PUBLICATIONS. Geochemistry, Geophysics, Geosystems

Transcription

1 PUBLICATIONS RESEARCH ARTICLE Key Points: Eruptive activity resumed at Turrialba volcano (Costa Rica) in 2010 with sporadic explosions that increased in frequency since October 2014 New data of 3 He/ 4 He in fumarole gases since September 2014 and chemistry of the products erupted between October 2014 and May 2015 Eruptive activity was triggered by the supply of the plumbing system by a 3 He-rich magma which led to an increasing juvenile component Eruptive activity at Turrialba volcano (Costa Rica): Inferences from 3 He/ 4 He in fumarole gases and chemistry of the products ejected during 2014 and 2015 Andrea Luca Rizzo 1, Andrea Di Piazza 2, J. Maarten de Moor 3,4, Guillermo E. Alvarado 5,6, Geoffroy Avard 3, Maria Luisa Carapezza 2, and Mauricio M. Mora 6,7 1 Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Palermo, Palermo, Italy, 2 Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Roma1, Roma, Italy, 3 Observatorio Vulcanologico y Sismologico de Costa Rica (OVSICORI), Apdo , Universidad Nacional, Heredia, Costa Rica, 4 Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque, New Mexico, USA, 5 Instituto Costarricense de Electricidad, Apdo , San Jose, Costa Rica, 6 Red Sismologica Nacional, Apdo. 2060, San Jose, Costa Rica, 7 Escuela Centroamericana de Geologıa, Universidad de Costa Rica, Apdo. 2060, San, Jose, Costa Rica Correspondence to: A. L. Rizzo, andrea.rizzo@ingv.it Citation: Rizzo, A. L., A. Di Piazza, J. M. de Moor, G. E. Alvarado, G. Avard, M. L. Carapezza, and M. M. Mora (2016), Eruptive activity at Turrialba volcano (Costa Rica): Inferences from 3 He/ 4 He in fumarole gases and chemistry of the products ejected during 2014 and 2015, Geochem. Geophys. Geosyst., 17, doi: / 2016GC Received 7 JUL 2016 Accepted 13 OCT 2016 Accepted article online 17 OCT 2016 Abstract A new period of eruptive activity started at Turrialba volcano, Costa Rica, in 2010 after almost 150 years of quiescence. This activity has been characterized by sporadic explosions whose frequency clearly increased since October This study aimed to identify the mechanisms that triggered the resumption of this eruptive activity and characterize the evolution of the phenomena over the past 2 years. We integrate 3 He/ 4 He data available on fumarole gases collected in the summit area of Turrialba between 1999 and 2011 with new measurements made on samples collected between September 2014 and February The results of a petrological investigation of the products that erupted between October 2014 and May 2015 are also presented. We infer that the resumption of eruptive activity in 2010 was triggered by a replenishment of the plumbing system of Turrialba by a new batch of magma. This is supported by the increase in 3 He/ 4 He values observed since 2005 at the crater fumaroles and by comparable high values in September 2014, just before the onset of the new eruptive phase. The presence of a number of fresh and juvenile glassy shards in the erupted products increased between October 2014 and May 2015, suggesting the involvement of new magma with a composition similar to that erupted in We conclude that the increase in 3 He/ 4 He at the summit fumaroles since October 2015 represents strong evidence of a new phase of magma replenishment, which implies that the level of activity remains high at the volcano. VC American Geophysical Union. All Rights Reserved. 1. Introduction Turrialba is the southernmost active volcano of the Central America Volcanic Front and is located about 35 km east of San Jose, the capital of Costa Rica (Figure 1). The last eruption prior to 2010 occurred during [Reagan et al., 2006]. After almost 150 years of quiescence, the volcano entered an unrest phase in 1996 that was initially characterized by anomalous seismicity [Martini et al., 2010]. Sporadic eruptions have occurred at the southwest crater since January 2010, with the opening of a new vent. Afterward the frequency has increased over time and has involved the eruption of fragments of altered preexisting material comprising a small percentage of a juvenile component [Reagan et al., 2011]. The reawakening of Turrialba volcano represents a serious hazard due to its proximity to both the capital city and main international airport of Costa Rica, which is only 39 km from the volcano. Preparedness is as important as adequate monitoring of this volcanic activity to minimizing its impact [van Manen, 2014; van Manen et al., 2015]. Nevertheless, forecasting the resumption of eruption at volcanoes is not obvious, above all during the initial vent-opening phases of eruptive activity at long dormant volcanoes. This is especially the case of Turrialba volcano, where vent-opening eruptions have been ongoing since Indeed, while large magmatic eruptions are often preceded by geophysical and geochemical precursory signals, there is still very little documented evidence worldwide of precursory geophysical (i.e., seismicity, deformation, and gravimetry) or RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 1

2 Figure 1. (a) Map of the southern sector of Turrialba volcano displaying the location of Falla Ariete and the gas sampling site. The top-left insert shows an image of the eruption that occurred on 13 March The area indicated in the top right is shown magnified in Figure 1b. (b) Summit crater area with the most important morphological features indicated: the position of the vents that have opened during the past 4 years and the sampling sites for gases and eruptive products. The white square shows the position of the CVTR seismic station (Red Sismologica Nacional [RSN: UCR-ICE]). geochemical (gas composition and gas flux) signals prior to phreatic or phreatomagmatic eruptions. Among the geochemical tools used for volcano monitoring, the 3 He/ 4 He measured in gases emitted from active volcanic area seems to have enormous potential for predicting nonmagmatic unrest. Sano et al. [2015] reported a decadal-scale increase in the 3 He/ 4 He toward more-magmatic values at Ontake volcano prior to the fatal phreatic blast that claimed 57 lives in 2014, which apparently represented the only long-tomedium-term precursory signal recorded by the local monitoring network. Similar increases in the 3 He/ 4 He but on a time scale of months to weeks have been documented at other active volcanoes worldwide, including Mt. Etna (Italy) [Caracausi et al., 2003; Rizzo et al., 2006; Paonita et al., 2012, 2016], Stromboli (Italy) [Capasso et al., 2005; Rizzo et al., 2009, 2015a], Ontake (Japan) [Sano et al., 2015], and Santorini (Greece) [Rizzo et al., 2015b]. These observations demonstrate the enormous potential of this tracer in predicting both magmatic and phreatic to phreatomagmatic volcanic unrest, so it could reveal fundamental also at Turrialba volcano. In addition, de Moor et al. [2016a] demonstrated using data obtained at Poas volcano, Costa Rica, that high-frequency gas monitoring is a powerful tool for identifying short-term precursory variations in gas emissions prior to phreatic eruptions. Most recently, de Moor et al. [2016b] show that plume gas compositions as measured by permanent Multi- GAS station located near the active crater at Turrialba show short-term precursory changes in gas composition prior to eruptive episodes at Turrialba. These variations in gas composition measured in the main gas plume reflect dynamic interactions between the hydrothermal system and magmatic gases derived from ascending magma bodies. However, more frequent eruptive activity is making maintenance of this Multi- GAS station risky and unfeasible. In this study, we focus on geochemical monitoring of fumaroles on the flank of the volcano, where gas samples can be collected safely. RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 2

3 After the onset of eruptive activity, the petrological study of the products that erupt from the volcano is fundamental, because it represents a complementary suite of information that can be used to identify the preeruptive conditions and processes. In fact, identifying a juvenile component in the erupted products is crucial to distinguishing magmatic or phreatomagmatic events from purely phreatic ones [e.g., Barberi et al., 1992; Nakada et al., 1995; Cashman and Hoblitt, 2004; Alvarado et al., 2016]. Cashman and Hoblitt [2004] identified a juvenile component in the precursory ash that erupted from Mount St. Helens 3 months before the onset of the eruption on 18 May However, Pardo et al. [2014] highlighted uncertainties in distinguishing phreatomagmatic from phreatic products due to the extreme difficulty of truly differentiating between the juvenile component from well-preserved material from previous eruptions. Those authors argued that excluding the presence of fresh magma input within an eruption from the study of ash deposits only is not straightforward, and that strong evidence that new magma has not erupted is the lack of particles with a distinct glass composition. The problem of distinguishing phreatomagmatic from phreatic ash is very relevant to the Turrialba eruptive activity, since a crucial question is whether the current activity will progress in a final magmatic stage. This article presents results from new 3 He/ 4 He measurements of fumarole gases collected in the summit area of Turrialba starting 1 month before the explosions resumed in October These data are compared with similar measurements carried out since 1999 with the aim of assessing the events that triggered the recent unrest and characterizing the current state of activity. We also present detailed petrological and geochemical analyses of the products that erupted between October 2014 and May 2015 and compare them with the scoriae and ashes that erupted during The main aim is to propose a likely evolutionary scenario for the current eruptive activity and to evaluate the related hazard implications. 2. Chronology of Turrialba Volcano Reactivation Turrialba volcano has experienced at least six magmatic eruptive periods during the past 3400 years. The most-recent eruption prior to 2010 occurred during and was characterized by a sequence of phreatic explosions that transitioned to phreatomagmatic activity and climaxed with Strombolian eruptions [Reagan et al., 2006]. After that eruptive phase, the volcano entered a quiescence state that lasted until the late 1990s. The recent reactivation of Turrialba was characterized by increasing seismicity, ground deformation, and fumarole activity during the mid-1990s to early 2000s, and during by the expansion of fumarole fields between and within the central and southwest craters, and along the western and southwestern outer flanks of the edifice [Martini et al., 2010; Vaselli et al., 2010]. These structural changes have been accompanied by increases in the temperature and concentration of magmatic fluids (SO 2, HCl, and HF) at the crater fumaroles since 2005 [Vaselli et al., 2010], and by a noticeable increase in the SO 2 flux by up to 2 orders of magnitude since late 2007 [Martini et al., 2010; Conde et al., 2013], which strongly indicate an enhanced degassing of magmatic fluids. The seismic activity and gas flux peaked during 2009 and 2010 [Conde et al., 2013], with the first explosive event occurring on 5 January 2010 [Reagan et al., 2011; Campion et al., 2012], which opened a vent inside the southwest crater. Ash emissions occurred in January 2012, opening new fractures with high-temperature fumaroles from which ash emissions took place again in May The activity at Turrialba escalated on 30 October 2014 with an energetic explosion at 11:35 P.M. (local time) that caused the collapse of the eastern side of the southwest-crater wall and the ejection of blocks up to 2 m in diameter near the eruptive vent. Ash emissions continued for 2 weeks, and a more-energetic explosion occurred on 9 December. This period was followed by 3 months of strong degassing and seismicity. Another eruptive phase started on 8 March 2015, with >32 intermittent ash emissions events occurring on at least 20 days up to 18 May. The most-energetic explosions during this period occurred on 12 March (Figure 1a), 7 April, 24 April, and on 6 May. Despite further weak ash emission events occurring in June and August, the volcano remained quiet until explosive activity resumed for 2 weeks during mid-october 2015 with ash columns (500 m high) and small pyroclastic surges affecting the summit area (OVSICORI-UNA, 2015). The activity then increased significantly from January to July 2016 (when this manuscript was submitted for publication), with the occurrence of several ash eruptions. RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 3

4 Figure 2. (a) The products that erupted during 2014 and 2015 are characterized by a light gray, heterogeneous basal component containing abundant hydrothermal minerals, while the top fine ash layer is darker, denser, and more homogeneous. The middle orange altered layer corresponds to the hiatus between the 2014 and 2015 eruptive episodes. The alteration occurred because of a long exposure to the gas plume. (b) The products of the eruption are well preserved in the crater region and show orange basal phreatic units covered by thinly bedded gray phreatomagmatic deposits and a massive scoriaceous unit. The scale bar reported at the right of the stratigraphic sketch is indicative and does not correspond with the height on the left column foe which the scale can be evaluated considering the hammer size. 3. Methods 3.1. Sampling of Gases and Ashes The high-temperature fumarole on the western rim of the southwest crater has been sampled twice (in September 2014 and February 2015, when its temperature was 2108C and 1648C, respectively), while a lowtemperature site (boiling temperature) along the active fault called Falla Ariete has been sampled 6 times (from January 2015 to February 2016) (Figure 1). We determined the 3 He/ 4 He in the gas samples at the noble gases isotope laboratory of Istituto Nazionale di Geofisica e Vulcanologia (INGV) in Palermo, Italy. The geochemical characterization of these fumaroles and their magmatic-derived signature have been widely investigated in previous studies [e.g., Tassi et al., 2004; Hilton et al., 2010; Vaselli et al., 2010; Di Piazza, 2014; Di Piazza et al., 2015]. We collected six samples of ashes and lapilli from the explosions that occurred during October December 2014 and on 12 March, 7 April, and 6 May These samples were collected in the summit area of the volcano within a few hundred meters of the eruptive vent by digging pits in fresh deposits, or by sampling ash from the soil immediately after the explosive events (Figure 2a) Analytical Procedures He/ 4 He Measurements in Gases The element and isotope compositions of both He and Ne in fumaroles were measured by admitting the gases into an ultra-high-vacuum (10 29 to mbar) purification line, in which all of the species in the gas mixture except noble gases were removed. Before isotope analysis, He ( 3 He and 4 He) and Ne ( 20 Ne) isotopes were separated from Ar by adsorbing the latter in a charcoal trap cooled by liquid nitrogen (77 K). He and Ne were then adsorbed in a cryogenic trap connected to a cold head cooled with an He compressor to <10 K. He was desorbed at 42 K and admitted into a GVI-Helix SFT mass spectrometer. After restoring the ultra high vacuum in the cryogenic trap, Ne was released at 82 K and then admitted into a Thermo-Helix MC Plus mass spectrometer. The same procedure was adopted for the He-isotope and Ne-isotope measurements of the air standard. RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 4

5 Table 1. 3 He/ 4 He (in Units of R/Ra, and in Units of Rc/Ra After Correction for Air Contamination), the 4 He/ 20 Ne, and the Temperature of Fumarole Gases Collected at the Turrialba Summit Area From 2005 to 2016 a Sampling Site Date (mm/dd/yy) Temperature (8C) R/Ra 4 He/ 20 Ne Rc/Ra Err 6 (Rc/Ra) This Study Southwest Crater 09/05/ Southwest Crater 02/04/ Falla Ariete 01/27/ Falla Ariete 05/27/ Falla Ariete 09/18/ Falla Ariete 10/28/ Falla Ariete 12/18/ Falla Ariete 02/10/ Di Piazza et al. [2015] West Crater (HT) 03/16/ West Crater b 03/20/ Central Crater 03/20/ Central Crater 03/20/ Central Crater 03/20/ Vaselli et al. [2010] West Crater b Feb West Crater b Apr West Crater b Mar West Crater (HT) Mar Central Crater Sep Ariete Fault Mar Hilton et al. [2010] West Crater c Jan West Crater c Mar West Crater c Jul West Crater d Jul Central Crater Jun Central Crater Jun Central Crater Jan a Data for are from Hilton et al. [2010], Vaselli et al. [2010], and Di Piazza et al. [2015]. HT indicates high-temperature fumarole collected in the inner rim of southwest crater and corresponds to our sampling site. b Fumarole collected at the top of the southwest-crater rim. c Fumarole collected at the top of the inner walls of the southwest-crater rim. d Fumarole collected at the top of the outer walls of the southwest-crater rim. Values of the 3 He/ 4 He are expressed in units of R/Ra (where Ra is the 3 He/ 4 He of air, which is equal to ) and are corrected for atmospheric contamination based on the 4 He/ 20 Ne ratio [e.g., Sano and Wakita, 1985]. Hereafter we report the 3 He/ 4 He ratio corrected for atmospheric contamination in units of Rc/Ra. In the gas samples collected in this study, the 4 He/ 20 Ne ratio ranges from 0.9 to 902 (air 4 He/ 20 Ne ), with samples collected from September 2015 onward showing the lowest ratios ( 4 He/ 20 Ne 2.3; Table 1). These samples display the largest difference between raw and corrected 3 He/ 4 He values (up to 2.3 Ra), but the relative variations observed between September 2015 and February 2016 (see section 4.1) can be considered accurate because they were recorded at comparable 4 He/ 20 Ne values (Table 1). The errors were generally within Ra. Typical blanks for He and Ne were < and < mol, respectively. Further details about the sampling and analytical procedures are available in Di Piazza et al. [2015] and Rizzo et al. [2015b] Petrological and Geochemical Analyses of Ashes Sample preparation, observation of polished sections, and chemical analyses were carried out at the INGV laboratories in Rome, Italy. Ashes and lapilli were dried at room temperature (258C), weighed, and sieved at half-/ intervals. Sieved samples were cleaned using several bath cycles in distilled water and thereafter in acetone. Under a binocular microscope, 50 particles of the freshest-looking component with a grain size of 2 < / < 1 were selected from each sample. Glassy particles were embedded in resin, ground down to expose the particles, polished (until a 0.1 lm-grain polycrystalline diamond slurry), and coated with carbon. Backscattered-electron (BSE) images of polished sections were obtained using scanning electron microscopy. Chemical analyses of glasses (from the 2014 and 2015 explosions; Table 2) and minerals (from the 2015 explosions only; Table 3) were performed using an electron microprobe (JXA8200, JEOL) combined with energy-dispersive and wavelength-dispersive spectrometer detectors (composed of 5 spectrometers RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 5

6 Table 2. Results From Electron Microprobe Analyses of Residual Glass of Turrialba Ashes That Erupted During 2014 and 2015 a Sample N SiO 2 TiO 2 Al 2 O 3 FeO MnO MgO CaO Na 2 O K 2 O P 2 O 5 SO 3 Total 6-May-15 cl cl cl cl cl cl cl cl cl Apr-15 cl cl cl cl cl cl Mar-15 cl cl cl October 2014 cl cl cl cl cl cl cl cl cl cl cl cl cl cl cl November 2014 cl cl cl cl cl cl cl cl cl cl cl cl cl cl cl cl cl cl cl cl December 2014 cl cl cl cl cl RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 6

7 Table 2. (continued) Sample N SiO 2 TiO 2 Al 2 O 3 FeO MnO MgO CaO Na 2 O K 2 O P 2 O 5 SO 3 Total cl cl cl cl cl cl cl cl cl a Major oxides are reported as wt %. N 5 number of analysis of which we reported the average values. with 12 crystals). We used a slightly defocused beam with a size of 5 lm, with a counting time of 5 s for the background and 15 s for the peak Daily Dominant Frequency of Seismic Signals Continuous seismic record from CVTR station of the Red Sismologica Nacional (RSN: UCR-ICE) was passed through a fourth-order Butterworth band-pass filter from 1 to 45 Hz. The signal was then divided in 10 minlong samples, to which the fast Fourier transform was applied and the dominant frequency was then calculated between 1 and 25 Hz. Each sample was tapered using a Hann function. The mean, standard deviation, and mode were calculated daily. 4. Results and Discussion He/ 4 He in Fumarole Gases Figure 3a shows the time series of the 3 He/ 4 He at Turrialba fumaroles located at the central and southwest craters as well as those located at Falla Ariete (Figure 1). Analytical data and fumarole temperatures are shown in Table 1. Data from 1999 to 2011 are from Hilton et al. [2010], Vaselli et al. [2010], and Di Piazza et al. [2015], and those from September 2014 to February 2016 were obtained in the present study. The 3 He/ 4 He at the fumaroles ranges from 7.32 to 8.01 Ra at the southwest crater ( ), from 6.85 to 7.88 Ra at the central crater ( ), and from 7.57 to 8.00 Ra at Falla Ariete ( ) (Figure 3). In detail, the 3 He/ 4 He at the fumaroles from the central crater decreased from 7.48 in 2001 to 6.85 Ra in 2005 and increased (with some variability) from 6.85 Ra in 2005 to 7.88 in At the southwest crater, the 3 He/ 4 He increased from 7.34 Ra in 1999 to 8.01 Ra in 2001, followed by a decrease to 7.32 Ra in Thereafter it increased again to 7.96 Ra in 2011 and 7.94 Ra in 2014, and then decreased to 7.49 Ra in the single sample collected in 2015 (Figure 3a). Increasingly dangerous conditions in the crater region since 2015 forced us to stop sampling the southwest-crater fumaroles. Thus, since February 2015 the sample collection focused on the Falla Ariete boiling-temperature fumaroles, which previously displayed 3 He/ 4 He of 7.84 Ra in 2008 and varied from 7.57 to 8.00 Ra during 2015 and 2016, with a clearly increasing trend from October 2015 onward (Figure 3b). The maximum 3 He/ 4 He were recorded in March 2011 ( Ra) [Di Piazza et al., 2015] and September 2014 (7.94 Ra) at the Southwest Crater, and in December 2015 (8.00 Ra) at Falla Ariete. The morphology of the summit area has been progressively modified since the beginning of the eruptive activity in January 2010, leading to the disappearance of the central-crater fumarole field in October The lower and more-variable 3 He/ 4 He measured in these fumaroles are consistent with a larger contribution of crustal 4 He at the area, possibly reflecting a contamination of magmatic gases by hydrothermal fluids or by interaction with crustal rocks [Hilton et al., 2010; Vaselli et al., 2010], or it could be related to the low flux that characterized this fumarole field. However, the general trend of the 3 He/ 4 He variation is similar to that at the southwest fumaroles (Figure 3a). Data collected over many years demonstrate that the magmatic 3 He/ 4 He signatures in fumarole gases at Turrialba volcano are strongest at the southwest crater, irrespective of whether the gases have a high or low temperature [Hilton et al., 2010; Vaselli et al., 2010; Di Piazza et al., 2015]. The fumaroles at Falla Ariete exhibit comparable or even higher 3 He/ 4 He values than the crater fumaroles (Figure 3), which allowed us to sample gases representative of the magmatic source in safer conditions (Table 1). RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 7

8 Table 3. Results of Electron Microprobe Analyses of Olivine, Pyroxene, and Plagioclase Crystals of the 2015 Samples Only a Date 03/12/15 03/12/15 03/12/15 03/12/15 03/12/15 03/12/15 03/12/15 03/12/15 03/12/15 04/07/15 04/07/15 04/07/15 04/07/15 04/07/15 04/07/15 04/07/15 04/07/15 Spot name cl21plg1 cl23plg1 cl23plg2 cl24plg1 cl24plg2 cl24plg3 cl25plg1 cl25plg2 cl25plg3 cl10plg1 cl10plg2 cl10plg3 cl11plg1 cl11plgmf cl12plg1 cl12plg2 cl12plg3 SiO Al 2 O Fe 2 O CaO Na 2 O K 2 O Tot Formula Based On 8 Oxygen Si Al Fe Ca Na K An Ab Or Ca 1 Na 1 K Si 1 Al Date 04/07/15 04/07/15 04/07/15 04/07/15 04/07/15 04/07/15 04/07/15 04/07/15 04/07/15 04/07/15 04/07/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 Spot name cl12plg4 cl13plg1 cl13plg2 cl13plg3 cl13plg4 cl15plg1 cl15plg2 cl16plg1 cl16plg2 cl16plg3 cl16plg4 cl2plg1 cl2plg2 cl2plg3 cl2plg4 cl2plg5 cl2plg6 SiO Al2O Fe 2 O CaO Na 2 O K 2 O Tot Formula Based On 8 Oxygen Si Al Fe Ca Na K An Ab Or Ca 1 Na 1 K Si 1 Al Date 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 05/06/15 Spot name cl4plg1 cl5plg2 cl5plg3 clplg2 clplg3 clplg4 clplg5 cl7plg1 cl7plg2 cl8plg1 cl8plg2 cl9plg1 cl9plg2 cl9plg3 cl9plg4 cl9plg5 cl9plg6 cl9plg7 SiO Al 2 O Fe 2 O CaO Na 2 O K 2 O Tot Formula Based On 8 Oxygen Si Al Fe Ca Na K An Ab Or Ca 1 Na 1 K Si 1 Al a Major oxides are reported as wt %. RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 8

9 Figure 3. Time series of the 3 He/ 4 He values (reported in units of Rc/Ra) in fumaroles at the central and southwest craters and at Falla Ariete during (a) and (b) The green line indicates the maximum Rc/Ra values measured in fluid inclusions (FIs) of basalticandesitic magmas that erupted during [after Di Piazza et al., 2015]. Gray circles indicate the explosive events that have occurred at Turrialba volcano since The black points show the daily dominant frequency mode of seismic records at the CVTR seismic station from January 2009 to April The study of noble gases in fluid inclusions hosted in rocks that erupted at Turrialba during the last 7 ka revealed that the maximum 3 He/ 4 He would be expected to be 8.1 Ra [Di Piazza et al., 2015]. This would be mainly determined by the signature of the most-mafic magma types found in the Turrialba volcanic complex. Thus, 3 He/ 4 He values in fumarole gases approaching 8.1 Ra should reflect the direct degassing of a mafic and uncontaminated magma. In addition, because the solubility of He in silicate melts is low, and comparable to that of CO 2 [Paonita, 2005, and references therein], the increase in the 3 He/ 4 He should reflect preferential degassing from a mafic magma deep-seated in the plumbing system, as observed in other magmatic systems worldwide [e.g., Caracausi et al., 2003; Rizzo et al., 2015a]. Therefore, the increase in the 3 He/ 4 He observed at the southwest-crater fumaroles from 1999 to 2001 indicates a phase of degassing at depth of a mafic and 3 He-rich batch of magma and the arrival of moremagmatic fluids at the surface. This is confirmed by the enhanced degassing rate observed in 2000 from the fumaroles at the central and southwest craters, with the formation of a large fracture (named Quemada) discharging fluids at 908C between the craters [Tassi et al., 2004; Vaselli et al., 2010]. Seismic swarms (>9000 events/year in 2001), radial inflation, and changes in the fumarole-gas composition were also detected during 2001 and 2002 [Barboza et al., 2003a, 2003b; Tassi et al., 2004; Martini et al., 2010]. The 3 He/ 4 He at the southwest-crater and central-crater fumaroles decreased after 2001 (i.e., from 8.01 to 7.32 Ra at the RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 9

10 Figure 4. BSE imaging of ash particles emitted by the explosions during (a, b) October December 2014, (c, d) March 2015, and (e, f) May (a, b) Porphyritic block and a particle with a molten surface. (c) A particle with strong crystallization. (d) A glassy shard. (e, f) Glassy and spongy shards found in the products of the most-recent explosions, both of which may represent fresh magma. southwest crater and from 7.48 to 6.85 Ra at the central crater) (Figure 3a), suggesting a reduced supply of gas from the primitive magmatic source and an enhanced release of hydrothermal fluids. Accordingly, the seismicity was significantly lower than the maximum exhibited in 2001 [Hilton et al., 2010, who reported seismic data from the Global Volcanology Program, There was a new increase in the 3 He/ 4 He after 2005 at the central-crater and southwest-crater fumaroles, indicating the recurrence of degassing at depth of a mafic and 3 He-rich batch of magma and the arrival of more-magmatic fluids at the surface. This may have persisted up to September 2014, when the same high 3 He/ 4 He ratio was found at southwest-crater fumaroles (Figure 3a), though there may have been unrecorded variations since no samples were obtained during Considering that the SO 2 flux from the craters remained high between the first ash emission in 2010 and the beginning of the eruptive cycle in October 2014, with average values above 500 t/d [Conde et al., 2013; OVSICORI-UNA, 2015], it can be inferred that the strong degassing of a primitive magma source continued during the period with no measurements, as suggested by the high 3 He/ 4 He at the fumaroles (Figure 3a). The 3 He/ 4 He at the southwest-crater fumaroles decreased in February 2015 (from 7.94 to 7.49 Ra) toward values falling between those measured during (Figure 3). This suggests a reduced supply of gas from the primitive magmatic source and an enhanced release of hydrothermal fluids. Samples collected at Falla Ariete in February, May, and September 2015 yielded similar 3 He/ 4 He values (Figure 3b), which indicates the same source of volatiles and that active magma degassing was continuing at depth. Since October 2015, the 3 He/ 4 He at Falla Ariete increased toward the maximum value (8.00 Ra, Figure 3b), indicating that fresh magma was again replenishing the plumbing system beneath Turrialba. Moreover, the eruptions at Turrialba escalated again during the first half of 2016, peaking in May (Figure 3b), and the level of activity of the volcano could increase further in the following months Petrographic and Chemical Analyses of Products That Erupted During 2014 and 2015 Six samples of ash that erupted during the sequence of explosive events that occurred between October 2014 and May 2015 were collected in the Turrialba summit area. BSE imaging of these samples (Figure 4) revealed that the shape of the particles vary from irregular to angular. Fresh glassy particles are not RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 10

11 Figure 5. Classification diagrams of the main mineralogical phases of Turrialba rocks contained in ashes erupted in 2015, compared with phenocrysts of Turrialba products erupted in : a pyroxene ternary diagram; the olivine forsteritic molar percentage; and a plagioclase ternary diagram. It was not possible to analyze mineral phases in the 2014 samples. abundant and are found mainly in the most-recent deposits (e.g., the sample collected on 6 May 2015). Most of the erupted particles are microlite rich, many are characterized by the presence of euhedral-tosubhedral phenocrysts and microlites of plagioclase and clinopyroxene (commonly zoned), and a few are olivine and magnetite. Samples from the explosions during October December 2014 are strongly dominated by highly altered particles, with sulfides, native S, and signs of hydrothermal alteration (Figures 4a and 4b); some particles from the 2014 explosions display fresh-looking glasses, but particles with signs of plagioclase alteration and molten surfaces have also been observed. De Moor et al. [2016b] recently showed that many of the fresh-looking particles in an ash sample from the eruption on 29 October 2014 show geochemical evidence for cryptic alteration. Ashes from the explosions during March and April 2015 show euhedral and commonly zoned plagioclase and clinopyroxene (Figures 4c and 4d), and an increased proportion of fresh-looking particles. The ashes from 6 May are the least altered of the entire suite (Figures 4e and 4f), not appearing to be modified by recycling or alteration processes such as those from the previous eruptions. Glassy morphologically unaltered particles present spongy outlines, with very well-rounded and clean vesicles, and are completely aphyric or display very small crystals of plagioclases (Figures 4e and 4f). The chemical characterization of glasses and minerals (Tables 2 and 3) was performed on only the freshest particles from each explosive event (similar to particles shown in Figures 4e and 4f). Figure 5 shows the mineral chemistry of the freshest ash particles that erupted at Turrialba volcano in These samples greatly differ in the degree of crystallinity, ranging from completely glassy to 70% of crystals content. Plagioclase is ubiquitous as a euhedral phenocryst phase ranging in composition between An 50 and An 67 ; clinopyroxenes are euhedral-to-subhedral with an augitic-to-diopsidic composition (Wo 4 47,En 43 72, and Fs 8 23 ) often zoned; olivine crystals (Fo ) and oxides (Mt ) occur as phenocrysts and microphenocrysts or microlites. The composition of mineral phases found in the fresh ashes of explosions during 2015 matches that of the eruptive products [Di Piazza, 2014; Di Piazza et al., 2015], though the crystals from 2015 typically show a narrower variability (Figure 5). Based on the SiO 2 and alkali contents (Table 2), microanalysis of glasses revealed that most of the samples are basaltic-andesite to trachy-andesite in composition (Figure 6), with a few anomalous exceptions displaying a more-evolved composition (dacite and rhyolite). The silica content varies from 55 to 65 wt % while total alkalies from 2.5 to 8.0 wt %, and only one sample displaying a rhyolitic composition (SiO 2 77 wt % and alkali 8 wt %; Figure 6). The compositional range of major elements in ashes is broader for October and November 2014 than from December 2014 to May 2015 (Table 2). Particularly, some of the clasts erupted in October and November 2014 have anomalously low Na 2 O(<2%), which are not observed in the following eruptions (Table 2; Figure 6). In particular, the ashes that erupted in April and May explosions show SiO 2 contents ranging between 55 and 60 wt %, alkali compositions between 4 and 7 wt%, and MgO contents between 3 and 5 wt %. We attribute the lower alkali content of the October and RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 11

12 Figure 6. Diagram of the total alkali versus silica content [after Le Bas et al., 1986] comparing the compositions of the samples collected in this study with those of bulk rocks and glasses of the eruption from Di Piazza et al. [2015]. November 2014 particles (Figure 6) to leaching by hydrothermal fluids or interaction with volcanic gases in the plume [Alvarado et al., 2016], which is a process commonly observed in ash particles that erupt from other volcanoes [Dellino et al., 2001; Delmelle et al., 2007; Moune et al., 2007; Islam and Akhtar, 2013, and references therein]. Similarly, de Moor et al. [2016b] found strong evidence for cryptic alteration in the freshlooking particles that erupted on 29 October It is notable that this loss of alkali did not occur in the particles from the samples that erupted later. This evidence is in accordance with the petrographic observations described above, in which we recognized that samples from the explosions occurred during October December 2014 are strongly dominated by highly altered particles, with sulfides, native S, and signs of hydrothermal alteration (Figures 4a and 4b). The absence of geochemical evidence for alteration in the 2015 magmatic particles supports the evolution of eruptive behavior to a more-open conduit system connecting the shallow magma body (or bodies) to the surface. We suggest that the 2014 fresh-looking particles were emplaced shortly (weeks to months) prior to the eruption onset in the shallow subsurface as breccia dikes deposits. They were then exposed to hydrothermal fluids and magmatic gases streaming through the system prior to reaching the surface as entrained material during the eruption on 29 October 2014, whose gas compositions were more hydrothermal in character than the 2015 eruptions [de Moor et al., 2016b]. In contrast, the eruptions on April and May 2015 were purely magmatic (based on their gas chemistry), suggesting that these eruptions formed conduits that directly connected the magmatic system to the surface. This would mean that the fresh-looking particles in the later eruptions did not interact with hydrothermal fluids prior to or during the eruption, thereby preserving their pristine magmatic compositions. The pristine particles in the 2015 ash are thus convincingly juvenile in character from both geochemical and petrographic standpoints. The samples from October 2014 provide ambiguous evidence, in that the fresh-looking particles appear pristine both morphologically and petrographically, whereas the geochemical data presented here (alkali leaching) and the F enrichment reported by de Moor et al. [2016b] show that these samples were actually altered by hydrothermal interactions. Figure 7 shows variation diagrams of selected major elements versus the SiO 2 content. The evolutionary trend observed in the truly pristine juvenile 2015 glasses resembles that of the glasses from the RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 12

13 Figure 7. Variation diagrams of major elements versus SiO 2 for Turrialba rocks. The black dashed lines indicate the chemical modification due to fractional crystallization (same legend as in Figure 6). Oxides are expressed in wt %. Green areas are from electron microprobe analyses of glasses of the eruption [Di Piazza, 2014]. eruptive products. The different degrees of evolution are also related to the various extents of crystallization observed in the ash samples Eruption Trigger and Composition of Involved Magma There was a transient increase in 3 He/ 4 He at the southwest-crater fumaroles between 1999 and 2001, which indicated an enhanced output of magmatic volatiles from a mafic magma deep-seated in the plumbing system of the volcano (Figure 3a). Periodic measurements of 3 He/ 4 He in fumarole gases since 2005 have identified a 6 year-long increasing trend related to the degassing of a mafic and 3 He-rich magma, which was replenishing the Turrialba plumbing system. Figure 3 also shows the dominant frequencies of the continuous seismic record at Turrialba during The trend of this spectral parameter displays periods of high (>5 Hz) and low frequencies (<3 Hz); the former are interpreted to be caused by shear fractures due to shallow pressurization or by an enhanced hydrothermal activity, while the latter indicate resonance in a fluid-filled cavity due to pressure changes induced by magma or gas transport from depth [e.g., Chouet, 1986], or a simple mechanical failure [Eyre et al., 2015]. The dominant frequency was generally moderate to low during 2009 and 2010 (Figure 3a), which appears to be consistent with the magma recharge from depth inferred by the 3 He/ 4 He values. We argue that the replenishments during and (up to 2011) were causally related to the resumption of the eruptive activity in Indeed, shallower levels of the plumbing system have been progressively involved, as testified by the reactivation of the summit area in 2001, with the opening of new degassing vents and the occurrence of significant morphological changes during [Martini et al., 2010; Vaselli et al., 2010]. The increases in the temperature and concentration of magmatic fluids (SO 2, HCl, and HF) at crater fumaroles [Vaselli et al., 2010] and the enhanced degassing of SO 2 from the craters [Conde et al., 2013] clearly indicated the arrival of magmatic fluids at the surface. The gradual replenishment of magma in the shallow portion of the plumbing system provided heat and gas to the overlying, sealed, and vapor-dominated hydrothermal system. This would have caused the pressurization and mobilization of fluids, partial displacement of the hydrothermal system beneath the craters, opening of the system following breakage of the sealing carapace by phreatic explosions, and the consequent release of fluids, ash, lithics, and rocks. From 2011 onward, there was a generally RIZZO ET AL. ERUPTIVE ACTIVITY AT TURRIALBA VOLCANO 13