Of all the volcanic calderas in the world the ∼ 12 km wide Campi Flegrei (CF) in southern Italy is arguably the one with the highest destructive potential, since it is in a state of unrest and located within an urban area of over 2 million residents, with Naples being the largest urban nucleus in the area (Fig. 1a). Its last eruption dates back to 1538 (Dvorak and Gasparini, 1991). Ever since, CF has undergone various series of new, rather swift uplifts (bradyseisms), indicating unrest followed by a decrease in ground level usually at a much slower rate (Chiodini et al., 2010; Troiano et al., 2011; D'Auria, 2015; De Natale et al., 2017). Since the last energetic unrest of 1982–1984, the caldera has been subject to intense geophysical and geochemical monitoring, with the greatest interest in the Solfatara crater, in the center of CF, and in the Pisciarelli area, on the eastern outer slope of Solfatara. Around 2005 a new net uplift, although at a relatively slow rate, commenced. At Pisciarelli, where the more recent low-energy seismic swarms are localized (D'Auria et al., 2011), the fumarole temperature increased from below 100 ∘C in 2005 to ∼ 115 ∘C in 2015. The amount of water vapor has increased visibly and the strongly degassing area has been considerably enlarged in the past few years (Chiodini et al., 2015). Given these major signs as well as other signs, mainly related to fluid geochemical variations at the fumaroles of Solfatara (Chiodini et al., 2015, 2016), national civil protection authorities have changed the state of CF from green (quiet) to yellow (scientific attention).

As all calderas, CF represents a complicated makeup that includes a magmatic plumbing system with a depth of up to ∼ 8 km (Bodnar et al., 2007; Zollo et al., 2008; Vitale et al. 2014; Moretti et al., 2017), feeding the overlying hydrothermal system (Chiodini et al., 2010; Troiano et al., 2011; De Siena et al., 2017a) through an intricate network of fractures (Zollo et al., 2008; De Siena et al., 2010; Byrdina et al., 2014). A clear picture of the feeding mechanisms and its dynamics is one of the central open questions regarding CF and is subject to ongoing debate. There is a broad consensus among researchers that injections of deep, hot, and oxidized fluids into the hydrothermal system of CF cause increased CO2 soil degassing (Cardellini et al., 2016), increased CO2 content in the fumarole discharges (with consequent decreasing trends of H2O / CO2 and H2S / CO2 ratios), and ground uplift (Caliro et al., 2007; Chiodini et al., 2012; Aiuppa et al., 2013). As a matter of fact, there is a fair correlation between soil / fumarole CO2 degassing strength and episodes of ground uplift (D'Auria et al., 2011; Chiodini et al., 2012, 2016) following this order: uplift and months later an increase in CO2 relative to other gases. There appear to be two logical main causes for this.

i.

An increase in the supply of fluids and associated thermal energy into the hydrothermal system for depressurization of the magmatic source (Allard et al., 1991; Chiodini et al., 2016). This increased supply is thought to stem from either the ∼ 8 km deep main magma reservoir (Bodnar et al., 2007; Zollo et al., 2008; Moretti et al., 2017) or from a contribution of a magma batch that intruded into the shallow subsurface (∼ 3–4 km depth) concomitantly with the 1982–1984 unrest episode (Chiodini et al., 2010; Caliro et al., 2014) and is periodically rejuvenated by arrivals of deep, more primitive magma (Bagagli et al., 2017).

Discriminating between these mechanisms is beyond the scope of this study, but any insights towards a better understanding of these processes are important to improve early warning and civil protection measures in the CF area. Measuring emission rates (fluxes) of CO2 provides an additional way to assess the hazard at CF. The fumarole area of Piscarelli, approximately located in the center of the CF caldera (Fig. 1a) and recently the scene of drastic changes in its activity, is a prime geochemical sampling spot to learn about the volcanic processes taking place beneath CF. A spatially integrated measurement of CO2 flux that accounts for all possible CO2 vents and diffuse degassing is desirable to obtain a quantitative picture of CO2 degassing, but it has only been done a few times after 2012 at Pisciarelli (Pedone et al., 2014; Aiuppa et al., 2015; Queißer et al., 2016a). To increase the number of observations it was decided to revisit CF 14 months after the last such measurement (Queißer et al., 2016a) and remeasure CO2 fluxes.

The CO2 concentrations needed to estimate the CO2 flux are commonly sampled at individual points, which may miss out particular sources, such as smaller fumarolic discharges (Chiodini et al., 2015). On the other hand, point measurements are very precise and valuable in characterizing local degassing elements, such as fractures. Path-integrating, scanning gas measurement techniques, on the other hand, may add value by providing a spatially comprehensive measurement. To attempt a spatially inclusive measurement of all possible sources of CO2, diffusive soil, and vented degassing, we used a laser remote-sensing spectrometer (LARSS), developed in the ERC proof-of-concept project CarbSens. Combined with point measurement techniques, such as accumulation chambers, LARSS may help to yield a more complete picture of degassing. It represents a further miniaturization of a similar system developed in the ERC project CO2VOLC. The instrument and its working principle are detailed elsewhere (Queißer et al., 2017). Only a brief overview is therefore given. LARSS consists of a main unit and a transmitter–receiver unit (TX/RX unit; Fig. 1b). The latter comprises the telescope, transmitter, and an integrating sphere for power reference measurement. It is portable (weight: 10 kg main unit + 6 kg TX/RX unit), which allows it to be transported easily and set up on any kind of surface, such as house roofs or airplanes.

The CO2 absorption line at 1572.335 nm (R16 transition) is sampled at 40 wavelengths by sweeping the emission wavelength of a diode laser. The laser light is amplified, transmitted, backscattered at a topographic target, and received by the telescope. After the detected signal is digitized, the optical transmittance of the telescope's viewing path is deduced for each of the 40 wavelengths. A model absorption spectrum is fitted to the 40 measured transmittances, resulting in a best estimate of the path-averaged CO2 column density (in m-2). The path length may be up to 2 km. Profiles of CO2 concentrations, i.e., CO2 concentrations versus angle, are attained by scanning the TX/RX unit across a degassing plume (see Queißer et al., 2016a, for details on scanning geometry). Along with the plume transport speed these profiles are then used to obtain CO2 fluxes, following ΦCO2=uMCO2NAΔβ∑plumeriNplcol(ri), where u refers to the component of the plume transport speed perpendicular to the plane of the CO2 concentration profile, i.e., the component perpendicular to the plane of the scan. MCO2 is the molar mass of CO2 (in kg mol-1) and NA is Avogadro's constant (in mol-1). Δβ is the constant scan angle increment. Nplcol, the background-corrected, or in-plume column density of CO2, is retrieved by subtracting the ambient CO2 column density measured outside the plume from the total CO2 column density, i.e., Nplcol(ri)=Ncol(ri)-Nbgcol(ri), where Ncol(ri) is the total column density as measured and Nbgcol(ri) represents the ambient column density. The ranges ri are measured with a range finder lidar aligned with the telescope. For convenience and display purposes, if meteorological data are available, column densities may be converted to path-averaged mixing ratios (in ppm) as detailed in Queißer et al. (2017). The plume speed is retrieved by digital video tracking of the plume of condensed water vapor as described in Queißer et al. (2016a).

The data presented here are a subset of data acquired during a campaign probing CO2 at the Pisciarelli-Solfatara area between 24 and 26 May 2017. LARSS was placed on the roof of the Tennis Hotel, located ∼ 320 m east of the Pisciarelli fumaroles and offering an unobstructed view of the complete fumarole degassing activity. Between 17:07 and 18:04 local time on the 24 May 2017, nine lateral angular scans were performed, of which six are displayed in Fig. 2. A step motor rotated the TX/RX unit between 257.4 and 243.4∘ (Fig. 1b) with a velocity of 2.5 mrads-1, corresponding to a lateral section of ∼ 80 cm at the fumarole area per data point. Δβ was retrieved by multiplying the scanning angular speed by the time between subsequent measurements as recorded in the time stamps of the raw data files. Each scan took around 90 s. Meteorological data (temperature, pressure, humidity) were recorded using a Kestrel portable meteorological station placed next to LARSS. Nbgcol was measured by a scan upwind, outside of any volcanic plume, using a hill range between 700 and 900 m distance as a target. The corresponding column-averaged CO2 mixing ratio was found to be 499 ppm. For comparison, two in situ measurements with a LI-COR analyzer were performed at points near the optical paths of LARSS, yielding CO2 mixing ratios of 550 and 560 ppm. These are remarkably high CO2 concentrations, given that the wind came from the sea (south). The proximity of the measurement points to the road and the dense network of roads in that area may well cause these values (Schmidt et al., 2014). Consequently, Nbgcol corresponding to an ambient CO2 mixing ratio of 499 ppm ± 61 ppm was considered.

The highest CO2 concentrations were usually detected near the center of the probed area (near 250∘; Fig. 2b to g). This main plume reveals a fine structure, suggesting three sub-peaks, which could be related to three main vents in very close proximity to each other identified by Pedone et al. (2014). The highest column-averaged CO2 mixing ratio measured was 1777 ppm (Fig. 2e), which is, however, associated with a relatively large uncertainty of 236 ppm (1 SD). Note that Pedone et al. (2014) measured a peak value of 1444 ppm in early 2013 at approximately the same location. Elevated concentrations also occurred towards the southern edge of the probed area (∼ 21 m south of the main plume), at the slope. The corresponding peak repeatedly arose near 246∘ (especially Fig. 2b, c, d, and g).

Uncertainties in path-averaged CO2 mixing ratios were usually between 2 % and 5 % or 10 to 30 ppm (associated with a path-averaged detection limit of ∼ 10 000 ppm m). The main source of uncertainty was the contribution of the instrument itself (baseline drift) and the fitting error. The latter had been significantly improved (roughly halved) by recently increasing the number of sampled wavelengths from 20 to 40. A detailed description of the influences of various error sources is provided in Queißer et al. (2017). The in-plume CO2 concentrations found were mostly between 500 and 4000 ppm, with peaks around 6000 ppm, and they agree well with those measured by the fixed in situ station (Fig. 2b–g). In-plume concentrations had associated uncertainties which were naturally larger than those of the column-averaged values, that is, typically between 4 and 15 % or around 150 ppm. Local wind eddies may lead to local maxima of CO2 concentrations and may also explain the shift in the global concentration maximum after Fig. 2d, suggesting a generally “wobbly” character of the CO2 plume.

The measured vertical plume speed component was 0.65 m s-1 (min 0.28 m s-1, max 1.05 m s-1) until 17:43:47 and 0.80 m s-1 (min 0.31 m s-1, max 1.37 m s-1) after that. The plume speed uncertainties were calculated from the Student t variance as detailed in Queißer et al. (2016a). Given the complex terrain and the fact that the measurement was performed close to the ground, the velocity field across the scanned plume was generally not constant, in addition to temperature variations causing different speeds across the plume. The corresponding variability has been accounted for by tracking different paths of propagating water vapor across the plume and using the variability in the error estimation. Plume speed is in fact one of the main sources of uncertainty, adding an uncertainty of the order of 30 % to the flux.

Table 1 shows the flux values computed using Eq. (1), with a mean value of 6.7 ± 2.9 kg s-1 (578 ± 246 t d-1). As noted in previous measurements at Pisciarelli, the measured fluxes fluctuate by over 100 % over the course of minutes (Aiuppa et al., 2015; Queißer et al., 2016a). However, an observational window of 1 h reflected the same variability as an 8 h long window (Aiuppa et al., 2015). The rigorous error assessment, i.e., taking all relevant error sources into account, including conservative systematic error estimates, led to a rather high uncertainty in the flux values. The conservatively chosen uncertainty in the ambient CO2 concentration, an order of magnitude higher than usual, accounts for between 20 % and 70 % (depending on the profile) of the flux uncertainties presented in Table 1. The other chief source of flux uncertainty is the plume speed, which, depending on the scan, caused an increase in error by the same magnitude.

The mean flux of 6.7 kg s-1 corresponds to the complete extension of the scan, that is, the vegetation-free zone of ∼ 70 m in lateral diameter (Fig. 1c). When integrating over the central area only (between 252.5 and 247.0∘), roughly including the aforementioned three major vents, the mean flux obtained is 284 ± 107 td-1 and is compatible with the estimated area-integrated value from the in situ automated flux measurement station FLXOV3 (Fig. 3). This may explain the offset between the fluxes of this work and FLXOV3. Focusing on the main vent area only, however, neglects persistent degassing features, such as at the southern edge of the fumarole area, as well as diffusive soil degassing taking place within the scanned sector (Caliro et al., 2007). This spatially comprehensive character of the measurement is one of the main merits of the remote-sensing technique applied here.

The soil CO2 flux at CF is known to have increased in magnitude and spatial extension since 2005 (Cardellini et al., 2016) as has the CO2 content in CF high-temperature fumaroles (Chiodini et al., 2010; 2016). Figure 3 suggests a slight acceleration in CO2 degassing from the soils of Pisciarelli since about 2009 (FLXOV3 series) confirmed by post-2012 CO2 measurements integrated over the whole exhaling area, which coincides fairly well with the observed acceleration in ground uplift. The similarity between the uplift and degassing trends suggests that both processes are intrinsically related. In fact, the preferential exsolution of CO2 from the deep magmatic body due to its low solubility at high pressure implies an associated release of H2O simultaneously to CO2 output (Chiodini et al., 2001) or when CO2 is completely exhausted in the magma (Chiodini et al., 2016). In any case, the participation of H2O in the degassing process results in a very efficient mechanism to convey heat from depth to the hydrothermal system and the overlying rocks, favoring thermally induced dilation (ground deformation) and enhancing the permeability of fluids flowing through them (greater degassing at the surface). Recent findings indeed point towards an impulsive influx of hot magmatic fluids into the hydrothermal system as a possible source mechanism at CF that eventually caused the observed geophysical and geochemical time series, including the present one (Chiodini et al., 2017).

The inspection of Fig. 3 confirms this general scheme although the CO2 fluxes measured at Pisciarelli in May 2017 (this study) and in March 2016 (Queißer et al., 2016a) seem to suggest an increase at a larger rate than the observed uplift would imply. In particular, the latest available data, up to April 2017, suggest a deceleration of ground uplift at CF as of 2016 (Fig. 3 and in more detail INGV, 2017), which, as far as the resolution of our data permits us to say, is not accompanied by a leveling out of degassing strength.

Our results related to the CO2 degassing are compatible with findings which state that the elastic rock matrix of CF is transitioning to inelastic behavior under long-term stress accumulation, accompanied by a permeability increase in the shallow crust, disguising any direct indicator of unrest, such as rapid ground uplift or enhanced seismicity (Bodnar et al., 2007; Di Luccio et al., 2015; Kilburn et al., 2017). In line with this prospect is a clear seismic velocity decrease since 2012 (Zaccarelli and Bianco, 2017), which could be due to, for instance, a softening bulk or increase in CO2 saturation in the CF aquifer (Queißer and Singh, 2012) that may also explain the strong seismic attenuation observed (De Siena et al., 2017b).

The aforesaid could justify the discrepancy recently highlighted by Moretti et al. (2017) between weak geophysical signals (moderate uplift and low seismicity) and drastic changes in geochemical indicators characterizing the present stage of the CF history.

Finally, it should be mentioned that heterogeneity in the subsoil (Montanaro et al., 2016) and dynamic alterations in subsoil rock matrix properties such as those due to seismic energy (Gresse et al., 2016) may modulate the emission of stored gas and therefore cause changes in degassing strength.

About 14 months after the last survey, we revisited the Pisciarelli area. The current CO2 flux was quantified using the portable remote-sensing spectrometer LARSS, which detects CO2 in a spatially integrated manner. Although associated with a fairly conservative uncertainty, the result, along with fluxes measured in 2016, implies an increase in CO2 flux in the last 2 years. Drawing solid conclusions based on our data is not possible. Nonetheless, given the slow, almost halted ground uplift since 2016, our result could indicate a release of deep magmatic gases towards the hydrothermal system, possibly accompanied by an increased bulk permeability of the shallow crust.

Our measurements, although reasonable, do not permit an unequivocal conclusion whether the origin of the gas emitted at the surface is purely hydrothermal or magmatic nor regarding the migration mechanisms from the bottom to the top of the CF plumbing system. Nevertheless, the spatially comprehensive values of CO2 flux acquired through LARSS may help constrain the degassing process as a whole and then provide clues about the strength of the CO2 source, for example via mass balance considerations (Allard et al., 1991), possibly adding to geochemical appraisals (Moretti et al., 2013). However, more measurements of this kind are needed (higher temporal resolution). Furthermore, point measurements should be added in the future to systematically test and verify the capability of LARSS to probe all degassing elements in its path comprehensively. For challenging degassing situations as at CF, integrating LARSS with point measurements may provide a powerful means to obtain a complete picture of degassing. Point measurements are able to draw detailed maps of the emission areas which LARSS is not capable of. However, using two instruments 2-D tomography can be performed (Queißer et al., 2016b), although much more improvement of this technique is needed to converge point measurements to degassing maps. Moreover, there is potential to further reduce uncertainty in the measured fluxes. To that end, the plume speed estimation will be further improved, especially with respect to resolving the plume speed variations (velocity field) across the scanned plume.