Carbon dioxide (CO₂) together with water vapor (H₂O) and sulfur dioxide (SO₂) are usually the most abundant gases released to the atmosphere during volcanic eruptions (Giggenbach, 1996; Fischer and Chiodini, 2015 and references therein). During non-eruptive periods, gas emissions are also frequent in many volcanic systems not only through the presence of permanent fumaroles, but also due to the existence of thermal and cold CO₂-rich springs, as well as soil diffuse degassing emissions (Allard et al., 1991; Chiodini et al., 1998). In the latter, the most studied gases released through volcanic soils are usually CO₂ and the radioactive gas radon (²²²Rn); in some cases the CO₂ diffusely emitted by the soils is of similar magnitude to the CO₂ released from fumaroles and crater plumes (e.g., Allard et al., 1991; Chiodini et al., 2010a; Viveiros et al., 2010; Pedone et al., 2015). Mantle-derived CO₂ can also be released away from active volcanoes through deep tectonic structures, as demonstrated by several studies (Chiodini et al., 1999, 2010b; Jung et al., 2014; Lee et al., 2016). In fact, in the last 30 years, CO₂ spatial distribution has been used worldwide to identify hidden tectonic structures, since faults/fractures act as preferential pathways (high permeability zones) for the escape of gases from the deep crust or mantle to the surface (Giammanco et al., 1999, 2006; Baubron et al., 2002; Hutchison et al., 2015; Liuzzo et al., 2015). Based on the release of soil gases in confined areas, Chiodini et al. (2001) named the anomalous CO₂ degassing areas, where hydrothermal/volcanic CO₂ is released, as diffuse degassing structures (DDS), whose shape depends on morphological, geological, and structural factors, such as the topography, existence of lithological heterogeneities, and presence of faults/fractures (Schöpa et al., 2011; Peltier et al., 2012; Pantaleo and Walter, 2014).

CO₂ has been widely used for volcanic monitoring due to its low solubility in silicate melts, being consequently one of the first gases released to the surface in case of replenishment of a magma chamber (e.g., Hernández et al., 2001; Carapezza et al., 2004; Giammanco et al., 2006; Aiuppa et al., 2010; Liuzzo et al., 2013; de Moor et al., 2016). Carbon dioxide is also an inert asphyxiant gas if present in high concentrations in the air, and above 10 vol.% can be lethal (Blong, 1984; Weinstein and Cook, 2005). Several hundreds of deaths have been reported in volcanic and non-volcanic environments due to deep-derived CO₂ emissions (Hansell and Oppenheimer, 2004) and high CO₂ concentrations have been reported in buildings located on CO₂ anomalous zones (Baxter et al., 1999; Viveiros et al., 2009, 2016a). This fact highlights the importance of identifying CO₂ anomalous areas by producing degassing maps, which should constitute valuable tools for land-use planners (Beaubien et al., 2003; Viveiros et al., 2009, 2010, 2015, 2016a). Building in high CO₂ degassing areas should be avoided, or, depending on the soil gas concentrations measured, buildings should follow few “gas-resistant” construction rules as the ones defined by Viveiros et al. (2016a).

Carbon dioxide is a colorless and odorless gas that is only detected with specific equipment. Several instruments have been developed in the last decades to detect this gas in soils in volcanic/tectonic environments, even if the first studies started to be applied in agricultural fields to measure soil respiration rates (e.g., Kanemasu et al., 1974; Parkinson, 1981). Relatively short-time and low-cost methodologies using infrared CO₂ detectors have been implemented to measure soil CO₂ emissions since the end of the eighties, which include soil CO₂ concentration measurements (Giammanco et al., 1999, 2006; Evans and Staudacher, 2001) and soil CO₂ fluxes (dynamic concentration and accumulation chamber methods; Gurrieri and Valenza, 1988; Chiodini et al., 1998). Studies performed in diffuse degassing environments show that not only CO₂ is permanently emitted to the atmosphere from soils, but also significant variations in the amount of gases emitted are observed. Several spike-like and long term variations have been interpreted as resulting from meteorological influences and seasonal effects (Granieri et al., 2003; Viveiros et al., 2008, 2014; Rinaldi et al., 2012). Gases released from soils in hydrothermal-volcanic environments may also result from organic matter decomposition and fauna respiration, the so-called soil respiration (Luo and Zhou, 2006). Discrimination of the different CO₂ sources has been done through the use of statistical methodologies (Chiodini et al., 1998), as well as the carbon isotopic composition of the CO₂ released (e.g., Chiodini et al., 2008).

Soil CO₂ concentration surveys were performed at Faial, Pico, and São Jorge islands in the period between 2001 and 2004, and preliminary results are available in Master theses (Faria, 2002; Marcos, 2006). In 2011, sporadic soil CO₂ flux measurements were also performed in the islands of Faial and Pico, with the main scope of evaluating the stability of the previously recognized anomalous CO₂ degassing areas and select an area to install a permanent soil CO₂ flux station. This equipment is necessary to integrate with other seismo-volcanic monitoring techniques and to complement the gas geochemistry network already installed in the Azores archipelago (Viveiros et al., 2008). This study reviews the preexisting database on soil CO₂ concentration applying more recent statistical tools in order to (1) evaluate the existence of different sources for the CO₂ diffusely released from soils, (2) produce CO₂ degassing maps, (3) identify the main DDS, and (4) correlate the soil CO₂ emitted with the main volcano-tectonic structures identified in the islands.

The Azores archipelago is composed of nine volcanic islands located in the North Atlantic Ocean where the American, Eurasian and the Nubian plates meet at a triple junction (Searle, 1980). The main tectonic features are (1) the Mid-Atlantic Ridge (MAR) that crosses the archipelago between the islands of Flores and Faial, (2) the East Azores Fracture Zone (EAFZ), which extends E-W from the MAR to south of Santa Maria, and the Azores-Gibraltar Fault Zone that includes the E-W trending Gloria Fault, and (3) its western segment, the Terceira Rift (TR), which extends from the MAR to the island of Santa Maria along a general WNW-ESE direction, and corresponds to the present-day EU-NU plate boundary (e.g., Searle, 1980; Madeira and Ribeiro, 1990; Vogt and Jung, 2004; Miranda et al., 2015; Figure 1). The main tectonic structures in the islands are normal dextral faults with a WNW-ESE trend, characteristic of the Terceira rift; the NNW-SSE conjugate fault system exhibits oblique normal left lateral displacement (Madeira and Brum da Silveira, 2003; Madeira et al., 2015). These main fault systems are also present in Faial, Pico, and São Jorge islands, with the dominant structures striking WNW-ESE and dipping 60–90° to the NNE or to the SSW controlling the general shape of some islands (Madeira and Brum da Silveira, 2003).

Most of the identified faults in the Azores archipelago are considered active (Madeira, 1998; Madeira and Brum da Silveira, 2003; Marques et al., 2013; Madeira et al., 2015) and the major faults usually have well-developed scarps; however the volcanic nature of the islands, together with differential erosion, can attenuate or amplify the tectonic slope (Madeira and Brum da Silveira, 2003) and even make the identification of tectonic structures difficult (due to scarp burial by pumice deposits, for instance). In addition, the exuberant vegetation of some of the Azorean islands covers most of the geological structures. Tectonic structures used in the present study were mapped by Madeira and Brum da Silveira (2003) and Madeira et al. (2015) through fieldwork and vertical air-photo interpretation. All the faults in the study islands are considered active by Madeira and Brum da Silveira (2003) as they displace volcanic sequences younger than 100 ka.

The cause for the volcanism in the islands is highly debated in the literature and most studies suggest the presence of a mantle plume (e.g., Schilling, 1975; Cannat et al., 1999; Moreira et al., 1999; Jean-Baptiste et al., 2009), even if some studies argue that the addition of H₂O to the mantle together with a small temperature anomaly could be enough to induce melting (Bonatti, 1990; Asimow et al., 2004; Métrich et al., 2014).

The complex geodynamic setting of the Azores explains the frequent seismicity and volcanism in the islands. Since the settlement of the islands, in the fifteenth century, at least 28 volcanic eruptions and more than 15 major earthquakes caused a large number of casualties (>6,500 deaths; Gaspar et al., 2015 and references therein).

Present seismo-volcanic activity in the archipelago is characterized by almost daily record of seismic events (Gaspar et al., 2015 and references therein) and the presence of secondary manifestations of volcanism in most islands. These manifestations include hydrothermal fumaroles, cold CO₂-rich, and thermal springs, as well as soil diffuse degassing areas (Ferreira et al., 2005; Cruz et al., 2010; Viveiros et al., 2010; Caliro et al., 2015; Silva et al., 2015). Fumarolic emissions are found at Faial and Pico islands, respectively, at the remnant of the 1957–58 eruption (Capelinhos) and at the summit of Pico Volcano. São Jorge manifestations are characterized only by the presence of cold CO₂-rich springs, which are also found out at Pico and Faial (Ferreira et al., 2005).

Visible secondary manifestations of volcanism were previously recognized in the islands of Faial, Pico, and São Jorge (Ferreira et al., 2005; Cruz et al., 2010). The present study highlights the additional presence of soil CO₂ diffuse degassing areas, most of them associated to faults crossing the study sites. The maximum soil CO₂ concentration (20.7 vol.%) and flux (339 g m⁻² d⁻¹) values were measured at Faial Island, in the Praia do Almoxarife area and are associated to the Espalamaca fault (DDS F4 in Figure 6). These measured soil CO₂ concentration values are comparable with the concentrations recorded at Mosteiros and Ribeira Seca villages (São Miguel Island), but significantly lower than at Furnas Volcano, where values as high as 100 vol.% were measured (Viveiros et al., 2015). The maximum soil CO₂ concentration values at Pico and São Jorge islands were similar (~13 vol.%), but lower than those recorded at Faial Island.

The measured ranges of CO₂ suggest different origins for the gases released from soils. Statistical methodologies applied to the recorded data showed different populations both for the CO₂ concentrations and fluxes, suggesting biogenic and volcanic-hydrothermal sources. Discrimination of these sources is particularly relevant in the Azores archipelago due to the abundant vegetation that covers the volcanic soils. Lowest CO₂ values (population “A,” Table 4) probably correspond to areas where biologic processes are responsible for the gas released from soils. Values very close to zero, or even zero, should represent areas where even the biologic production is scarce. Population “C,” only present in the soil CO₂ flux dataset of Faial Island, is representative of CO₂ fed by an endogenous source (volcanic-hydrothermal origin). Data associated with population “B” probably represent the mixture of biogenic and hydrothermal CO₂ contributions. Most of the measured soil CO₂ concentration data fall within populations “A” (65–75% of the data), corresponding to biogenic contribution for the emitted CO₂ (Table 4). In what concerns the soil CO₂ flux determinations, most of the data are interpreted as a mixture of biogenic and hydrothermal/volcanic contributions. The high percentage of CO₂ flux data in population “B” is probably explained by the fact that the flux surveys were carried out in the areas where anomalous CO₂ values were previously detected and mostly related with the presence of faults. The main objective of the 2011 campaigns was to select an area to install permanent soil CO₂ flux stations in the islands of Pico and Faial.

The cut-off value defined for each dataset according to the criteria suggested by Sinclair (1974), i.e., the 95th percentile of a determined population, allows to establish thresholds to discriminate the biogenic contribution from a deep source (Table 5). The 95th percentile for population “B” is selected in this study to define the DDS (Chiodini et al., 2001). For the particular case of Faial Island, the 95th percentile of populations “A” and “B” (soil CO₂ concentrations) is similar. In what concerns the limits defined for the soil CO₂ fluxes, the thresholds are the same for Pico and Faial islands (12.6 g m⁻² d⁻¹ for the biogenic contribution and 39.8 g m⁻² d⁻¹ for the definition of DDS). Scales of the distribution maps presented in Figures 3, 4 were defined according to these limits. The final DDS proposed in the present study resulted from the integration of soil CO₂ flux and concentration values.

The thresholds proposed here are similar to data available in the literature. For instance, available data on CO₂ production from a wide variety of ecosystems show CO₂ fluxes ranging from ~0.5 to ~19 g m⁻² d⁻¹ (e.g., Raich and Schlesinger, 1992; Raich and Tufekcioglu, 2000) with maximum flux values attributed to grassland (~50 g m⁻² d⁻¹; e.g., Norman et al., 1992; Bajracharya et al., 2000; Nakadai et al., 2002). The biogenic threshold for the soil CO₂ fluxes at Faial and Pico islands is thus within the values published for other areas. At Furnas Volcano (S. Miguel Island) a value of 25 g m⁻² d⁻¹ was selected as the limit, which integrated both statistical methodologies and carbon isotopic data (Viveiros et al., 2010). The threshold defined for the DDS at Furnas Volcano was 50 g m⁻² d⁻¹, not far from the value suggested in this study (39.8 g m⁻² d⁻¹). The CO₂ sources should be better constrained by the δ¹³C_CO2 data, which is not available for this study. For this reason, the evaluation of different feeding sources was only based on the cumulative probability plots.

The CO₂ degassing map for Faial Island shows an absence of CO₂ emission in the youngest part of the island (Capelo Volcanic Complex), including the areas where the recent volcanic eruptions occurred (Figures 2, 6). Even in areas where thermal anomalous values were measured, which are associated to the only fumarolic (steam) field in Faial, in Capelinhos Volcano, no soil CO₂ degassing (both concentration and flux) was detected. In the fumarolic field the maximum temperature measured was 91°C and the main gases released are the atmospheric constituents (N₂, O₂, and Ar), with minor amounts of CO₂ and H₂. These emissions are probably due to some remnant heat from the 1957–58 volcanic eruption that still heats some water and air in the system. Higher soil CO₂ concentrations were measured in the floor of the summit caldera (Caldeira Volcano; DDS F1, Figure 6), where a phreatic eruption occurred during the Capelinhos 1957–58 volcanic eruption, however no thermal anomaly was identified in the area.

Two main fault systems trending WNW-ESE and NNW-SSE were recognized in the study islands by previous works (Madeira and Brum da Silveira, 2003; Madeira et al., 2015), and most of the DDS identified in this study occur associated with the previously mapped tectonic structures. A clear correlation between distribution of the CO₂ degassing anomalies and one of the main trending pattern is not observed; instead, and similarly to the observed for volcanism that is tectonically controlled (Madeira et al., 2015 and references therein), the main degassing anomalies seem to be found out in the intersection of different fault systems.

Based on soil CO₂ concentrations coupled with the soil CO₂ flux values, seven DDS were then identified in the Faial island (Figure 6). Main CO₂ anomalies are located in the eastern side of the island, essentially delimited by the Pedro Miguel graben structure that affects products from the Ribeirinha and Caldeira central volcanoes. The large DDS F3, which comprises more than half of the graben floor, lays between the important WNW-ESE tectonic structures that define the Pedro Miguel graben: between the Lomba Grande Fault (LG, Figure 2) that dips to the south and the north-dipping Espalamaca and Flamengos faults that define the southern part of the graben structure. The presence of deeply-derived CO₂ in the Pedro Miguel graben suggests that the graben faults correspond to deep structures that link the surface to deep areas where CO₂ is trapped and then channeled to the surface. DDS F5 probably results from the intersection of the graben faults with the mapped NNW-SSE structure identified in the area. In addition, this DDS contributes to define a degassing lineament (1′, Figure 6) that can be the prolongation of the Chã da Cruz Fault.

The graben faults do not constitute barriers to the gas traveling to the surface since diffuse degassing structures are not limited to the graben floor and, for instance, DDS F2 is identified in the upthrown block of the Flamengos Fault. Two degassing lineaments are also associated with this DDS, which argue to confirm the presence of the hidden N-S radial fracture (lineament 2′; Figure 6).

The DDS F6 is associated with Flamengos and Lomba de Baixo faults, which dip to the north. The main CO₂ anomaly is however observed in the top of the scarp of Lomba de Baixo Fault and should result from the high permeability of the area that is crossed by several tectonic structures. Similarly, DDS F7 is also found out in the intersection of different trending tectonic structures (WNW-ESE, NNW-SSE, and NE-SW). As mentioned before, the CO₂ anomalous areas occur, in general, where a network of faults/fractures are identified, and thus seem to result from the intersection of permeable structures that allow the gas to travel from depth and escape to the surface (see DDS F2, F5, F6, and F7 in Figure 6).

The highest soil CO₂ flux and concentration values, representing a clear volcanic-hydrothermal origin, were measured at Praia do Almoxarife village (DDS F4) and were detected close to a building where in 2007 lethal indoor CO₂ concentrations (>15 vol.%) were measured (Ferreira and Viveiros, 2007). This very localized DDS is aligned with submarine gas emissions offshore of Faial, along the submerged Espalamaca Fault scarp, which are mainly constituted by cold CO₂ (Viveiros et al., 2016b).

From a public health risk assessment perspective, the DDS F4, located at Praia do Almoxarife, highlights the need to perform a detailed CO₂ flux map in that village to identify which buildings may be located over anomalous CO₂ degassing sites, with the consequent impacts that it may have on the population. Even if the other DDS are not directly associated with inhabited areas, detailed CO₂ degassing maps should also be performed to evaluate possible risks for future constructions.

A total of four DDS were identified at Pico Island. The main CO₂ degassing anomalies are located in the summit area of the youngest volcano (DDS P2), Pico Volcano, where the only fumarolic emissions of the island are found (Figure 7). DDS P1 and P2 are associated with the Lomba de Fogo-S. João NNW-SSE trending fault that crosses the summit of Pico Volcano and fed the 1718 eruption. Similarly to the central volcano, the fumarolic field occurs in the intersection of the NNW-SSE, WNW-ESE, and NE-SW structures. Chemical analyses performed on Pico fumaroles in the summer of 2016 are similar to those previously obtained in 1994, showing the stability of its feeding system across two decades. The gases emitted are mainly composed of N₂ and CO₂, with minor amounts of O₂ and Ar, whose composition suggests air contamination during the ascend of gas from the deep reservoir to the surface. This fact is probably correlated with the location of the fumaroles in the volcano's summit, allowing atmospheric air to be introduced in the system, interfering with the original gas composition in the hydrothermal reservoir.

Despite the DDS P1 and P2 the remaining anomalous CO₂ degassing sites are also associated with tectonic structures previously identified by Madeira and Brum da Silveira (2003) in older volcanic areas. These important CO₂ degassing anomalies are correlated with the Lagoa do Capitão and Topo areas (DDS P3 and DDS P4, respectively, Figure 7), where several faults intersect. DDS P3 is correlated with the important WNW-ESE trending Lagoa do Capitão Fault. This structure dips to the south, but CO₂ anomalies are observed also in the upthrown block of the fault. This behavior may be associated with a higher permeability in the top of the scarp (due to fracture along the scarp) or with the presence of not mapped tectonic structures, such as the suggested lineament 1' that can be the elongation of the probable NNW-SSE fault defined by Madeira and Brum da Silveira (2003). In what concerns DDS P4, identified in the southwestern side of the WNW-ESE Topo trending fault, it probably results from the intersection of hidden NNW-SSE and NE-SW tectonic structures (Figure 7).

Similarly to Faial, faults/fractures intersection seems to favor higher permeability for the gas to travel from depth to the surface. Madeira and Brum da Silveira (2003) suggested several probable/hidden faults for the island trending WNW-ESE and NW-SE, and the present study shows that significant CO₂ anomalies correlate well with the areas of intersection of hidden/probable NNW-SSE structures with the dominant WNW-ESE faults.

The areas associated with historical eruptions in Pico Island (1562–64, 1718 and 1720, Figure 2), similarly to what was observed for Faial (Capelo Peninsula), do not show CO₂ emissions, what can be probably explained by the low or absent coverage of soils together with the high permeability of the recent lava flows that allow ingress of atmospheric air in the sub-superficial layers and dilution of the deep gases in their traveling to the surface. Nevertheless, it cannot be excluded that lava flows may act locally as barriers to the gas ascent to the surface when they are compact and low fractured. This behavior was also observed by Padrón et al. (2013) at Timanfaya Volcano (Lanzarote, Canary Islands), where the absence of CO₂ emission along the eruptive fissures was justified by the capping nature of the lava flow fields of the 1730–36 eruptions. Even in the vent areas or in the eruptive fissures the CO₂ degassing is low. Studies performed by Giammanco et al. (1999) on Mt. Etna (Italy) justified the absence of soil degassing in recent eruptive fissures due to obstruction after magma solidification or to sealing from hydrothermal alteration.

From a hazard perspective, none of the main anomalous CO₂ sites are located in areas with buildings; nevertheless, and considering the distances between sampling sites, a detailed survey should be carried out in the villages located close to the identified DDS.

The comparison of the two surveys carried out at Pico fumarolic emissions shows that the gas composition remained quite stable in the last two decades; this fact argues for the stability of the feeding systems, and consequently can be useful for any seismo-volcanic monitoring programme.

Four main DDS were identified in São Jorge Island, where no thermal anomalous zones are known. SJ3 is the only DDS located in the intersection of different WNW-ESE trending faults (Figure 8) and probably corresponds to an area of increased permeability that allows the gas to escape at the surface. The DDS SJ1 is aligned with a WNW-ESE trending probable fault located to the west and east of the degassing anomaly and mapped by Madeira and Brum da Silveira (2003). The alignment of the soil CO₂ degassing anomalies (1′, Figure 8), once again, may contribute to define the location of hidden faults. The main WNW-ESE fault zones that dominate the younger western part of the island do not show any CO₂ degassing, what can be explained by the superimposed lithological control, as mentioned also for Pico and Faial islands. In the eastern part of the island, and along the WNW-ESE north dipping Urze-São João Fault (Madeira et al., 2015) few anomalous soil CO₂ concentrations are observed, even if not directly associated with the dipping direction, and were identified as DDS SJ4.

No tectonic structure has been identified in the area surrounding the SJ2 anomaly, thus further studies are needed to better understand the permeability in the area. These future works should include soil CO₂ flux surveys that could contribute to better define the DDS here proposed and eventually identify other anomalous areas. This study does not show significant anomalous CO₂ emissions close to the probable Ribeira Seca Fault, and thus cannot contribute to support its existence, that is still under debate.

Considering that CO₂ may act as asphyxiant, a detailed survey at Ribeira Seca village is needed for land-use planning. Three of the anomalous zones (SJ2, SJ3, and SJ4) are located in the oldest volcanic system of the island (Topo Volcanic System) and, similarly to what is observed on Pico and Faial islands, no soil CO₂ degassing was detected on the areas where historical eruptions (1580 and 1808) occurred (Figure 2), fact that should be explained by the same factors as mentioned above for Faial and Pico islands (absence of well-developed soils, permeability of the lava flows and existence of natural barriers for the gas migration and release at the surface).

Degassing phenomena during periods of quiescence may be a permanent risk in any volcanic area (Viveiros et al., 2009) and, even if no incidents were reported for the islands of Pico and São Jorge, a family was relocated in Faial Island in 2007 due to high indoor CO₂ concentrations (>90 vol.%) (Ferreira and Viveiros, 2007). As mentioned above, additional detailed surveys need to complement these studies, mainly focusing in the inhabited areas close to the DDS. For the specific case of São Jorge, and considering that no recent surveys are available, the stability of the degassing areas cannot be confirmed and as so a soil CO₂ flux survey is highly recommended. In addition, Mendes et al. (2013), based on GPS data, suggested possible magma movement in the youngest part of the island, highlighting the need to obtain complementary information from a geochemical point of view.

Anomalous soil CO₂ degassing areas were identified at Faial, Pico, and São Jorge islands and a total of 15 diffuse degassing structures (DDS) were defined based on the integration of soil CO₂ flux and concentration measurements. Statistical methodologies applied to the measured values show distinct populations for the CO₂ emitted at the surface, which probably represent different origins (biogenic and hydrothermal-volcanic). Thresholds were defined for each population highlighting the existence of deeply-derived CO₂ that is transported to the surface through active faults. In fact, main WNW-ESE and NNW-SSE trending faults in the three study islands are usually well correlated with CO₂ degassing and constitute the pathways for the migration of CO₂ from deep reservoirs to the surface; few exceptions may be associated with the superimposed effect of the lithology. The main CO₂ emissions seem to occur in the older volcanic systems, where the development of soils reduces possible air contamination and the gases migrate to the surface through the high permeability zones that correspond to the intersection of fractures/faults. The absence of soils and the permeability of the recent lava flows, which allow the dilution of the deep gases with atmospheric air, may explain the lack of CO₂ emission in the younger volcanic systems. In addition, locally, some low permeable lava flows may act as barriers for the deep gas emissions.

For Pico and Faial islands two different methodologies were used in the present study to measure the soil CO₂diffuse degassing. Even if no simultaneous measurements were performed, higher soil CO₂ flux values were essentially coincident with areas with anomalous soil CO₂ concentrations. The generally good spatial correlation observed between the soil CO₂ flux and concentration values in different surveyed periods shows that the anomalous CO₂ degassing areas remain stable along time, arguing for the use of these techniques as useful tools for volcanic monitoring programmes. Considering the need to establish a permanent gas monitoring network in these islands, several DDS are here suggested as potential sites to install a permanent soil CO₂ flux station, taking into account the state of activity of the different volcanic systems. DDS F2 or F3 at Faial Island are two potential areas to install a permanent soil CO₂ flux station since they correspond to the flank of the active Caldeira central volcano. At Pico Island both DDS P1 and P2 seem to be adequate to install a permanent equipment, even if detailed soil CO₂ flux measurements should be performed to select the highest CO₂ emission site.

Considering that (a) about 32% of the historical volcanic eruptions that affected the Azores archipelago occurred inland or offshore the islands of Pico, Faial, and São Jorge (Gaspar et al., 2015), (b) some of the events were lethal causing the death of a high number of persons, even when the type of volcanism was less explosive (Hawaiian or Strombolian activity), (c) the state of activity of the volcanoes that form these three islands, and (d) the number of inhabitants (38,180, according to the Census, 2011) that can be vulnerable to a future volcanic event, the existence of a volcano monitoring programme that involves different techniques (geophysics, geochemistry) is crucial to identify periods of unrest. The CO₂ degassing maps here presented are thus important tools that characterize the CO₂ emission in quiescent periods of activity and are important to identify future periods of reactivation. Besides, several studies performed in these islands (e.g., Madeira and Brum da Silveira, 2003; Hildenbrand et al., 2012; Zanon and Frezzotti, 2013; Zanon et al., 2013; Trippanera et al., 2014) have highlighted the importance of the tectonic control on the plumbing systems of the volcanoes, and most of the main degassing areas are also tectonically controlled, what reinforces the need to integrate different geological studies to better understand the volcanic systems. On another hand, the location of some of the identified DDS close to inhabited areas highlights the need to identify hazardous CO₂ zones both for the existing and for future constructions.

FV revisited/processed the data and drafted the manuscript. MM, CF, FV, and CS carried out the soil diffuse degassing field surveys and contributed for the elaboration of the data. JG and TF provided important suggestions during the processing. All the authors have read and approved the final manuscript.