The fumarolic CO₂ output from Pico do Fogo Volcano (Cape Verde)

Together with tectonic degassing, subaerial volcanism is the primary outgassing mechanism of mantle-derived CO₂ to the atmosphere (Werneret alii, 2019; Fischeret alii, 2019). Over geological time, tectonic and volcanic degassing have been the primary mechanisms for carbon exchange in and out our planet (Dasgupta and Hirschmann, 2010; Dasgupta, 2013; Wonget alii, 2019), ultimately playing a control role on pre-industrial atmospheric CO₂ levels and the climate (Van der Meeret alii, 2014; Bruneet alii, 2017). Although attempts to estimate the global volcanic CO₂ output started early back in the 1990s (e.g., Gerlach, 1991), substantial budget refinements have only recently arisen from the 8-years (2011-2019) DECADE (Deep Earth Carbon Degassing; https://deepcarboncycle.org/about-decade) research program of the Deep Carbon Observatory (https://deepcarbon.net/project/decade#Overview) (Fischer, 2013; Fischeret alii, 2019).

One key result of DECADE-funded research has been the recognition that the global CO₂ output from subaerial volcanism is predominantly sourced from a relatively small number of strongly degassing volcanoes. Aiuppaet alii, (2019) showed that the top 91 SO₂ volcanic emitters in 2005-2015 (those systematically detected from space; Carnet alii, 2017) produce a cumulative CO₂ release of ~39 Tg/yr, nearly half of which (~19 Tg CO₂/yr) is produced by only 7 top-degassing volcanoes. It has also been found, however, that a non-trivial CO₂ output is additionally sustained by fumarolic degassing (Fischeret alii, 2019; Wernerat alii., 2019) and groundwater transport (Taran, 2009; Taran and Kalacheva, 2019) at hydrothermal volcanoes in quiescent stage. These low-temperature (hydrothermal) fumarolic emissions typically release CO₂ in the absence of easily detectable (ultraviolet - UV - spectroscopy) spectroscopy) SO₂, implying that traditional “indirect” CO₂ flux quantification using the volcanic gas CO₂/SO₂ ratio proxy in tandem with remotely sensed SO₂ fluxes (e.g. Werneret alii, 2019) cannot be employed; more challenging airborne (Werneret alii, 2009) or ground-based (Pedoneet alii, 2014; Aiuppaet alii, 2015; Queisseret alii, 2016) “direct” CO₂ flux measurements are required instead. These technical limitations have prevented us from establishing a robust catalogue for fumarolic CO₂ outputs, as <50 of the several hundred degassing volcanoes in “hydrothermal-stage” worldwide have been measured for their CO₂ flux (Werneret alii, 2019). As a consequence, the extrapolated current inventories for the global fumarolic hydrothermal CO₂ flux (from 15 to 35 Tg CO₂/yr; Fischeret alii, 2019; Werneret alii, 2019) still involve very large uncertainties. In addition, most of the available information is for low-temperature arc volcanic gases, while much less is known for the fumarolic CO₂ output for non-arc settings (divergent, intra-plate or continental rift; e.g., Ilyinskayaet alii, 2015, 2018).

Pico do Fogo, in the Cape Verde Archipelago, is part of the Macaronesia region, an area of the Atlantic Ocean off the western coasts of Africa, also including the archipelagos of the Azores, Madeira and Canary (Fig. 1). This 2829 m a.s.l high strato-volcano (Fig. 2a), located on the island of Fogo, has been the most frequently erupting volcanic centre of the Macaronesia region in the last 500 years (Ribeiro, 1960). All historical eruptions occurred on its upper flanks or at its summit crater. Between eruptions, the summit crater of Pico do Fogo hosts a persistent fumarolic field (Fig. 2b-e), with several gas vents ranging in temperature from boiling to >200°C (Dioniset alii, 2014; Meliánet alii, 2015). The CO₂ output sustained by diffuse degassing across the crater floor was estimated in the range 147±35 (in 2009) to 219±36 t/d (in 2010) (Dioniset alii, 2014, 2015), but no comparable data yet exists for the fumarolic CO₂ output itself.

Here we fill this gap of knowledge by presenting the very first results for the fumarolic output of CO₂ and other volatiles from Pico do Fogo. These results were obtained from a gas survey on February 5, 2019, during which we combined real-time in-situ measurement of the crater gas compositions (Multi-GAS), direct sampling of the hottest fumarole, and near-vent remote sensing of the SO₂ flux with an UV Camera. Our new data set contributes to improved quantification and understanding of Fogo’s quiescent degassing during the multi-decadal phases separating eruptions, and offers an interesting comparison with the gas output measured during the recent 2014-2015 eruption (Hernándezet alii, 2015). More broadly, our results for Pico do Fogo add a novel piece of information to the still fragmentary data base for fumarolic CO₂ emissions from global volcanoes in hydrothermal stage.

The Cape Verde Archipelago, extending between 15 and 17°N latitude 500 km to the west of Senegal, is composed of 10 main islands that are the emerged portions of a high oceanic plateau (2 km above the sea floor). Fogo Island is located at the south-western edge of this system (Fig. 1). The Cape Verde oceanic Rise, the world’s largest geoid and bathymetric seafloor anomaly (Courtney & White, 1986), has been interpreted as due to a hot-spot mantle swell centred north-east of the Sal Island (Crough, 1978, 1982; Holmet alii, 2008). The presence of an active mantle plume beneath the northern part of Cape Verde at least has been suggested by some authors based on seismic imaging (Montelliet alii, 2006; Liu & Zhao, 2014; Sakiet alii, 2015). A mantle plume contribution is also consistent with high primordial ³He (³He/⁴He ratios up to 12.3-15.7 Ra) in volcanics from São Vicente and São Nicolau islands (Christensenet alii, 2001; Doucelanceet alii, 2003; Mataet alii, 2010; Mourãoet alii, 2012). However, a plume origin for Macaronesian volcanism is still matter of debate (Bonatti, 1990; Asimovet alii, 2004), and the role of decompressional melting (Métrichet alii, 2014) favoured by extensional lithospheric discontinuities (Marqueset alii, 2013) has received increased attention recently. Volcanism at the Cape Verde Islands is thought to have started 24–22 Ma ago on the northeastern islands, followed by a more recent westward migration of volcanic activity (both in the northern and southern branches of the archipelago) during the Pliocene-Pleistocene (Holmet alii, 2008). Erupted products spread a large compositional range but mafic, silica-undersaturated lavas (basanites, tephrites, and nephelinites) prevail (Gerlachet alii, 1988; Davieset alii, 1989; Holmet alii, 2006), eventually associated with rarer carbonatites (Kogarkoet alii, 1992; Hoernleet alii, 2002). Trace-element and isotope geochemistry of the erupted volcanics are extremely heterogeneous, with significant differences between the northern and southern islands, implying the probable involvement of several distinct mantle sources: a lower mantle plume containing both mixed HIMU (High μ = ²³⁸U/²⁰⁴Pb at zero age) and EM1 (Enriched Mantle 1) end-members, possibly a 1.6-Ga recycled oceanic crust, plus the depleted upper mantle (northern islands) and the subcontinental lithospheric mantle (southern islands) (Gerlachet alii, 1988; Davieset alii, 1989; Holmet alii, 2006; Christensenet alii, 2001; Doucelanceet alii, 2003; Milletet alii, 2008). The actual relative proportions of each of these sources are still debated however.

Fogo Island (Fig. 1b), formed during the last 3-4.5 Ma, has been the single site of historical volcanic activity (27 reported eruptions) since the discovery of the archipelago in the XV^th century. The dominant structure of the island is Monte Amarelo Volcano whose summit was truncated by three massive flank collapses between ca. 60 and 43 ka (Fig. 1b) (Dayet alii, 1999; 2000; Marqueset alii, 2020). The post-collapse (62 ka to present) activity has been primarily concentrated within the Chã das Caldeiras depression (Fig. 1b), leading to progressive infilling of the collapse scar and the formation of the Pico do Fogo cone. The cone itself (Fig. 2a) has remained the primary eruptive centre until 1785 (Ribeiro, 1960), when fissure-fed effusive eruptions became concentrated along the flanks of the volcano, occurring at an average frequency of one every ~50 years. The most recent eruptions happened in 1951 (Hildneret alii, 2012), 1995 (Hildneret alii, 2011) and 2014-2015 (Carracedoet alii, 2015; Cappelloet alii, 2016; Richteret alii, 2016; Mataet alii, 2017). Eruptive products of the Amarelo-Fogo volcanic complex are primarily alkali-rich tephritic to basanitic lavas (with rarer foidites and more evolved phonolites). They are thought to ascend from a 16–28 km deep magma storage zone, emplaced in the underlying lithospheric mantle (Gerlachet alii, 1988; Doucelanceet alii, 2003; Hildneret alii, 2011, 2012; Mataet alii, 2017).

On February 5, 2019 we realized extensive field investigations and measurements of the summit crater fumarolic emissions of Pico de Fogo volcano (Fig. 2a-e). We used a portable Multi-component Gas Analyser System (Multi-GAS) to analyse in real-time the fumaroles’ compositions during walking traverses across the fumarolic field (see the track shown in Figure 2e). The walking traverse mode, first used on Vulcano Island, in southern Italy (Aiuppaet alii, 2005a), is ideal to explore the chemical heterogeneity of a fumarolic field as a high number of fumarolic vents can sequentially be analysed while slowly moving along the path. During the traverse, the Multi-GAS continuously acquired data at 0.5 Hz, and its position was synchronously geo-localized with an embedded GPS. In addition to areas of diffuse soil degassing, 17 main fumarolic vents, showing the strongest emissions, were identified during the traverse (Fig. 2e). Gas composition at each of these vents was determined (Tab. 1) by keeping the MultiGAS inlet at a constant position (and for a few minutes) at about ~50 cm height above the fumarolic vent. Our Multi-GAS instrument comprised the following sensor combination (e.g., Aiuppaet alii, 2016): a Gascard EDI030105NG infra-red spectrometer for CO₂ (Edinburgh Instruments; range: 0-30,000 ppmv); 3 electrochemical sensors for SO₂ (T3ST/F-TD2G-1A), H₂S (T3H-TC4E-1A) and H₂ (T3HYT-TE1G-1A), all from City Technology; and a KVM3/5 Galltec-Mela temperature (T) and relative humidity (Rh) sensor. H₂O concentration in the fumarolic gases was calculated from co-acquired T, Rh and pressure readings using the Arden Buck equation (see Aiuppaet alii, 2016). Reading from the H₂S sensor were corrected for 14% cross-sensitivity to SO₂. Gas ratios in each of the main fumaroles (Tab. 1) were derived from scatter plots of the gas concentrations using the Ratiocalc software (Tamburello, 2015). Uncertainties in all derived ratios are <15%, except for H₂O/H₂S (≤ 25%).

The fumarole 15, displaying the highest emission temperature (T = 315°C), was sampled for dry gases only by inserting a titanium tube 50 cm-long into the vent. This tube was connected to both a quartz line equipped with a condenser in order to remove water vapour and a three-way valve with a syringe allowing to force gas flow into the line. Three dry gas samples were stored in glass bottles equipped of two stopcocks and then moved to the INGV laboratory in Palermo for chemical analysis. Concentrations of He, H₂, O₂, N₂, CO, CH₄, CO₂ and H₂S were determined using a gas chromatograph (Clarus 500, Perkin Elmer) equipped with a 3.5-m column (Carboxen 1000) and a double detector (hot-wire detector and flame ionization detector [FID]). SO₂ was not measurable with this sampling/analytical setup. Analytical errors were <3%. The results are reported in Tab. 2.

Simultaneously to our Multi-GAS traverse, we also operated a portable dual UV camera system for measuring the volcanic SO₂ flux. The camera system registered at 0.5 Hz for ~100 minutes from a fixed position on the inner crater terrace’s rim, deep inside the summit crater (see Figs. 2b, 2e). The system used two co-aligned cameras (JAI CM-140GE-UV), both fitted with optical lenses of 45° Field of View (FoV), and mounting two different band-pass optical filters with Full Width at Half Maximum (FWHM) of 10 nm and central wavelengths of 310 and 330 nm, respectively. The filters were applied in front of the cameras so to achieve differential UV absorption in the SO₂ band (Kantzaset alii, 2009; Kernet alii, 2010; Delle Donneet alii, 2019). The system, housed in a peli case and powered by a 12V LiPo battery, was mounted on a tripod and rotated to look upward to image the crater’s inner northern slope (where the fumarolic field is located) and a portion of the background sky (Figs. 2b, 2d). Data acquisition was commanded via PC using the Vulcamera software (Tamburelloet alii 2011). The acquired images (520x676 pixels at 10-bit resolution) were post-processed using standard techniques (Kantzaset alii, 2009; Tamburelloet alii, 2011, 2012): sets of co-acquired images were combined into absorbance images and were then converted into SO₂ slant column amount (SCA) images by successively using three different calibration cells. Finally, we derived an Integrated Column Amount (ICA) time-series by integrating the SCA along the cross-section shown in Fig. 2b and then the SO₂ flux by multiplying the ICA with the plume speed. The plume speed (1.9±0.6 m/s) was obtained by processing image sequences acquired at 0.2 Hz using a LifeCam Cinema HD (Microsoft) USB visible camera, integrated in the UV Camera system. Processing involved quantifying the rising speeds of ~50 individual gas puffs of well-resolved structure, moving upward from the fumarolic field toward the crater edge (Fig. 2d).

Finally, from the same position as the UV camera, we used a portable handheld thermal camera (model FLIR E5) in order to acquire a thermal map of the fumarolic field (see Fig. 2b). This map allowed us to verify that the hottest degassing areas were in large part covered by the Multi-GAS traverse. Temperatures of fumaroles 5 and 14-15, the hottest vents in the field (Fig. 2b), were also directly measured in situ with a portable thermocouple.

As a whole, during the ~74-minute duration of our Multi-GAS traverse, we obtained 4446 simultaneous measurements of H₂O, CO₂, SO₂, H₂S and H₂ concentrations in Fogo gas emissions (one analysis every 2 seconds). The entire dataset is illustrated in Figure 3 where the gas concentrations in the near-vent fumarolic plumes are displayed as scatter plots. The concentrations of H₂O, CO₂ and H₂ were corrected for the respective air background values of ~12,000, ~600 and ~0.5 ppmv measured upwind (outside) the fumarolic field (Fig. 2e). The high background CO₂ concentration compared to “normal” atmosphere (~400 ppmv) is explained by the high diffuse soil CO₂ emission through the inner crater floor (Dioniset alii, 2014, 2015).

The absolute gas concentrations measured along our traverse display quite large variations (Fig. 3), indicating chemical heterogeneity in the fumarolic field emissions. This is especially evident in the SO₂ vs. H₂S scatter plot (Fig. 3). Otherwise, one observes broad co-variations among most gas species, even though with some spread. The maximum peak values reached ~23,000 (H₂O), ~20,000 (CO₂), 118 (H₂S), 62 (SO₂) and 30 (H₂) ppmv.

The molar compositions of fumarolic gases from the 17 individualized vents (Tab. 1) confirm this spatial heterogeneity. Each fumarole actually exhibited stable, well-resolved composition (see the fumarole 15 example in Figure 3). Instead, the SO₂/H₂S ratios in all fumaroles span more than three orders of magnitude, from 0.001 to 1.5 (Tab. 1 and Fig. 3). The H₂O/H₂S, CO₂/H₂S, and H₂/H₂S also varied considerably within the fumarolic field, with respective ranges of 98-480, 108-240 and 0.05-0.24 (Tab. 1 and Fig. 3).

Table 2 shows the chemistry of dry gases collected from the hottest (315°C) F15 fumarole (Fig. 2d, e). CO₂ is the overwhelming component (up to 97%), followed by H₂S (around 1%), H₂ (952-979 ppm), CO (15-17 ppm) and CH₄ (around 1-2 ppm). N₂ and O₂ contents reflect air contamination of the samples, with minimum values of 0.5% and 0.1%. The concentration of helium is around 8 ppm in our less contaminated sample. Whatever the degree of air contamination, our samples from the hottest F15 fumarole reveal CO₂/H₂S (94-107) and H₂/H₂S (0.09-0.10) ratios (Tab. 2) that are very comparable to the corresponding ratios determined with Multi-GAS.

The SO₂/H₂S ratio is a commonly used marker to distinguish the magmatic (SO₂-rich) vs. hydrothermal (H₂S-rich) nature of volcanic gas (e.g. Aiuppaet alii, 2005b). Figure 4 shows that Pico do Fogo fumaroles define a nearly continuous trend from two end-members:

• a magmatic end-member, represented by the hottest gas from fumaroles 14-15 (T = 315-316 °C), characterized by H₂O/CO₂ of ~ 2, CO₂/S_t of ~ 100, high SO₂ (~0.2 mol. %) and relatively low H₂S, and oxidised (redox conditions of about 1 log unit above the Nickel-Nickel Oxide buffer at ~500°C, estimated from the measured SO₂/H₂S ~ 0.9-1.4 and H₂/H₂O ~ 0.0004; see methodology in AiuppAet alii, 2011); and,

• a hydrothermal end-member, represented by fumaroles 3-8, that is H₂S-dominated (~0.35-0.43 mol. %; SO₂/H₂S of ~ 0.01-0.2), relatively richer in CO₂ (CO₂/S_t > 130 and H₂O/CO₂ < 1) and more reduced (H₂/H₂O > 0.0015) (corresponding to redox conditions close to the FeO-FeO1.5 buffer; Giggenbach, 1987).

The red star in Figures 4a-d represents the spatially integrated composition of Pico do Fogoś fumarolic emission, calculated as the arithmetic mean of compositions of the 17 main fumaroles. It is characterized by the following ratios, normalized to H₂S: SO₂/H₂S = 0.3±0.4, H₂O/H₂S = 299±109, CO₂/H₂S = 153±33 and H₂/H₂S = 0.2±0.04 (Tab. 1). The mean SO₂/H₂S ratio of ~0.3 is not much different from the SO₂/H₂S ratio of 0.12 of the bulk volcanic plume (Tab. 1 and Fig. 4) determined after 30-min continuous Multi-GAS measurements made on the outer crater rim (see “bulk plume Multi-GAS site” in Fig. 2b, e). At that Multi-GAS site, we could intercept only a very dilute plume, rising buoyantly from the fumarolic field inside the crater floor (Fig. 2d). Only small concentrations of H₂S (~ 1 ppmv) and SO₂ (~ 0.15 ppmv) could be detected, no volcanic H₂O, CO₂, or H₂ being resolvable from the air background. Given these very low H₂S and SO₂ concentrations, well below our calibration range (10-200 ppmv), the inferred bulk plume SO₂/H₂S ratio of 0.12 must be considered with caution; we just take it as indication that hydrothermal H₂S-rich fumaroles prevail over the more magmatic end-member fumaroles in the bulk gas emission from Pico do Fogo, in agreement with indications from the arithmetic mean of fumarolic compositions.

Figure 5a presents the SO₂ flux time-series obtained by the UV Camera on February 5, 2019. A plot of SO₂ column amounts along the UV cross-section of Fig. 5b shows that, thanks to the short distance (~200 m) between the camera and the targeted plume, a feeble but continuous SO₂ emission (<400 ppm·m; mean, 140±110 ppm·m) was detected by the UV Camera in the leftmost portion of the camera FoV (Fig. 5c), and persisted throughout the ~100 minutes of recording (Fig. 5a). During our measurement interval the SO₂ flux varied between 0.3 and 2.3 tons/day (or 0.009 to 0.06 kg/s) and averaged at 1.4±0.4 tons/day (0.016±0.004 kg/s).

The molar gas ratios determined by Multi-GAS measurements allow to compute the molar percentages of H₂O, CO₂, H₂S, SO₂ and H₂ in each fumarole and in the mean gas composition (Table 1). These percentages for only the 5 above species are upper bounds since we did not determine other possible minor species (N₂, HCl) in the gases. Otherwise, they are not affected by the presence of reduced carbon species, whose amount was verified to be very low in F5 fumarole this study and (Meliánet alii, 2015). According to our results, the Pico do Fogo fumaroles are moderately hydrous (41-73 % H₂O; mean, 64 %), CO₂-rich (27-59 %; mean, 36 %), and contain about ~0.3 % S_t and 0.04 % H₂ (Tab. 1). These mean values match well the composition of the F15 fumarole, directly sampled and analysed in laboratory, as regards the H₂/H₂S and CO₂/H₂S molar ratios (Tab. 2).

The triangular plot in Figure 6 puts the H₂O-CO₂-S_t compositions of our Pico do Fogo fumaroles in a wider context, by comparing them against the compositions of (i) the 2014 Fogo eruption plume (Hernándezet alii, 2015), which represents the only available datum for the Fogo magmatic gas signature to date; (ii) magmatic gases from other intraplate, rift and/or divergent-plate volcanoes (see Aiuppa, 2015 for data sources); and (iii) fumaroles from other volcanic systems in the Macaronesia region, including the Azores (Caliroet alii, 2005; Ferreira & Oskarsson, 1999; Ferreiraet alii, 2005; Mares project, this study) and Teide in the Canary (Meliánet alii, 2012; Mares project, this study).

The Pico do Fogo summit fumaroles are compositionally distinct from the magmatic gases released during the 2014 eruption (Hernándezet alii, 2015), this latter falling well within the range of measured magmatic gas compositions at other intraplate volcanoes (yellow field, from Aiuppa, 2015). More specifically, the summit Fogo fumaroles are evidently S-depleted relative to the 2014 magmatic gas, which strongly suggests intense sub-surface scrubbing of reactive S compounds under the “hydrothermal” conditions of the fumarolic field, where surface temperatures (≤ 315 °C) are well below the boiling temperature of liquid sulfur (455 °C; above which S scrubbing become minimal, if any; Aiuppaet alii, 2017). Extensive S deposition in the sub-surface environment of the summit fumaroles is further supported by CO₂/S_t ratios being far higher in the fumaroles (93-162) than in the 2014 eruption gas (1.5; Hernándezet alii, 2015) (Figs. 6, 7). The two hottest summit fumaroles (F14 and F15) consistently display the lowest CO₂/S_t ratios (93-97), but these are still two orders of magnitude higher than in the eruptive gas, confirming the importance of sulfur scrubbing (Fig. 7). This is also verified for the dry gases directly sampled from fumarole F15, whose CO₂/H₂S ratio is 94-107 (Tab. 2).

Fogo summit fumaroles are also less hydrous (or more CO₂-rich) than the 2014 eruptive gas (Fig. 6). If the 2014 gas is representative of the magmatic gas feeding the summit fumaroles (a magmatic gas supply is indeed supported by the low but measurable SO₂ output; Fig. 5), then the simplest explanation of H₂O depletion in the fumaroles is extensive steam condensation in the fumarolic conduits due to low temperature conditions. Because our Multi-GAS measurements were made in air-diluted (and cooled) fumarolic plumes, we cannot entirely exclude that partial H₂O condensation could have also occurred during plume transport and/or in the Multi-GAS inlet system (tubing + filter), such as previously observed at other volcano-hydrothermal systems (e.g., Allardet alii, 2014; Lopezet alii, 2017; Tamburelloet alii, 2019). However, we note that our Multi-GAS-derived H₂O range (41-73 %) partially overlaps with the H₂O range (52-92 %) for the summit Fogo fumaroles previously determined from direct gas sampling (Meliánet alii, 2015). We thus conclude that both subsurface and within-plume H₂O condensation may combine to drive the summit fumaroles toward a less hydrous and correspondingly CO₂-enriched composition compared to the 2014 eruptive gas. We cannot exclude, however, that the magmatic gas that feeds the persistent summit fumaroles is compositionally different from the 2014 eruptive gas. If for example the magmatic gas source is the Pico do Fogo magma reservoir located in the uppermost mantle at 16–28 km depth (Hildneret alii, 2011, 2012; Mataet alii, 2017), then it is well possible that its composition has deeper (CO₂-richer, H₂O-S-poorer) signature than that of eruptive 2014 gas (derived from shallow degassing).

The Pico do Fogo fumaroles plot at the CO₂-rich end of the compositional array defined by volcanic hydrothermal fluids in the Macaronesia region (Fig. 6). The majority of volcanic fumaroles from the Azores (São Miguel, Terceira and Graciosa islands) and from Teide volcano in the Canari are shifted toward the H₂O corner. This is a typical (but not exclusive) feature of most hydrothermal steam vents worldwide (Chiodini & Marini, 1998), which reflects their derivation from the boiling of meteoric groundwater-fed hydrothermal systems (Caliroet alii, 2015). The less hydrous compositions of Pico do Fogo fumaroles suggest the absence of a shallow boiling hydrothermal aquifer underneath Fogo summit, and consequently a weaker (relative to Azores and Teide) hydrothermal fingerprint (greater magmatic signature), especially in the hottest fumaroles (F14 and F15) that also exhibit lower CO₂/S_t ratios (Fig. 7) and higher SO₂/H₂S ratios (Fig. 4). These SO₂-bearing F14-F15 fluids appear as formerly magmatic gases that have undergone partial H₂O-S_t loss (via condensation + scrubbing) during cooling and hydrothermal re-equilibration (Fig. 6). Instead, the most SO₂-poor, H₂S-dominated fumaroles (e.g., F3-F8) have suffered more significant hydrothermal processing, as testified by their lower H₂O/CO₂ (< 1), higher CO₂/S_t (> 130), and more reduced (H₂-rich) redox conditions, typical of hydrothermal fluids (Fischer & Chiodini, 2015) (Figs. 4, 7).

To conclude, we attribute the CO₂-rich compositions of the Pico do Fogo fumaroles to a combination of (i) hydrothermal interactions (partially removing magmatic sulphur and water) and possibly (ii) a deep magmatic gas source.

Combining the compositional data described above with the UV camera-based SO₂ flux record depicted in Figure 5, we can reliably estimate the output of CO₂ and other volatiles from the summit crater fumarolic field of Pico do Fogo (Table 3). To do this calculation, we combine the measured mean SO₂ flux (1.4±0.4 tons/day) and the mean molar composition of the summit fumaroles (64.1±9.2 % H₂O, 35.6±9.1 % CO_2, 0.2±0.08 % H₂S, 0.06±0.06 % SO₂, and 0.04±0.02 % H₂; red star in Figs. 4, 6 and 7), the S_t (0.26±0.14 %) of which is scaled to the bulk plume SO₂/H₂S ratio of 0.12 (Tab. 1 and Fig. 4) to infer the bulk plume mass ratios at 558 (H₂O/SO₂), 756 (CO₂/SO₂), 4.2 (H₂S/SO₂) and 1.1 (H₂/SO₂), respectively. This procedure allows us to smooth the effect of the large compositional heterogeneity of the fumarolic vents. We just note that the bulk plume SO₂/H₂S ratio of 0.12 characterizes the predominance of H₂S-dominated (F3-F8-like) hydrothermal fluids over more SO₂-rich (F14-F15-like) “more magmatic” fumaroles.

We obtain a daily fumarolic CO₂ output of 1060±340 tons (Table 3). We also estimate a daily release of 780±320 H₂O, 6.2±2.4 H₂S and 0.05±0.022 H₂. These results demonstrate that the fumarolic gas output is larger, for all volatiles, than diffuse degassing through the crater floor (Dioniset alii, 2014, 2015) (Fig. 8). For example, the latter has been estimated to produce 147-219 (±35) tons/day of CO₂ (Dioniset alii, 2014, 2015), which is only 14-20% of the inferred fumarolic CO₂ output. Even considering the soil CO₂ output estimated at the scale of the entire island (828±5 tons/day; Dioniset alii, 2015), the contribution of diffuse degassing remains less than a half (~ 43%) of the total Fogo island CO₂ degassing budget (~1890 tons/day; this study and Dioniset alii, 2015).

In contrast, the daily fumarolic gas output is far lower than the eruptive gas output (Fig. 8) for the 2014 eruption derived by Hernándezet alii, (2015) by combining SO₂ flux measurements with a scanning UV spectrometer (using the Differential Optical Absorption Spectroscopy – DOAS - technique) and a Multi-GAS-derived plume composition. Our fumarolic SO₂ output, for example, is a factor ~7000 lower than the large (~10 ktons) daily eruptive release (Hernándezet alii, 2015). Let emphasize, however, that while summit fumarolic emissions at Fogo have persisted as a stable degassing feature over the past few centuries (Ribeiro, 1960), eruptive degassing has been restricted to the relatively infrequent eruptions. There are only 10 reported eruptions since 1785 (Ribeiro, 1960), of which only 3 since 1951 (Hildneret alii, 2011, 2012; Carracedoet alii, 2015; Mataet alii, 2017). Between June 12, 1951 (the onset of the first, well recorded XX century eruption; Hildneret alii, 2012) and February 8, 2015 (the end of the last eruption), Fogo has been in eruption for only 200 days (e.g., 0.008 % of the 24710 elapsed days). If we take the November 30, 2015 gas output (Hernándezet alii, 2015) as typical for Fogo eruptive daily degassing rate, we can roughly compute a cumulative eruptive release for 1951-2015 (200 days of eruption) of ~4 Mtons of H₂O, ~2 Mtons of CO₂ and SO₂, 11 ktons of H₂S and 0.04 ktons of H₂. These masses, when scaled to (integrated over) the 24710 days elapsed from June 12, 1951 to February 8, 2015, correspond to daily eruptive outputs of only 196, 86, 82, 0.5 and 0.002 tons/day for H₂O, CO₂, SO_2, H₂S and H₂, respectively (Fig. 8). Our back-of-the-envelop calculations demonstrate that, when examined on longer-term perspective, eruptive emissions at Fogo are significant for only SO₂, while they do make a relatively small contribution to the emission budget of other volatiles (Fig. 8).

We therefore conclude that summit crater fumarolic emissions at Pico do Fogo are the dominant source of volcanic CO₂ (and most other volatiles) over multi-decadal scale.

On a broader perspective, our results for Pico do Fogo in Cape Verde archipelago add a new piece of information to the global catalogue of volcanic CO₂ emissions. Recent work (Fischeret alii, 2019; Werneret alii, 2019) has attempted at refining the global volcanic CO₂ emission inventory, by reviewing, cataloguing and synthesizing the volcanic CO₂ output information available in the international literature. It was found that, by late 2019, CO₂ flux measurements have become available for 102 of the ~500 degassing subaerial volcanoes worldwide (Fischeret alii, 2019; Werneret alii, 2019; Fischer & Aiuppa, 2020 submitted). Different strategies have been used to extrapolate the cumulative CO₂ output “measured” for the 102 volcanoes (~44 Tg/yr) to CO₂ emissions from the several hundred “unmeasured” subaerial degassing volcanoes. These have included the use of independent rock-chemistry information (Aiuppaet alii, 2019) and/or the identification of statistical properties (mean CO₂ output and confidence intervals) for different categories of volcanoes. On the latter basis, it was proposed that the present-day global volcanic CO₂ budget is dominated by the category of Strong Volcanic Gas Emitters (Svge) – which includes the ~100 top degassing volcanoes whose SO₂ emissions are systematically detected from space-borne and/or ground-based spectrometers (Carnet alii, 2017; Fischeret alii, 2019). S_vge have an inferred total (extrapolated) CO₂ output of ~ 36-39 Tg/yr (Aiuppaet alii, 2019; Fischeret alii, 2019). It was additionally found that a group of Weak Volcanic Gas Emitters _(Wvge), although degassing in a more subtle manner (this category includes volcanoes with no visible plumes and/or minor to absent SO₂ emissions), may still contribute between 15 (Fischer et alii, 2019) and 35 (Werneret alii, 2019) Tg CO₂/yr, simply because they are numerous (~400) globally. Unfortunately, however, these results are subject to very large uncertainties because measuring the CO₂ output from quiescent/hydrothermal volcanoes is especially challenging from a technical viewpoint (indirect SO₂ flux-based estimates are hampered by low to absent SO₂; Werneret alii, 2019), making the CO₂ flux catalogue particularly incomplete for W_vge.

Pico do Fogo falls within the W_vge category, as no plume is visually observable (Fig. 2) and no SO₂ is detectable by satellite except during the infrequent eruptions (Global Volcanism Program, 2017). Our results show, however, that SO₂ is present in tiny but measurable quantities in the fumaroles (Table 1), making both the SO₂ flux and, indirectly, the CO₂ flux (Table 3) measurable from a very proximal location on ground (Fig. 2; note that a test made with UV-Camera from the base of the volcano were unable to detect any SO₂ release).

When put in the context of global volcanic CO₂ fluxes (Fig. 9; data from Fischeret alii, 2019), the fumarolic CO₂ flux from Pico do Fogo (ca. 1000 tons/day) confirms that Wvge volcanoes can emit CO₂ in quantities that, in some cases, can rival the emissions of S_vge volcanoes. High CO₂ emission from such W_vge systems, despite negligible (hydrothermal-dominant) to weak (magmatic-hydrothermal) SO₂ emission (Fischeret alii, 2019), result from their exceptionally high CO₂/S_t signature (Aiuppaet alii, 2017). Pico do Fogo fumaroles are not an exception, but owing to their high CO₂/S_t compositions they can sustain a CO₂ output of order 1000 tons/day, at the upper range of the global W_vge and Svge populations (Fig. 9). Therefore, our present results further demonstrate that refining the global inventory for volcanic CO₂ output will require enhanced quantification of the weaker, poorly visible emissions sustained by quiescent hydrothermal volcanoes, the majority of which still lack CO₂ flux quantification.

We have shown here that fumarolic activity on-top of Fogo Volcano, in the Atlantic Cape Verde Archipelago, is currently a poorly visible but substantial source of volcanic volatiles to the atmosphere. The fumarolic CO₂ output (~1060 tons/day), in particular, is found to exceed by far the time-integrated eruptive CO₂ flux (~86 tons/day) from the volcano, as well as the estimated total CO₂ budget from soil degassing across Fogo Island (147-828 tons/day). On a broader scale, our results confirm that quiescent volcanoes characterized by hydrothermal activity during quiescent stages can produce CO₂ emissions that rival those of more manifestly degassing (Strong Volcanic Gas Emitters, S_vge) owing to their CO₂-enriched fumarole compositions (CO₂/S_t ratios of 93-163 at Pico do Fogo in 2019). At Pico do Fogo, these CO₂-enriched compositions likely result from the interactions (scrubbing of magmatic sulphur, and water condensation) of a deep magmatic gas supply (perhaps sourced from a 16–28 km deep magma reservoir in the uppermost mantle; Hildneret alii, 2011, 2012; Mataet alii, 2017) with a shallow hydrothermal system.

This research was funded by the Portuguese Fundacão para a Ciência e a Tecnologia (MARES project - PTDC/GEO-FIQ/1088/2014), the DECADE project of the Deep Carbon Observatory, and the Italian Ministero Istruzione Università e Ricerca (Grant n. 2017LMNLAW). We thank Francesco Salerno and Manfredi Longo from INGV-Palermo for providing support with gas chromatographic analysis. The manuscript benefited from constructive reviews from Taryn Lopez, Yuri Taran and from the Associate Editor Orlando Vaselli.