Volcanic gases can pose a permanent threat to the population not only during eruptive episodes but also in quiescent volcanic phases, since gases may be continuously released from the volcanic edifice (Blong, 1984; Hansell and Oppenheimer, 2004; Edmonds et al., 2015; Fischer and Chiodini, 2015; Williams-Jones and Rymer, 2015). Some volcanic gases are toxic even in low concentrations (e.g., H₂S, HF, HCl, and SO₂), while others, such as CO₂, are dangerous only if present in such concentrations that act as an inert asphyxiant and displace oxygen in the air down to dangerously low levels (Weinstein and Cook, 2005).

Carbon dioxide, an odorless and colorless gas, is immediately dangerous to human life above 10 vol%, causing rapid loss of consciousness, asphyxiation, and death (NIOSH, 1976; Le Guern et al., 1982; Blong, 1984; Wong, 1996; NIOSH, 2019; IVHHN, 2024). Symptoms associated with CO₂ exposure include breathing acceleration, dyspnea, increased heart rate, headaches, sweating, dizziness, ringing in the ears, vertigo, vomiting, and muscular weakness (e.g., Blong, 1984; Wong, 1996; IVHHN, 2024). A CO₂ concentration of 3 vol% has been defined as the Short-Term Exposure Limit (STEL), and 0.5 vol% as the limit for an 8 h exposure (TWA—Time-Weighted Average) (NIOSH, 2019).

CO₂ is one of the most abundant volcanic volatiles and it is the main gas released in diffuse degassing areas. It may accumulate hazardously in poorly ventilated or depressed zones since it is denser than air at standard temperature and pressure (STP) (NIOSH, 1976; Chiodini et al., 1998; Viveiros et al., 2016). Together with the fumarolic fields, soil diffuse degassing areas are potentially one of the main hazardous zones in quiescent volcanic regions since the gases (usually CO₂, H₂S, and ²²²Rn) are continuously released from soils and can ingress into buildings without being acknowledged by the population (e.g., Barberi et al., 2007; Edmonds et al., 2015; Viveiros et al., 2016; Barberi et al., 2019; Carapezza et al., 2020; Carapezza et al., 2023).

Even considering that CO₂ is often a neglected natural risk (D'Alessandro, 2006), in the last decades more than 2,000 deaths were reported in volcanic areas associated with this gas (Hansell and Oppenheimer, 2004; Weinstein and Cook, 2005; Hansell et al., 2006; Edmonds et al., 2015). The most tragic events were the Dieng Plateau (Indonesia) gas cloud emission, which caused the death of at least 142 persons (Le Guern et al., 1982; Allard et al., 1989), and the gas release from lakes Monoun (1984) and Nyos (1986) in Cameroon, responsible for the death of about 39 and 1700 persons, respectively (Barberi et al., 1989; Baxter and Kapila, 1989; Brown et al., 2017). CO₂ has been also responsible for fatal incidents in quiescent areas of Central Italy (Annunziatellis et al., 2003; Beaubien et al., 2003; Carapezza et al., 2003; Barberi et al., 2007; Costa et al., 2008; Barberi et al., 2019; Carapezza et al., 2023), United States (Mammoth Mountain) (Farrar et al., 1995; Sorey et al., 1998) or Japan (Hakkoda volcanic complex) (Hernández et al., 2003). Deaths are also commonly associated with the so-known “mazuku” in DR Congo, associated with the diffuse degassing processes of the Nyiragongo and Nyamulagira active volcanoes. The number of fatalities in these areas is difficult to quantify (Smets et al., 2010; Balagizi et al., 2018; Boudoire et al., 2022).

In what concerns the Azores archipelago (Figure 1), during non-eruptive phases, in 1992, two visitors of Furna do Enxofre lava cave (Graciosa Island) died due to the silent and permanent emission of CO₂ occurred. CO₂ concentrations above 15 vol% were measured on the day after the incident in the deepest and non-ventilated area of the cave (Gaspar et al., 1998; Viveiros, 2003). A fatality inside an abandoned water well at Mosteiros village (São Miguel Island) at the end of the 1980 s seems to be also attributed to high CO₂ concentrations, even if no measurements were performed at that time. In addition to the deaths associated with the diffuse degassing areas, incidents in the hydrothermal fumarolic fields also occurred in the last decades causing the death of at least two persons due to severe burns. Five of the 250 syn-eruptive fatalities were also associated with inhalation of volcanic gases (Viveiros, 2003; Gaspar et al., 2015). In the last 25 years, several families in the islands of S. Miguel and Faial were dislodged due to hazardous indoor CO₂ concentrations (Viveiros et al., 2015).

The discomfort higher indoor CO₂ may cause on the population is usually associated with the so-called “sick building syndrome” (e.g., Wittczak et al., 2001) and it is not necessarily applied to volcanic environments, but mainly due to occupation coupled with reduced ventilation. The impact higher indoor CO₂ levels may have on human health has been already highlighted (e.g., Erdmann and Apte, 2004; Satish et al., 2012) in several non-volcanic environments. Recently, Stewart et al. (2022) carried out a review on the impacts volcanic air pollution may have on human health, and most of the effects were associated with the acidic SO₂, and the only studies focusing on CO₂ were carried out in the Azores archipelago, specifically at Furnas Volcano (Amaral and Rodrigues, 2007; Linhares et al., 2015). More recently, Carapezza et al. (2023) also discriminated potential health impacts (increased risk of mortality and diseases of the central nervous system) due to significantly higher CO₂ emission in an Italian residential area.

Given the possible impact CO₂ may have on human health, and the silent ingress of the gas in buildings located in diffuse degassing areas, identification of anomalous soil CO₂ is crucial to reduce the risk of exposure on the volcanic degassing sites (Viveiros et al., 2009; 2010). In what concerns outdoor environments, CO₂ usually dilutes in the atmosphere, as shown by several gas dispersion models applied in diffuse degassing sites (Pareschi et al., 1999; Costa et al., 2008; Chiodini et al., 2010; Granieri et al., 2013; Massaro et al., 2021; Massaro et al., 2022; Rave-Bonilla et al., 2023; Viveiros et al., 2023). However, under certain circumstances and in confined spaces, hazardous concentrations may still be detected (Diliberto et al., 2021; Viveiros et al., 2023). Hazard maps produced based on the dispersion models do not account for indoor CO₂ exposure, where lethal concentrations may be detected (Viveiros et al., 2009; Viveiros et al., 2015; Barberi et al., 2019; Carapezza et al., 2023).

Indoor CO₂ measurements are difficult to assess due to permits and the general unavailability of the population. Literature shows indoor measurements, up to the moment, mainly in the Azores archipelago (Viveiros et al., 2009; Silva et al., 2015a; Viveiros et al., 2015; Viveiros et al., 2016), and in some Italian areas (Carapezza et al., 2003; Barberi et al., 2019; Carapezza et al., 2023; Carapezza et al., 2015). Few studies (Baxter et al., 1999; Annunziatellis et al., 2003; Viveiros et al., 2015; Barberi et al., 2019; Carapezza et al., 2023) show that soil CO₂ flux/concentration maps have been used to identify anomalous zones in volcanic/hydrothermal areas and infer potential indoor CO₂ exposure. Nevertheless, as far as we are aware, criteria to produce this indoor hazard and/or risk maps based on diffuse degassing are not found in the literature.

This study thus aims to define criteria to produce indoor CO₂ risk maps based on the Furnas Volcano study case, which can be extrapolated to other diffuse degassing areas. The criteria are discussed and the results are validated based on indoor CO₂ measurements.

Furnas Volcano is a trachytic central volcano located in the eastern part of S. Miguel Island (Azores archipelago, Portugal) (Guest et al., 1999). Two subplinian volcanic eruptions occurred since the settlement of the island in the 15th Century, one in 1,439-43 and other in 1,630 (Guest et al., 1999). Two parishes from Povoação County, Furnas and Ribeira Quente, are located, respectively, inside the caldera and in the southern flank of the volcano. Currently various hydrothermal manifestations are observed in this volcanic system, which include low temperature fumaroles, thermal and cold CO₂-rich springs (Figure 2A) (Ferreira et al., 2005; Viveiros et al., 2010; Silva et al., 2015a; Silva et al., 2015b; Caliro et al., 2015). Despite the visible gas emissions, an important CO₂ diffuse degassing area was recognized below Furnas village in the early nineties (Baxter et al., 1999). Studies carried out in the last two decades showed that the anomalous CO₂ degassing areas remained stable (Sousa, 2003; Viveiros et al., 2010; 2012; Pedone et al., 2015; Bagnato et al., 2018; Girault et al., 2022). Viveiros et al. (2010) also identified an important mantle-derived CO₂ degassing zone at Ribeira Quente village. A value around 954 t d⁻¹ was estimated for the hydrothermal CO₂ diffusely released by the soils at Furnas Volcano (Viveiros et al., 2010; Viveiros et al., 2012). The diffuse degassing studies showed that anomalous zones are essentially associated with the WNW-ESE tectonic structures found out at Furnas Volcano (Viveiros et al., 2010; Carmo et al., 2015). Silva et al. (2015a), Silva et al. (2015b) mapped soil radon (²²²Rn) anomalies in both villages showing also a good correlation between the anomalous CO₂ and ²²²Rn areas. Hazardous indoor CO₂ and ²²²Rn concentrations were detected in several buildings (Baxter et al., 1999; Viveiros et al., 2009; Silva et al., 2015a; Viveiros et al., 2015; Viveiros et al., 2016), and most of the anomalous indoor CO₂ were associated with extreme meteorological conditions, namely decreases in the barometric pressure and periods of rainfall (Viveiros et al., 2009; Viveiros et al., 2016).

According to the 2011 survey of population and housing, the two villages located in the Furnas geographical area had 2,206 inhabitants and slightly decreased to 2,081 in the last survey (Censos, 2021). Nevertheless, these numbers, the significant increase of tourists in the last decade in the Azores call the attention that these numbers may be significantly higher during the summer season. Resendes (2004) surveyed Povoação municipality to evaluate the vulnerability of buildings to several geological hazards (explosive volcanic eruptions, earthquakes).

Two villages (Furnas and Ribeira Quente) (Figure 2A) located on Furnas Volcano were used as study cases to show the implementation of the above-described method and evaluate its adequacy.

This study constitutes the first approach to produce risk maps of human exposure to anomalous deep-seated (volcanic/hydrothermal) CO₂ with criteria that can be extrapolated to any diffuse degassing environment. As far as we are aware, no other studies are found in the literature that attempt to propose risk maps for indoor environments based on soil degassing mapping. Considering the recent unrest period at Vulcano Island (Italy) (Diliberto et al., 2021; Di Martino et al., 2022) as well as the anomalous CO₂ degassing detected in La Bombilla and Puerto Naos (La Palma, Spain) after the 2021 volcanic eruption, the application of these methodologies to define susceptibility and risk maps may be helpful tools to be used by authorities and land-use planners.

In the last three decades, a significant amount of diffuse degassing areas have been mapped with various methodologies, and numerous anomalous zones recognized (Werner et al., 2019 and references therein). Soil CO₂ fluxes measured with the accumulation chamber method have been the prevalent methodology and these studies are mainly used for volcano monitoring, identification of tectonic structures, and/or quantification of the Earth’s carbon budget (e.g., Chiodini et al., 1998; Cardellini et al., 2017; Tamburello et al., 2018; Werner et al., 2019). In addition to these applications, when buildings are located above CO₂ anomalous zones, gases may ingress indoors and frequently reach hazardous concentrations. Soil CO₂ concentration/flux maps have been in some cases used as hazard maps (Baxter et al., 1999; Annunziatellis et al., 2003; Carapezza et al., 2011; Carapezza et al., 2012; Viveiros et al., 2015; Diliberto et al., 2021; Di Martino et al., 2022). However, no criteria to define the susceptibility/hazard levels have been discussed in the literature and no attempt to infer the indoor risk of exposure was made.

An ideal approach would be to measure continuously and in real-time indoor and outdoor CO₂ concentrations in areas that are recognized as degassing deep-derived CO₂. However, together with the cost of such systems, difficulties in having permits to record indoor CO₂ data, especially in private buildings, make this task challenging. This study thus aims to obviate this problem by using the already available CO₂ degassing maps.

The criteria here defined were tested at Furnas Volcano residential areas, where several soil CO₂ degassing maps were available (Baxter et al., 1999; Viveiros et al., 2010; 2012; Pedone et al., 2015; Bagnato et al., 2018; Girault et al., 2022), and showed that spatial CO₂ anomalies have remained stable during the last three decades. In addition, hazardous CO₂ concentrations have been recorded indoors (Viveiros et al., 2009; Viveiros et al., 2014a; Silva et al., 2015b; Viveiros et al., 2015; Viveiros et al., 2016). The study carried out by Viveiros et al. (2010) accomplishes with the larger areas and, for this reason, the CO₂ degassing map was reclassified and converted in a susceptibility map. The three susceptibility classes were allotted based on the CO₂ origin and the existence of DDS, criteria that can be easily extended to any other degassing areas. The definition of the biogenic contribution and the DDS threshold is site-dependent and must be set considering for instance the type of vegetation, CO₂ fluxes, state of volcanic activity, and existence of impermeable structures (faults, fractures).

Previous studies (Baxter et al., 1999; Sousa, 2003) have estimated that about one-third of Furnas village buildings were located over important deep-derived degassing and, consequently, residents were at risk of exposure to volcanic/hydrothermal CO₂. This study encompasses a larger area and not only increases the number of buildings at high risk of exposure to anomalous CO₂ at Furnas village (58%) but also shows that approximately 98% of the buildings at Ribeira Quente village have a high or very high risk of anomalous indoor CO₂. These results suggest the importance of accounting with CO₂ degassing maps for land-use planning and the need to evaluate the presence of diffuse soil gases previously to setting edifices in degassing areas.

The vulnerability classes represent both the exposure and the vulnerability of the exposed elements and were here defined based on the indoor/outdoor environments, as well as on the existence of “gas-resistant” construction strategies in the buildings. Crosscheck tests were done at Furnas village to validate the methodological approach and a good correspondence was obtained between indoor gas concentration and the risk levels assigned (Table 3). Nevertheless, the moderate risk level assigned to a building studied by Viveiros et al. (2009), where indoor CO₂ values as high as 20.8 %vol. were measured at the ground floor level, which may put into question the adequacy of the method. The susceptibility level in this case was assigned as high, but the lower vulnerability of the building associated with the vented space between the soil and the ground floor reduced the risk to moderate. As demonstrated before (Viveiros et al., 2009), the existence of these vented spaces reduces and delays the ingress of soil gases into the building, but during persistent and extreme meteorological conditions hazardous, CO₂ concentrations can still be measured indoor if no natural/artificial ventilation system is activated. For this reason, we mentioned the need to evaluate the efficiency of the mitigation strategies implemented in the buildings and in some cases, as the one mentioned above, it can be necessary to set up more than one “gas-resistant” code. Further studies need to focus on the vulnerability classes and better characterize the buildings as well as evaluate the adequacy of each mitigation strategy implemented. The vulnerability classes will probably need to consider more than one mitigation strategy but, even considering these limitations, we still decided to present this proposal based on the recommendations from the Disaster Risk Management Knowledge Centre that mentions that it is better to start performing a risk assessment and analysis than wait until better data become available (Poljanšek et al., 2017).

We would like to highlight, however that, based on the CO₂ degassing map, all the checked buildings resulted in adequate CO₂ susceptibility levels, which suggests the appropriateness of not only the criteria and levels defined but also the survey and methodological approaches used to elaborate the maps. Considering that CO₂ fluxes may highly variate at short distances (e.g., Chiodini et al., 1998; Chiodini et al., 2001; Cardellini et al., 2003), it is important that the grid design is detailed enough to account for the variability and well represents the anomalous areas. The study used as a basis to carry out the susceptibility maps (Viveiros et al., 2010) seems adequate for this purpose. In other degassing areas, and even if there is no information about the exposure and/or vulnerability, we suggest that the susceptibility maps can still be used as tools for land-use planners. We are aware of the low number of buildings used in the current study to validate the method, but as mentioned earlier, indoor measurements may be challenging to obtain. It will thus be important to test this methodology in several other degassing areas, and Colli Albani area or Vulcano (Italy), or even the recent degassing episodes at La Palma (Spain) are good candidates for testing.

Due to the characteristics of the CO₂, which is denser than air at STP, any underground structure (basements, pits, depressions, excavations, mines) increases the vulnerability, as previously discussed. Consequently, CO₂ risk may increase with the presence of any of these underground structures (vulnerability class set up as very high). In some cases, and as a consequence of the increased vulnerability, the high risk of CO₂ zones may be associated with moderate susceptibility levels. For this reason, we suggest that construction should be allowed only in the areas defined as low-risk zones. This is particularly relevant if one considers not only the significant increases in the CO₂ flux due to meteorological variations (Viveiros et al., 2009; Viveiros et al., 2015; Viveiros et al., 2016) but also that seismic events may cause sudden rises in the soil gas flux, as recently showed by Gresse et al. (2016).

Previous studies carried out in various degassing areas of the Azores (Viveiros et al., 2015) showed hazardous indoor CO₂ concentrations in areas identified as “moderate” risk zones (defined based on soil CO₂ concentrations >1.5 vol% and <5 vol%), highlighting the need to be restrictive in the authorizations to new constructions. Levels defined for the Azores in what concerns the soil CO₂ concentrations (high risk >5 vol%) are in agreement with an Italian regional law from 2012 (N. A00271 19/01/2012). This decree considered that only areas with soil CO₂ concentration below 1 vol% are considered suitable for construction, and areas above 5 vol% should be classified as non-building areas (Barberi et al., 2019 and references therein). In between these two CO₂ limits, those authors suggested the possibility of applying several mitigation actions. We suggest that similar laws are applied to any degassing area and that limits are set up based not only on the CO₂ concentrations but also on the soil CO₂ fluxes, which most of the time constitute the available gas data.

Land-use planning legislation must account both with the susceptibility maps as mentioned, but also consider the vulnerability of the buildings and, for this reason, “gas-resistant” construction rules need to be taken into consideration by civil engineers, architects, and any decision-making responsible.

Other hazardous gases, such as ²²²Rn and H₂S, with severe impact on human health (Durand and Scott, 2005; Kim et al., 2016; Kristbjornsdottir et al., 2016), are released from volcanic soils highlighting the need to produce exposure risk maps also for these volatiles. A resulting risk of exposure map for various gases should account for the integration of the individual gas species. For the particular case of Furnas Volcano, ²²²Rn studies have already been developed on Furnas Volcano (Silva et al., 2015a) showing that 87% of the buildings at Furnas and 99% at Ribeira Quente villages are over anomalous radon degassing. In what concerns H₂S, no detailed soil survey was performed, but some punctual soil H₂S concentration measurements were done along the main soil diffuse degassing areas showing that this gas is detected essentially near the fumarolic fields. Future studies also need to identify the thermal anomalous zones, since soil gases may show different behaviors associated with the temperature: when soil temperature is high, the CO₂ may migrate to the upper floors of the buildings due to the decrease in density, and this should be tested in future studies too.

For outdoor environments, we consider that the best approach is the application of gas dispersion models (e.g., Costa et al., 2008; Massaro et al., 2021; Massaro et al., 2022; Rave-Bonilla et al., 2023; Viveiros et al., 2023), which may identify, based on probabilistic evaluation areas, where CO₂ may accumulate in hazardous concentrations. However, Viveiros et al. (2023) recently showed that even if the dispersion models do not display hazardous outdoor CO₂ concentrations, some dead animals and high CO₂ concentrations were still measured. For this reason, we suggest that the final CO₂ risk map for diffuse degassing areas should intersect the methodology now applied with the dispersion models, essentially to identify potential hazardous areas that the dispersion models may miss due to the scale used.

Risk assessment is a complex task that accounts for several variables and criteria. As far as we know this study presents for the first time an approach to estimate the indoor deep-seated (volcanic/hydrothermal) CO₂ risk in diffuse degassing areas by combining susceptibility and exposure/vulnerability maps. The tests carried out at Furnas and Ribeira Quente villages (São Miguel, Azores) show that residents from the majority of buildings are at least at high risk of exposure to anomalous CO₂ concentrations Susceptibility classes were defined based on the CO₂ sources and the existence of diffuse degassing structures (DDS). In order to establish the susceptibility maps based on the degassing CO₂ values, it is important to carry out detailed surveys that result in a map as much closer as possible to the CO₂ distribution.

Vulnerability classes were performed based on human exposure in outdoor/indoor environments, considering the structure of buildings and the existence of “gas-resistant” codes. By combining the above-mentioned maps, a risk exposure to volcanic-hydrothermal CO₂ is estimated. The results need to be validated by performing indoor CO₂ measurements. In addition, this methodological approach should be tested in several other degassing areas that may affect edifices.

Some mitigation actions may be applied to reduce the risk of CO₂ exposure as mentioned before. Forbidden construction in anomalous degassing areas, usually with susceptibility higher than moderate, should be the more advisable recommendation. However considering the areas where construction already exists, as Furnas and Ribeira Quente villages, some mitigation strategies to reduce the vulnerability can be implemented. Dislodgement of the residents in the high and very high-risk zones may also be a solution difficult to apply due to economic, social, cultural, and political reasons. However, depending on the recorded indoor CO₂ concentrations, displacement may be mandatory. “Gas-resistant” codes should be implemented in buildings located in diffuse degassing zones to reduce the gas entry and/or to ventilate the ambient air. Some of these measures include the introduction of natural and/or artificial ventilation systems, such as the installation of under-floor ventilation systems or positive-pressure air-conditioning, for example, and the implementation of impermeable membranes on the ground floor to reduce gas ingress. Other mitigation actions include the installation of permanent real-time monitoring and alarm systems in the buildings located in high and very high-risk zones, as it is already implemented in Caldeiras da Ribeira Grande degassing area (Fogo Volcano, Azores) or Puerto Naos (Cumbre Vieja, Canaries).