oversaturation of water (Baker and Alletti 2012) for a basaltic andesite magma, although amphiboles are only present in the white pumices that represent a very small (<1 vol.%) fraction of the eruptive products. Volatile oversaturation can be caused by crystallization of anhydrous minerals (e.g. Stix et al. 1997), by the exsolution of volatiles from a cooling injection at the base of a magma chamber (Folch and Martí 1998) or by a decompression event such as volcano edifice failure (Scandone and Giacomelli 2001).

There was no obvious decompression event, so the most probable options are the first two. Eruption triggering by differentiation and volatile oversaturation in a closed system was first investigated by Blake (1984) for silica-rich systems. Later, Tait et al. (1989) and Stix et al. (1997) analyzed basaltic and andesitic magmatic systems respectively, showing that low degrees of crystallization (< 10%) can generate the necessary overpressure, depending on the depth of the magma chamber.

The second option is that the triggering mechanism was a small intrusion of magma that did not significantly mix with the host magma, or that had a composition close to that of the magma resident at the base of the chamber. This intrusion would have not led to extensive magma mixing but may have heated the base of the chamber, as shown by rare disequilibrium textures and reverse zoning in minerals (Fig. 11a). The arrival of magma at the base of the reservoir generated overpressure because volatiles were exsolved from the intruded magma (Folch and Martí 1998). This intrusion process may provoke convection and mobilization of different parts of the chamber with different degrees of crystallinity (Fig. 13a) as evidenced by the eruption of clasts with different textures such as white pumices and the very crystalline white blocks from the deposit C of the PDCs.

Fig. 13
figure 13

Cartoon (not to scale) showing the evolution of the April 2015 eruption. a Crystallization or a small intrusion of magma produces exsolved volatiles, with generation of convection and vesiculation of the intruded magma that increase the overpressure. b The overpressure triggers the first eruption, decompressing the chamber and producing strong vesiculation of the upper part of the chamber. c After the end of the first phase, strong vesiculation and crystallization occur in response to depressurization after evacuation of first-phase magma, increasing the overpressure again. d The second phase starts with higher eruption rate than the first phase. e As vesiculation continues, dark-grey dense clasts begin to be erupted, with a lower eruptive column and major occurrence of PDCs. f Eruption rate increases again and a stable eruptive column is reestablished. Dark-grey clasts dominate the erupted products. Sudden stop of the eruption. Strong VT seismicity signals the readjustment of the magma chamber

Full size image

In either case, the exsolution of volatiles generated an overpressure inside the magma chamber (Fig. 13a), triggering the first phase of the eruption. The eruption of the first batch of magma (Fig. 13b) started to decompress the chamber, causing further vesiculation (Scandone and Giacomelli 2001). However, because of the short duration of the first phase, we suggest that the decompression due to the evacuation of magma was faster than the nucleation of bubbles, causing the first eruption phase to stop. During the 5.5 hours hiatus between phases, vesiculation inside the chamber became more vigorous (Fig. 13c), signaled by strong harmonic tremor (Scandone and Giacomelli 2001; McNutt et al. 2015). The tiny, Na-rich overgrowth rims observed on plagioclase phenocryst suggest a decompression-induced degassing process occurring just before the eruption (Crabtree and Lange 2011). Experimental studies suggest that a 1-2 wt.% decrease in water contents of siliceous magmas can result in lowering the An content of plagioclase by tens of mole percent. (Couch et al. 2001, Waters and Lange 2015). This vesiculation increased the overpressure inside the chamber up to a threshold value that triggered the second phase. This second phase started with a stronger eruption rate, as vesiculation reached higher degrees (Fig. 13d). As the eruption continued, higher degrees of exsolution promoted the crystallization of microlites, increasing the viscosity of magma and inhibiting the growth of vesicles, hence provoking an increase in the proportion of dark-grey dense juveniles. The eruption rate during this transitional stage was lower (Fig. 13e), but the continuous vesiculation of magma in the reservoir favored a new increase in eruption intensity (Scandone and Giacomelli 2001) towards the end of this phase, with a return to a sustained column again (layer 3, Fig. 13f). VT events towards the end and after the second phase probably reflected readjustment of the magma chamber. After the end of the second phase, the magma chamber was highly degassed and (partially) emptied, generating only small eruptions that lasted for a couple of weeks.

Conclusions

The Calbuco volcano eruption of April 22-23, 2015 was marked by two sub-Plinian events that generated eruptive columns of 15-17 km height with a NE dispersion axis and a total erupted volume of 0.15 km3 (DRE). Pyroclastic density currents produced only ~10% of the total deposit volume and reached distances of up to 6 km. Despite the wide range in textures of clasts emitted during the eruption, most of the eruption products are basaltic andesite in composition. The eruption was preceded by scarce seismicity, which together with the homogenous composition and lack of reverse zoning and disequilibrium textures in minerals, argues against magma mixing as a triggering mechanism of the eruption. Instead, we propose that the eruption could have been triggered by volatile oversaturation, probably enhanced by crystallization or volatile addition from a small intrusion without significant mixing or with the same composition as the main body of magma. This caused over-pressurization of the system, which favored fracturing of the surrounding rock and opening of a conduit to the surface, thus causing the onset of the first eruptive phase. This phase lasted 1.5 hours. After a pause of 5.5 hours, a second phase began and was marked by a stronger intensity at its beginning. Nearly 2 hours after the start of the second phase, the eruption rate decreased and widespread PDCs were generated possibly due to the abundance of denser fragments (dark grey dense juvenile clasts) in the column. This second phase caused more decompression and vesiculation inside the chamber, thus increasing microlite crystallization and bubble nucleation, and culminated with a sustained eruptive column that reached a height of 17 km at the end of the eruption. The eruption ended abruptly after 6 hours of almost continuous activity.