Lastarria sensu stricto (ss) is an active stratovolcano located in the southern portion of the Central Volcanic Zone of the Andes (CVZA) which extends along the Chile-Argentina border (Figure 1). Lastarria ss is part of the Lastarria Volcanic Complex (LVC), which also includes the Negriales lava field and Espolón Sur volcano (Naranjo, 2010). Negriales volcano (from 400 ± 60 to 116 ± 26 ka) is composed of andesitic-to-dacitic lava flows, coulees, and domes, whereas Espolón Sur volcano (150 ± 50 ka) is constituted of andesitic lava and pyroclastic flow deposits. Lastarria ss is the only active structure of the complex and has evolved through 10 eruptive stages (from 260 ± 20 to 2.46 ± 0.050 ka; Naranjo, 2010). Their volcanic products correspond to andesitic-to-dacitic lava flows and domes, pyroclastic flows and fallout deposits, and debris avalanche deposits (Naranjo, 1992; Naranjo, 2010). Although there are no records of historical eruptive activity, a persistent degassing has been observed since the early 19th century (Casertano, 1963; González-Ferrán, 1995), which is concentrated in four fumarolic fields (Figure 1) located in the northwestern flank (fumarolic field 1), in the eastern and western rim of crater IV (fumarolic fields 2 and 3) ,and inside of crater V (fumarolic field 4) (Aguilera et al., 2012).

Large-scale deformation, as detected by InSAR (Interferometric synthetic-aperture radar) and geodetic measurements, has been observed since 1998 in the Lazufre area, which consists of, from north to south, LVC, Cordón del Azufre and Bayo volcanoes. The zone of uplift has been extending over a ∼45 × 37 km NNE-oriented elliptical area (Pritchard and Simons, 2002; Pritchard and Simons, 2004; Froger et al., 2007; Ruch et al., 2008; Anderssohn et al., 2009; Pearse and Lundgren, 2013; Henderson et al., 2017; Pritchard et al., 2018). The maximum average uplift was observed until 2010 with a rate of 3 cm/yr (Pearse and Lundgren, 2013), whilst between 2011 and 2016 the uplift rate was reduced to <1.5 cm/yr, and should continue to decrease over the next decades, until uplift is anticipated to stop (Henderson et al., 2017). Six alternative models have been proposed to explain this large-scale deformation: 1) magma injection; 2) thermal expansion due to assimilation of wall rock; 3) thermally induced volatile exsolution produced by magma crystallization; 4) lateral expansion due to the intrusion of a sill-like magmatic body (Pritchard and Simons, 2004; Froger et al., 2007; Ruch et al., 2008; Anderssohn et al., 2009; Pearse and Lundgren, 2013); 5) a pulse of heat from the magmatic system into an overlying hydrothermal aquifer without changes in magma reservoir volume (Froger et al., 2007); 6) exsolved volatiles from a melt and temporally trapped at relative deep levels (>10 km) that cause pressurization of a deep reservoir, and consequently ground uplift (Pritchard et al., 2018). The source of the large-scale deformation has been located in a range between 7 and 15 km deep (Pritchard and Simons, 2004; Froger et al., 2007; Ruch et al., 2008; Anderssohn et al., 2009; Pearse and Lundgren, 2013; Henderson et al., 2017; Pritchard et al., 2018). A secondary small-scale deformation affecting an area of 6 km² was observed below Lastarria ss edifice between 2003 and 2005, with a deformation rate of ∼9 mm/yr, which has been related to a pressurized shallow source (hydrothermal aquifer) lying <1,000 m below Lastarria ss edifice (Froger et al., 2007).

Spica et al. (2015), using seismic tomography, further identified a shallow magmatic source located between 3 and 6 km depth (not previously identified), and a shallow hydrothermal source located at 0.4 km depth. Stechern et al. (2017) using petrological approaches (mostly related to Holocene explosive events), identified two magmatic sources, the shallower located at 5–8 km depth, as well as a deeper reservoir between 11 and 15 km depth. Díaz et al. (2015) using resistivity measurements identified three resistivity zones, which are coincident with the shallow hydrothermal system, and the two magmatic reservoirs previously described.

Gas samples from fumarolic emissions were collected using a 50 cm in length and 25 mm in inner diameter titanium tube and pyrex glass dewared pipes connected with metallic clips and Teflon plugs. A 60 mL pre-evacuated glass flask equipped with a Thorion^® valve and partially filled (20 mL) with an alkaline suspension (4 M NaOH and 0.15 M Cd(OH)₂) was assembled in the titanium-glass line (Montegrossi et al., 2001). Incondensable species (He, Ar, N₂, O₂, CH₄, H₂, CO, and light hydrocarbons) remained in the headspace, soluble (water vapor), acidic species (SO₂, HCl, HF) and CO₂ dissolved into solution, whereas H₂S precipitated as CdS.

The chemical compositions of incondensable, dissolved, and precipitated species were determined at the Laboratory of Fluid Geochemistry of the Department of Earth Sciences, University of Florence, Italy. Incondensable species from the headspace were analyzed by gas chromatography (GC) using the Shimadzu 15, 14 A and Thermo Focus equipment. The liquid phase was analyzed by ionic chromatography (IC; Metrohm 761), where the concentration of Cl⁻ (from HCl dissolution), F⁻ (from HF dissolution), and SO₄ ^2- (from SO₂ dissolution) were determined. Acidic titration (AT; Metrohm Basic Titrino) was used to determine the concentration of CO₂ ^3- (from CO₂ dissolution). Solid CdS was dissolved using H₂O₂ to analyze H₂S as SO₄ ^2- by IC (Montegrossi et al., 2001). Analytical error for GC, AT, and IC analysis was <5%.

SO₂ emission rates of Lastarria volcano were measured using a PiCam, a low-cost UV camera, developed at University of Sheffield, United Kingdom (Wilkes et al., 2016; 2017). This instrument consists of two cameras with a field of view (FOV) of 23.1° × 17.3°. Each camera has a bandpass filter with a transmission wavelength on-band and off-band for SO₂ absorption, corresponding to 310 and 330 nm, respectively. The instrument is connected to a laptop via WiFi, with the Picam then controlled through Python 3 code (Aguilera et al., 2020). Prior to each measurement sequence the Picam was calibrated with gas cells with column densities of 100, 467, and 1,989 parts per million by meter (ppm m). Additionally, dark and clear sky corrections were performed, in order to correct the residual noise with a black image, and correction of vignetting through a mask in the clear sky, respectively.

The measurements were carried out on 12, 13 May and 1, 2 November 2018, on 17, 18 January and 21, 22 November 2019 at a distance of 2–5 km from the plume, with a range of geographic altitude from 4,600 to 5,000 m above sea level (m a.s.l. Figure 2; Table 1). On each day of measurement, calibration was performed every hour, with measurements possible for up to 6 h each time.

The sequences were processed using a python-based code in order to determine SO₂ emission rates (Wilkes et al., 2017; Aguilera et al., 2020). Plume velocity was calculated using cross-correlation technique, because the error with this technique is minor, considering the high temporal and spatial resolution of UV camera (McGonigle et al., 2005; Kantzas et al., 2010).

All the measurements presented in this article have uncertainties inherent to field conditions and characteristics of the instrument, which could affect the quality of data. Following Ilanko et al. (2020); Aguilera et al. (2020), uncertainties were determined (Table 2) associated with plume velocity, gas cell concentration and calibration drift, column density across line integration, light dilution. Uncertainties due to plume direction were negligible given near perpendicularity to the plume (Klein et al., 2017), while the plume was near transparent meaning errors from in-plume scattering were also negligible. Despite some uncertainty related to grounding plumes affecting 19% of the data, in this work we have processed UV camera data using integration lines which include the border between ground and plume, and in several positions along the plume, in order to reduce to the minimum that uncertainty. An estimation of this uncertainty has not been carried out.

The plume speed uncertainty depends on the distance between the PiCam and the plume (and consequently the distance between the two integrated columns amounts (ICA)), and the difference in time of the plume movement between one ICA and another during cross-correlation. Considering the distance errors of 200 m, the uncertainty related to the plume speed of our measurements is of ±27%. In the case of calibration drift, we considered variations in the illumination because of the changes in the sun position along 1 day of measurements, selecting and comparing all calibrations carried out and used for the processing of images, in the same site and same day, as per Ilanko et al. (2019). Low uncertainties were calculated for calibration drift, with values ranging from < ±0.46% to ±10%. Uncertainties in column density across line integration were low, with a minimum of ±0.25% and a maximum of ±20%, which were calculated according to Wilkes et al. (2017). In the case of uncertainties related to the light dilution, which is generated by scattering of photons into the FOV of the camera between the plume and the PiCam, we have considered the specific atmospheric conditions in the northern Chile Altiplano, as relatively low atmospheric pressure, very low humidity and scarce aerosol concentrations. According to these conditions; Aguilera et al. (2020) based on Campion et al. (2015) suggest that measurements carried out at distance <3 km, light dilution is negligible, whereas for distances between 3 and 6 km, this will be at least up to +20%. In relation to the above, at Lastarria volcano, we have estimated a negligible light dilution error in the case of 13 May 2018 measurements (distance from the plume <3 km), 1 November 2018, 17, 18 January 2019, and 21 November 2019 an uncertainty variable between +13 and +15% (distance of 4–4.5 km from the plume) and +25% for 2 November 2018 measurements (5 km distance from the plume), the above as a proportion in relation to the indicated by Aguilera et al. (2020). Uncertainty related to gas cell calibration is provided by the manufacturer (Resonance Ltd.), and is ±10% for 100, 467 and 1,989 ppm m cells.

Finally, the total error percentage calculated by the root mean square error (RMS) varies in a range of −19% and +17% (Table 2).

In this work we have used two different Differential Optical Absorption Spectroscopy (DOAS) instruments, the first one developed by The University of Sheffield, UK (Linear Scanning Spectrometer- LSS), whilst the second one was a portable scanning Mini-DOAS station of the NOVAC-type, which was equipped with conical scanning optics (see Galle et al., 2010; Conde et al., 2014 for details on the instrument). Both instruments were used for scanning DOAS measurements.

LSS instrument allowed us to determine the SO₂ column densities, from the light spectra scanned by the spectrometer, whose slit entrance is 50  μ m wide, with the optical bench set up to provide a spectral range of ∼240–400 nm and a spectra resolution of ∼0.65 nm. It uses a Sony ILX511 detector with a 2,048 element linear array CCD and 16- bits ADC. The scanning is performed by a mirror mounted to a stepper motor, which is covered by a 3D-printed scanner head and UV-transmissive curved Perspex window. Python software “SpecScan”, also developed by The University of Sheffield, is used to control the spectrometer and the scanning instrument, and subsequently retrieve column densities of SO₂ from the spectra. The PiSpec measurements were performed on 20, 21, and 22 November 2019, at 4 km N and 3.1 km NW from the summit of Lastarria volcano, 4,558 and 4,566 m a.s.l., respectively (Figure 2; Table 3). Scanning sequences lasted between 31 min, and 4 h and 54 min, being carried out between 9:44 and 14:39 h local time (UTC -3). Integration time used varied in a range of 550–700 ms, whilst the scanning range varied from 90° to 140°. The wind speed data was measured 1) in situ by a handheld anemometer measurements at ground level, and 2) obtained from the plume speed estimated by the UV Camera measurements. The spectra were processed post-acquisition using “SpecScan” for determination of SO₂ emission rates.

The SO₂ emission rate measurements from the NOVAC-type instrument were carried out on 17, 18 January 2019. On each acquisition day, the instrument was deployed for several hours at accessible locations roughly beneath the gas plume and about 4–8 km downwind of the emission sources on Lastarria (Figure 2; Table 3). For data acquisition, the DOAS instrument scans across the sky collecting spectra of incoming scattered UV light at 51 angular steps of 3.6° along a semi-conical surface in order to obtain SO₂ column density profiles perpendicular to the transport direction of the volcanic gas plume. Prior to each scan, exposure times (50–1,000 ms) used for the recording of individual spectra were automatically adjusted by the instrument depending on previously measured UV light intensity. Resulting acquisition times ranged between 5 and 15 min for a complete scan from horizon to horizon.

Scans were evaluated using the DOAS method (Platt and Stutz, 2008) in order to retrieve SO₂ column densities from the sunlight spectra measured across the gas plume. DOAS fits were performed in the 310–325 nm wavelength range of the measurement spectra, additionally including reference absorption spectra of SO₂ (Vandaele et al., 1994) and O₃ (Voigt et al., 2001), the latter being the main interfering gaseous UV absorber relevant for SO₂ retrieval in this spectral range. Furthermore, a Ring reference spectrum (Chance and Spurr, 1997) was included in the DOAS fit to reduce the undesired effects of rotational Raman scattering, and spectral shifts of measured spectra were determined and corrected for by means of comparison with a solar reference spectrum (Chance and Kurucz, 2010). Reference spectra were calibrated to match the spectral resolution of the two spectrometers by means of convolution with the slit function of respective spectrometers.

SO₂ emission rates were calculated using auxiliary information on plume transport height and direction constraining location of the gas plume within the measurement geometry of the scanning DOAS, and information on the speed of gas transport. Plume heights were estimated by visual inspection of plume photographs. Knowing the locations of both emission source and DOAS instrument, the plume transport directions were also determined by means of triangulation using the center of SO₂ mass in measured plume cross-sections and estimated plume heights as reference. For evaluation of the data, plume speeds required to eventually calculate emission rate were approximated using modeled wind speeds obtained from archived data of the Global Forecast System (GFS) provided by National Oceanic and Atmospheric Administration (NOAA), and by cross-correlation of plume features in image sequences obtained from contemporary UV-camera measurements.

H₂O, CO₂, H₂S, and HCl emission rates were calculated using Eq. 1 (López et al., 2013) combining determined daily average SO₂ emission rate ( F S O 2 ), measured by ground-based remote techniques, and the concentrations of fumarolic compounds obtained from direct sampling technique. F v o l a t i l e s = F S O 2 M v o l a t i l e s × X v o l a t i l e s M S O 2 × X S O 2 (1) where M v o l a t i l e s is the molecular weight of each major species, in the case of SO₂ ( M S O 2 ) is 64 gr/mol, X v o l a t i l e s is the average mole percent of each volatile species present in the fumarolic gases. Summing up the emission rates of all major gaseous compounds finally yielded the total volatile inventory of Lastarria.

Here we present a large database of 63 samples, 29 corresponding to the period May 2006–April 2009 (Aguilera et al., 2012), 1 sample from June 2010 (this work), 4 samples of November 2014 (2 from Lopez et al. (2018) and 2 this work), and 29 samples between May 2011 and November 2019 (this work). Altitude (m a.s.l.), temperature (°C) and chemical composition (mol%) of fumarolic gas discharge at Lastarria volcano, are presented in the Table 4. Fumaroles for the period May 2006-November 2019 presented outlet gas temperatures between 80.1°C and 408°C. Gas composition can be described in two different periods:

1. May 2006-November 2012: Water vapor and CO₂ were the predominant compounds whose concentrations were 81–91 and 7.2–17 mol%, respectively. SO₂ concentrations were within the range of 0.01 and 0.68 mol%, while H₂S concentrations varied from 0.01 to 0.3 mol%. Significant amounts of HCl (0.001–0.12 mol%), HF (0.0001–0.05 mol%), N₂ (0.03–0.51 mol%) and H₂ (0.0005–0.24 mol%) were also found in fumarolic gases. Minor amounts of carbon-bearing compounds such as CH₄ (≤0.001 mol%) and CO (≤0.001 mol%) were found, whereas noble gases (He and Ar) have values ≤0.0002 mol%, and concentrations of O₂ were below 0.01 mol%.

The results of the UV camera measurements were determined for individual and/or combined fumarolic fields, and corresponding to SO₂ emission rates ±1 standard deviation, the last representing the variability of the data during the measurement process, are presented in the Table 5, and summarized as follows:

On 13 May 2018 we measured SO₂ sequences with a total of 1,070 images obtained, where fumarolic fields 1 and 2 were measured individually, obtaining rates of 113 ± 25 and 76 ± 15 t/d, respectively, whereas the rates by combination of the emissions from the fumarolic fields 3 and 4 was 93 ± 21 t/d. The overall average emission rate considering the four fumarolic fields for 13 May 2018 was 282 ± 20 t/d.

SO₂ measurements on 1 November 2018 averaged 786 ± 40 t/d, which is an increase in comparison with SO₂ measurements carried out in May 2018, for all fumarole fields combined. Considerable increases were observed in field 1 (234 ± 54 t/d), and the plume produced by combination of fumaroles from fumarolic fields 1, 3, and 4 (500 ± 94 t/d).

On 17 January 2019 the overall emission rate measured was 726 ± 74 t/d, similar to those measured on 1 November 2018, and with an emission rate for fumarolic fields 1, 2, 3, and 4 of 298 ± 105, 282 ± 69, 73 ± 22, and 72 t/d, respectively. On 18 January 2019, the plume moved in an atypical direction (west), and only the fumarolic field 1 could be measured, obtaining a minimum emission rate of 137 ± 38 t/d.

The overall SO₂ emission rate estimated for 21 November 2019 (1 sequence and 890 images) was 359 ± 29 t/d, with a low SO₂ emission rate in the fumarolic field 1 (24 ± 23 t/d) in comparison to the measurements of November 2018 and January 2019.

If we consider the root mean square error (RMS) expressed in percentage (Table 2), the final emission rates for the same dates, and expressed as a range, where minimum and maximum values correspond to the overall average emission rate plus negative and positive maximum RMS error, the SO₂ emission rates correspond to 240–319 t/d (13 May 2018), 645–935 t/d (1 November 2018), 646–813 t/d (17 January 2019), and 291–420 t/d (21 November 2019).

The results of the measurements carried out with mini-DOAS in bulk plume are presented in the Table 3. On 17 January 2019, using wind speed values of 3.9 and 4.7 m/s, both obtained from GDAS1 at 500 hPa (corresponding at ∼5,600 m a.s.l.), the total SO₂ emission rate obtained was 172 ± 78 t/d. Recalculating the total SO₂ emission rate using the plume speed average values obtained from simultaneous measurements with the PiCam (9.8 m/s), the total SO₂ emission rate increased to 418 ± 217 t/d. On 18 January 2019, measurements were carried out at two different locations. At point 7, using a wind speed of 6.1 m/s (GDAS1 at 400 hPa), the total SO₂ emission rate obtained was 282 ± 64 t/d. At the point 10, using a wind speed 7.4 m/s (GDAS1 at 400 hPa), the total SO₂ emission rate obtained was 626 ± 85 t/d. The overall SO₂ emission rate for both days was 376 ± 111 t/d.

On 20–22 November 2019, we used average ground level wind speed (by use of an anemometer) of 7.3, 10.8, and 3.5 m/s, respectively, because the mean wind speeds obtained by the GFS model (GDAS1 at 500 hPa) were on average 40% lower than those observed in the field. The SO₂ emission rate (average ± 1SD) was 506 ± 241, 1,416 ± 568, and 467 ± 235 t/d, respectively. Whereas using the PiCam plume speed (11.7 m/s), the recalculated measurement increased to 1,534 ± 485 t/d on 21 November 2019 (the only day with a reliable plume speed obtained by PiCam). The overall SO₂ emission rate for 3 days was 981 ± 382 t/d.

The results of emission rates of gases for the years 2014, 2018, and 2019 are presented in Table 6, where only high temperature fumaroles were considered. Here we present the results for each period considering the average of direct sampling data obtained from the different fumaroles, and using the variations of the SO₂ emission rates based on the RMS error presented in the Section 4.2. For November 2014 we estimated the emission rates of main species using the average SO₂ emission rate measured by Lopez et al. (2018) corresponding to 604 t/d, and we used the ±50% error reported for remote measurements, corresponding to a range between 302 and 906 t/d. Consequently, for November 2014, the emission rates of H₂O, CO₂, H₂S, and HCl ranged 7,208–21,626, 747–2,241, 100–299, and 46–137 t/d. In the following years, the H₂O, H₂S, and HCl decrease, reaching the lowest ranges during January 2019, corresponding to 3,351–4,217, 42–52, and 48–60 t/d, whereas CO₂ increases to its highest range in the same period (2,158–2,716 t/d). During November 2019, the emission rates of H₂O and H₂S increase to values similar to the November 2018 (5,823–8,405 and 90–130 t/d, respectively), whereas CO₂ and HCl reached the lowest flux along the whole revised period (746–1,077 and 15–22 t/d, respectively).

The chemical composition of the fumarolic discharges from Lastarria volcano, which were investigated by Aguilera et al. (2012), covering a period between May 2006 and June 2009 (Table 4), showed that fluids have two origins: 1) a deep magmatic and 2) a shallow hydrothermal source. Along the whole range of fumarolic discharge temperatures (80°C–408°C) acid species such as SO₂, HCl, and HF were present, indicating a significant fluid contribution from magma degassing. Additionally, water isotopic composition (δD and δ¹⁸O) showed a pristine magmatic origin, poorly affected by meteoric contributions (Aguilera et al., 2012). However, concentrations of species like SO₂ and HF are lower than expected for magmatically dominated arc volcanoes (Fischer et al., 1998; Lewicki et al., 2000), especially if only fumaroles with temperatures above the saturated steam vapor at the local atmospheric pressure (120°C–408°C; Table 4) are considered. The latter suggests that secondary processes modified the original composition of magmatic volatiles, which include gas-rock interaction and salt deposition. The hydrothermal reservoir corresponds to a shallow boiling aquifer producing gases such as H₂S and CH₄. Magmatic volatiles are cooled and scrubbed at shallower depth, decreasing the concentration of highly soluble gases and the temperatures down to levels where condensation can occur. According to the model presented by Aguilera et al. (2012), the mixed distribution of low (<96°C) and high (>120°C) temperatures is explained by the occurrence of a discontinuous shallow aquifer, where water vaporization is catalyzed by the rising of hot magmatic volatiles. Consequently, for the period May 2006-June 2009, Lastarria can be considered as a volcano dominated by hydrothermal discharges (Figure 3), with some contributions of magmatic volatiles. According to our new data presented here (Table 4), the gas composition between May 2011 and 11 November 2012 remained stable without major changes and was dominated by hydrothermal fluids, similar to observations of the previous years (Figure 3). A major change in gas composition was detected only 2 weeks later by Tamburello et al. (2014) who used a combination of MultiGAS, filter packs, and ground-based remote measurements of SO₂ emission rates on 27–29 November 2012. Their data in comparison to previous measurements showed a clear increase of SO₂ and HCl contents, and a decrease of CO₂, which is considered compatible with an increasing magmatic signature. Tamburello et al. (2014) explained these changes by (1) dissimilar sampling conditions, where MultiGAS were carried out in the “bulk” plume, whereas previous direct sampling was concentrated in few fumaroles which could be more affected by secondary processes, therefore not representative of the bulk emissions; or (2) a consequence of a real evolution of the volcanic system from hydrothermal-to-more magmatic degassing. Lopez et al. (2018) confirmed the changes in the gas composition by combination of direct gas sampling, MultiGAS, filter pack, ground-based remote, and CO₂ diffuse measurements, describing an evolution of fluids to a stronger magmatic signature, as the increasing SO₂ and HCl concentrations suggest (Figure 3; Table 4). Unpublished data from February 2017 to November 2018 (Table 4) showed similar concentrations to the November 2014 data, confirming the increasing of SO₂ and HCl, although an increase of N₂ by one order of magnitude was also observed. In January 2019, new changes were observed in the gas compositions (Table 4), H₂O decreased to 68.2–70.3 mol%, whereas CO₂, HF and H₂ increased one order of magnitude in comparison with 2014–2018 data, corresponding to the highest concentrations throughout the studied period. Between April and November 2019 gas concentrations returned to the values observed during February 2017 and November 2018 (Table 4).

According to the evolution of gas composition previously described a noticeable evolution from hydrothermal-dominated to magmatic-dominated volatiles is observed in Lastarria volcano since May 2006 up to November 2019 (Figure 3). Despite the presence of magmatic gases like SO₂ and HF, hydrothermal compounds predominated between May 2006 and early November 2012, which have been attributed to secondary water-rock interaction processes such as interaction of magmatic gases with rocks and/or salt deposition/precipitation with decreasing temperatures, promoting the scrubbing of highly soluble species such as SO₂, HCl, and HF (Aguilera et al., 2012; Lopez et al., 2018). The magmatic-dominated compositions observed since late November 2012, are compatible with a shallow degassing process, due to the decreasing of CO₂, and the increasing of SO₂ and HCl concentrations (Figure 3). The increase of the more soluble species was likely caused by a more restricted scrubbing in the shallow hydrothermal system. The decreasing of the scrubbing process could be attributed to the increased input of magmatic volatiles, followed by the acidification and partial consumption of the hydrothermal system. Increase of SO₂ emission rate has been observed in the period between November 2018 and January 2019, which will be discussed in the following sections. Acidification of the hydrothermal system explains the decrease of H₂O/SO₂ and H₂O/HCl, which started in late November 2012 (Figure 4). In January 2019 the lowest concentrations of water vapor were observed (68.2–70.3 mol%; Table 4), which could be related to consumption of the hydrothermal system caused by the enhanced hot magmatic fluid inputs that led to an increase of the reservoir ebullition. The input of magmatic volatiles caused an increase in the temperature of the hydrothermal reservoir increasing the concentration of the temperature dependent gases, such as H₂ and CO (Table 4). This process was concordant with the first ever witnessed active sulfur flows at this volcano. In April and November 2019, the returning of gas composition to values close February 2017-November 2018 seems to represent a subtle recovery of the previously depleted hydrothermal system.

Similar to fumarolic gas composition, changes have been observed in the SO₂ emission rates in the period between November 2012 and November 2019 (Figure 5). The first SO₂ emission rates reported for Lastarria volcano were acquired using mini-DOAS from the summit of Lastarria (Tamburello et al., 2014). The daily rates (expressed as average ±1 standard deviation) were 1,917 ± 607, 473 ± 188, and 433 ± 314 t/d for 27, 28, and 29 November 2012, with an overall value for the 3 days of 884 ± 779 t/d and a much lower median of merely 538 t/d, the latter reflecting the outlier nature of the much higher emission rates obtained for 27 November 2012. SO₂ emission rates of 27 November 2012 generally were a factor ∼5 larger than those of 28 and 29 November 2017, and temporarily peaked at 36.14 kg/sec (which would correspond to 3,123 t/d). These observations were carried out in stormy and gusty conditions (wind speeds of up to 17 m/sec) during the passage of a low-pressure frontal system causing air pressure to drop by 10 hPa in 12 h (9:00–21:00) which may have enforced gas release on that day, and thus these emission rates may be considered as exceptionally high and likely not representative in the long-term. As to not overrate the measurements of November 27, we consider the median SO₂ emission rate of 538 t/d of the November 2012 campaign as representative of this period. Lopez et al. (2018) present data on 22 November 2014 based on UV camera and DOAS measurements, with an overall SO₂ emission rate of 604 ± 296 t/d. Although the SO₂ emission rate in November 2014 probably was slightly higher than November 2012, the high standard deviation and high error reported (±50%; Lopez et al., 2018) make it unclear whether this is related to increasing of degassing caused by rising of deep magmatic volatiles. Our UV camera data (Table 5) show clearly an increasing of SO₂ emission rate on 1 November 2018 and 17 January 2019 (786 ± 40 and 726 ± 74 t/d) (Figure 6) in comparison with May 2018 (282 ± 20 t/d), decreasing subsequently on 21 November 2019 (359 ± 29 t/d) (Figures 5, 6). These data show high stability (low standard deviation) and lower errors (−19 and +17%; Table 2) compared with Tamburello et al. (2014); Lopez et al. (2018) data, which can be considered as highly reliable. This behavior is compatible with the changes observed in the fumarolic gas composition, where the most magmatic signature of gas is coincident with the highest SO₂ emission rates in November 2018 and January 2019. Consequently, the increase of SO₂ emission rates could be explained by rising of magmatic volatiles, which caused a decrease of scrubbing, as argued in the previous section. The decrease of SO₂ emission rate in November 2019 is also coincident with the returning of the fumarolic gas composition to values of 2017. This decrease led to an increase of magmatic gas scrubbing and, consequently, a partial recovery of the hydrothermal system, as was previously discussed. We have not considered in this analysis the DOAS data from January and November 2019, because the scanning was done in a geometry not able to cover the total of the plume during January 2019, which produced an underestimation of the SO₂ emission rate. In the case of measurements carried out in November 2019, the overestimation of the emission rates is also related to the scanning geometry, where semi-horizontal scanning covered an “excess” of plume.

We have calculated the volatile budgets combining the emission rates of main species (H₂O, CO₂, SO₂, H₂S, and HCl; Table 6) for four different periods (November 2014, November 2018, January and November 2019), considering the species measured by direct sampling (Table 4). We have not combined our direct sampling data with SO₂ emission rates from Tamburello et al. (2014) due to large differences between our gas composition (hydrothermal-dominated) and that reported during late November 2012, which show magmatic-dominated composition. Consequently, here we assume that the enhanced SO₂ emission rates which were measured during late November 2012 were related to the magmatic signature. The volatile budget of November 2014 averaged at 16,804 t/d (with a minimum of 8,402 t/d and maximum 25,206 t/d), which is close to that calculated by Lopez et al. (2018) for the same date (average 12,401 t/d), and the budget calculated for November 2012 (13,480 t/d; Tamburello et al., 2014). Lopez et al. (2018) present large differences between minimum and maximum emission rates in their work—3,579 and 38,311 t/d, respectively—whereas in our case the differences are more restricted, with a minimum flux of 8,402 t/d and a maximum of 25,206 t/d. Lopez et al. (2018) calculated the volatile fluxes including the variations in the SO₂ emission rates and the changes in the gas compositions, whereas in our case, we used the average of the gas composition and the SO₂ emission rate average and their related RMS error. We consider our data more reliable due the very low variability of the SO₂ emission rates, and because our gas composition was standardized by the same sampling and analyzing methods, in the last case even using the same laboratory. On the contrary, Lopez et al. (2018) combined data from different direct sampling methods applying analyzing methods from different laboratories. The volatile budget decreased in November 2018 and January 2019 to 9,773 and 7,017 t/d, respectively, and in November 2019 it further increased subtly to 8,594 t/d. Here we have considered only our UV camera data, due to the under and overestimation observed for our DOAS data in January and November 2019, and discussed in the Section 5.2. The same pattern was followed by H₂O and H₂S, which is compatible with a progressive depletion of the hydrothermal system. On the contrary, CO₂ emission rates increase up to January 2019 when their highest values were reached, descending again in November 2019. The progressive increase of CO₂ emission rates could be related to the incorporation of deep magmatic volatiles, similar to the trend in direct sampling data.

In the regional context of the CVZA, Ubinas and Sabancaya volcanoes presented average total volatile fluxes of 23,898 and 5,469 t/d respectively, whereas average CO₂ and SO₂ emission rates ranged 1,222–1,366 and 988–1,325 t/d, respectively (both in November 2015; Moussallam et al., 2017). In the case of Ubinas volcano, the measurements were carried out during an eruptive period, and in the case of Sabancaya measurements were a few months before the beginning of a new eruptive cycle. According to Tamburello et al. (2014), Lascar volcano had a total flux of 6,517 t/d, while CO₂ and SO₂ emission rates were 534 and 554 t/d, measurements carried out during a passive degassing period without eruptive activity involved. Lastarria volcano average total fluxes varied in a range of 7,017–16,804 t/d, CO₂ emission rates from 921 to 2,425 t/d, and SO₂ emission rates between 359 and 786 t/d, making Lastarria volcano one of the most strongly degassing volcanoes of the CVZA, especially in the case of CO₂.