- © 2013 Mineralogical Society of America
INTRODUCTION: VOLCANIC CO2 EMISSIONS IN THE GEOLOGICAL CARBON CYCLE
Over long periods of time (~Ma), we may consider the oceans, atmosphere and biosphere as a single exospheric reservoir for CO2. The geological carbon cycle describes the inputs to this exosphere from mantle degassing, metamorphism of subducted carbonates and outputs from weathering of aluminosilicate rocks (Walker et al. 1981). A feedback mechanism relates the weathering rate with the amount of CO2 in the atmosphere via the greenhouse effect (e.g., Wang et al. 1976). An increase in atmospheric CO2 concentrations induces higher temperatures, leading to higher rates of weathering, which draw down atmospheric CO2 concentrations (Berner 1991). Atmospheric CO2 concentrations are therefore stabilized over long timescales by this feedback mechanism (Zeebe and Caldeira 2008). This process may have played a role (Feulner et al. 2012) in stabilizing temperatures on Earth while solar radiation steadily increased due to stellar evolution (Bahcall et al. 2001). In this context the role of CO2 degassing from the Earth is clearly fundamental to the stability of the climate, and therefore to life on Earth. Notwithstanding this importance, the flux of CO2 from the Earth is poorly constrained. The uncertainty in our knowledge of this critical input into the geological carbon cycle led Berner and Lagasa (1989) to state that it is the most vexing problem facing us in understanding that cycle.
Notwithstanding the uncertainties in our understanding of CO2 degassing from Earth, it is clear that these natural emissions were recently dwarfed by anthropogenic emissions, which have rapidly increased since industrialization began on a large scale in the 18th century, leading to a rapid increase in atmospheric CO2 concentrations. While atmospheric CO2 concentrations have varied between 190–280 ppm for the last 400,000 years (Zeebe and Caldeira 2008), human activity has produced a remarkable increase in CO2 abundance, particularly in the last 100 years, with concentrations reaching ~390 ppmv at the time of writing. This situation highlights the importance of understanding the natural carbon cycle, so that we may better determine the evolution of the anthropogenic perturbation.
The principle elements of the multifaceted and complex geological carbon cycle are summarized in Figure 1. The main sources of carbon are active and inactive volcanism from arcs and rift zones and metamorphism of crustal carbonates. The main sinks for geological carbon are silicate weathering and carbonation of oceanic crust. Knowledge of both the total magnitude of carbon ingassing during subduction and carbon released from volcanism and metamorphism would allow quantification of the evolution and relative distribution of volatiles in the crust and mantle (Dasgupta and Hirschmann 2010).
The main focus of this work is the role of volcanism in producing CO2 in the atmosphere and oceans. Volcanic CO2 sources can be divided into several categories, direct and diffuse degassing from active arc and rift volcanoes, diffuse degassing from inactive volcanoes and regional diffuse degassing from intrusive plutonic structures with associated crustal metamorphism. We focus here on non-eruptive degassing because, as shown below, continuous emission of CO2 from multiple sources appears to dominate short-lived eruptive emissions from point sources.
CO2 released directly from active volcanoes has three main sources, CO2 dissolved in the mantle, recycled CO2 from subducted crustal material (e.g., Marty and Tolstikhin 1998) and decarbonation of shallow crustal material (e.g., Troll et al. 2012). Separating the relative proportions of mantle and crustal carbon is possible through investigation of the isotopic composition of emitted carbon (e.g., Chiodini et al. 2011) and is increasingly important given that during eruptions magmatic intrusions may interact with crustal material, strongly enhancing the CO2 output of the volcanic system (Troll et al. 2012), at least temporarily. The magnitude of diffuse mantle CO2 can also be identified isotopically in mixed metamorphic and magmatic gases using Carbon (Chiodini et al. 2011) or Helium isotopes as a proxy for deep mantle sources in both major fault systems (Pili et al. 2011) and crustal tectonic structures (Crossey et al. 2009).
Our current estimates of volcanic carbon emissions are poorly constrained due to a lack of direct measurements. Measuring CO2 in subaerial volcanic plumes is a challenge, because while CO2 typically makes up ~10 mol% of volcanic gas emissions (the majority of which is normally water vapor), mixing with the atmosphere rapidly dilutes the volcanic CO2 signature. Nevertheless, technological advances and an increase in the number of volcanoes studied have greatly increased our knowledge of volcanic CO2 fluxes over the last 10 years. One of the main goals of this work is to update global geological CO2 flux estimates (e.g., Kerrick 2001; Mörner and Etiope 2002; Fischer 2008) using the recently acquired volcanic CO2 flux data.
Diffuse CO2 degassing from both volcanic and tectonic structures is a large contributor to the global geological CO2 emission, but is difficult to measure due to the large areal extent that may be in play, and the large number of degassing sites throughout the globe. Measuring the CO2 degassing rates into volcanic lakes and from submarine volcanism have significant technical challenges. In the following we review the state of the art of volcanic CO2 measurements and present a catalogue of reported, quantified emissions from geological sources. These measurements are then extrapolated to produce estimates of the global volcanic CO2 flux. These estimates are compared with previously published estimates of total CO2 emissions, silicate weathering rates and the rate of carbon consumption during subduction. We then examine the dynamic role of CO2 within magmatic systems and the magnitude of CO2 released during eruptions.
Carbon species in Earth degassing
CO2 is not the only carbon-containing molecule emitted from the Earth. In order of decreasing emissions, CO2, CH4, CO and OCS all contribute to the total carbon budget. Mörner and Etiope (2002) estimated that the global emission of CO2 from Earth degassing was ~600 million tonnes of CO2 per year (Mt/yr, 1 Mt = 1012 g), with ~300 Mt/yr produced from subaerial volcanism, and another 300 Mt/yr produced from non-volcanic inorganic degassing, mostly from tectonically active areas (Chiodini et al. 2005). For comparison, Cadle (1980) estimated that volcanic activity produces 0.34 Mt/yr of CH4. Mud volcanoes in Azerbijan were estimated to produce ~1 Mt/yr of CH4, however the global flux from mud volcanism is not known. Hydrocarbon seepage of CH4 globally is estimated to produce between 8 and 68 Mt/yr (Hornafius et al. 1999). Etiope et al. (2008) estimated that global CH4 emissions from geological sources to be 53 Mt/yr, a significant proportion of the geological C output.
CO is emitted directly from volcanoes, with a CO2/CO ratio that varies between ~10 and ~1000 depending on the oxygen fugacity and temperature of the fluid co-existing with melt prior to outgassing. OCS is also directly emitted within volcanic plumes, but in even smaller relative amounts than CO, typically 1000–10,000:1 for CO2:OCS (Mori and Notsu 1997; Burton et al. 2007a; Oppenheimer and Kyle 2008; Sawyer et al. 2008a). OCS is the most abundant S bearing gas species in the atmosphere, contributes to stratospheric sulfuric acid aerosol generation (Crutzen 1976) and is an efficient greenhouse gas (Brühl et al. 2012). Its budget is dominated by emissions from oceans and anthropogenic processes. From a total global output of ~1.3 Mt/yr of OCS only 0.03 Mt is estimated to arise from volcanism (Watts 2000). CS2 is the final trace carbon gas emission from volcanoes, with a similar flux and chemistry to OCS.
While recent estimates of geological emissions of CH4 (Etiope et al. 2008) clearly indicate that these emissions are significant compared with geological CO2 on the global scale, in this work we focus on CO2 emissions, and use the most recent volcanic CO2 flux data to update the measured global volcanic CO2 flux.
METHODS FOR MEASURING GEOLOGICAL CO2 EFFLUX
Ground-based measurements of volcanic plumes
Directly quantifying volcanic CO2 fluxes in the atmosphere is challenging due to the relatively abundant concentration of background CO2, currently ~390 ppm. To put this in context, a strong volcanic CO2 source such as Mt. Etna, Italy, produces a gas plume where 1 km downwind the average concentration of volcanic CO2 is only ~4 ppm (based on calculations with VOLCALPUFF plume dispersal model, Barsotti et al. 2008). Thus, an in situ measurement 1 km downwind needs to resolve a mere 1% excess CO2 compared with the background concentration. Most volcanoes emit less CO2 than Mt. Etna, so this is an optimistic scenario. This difficulty has led most researchers to focus on measurements of the volcanic emissions close to the source, using in situ and proximal remote sensing techniques. In such measurements, volcanic CO2 fluxes are normally determined by measuring the ratio of volcanic CO2 to another volcanic gas, typically SO2 due to the ease with which its flux can be measured (Oppenheimer et al. 2011), and then calculating the CO2 flux as the product of the CO2/SO2 ratio and the SO2 flux. The objective of the majority of the following techniques in the context of quantifying CO2 fluxes is therefore the determination of the CO2/SO2 ratio in the volcanic gas.
An exception to this combined CO2/SO2 and SO2 flux approach was adopted by Marty and Le Cloarec (1992) who utilized global volcanic flux estimates of 210Po and 210Po/CO2 ratios measured in direct sampling (see below) to derive estimates of global CO2 fluxes.
Direct sampling of a volcanic gas can be achieved with the use of Giggenbach bottles (Giggenbach and Goguel 1989), where high temperature fumarolic gases are collected in an alkaline solution for later laboratory analysis. This approach allows both bulk and trace gas species to be quantified, but the fact that the most abundant gas component, H2O, can condense in the tube leading to the alkali solution bottle means that H2O is challenging to quantify, and therefore absolute concentrations of the other species can be difficult to define. Air contamination is difficult to avoid, and can further increase the difficulty in determining the original volcanic gas concentrations. Such measurements require working in extremely close proximity to the degassing vent, and are ideally performed on the hottest and most highly pressurized emissions (to avoid air contamination), making their collection challenging and potentially hazardous. In addition, some of the most voluminous volcanic gas sources release very little volatiles from fumaroles, instead the bulk of the emission is open-vent degassing from craters. These plume emissions are impossible to sample without air contamination with such an approach.
The MultiGas approach (Shinohara 2005) has greatly simplified the measurement of volcanic CO2/SO2 ratios, allowing automatic, unattended analysis of volcanic plumes for extended periods of time (Aiuppa et al. 2007). This instrument combines a near-infrared spectroscopic measurement of CO2 and H2O concentrations with a solid-state chemical sensor for quantification of SO2 and H2S. The relatively low cost, low power requirement and ease of use of the instrument make it probably the most convenient and cost-effective way of determining in situ CO2/SO2 ratios available today. Some potential errors can arise, however, due to the different response time of the CO2 and SO2 sensors. Typical response times for the near-infrared optical technique used to measure CO2 is ~1 s, while response times for SO2 chemical sensors is typically longer, e.g., ~13–31 s for an Alphasense chemical sensor (Roberts et al. 2012). This means that fast changes in chemical composition or concentration are challenging to capture, however through the use of longer integration times problems arising from diverse sensor response times can be avoided. Quickly changing gas concentrations could instead be captured in theory with an optically based SO2 measurement.
Remote sensing measurements of CO2 amounts can be performed with infrared spectroscopy if the volcanic gas concentration is sufficiently high above the background atmospheric CO2 amount. The first such pioneering infrared measurements of volcanic CO2 amounts were conducted remarkably early, in 1969 by Naughton et al. (1969), during a lava fountain on Kīlauea, Hawai’i. Since then, open-path Fourier transform infrared (OP-FTIR) spectroscopy (Mori and Notsu 1997) has become a well-utilized tool to measure in situ volcanic gas compositions. Modern FTIR spectrometers are light (~8 kg), relatively low power (~30 W) and require no cryogenic cooling (e.g., La Spina et al. 2010), yet allow the simultaneous measurement of many volcanic gases, including H2O, CO2, SO2, HCl, HF, CO OCS and SiF4. Measurements of volcanic CO2 require an infrared light source, either an infrared lamp (Burton et al. 2000) or hot volcanic rocks (Allard et al. 2005; Burton et al. 2007a; Sawyer et al. 2008a). Infrared radiation is absorbed by volcanic gases before being measured with the infrared spectrometer and the resulting spectra can be analyzed (Burton et al. 2000, 2007a) to produce relative abundances of volcanic H2O, CO2, SO2 and several other gases. Smith et al (2011) concluded that such analytical approaches could deliver accuracies of ~5% in CO2 amounts. The greatest challenge in performing OP-FTIR measurements is obtaining a suitable source of IR radiation, with sufficient volcanic gas between the source and the spectrometer. Recent innovations in applying the remote-controlled mode at the summit of Stromboli (La Spina et al. 2013) show that it can be used for high temporal resolution monitoring of multiple gas sources.
CO2 emissions from fumarole fields can be determined through the introduction of a known flux of a tracer gas, such as SF6, and then measuring the volcanic CO2/SF6 ratio in the downwind gas emission. Mori et al. (2001) successfully used this approach to measure CO2 flux emissions from fumarolic vents on Izu-Oshima (Japan), Kirishima (Japan) and Teide (Canary Islands, Spain).
A recent innovation has been the use of portable mass spectrometers to determine in situ volcanic gas compositions (Diaz et al. 2010). This approach was used successfully before and after the 5th January 2010 eruption of Turrialba (Costa Rica), revealing significant changes in CO2, SO2 and He concentrations.
Volcanic SO2 flux measurements
As described above the determination of CO2 flux requires a further step after measurement of CO2/SO2 ratios, multiplication with an SO2 flux. Several assessments of arc volcanic CO2 emissions have been produced (e.g., Hilton et al. 2002; Fischer and Marty 2005) by interpreting volcanic SO2 inventories (e.g., Andres and Kasgnoc 1998; Halmer et al. 2002) Since errors on the SO2 flux propagate into the CO2 flux we briefly examine here the methods used to measure SO2 flux, together with their associated errors. The SO2 flux is much easier to measure directly than the CO2 flux for two reasons: firstly, SO2 is not present in the unpolluted troposphere and secondly, SO2 has a convenient, relatively strong absorption band in the ultraviolet, easily accessible using scattered sunlight as a source. This has allowed the creation of automatic networks of UV scanners, that permit volcanic SO2 fluxes to be monitored (Edmonds et al. 2003; Burton et al. 2009; Galle et al. 2010; Oppenheimer et al. 2011). Typically in the literature the greatest quoted source of error in SO2 flux measurements derives from wind speed estimates, and this error is normally indicated to be ~20–30%. A recent innovation in ground-based measurements of SO2 fluxes is the SO2 imaging camera, which uses an imaging sensor sensitive to the UV and optical filters to produce specific sensitivity to SO2 (e.g., Mori and Burton 2006). This approach has the potential to correct implicitly for wind velocity but it suffers potentially from cross-talk between SO2 and volcanic ash or aerosol.
Recent work on subtle radiative transfer issues relating to ground-based UV SO2 flux measurements highlight that there could be large, previously ignored, errors associated with light dilution (Kern et al. 2012). This is a process where light scattering from below the volcanic plume enters the instrument, diluting the light which passed through the volcanic plume from above, resulting in a net underestimation in the SO2 flux, of up to 90%. The true significance and importance of this effect, and the number of published SO2 fluxes that require re-analysis, has yet to be evaluated. It should be noted therefore that estimates of global CO2 flux based on published SO2 flux data may be subject to revision, depending on the impact of light dilution on SO2 flux measurements.
In addition to ground-based measurements of SO2 flux, satellite-based measurements are often used, working in both the ultraviolet (e.g., OMI, SCIAMACHY) and infrared (e.g., MODIS, ASTER, IASI) wavelengths. These instruments can produce maps of SO2 abundance with a repeat time of ~days. With such a repeat rate the same plume is normally not observed in two different images, and it is therefore not possible to perform a simple cross-correlation to determine plume velocity and therefore flux from images of SO2 abundance. The conversion from SO2 abundance image to a quantitative degassing rate is therefore not trivial, because both the age and velocity of the plume at each point in the image must be derived using an independent method (e.g., Merucci et al. 2011). An additional challenge with satellite-based measurements of SO2 flux is that the sensitivity to SO2 decreases in the lower atmosphere, such that low-lying volcanoes are difficult to measure unless they are in eruption, and even higher altitude volcanoes require a relatively large SO2 degassing rate to be reliably quantified (Carn et al. in press).
Measurements of CO2 flux using the combination of CO2/SO2 ratios and SO2 fluxes therefore reflect uncertainties in both measurements, typically estimated to be 10–20% and 25–30% respectively. As mentioned above, recent work in radiative transfer analysis highlights the potentially important, but largely overlooked, role that light dilution may play in producing potentially significant underestimates of SO2 fluxes from UV measurements (Kern et al. 2012). Notwithstanding these drawbacks, at the current time these data are the main constraints available for subaerial deep CO2 output from volcanoes.
Recent technological advances have allowed such integrated CO2/SO2, SO2 flux measurements to be fully automated for the first time, allowing real-time monitoring of CO2 fluxes. This is particularly important because the low solubility of CO2 in magmas means that deep, pre-eruptive, magmatic intrusions may be heralded at the surface by increases in CO2 flux. Aiuppa et al. (2011) reported the first time series of CO2 flux collected using combined MultiGas and SO2 flux networks (Burton et al. 2009) on Stromboli volcano. These revealed distinct oscillations in CO2 emissions, with periods of relatively high CO2 degassing followed by periods of low CO2 degassing, producing a steady average CO2 degassing rate of ~550 t/d. Such a pattern suggests a steady state supply of CO2 which is modulated by gas accumulation/permeability/magma supply processes within the magma feeding system. Interestingly, more intense explosive activity was observed after a period of intense CO2 degassing, opening the possibility of using such observations to forecast explosive volcanic activity at this volcano.
Airborne measurements of volcanic plumes
Two main approaches have been used to measure volcanic CO2 fluxes from the air. A direct method consists of flying an in situ CO2 analyzer (typically a closed-path near-infrared spectrometer) in a raster or ladder traverse across the cross-section of a volcanic plume (Gerlach et al. 1997; Werner et al. 2008). The resulting data can be interpolated to produce a CO2 concentration map, which can be integrated over the cross-sectional area of the plume and multiplied with wind speed to produce a CO2 flux. This approach has the advantage of being a direct measurement of the CO2 emissions, however each measurement requires ~1 hour and therefore very stable wind conditions are required in order to avoid errors in the flux calculation.
The second airborne approach combines plume traverse measurements with an ultraviolet spectrometer to derive SO2 flux with in situ measurement of the CO2/SO2 ratio provided with a closed-path FTIR spectrometer, sampling ambient air as the plane flies through the plume (Gerlach et al. 1998). This is probably the most robust methodology currently available for measuring CO2 fluxes, as the SO2 flux analysis can be performed close enough to the plume that light dilution is insignificant. Flying has its own challenges, however, due to the technical constraints involved in performing measurements on a vibrating platform, as well as the costs associated with flight time and difficulties presented from flying within the volcanic plume.
Airborne measurements of volcanic CO2 may also be achieved by viewing infrared radiance from the ground through a volcanic plume with a hyperspectral radiometer, and this has been successfully demonstrated on Kīlauea, Hawai’i using the AVIRIS hyperspectral imager (Spinetti et al. 2008). Results obtained with AVIRIS agreed well with ground-based measurements of CO2 emissions, validating the method.
More recently, similar measurements to those performed by Gerlach et al. (1998) have been conducted from an unmanned aerial vehicle platform (UAV) (McGonigle et al. 2008). Such an approach is appealing due to the significantly reduced cost and risk, as well as increased accessibility. However, the legal framework for conducting such measurements is complex and varies greatly between countries, making general take-up of such methodologies so far quite limited. Future longer distance stand-off measurements may be possible using larger UAVs offering an intermediate option to space-based satellite retrievals of air column soundings.
Space-based measurements of volcanic plumes
Global measurements of volcanic CO2 emissions would be the ideal approach to quantifying the global volcanic deep CO2 budget. The most promising observing platform for volcanic CO2 is the aptly named Orbiting Carbon Observatory (OCO) (Crisp et al. 2004). The rocket carrying the OCO failed to reach orbit when launched in 2009, and a new launch with a replacement satellite is currently planned for 2014. The OCO utilizes a single telescope to feed light to spectrometers which measure the columnar abundance of both O2 and CO2, using absorption bands at wavelengths of 0.67 micron for O2, and 1.61 and 2.06 micron for CO2. The purpose of the O2 column amount measurement is to normalize the CO2 column measurement to an average CO2 mole fraction in ppmv. The final error on the average CO2 concentration in the column is 0.3 wt%. The footprint of the OCO will be 1 km by 1.5 km at nadir, and will have a repeat observation period of 16 days.
A simple calculation of the CO2 emission from a strong emitter such as Mt. Etna allows a direct estimation of the feasibility of such measurements with OCO. Etna degasses CO2 at an average rate of 16,000 t/d or ~190 kg s−1 (Allard et al. 1991; Aiuppa et al. 2006, 2008; La Spina et al. 2010). Assuming an optimal geometry in which the gas source was at one edge of a single OCO footprint and the entire plume contained within the 1.5 km length of the pixel, with a windspeed of 5 m s−1 the maximum age of CO2 in the footprint would be ~300 seconds, and the total volcanic CO2 mass would be ~57,000 tonnes. Converting this CO2 mass to molecules and averaging over the OCO nadir footprint area produces a vertical column amount of ~5 × 1019 molecules cm−2. The atmospheric vertical column amount of CO2 at the average altitude of Etna is ~6 × 1021 molecules cm−2 and therefore the volcanic signal would be ~0.8 % of the atmospheric column, which is above the 0.3% error limit of OCO. A slower wind would produce a higher volcanic CO2 amount, while a less optimal geometry would decrease the relative contribution from the volcano. OCO therefore has the potential for measuring passive CO2 emissions from Mt. Etna, in optimal conditions. During eruptions the CO2 emission rate would increase, allowing for easier detection. The majority of degassing volcanoes are less productive than Mt. Etna, however, and would present a challenge for detection from OCO, unless they were undergoing an eruption.
Ground-based measurements of diffuse deep CO2
Significant amounts of diffuse CO2 are released from active volcanic areas, not only during eruptions but also during quiescent periods. This volcanic CO2 discharge occurs over the flanks of the volcanic edifice as diffuse soil emanations (Allard et al. 1991; Baubron et al. 1990), and adds to the voluminous and more obvious degassing from fumaroles and summit craters. Many CO2 soil flux measurement techniques have been applied to quantify these gases and include both direct and indirect methods (e.g., Reiners 1968; Kucera and Kirkham 1971; Kanemasu et al. 1974; Parkinson 1981).
The indirect methods are based on the determination of the CO2 concentration gradient in the soil (Camarda et al. 2006). These methods can be applied only if the transport of the gas is dominated by the diffusion and some properties of the medium are known. Direct methods require dynamic or static procedures, whether or not a flux of air is used to extract gas from the soil. The dynamic procedures require some corrections depending on the physical properties of the soil in the measurement point and on the design of the instrumental apparatus. Furthermore, all dynamic procedures are affected by overpressurization or depressurization depending upon the magnitude of the air flux chosen by the operator and according to Kanemasu et al. (1974) results are strongly affected by the physical modifications induced by pumping under different flux regimes. Camarda et al. (2006) present a demonstration of the indirect method applied to measuring CO2 fluxes from Vulcano (Italy), and show that with low pumping rates the sensitivity of the method to soil permeability is reduced.
The accumulation chamber method (or closed-chamber method) is a direct, static method originally used in agricultural sciences to determine soil respiration (Parkinson 1981) and then successfully adapted to measure CO2 soil flux of volcanological interest by Tonani and Miele (1991). This method is based on the measurement of the CO2 concentration increase inside an open-bottomed chamber of known volume, inverted on the soil surface. The initial rate of change of the concentration is proportional to the CO2 flux (Tonani and Miele 1991; Chiodini et al. 1996). The method does not require either assumptions about soil characteristics or the regime of the flux (advective/diffusive). The method has been tested by several authors under controlled laboratory conditions and provides reproducibility of 10% (Chiodini et al. 1998). In a field reproducibility test of the method, carried out at two points with high and low CO2 flux, Carapezza and Granieri (2004) found an uncertainty of 12% for high fluxes and 24% for low fluxes. Multiple measurements performed by ground-based methodologies allow a mapping of CO2 flux and an estimation of the total CO2 release by use of interpolation algorithms (Cardellini et al. 2003; Chiodini et al. 2005).
Eddy covariance or alternately Eddy correlation (EC) is a micrometeorological technique (e.g., Baldocchi 2003) recently proposed as a method to monitor volcanic CO2 emissions (Werner et al. 2000, 2003; Anderson and Farrar 2001; Lewicki et al. 2008). The basis of the EC is the calculation of the flux at the surface through the covariance between the fluctuations of the vertical component of the wind and the fluctuations of the gas concentration in atmosphere. The EC provides advantage of being an automated, time-averaged and area-integrated technique with a spatial scale significantly larger (square meter to square kilometer) than that of the ground-based methods (e.g., accumulation chamber). However the volcanic environment is often too heterogeneous for EC application, as suggested by the theory underlying EC, in terms of spatial and temporal variability of surface fluxes and morphology of the measuring field.
Diffusive degassing of deep CO2 in tectonically active areas
Since the early work of Irwin and Barnes (1980), it has become clear that a close relationship exists between active tectonic areas and anomalous crustal emissions of CO2. Due to their high crustal permeability, faults act as preferential pathways for the upward migration and eventual release of deep gases to the aquifers or directly to the atmosphere. Regional aquifers located in areas of high CO2 flux can dissolve most or part of the deeply generated gas because the relatively high solubility of CO2 in water. A carbon mass balance in the involved aquifer can be used to obtain an estimation of the amount of CO2 dissolved by groundwater. However, the large range of 13δCCO2 observed in such aquifers suggests that carbon can derive from multiple sources: atmospheric C, biogenic C, carbonate minerals derived C and deeply derived C. Therefore, an approach by coupling groundwater chemistry with hydrologic and isotopic data has to be applied in order to differentiate shallow versus deep sources. Chiodini et al. (2004) showed that in the tectonically active area of the Italian Apennines, approximately 40% of the inorganic carbon in the groundwater derives from magmatic sources. This observation suggests that there may be significant amounts of magmatic CO2 released in tectonic areas, perhaps a similar order of magnitude as subaerial volcanic degassing (Chiodini et al. 2004). Indeed, in volcanic areas, the dissolved CO2 in groundwater can be a significant component of the total CO2 flux at the volcano (e.g., Rose and Davisson 1996; Sorey et al. 1998; Inguaggiato et al. 2012).
Volatile emissions from the axis of mid-ocean ridges (MORs) in the form of black smokers are dramatic examples of submarine deep carbon emissions, however they tend to be short-lived and unpredictable, making collection of gas samples challenging. Nevertheless measurements of the composition and flux of such emissions have been conducted at dozens of sites (Kelley et al. 2004), using a wide range of gas collection techniques from piloted and remotely controlled submersible craft, allowing later analysis of gas samples in the laboratory. Direct measurements have been performed on only a fraction of the world’s MOR, and therefore previous work on the fluxes of CO2 from MOR has focused on quantifying emissions relative to a better-constrained global production parameter. These have included crustal production rates and the C content of the mantle (Gerlach 1989, 1991; Javoy and Pineau 1991; Holloway 1998; Cartigny et al. 2001; Saal et al. 2002), global mantle 3He flux (Corliss et al. 1979; Des Marais and Moore 1984; Marty and Jambon 1987; Sarda and Graham 1990; Graham and Sarda 1991; Marty and Zimmerman 1999), hydrothermal fluid flux (Elderfield and Schultz 1996) and the CO2/3He ratio in hydrothermal plumes. The latter two require estimates of global 3He fluxes, which are produced primarily at MORs (Allard et al. 1992).
An important aspect of MOR volcanism is that during the process of formation CO2 reacts with hot rock, sequestering CO2. In addition, dissolved carbonate in seawater reacts progressively within the shallowest ~60 m oceanic crust, producing steadily higher carbonate concentrations with increasing crustal age. Measurements of drill cores of the upper oceanic crust allowed Alt and Teagle (1999) to quantify the magnitude of the CO2 sink produced by crust reactions as of the magnitude of 150 Mt/yr CO2. This is of similar magnitude to the MOR CO2 flux of 97 ± 40 Mt/yr CO2, indeed it is probable that reactions in the oceanic crust absorb more CO2 than is emitted from MORs.
As well as emissions from the main axis of the MOR, degassing takes place on the flanks of the ridge, driving circulation of seawater through the crust. Sansone et al. (1998) sampled gas emissions from the eastern flank of the Juan de Fuca ridge with the Alvin deep sea vessel, both directly and through inverted funnels to concentrate the gas flow into titanium gas-tight samplers (Massoth et al. 1989). Further samples were collected using titanium syringe samples (Von Damm et al. 1985). Gas samples were acidified and then vacuum-extracted at sea with a glass/stainless-steel vacuum line. The total gas volume was determined with high precision capacitance manometers, and the extracted gas from each sample was sealed in break-seal glass ampoules for analysis ashore with gas chromatography and mass spectrometry.
Gas emissions also occur from active submarine arc volcanoes. Lupton et al. (2008) measured gas output from eleven volcanoes along the Mariana and Tonga-Kermadec arcs with remote controlled vehicles during three expeditions. Four of these volcanoes were found to produce distinct gas and liquid CO2 gas emissions, together with hydrothermal emissions from the main vents. Vent emissions were sampled using seawater-filled titanium alloy gas-tight bottles connected via tubing to a Ti sampling snout inserted directly into the vent. A valve was then opened and hydraulic pressure filled the bottle with a sample of vent gas. On the ship, samples were acidified and transferred under vacuum ampoules made of Pyrex and low-He permeability alumino-silicate glass for later laboratory analysis. Collection of liquid CO2 droplets was challenging due to the ~1000 fold expansion of liquid CO2 when converted to CO2 gas at 1 atm pressure, necessitating the use of a small volume Ti gas-tight bottle. Gas bubbles were collected with a plastic cylinder with relief valve that was placed over the emission until filled.
Understanding of the carbon balance in a subduction zone requires knowledge of the amount of carbon entering the zone within the subducting slab, CO2 loss from main arc volcanism and back arc, and the submarine fore-arc (see Fig. 1). This latter was measured from seeps in the Central America subduction zone by Füri et al. (2010). Seep fluids were collected over a 12 month period at the submarine segment of the Costa Rica fore-arc margin using 1/8 inch diameter copper tubing attached to a submarine flux meter operating in continuous pumping mode to measure CO2 and CH4 fluxes. Temporal variations during the sampling period were revealed by cutting the copper tubing in 0.4 m sections under vacuum and extracting the stored volatile samples in the laboratory for isotope ratios and compositions.
Volcanic lakes are significant, but previously unrecognized (Pérez et al. 2011) contributors to global deep CO2 budgets. CO2 gas emissions from volcanic lakes are in the form of both diffuse degassing from the lake surface and bubbling (Mazot and Taran 2009). Lake CO2 emissions were therefore measured with a floating gas accumulation chamber with an in-built NIR sensor to measure CO2 concentrations. Conversion of CO2 concentrations to fluxes was made using simultaneous measurements of pressure and temperature.
REPORTED MEASUREMENTS OF DEEP CARBON FLUXES
During the Holocene, ~1500 volcanoes on land erupted and in recorded history there have been 550 known eruptions. Typically 50–70 volcanoes erupt explosively each year and ~500 produce a gas emission either through hydrothermal systems or open-vent degassing (Siebert and Simkin 2002). Of these, only a small fraction have had their CO2 flux measured directly, however the number of measured volcanoes has greatly increased in recent years. In this section we first present published data on measured volcanic CO2 emission rates from active volcanoes, diffuse degassing of volcanic areas and tectonically active areas, followed by volcanic lakes and submarine emissions. We conclude by producing a global sum of CO2 emission rates.
We report in Table 1 all known volcanic plume CO2 flux measurements from persistently degassing volcanoes. We have chosen data in which CO2 fluxes were measured either by near simultaneous measurement of CO2/SO2 ratios and SO2/gas flux, or direct measurement of the CO2 flux. This was done because the number of volcanoes for which CO2 flux has been measured accurately has greatly increased in the last years, with 40 new measurements reported since 2000. We performed a simple average of the reported CO2 flux measurements for each volcano to produce Table 2, a summary which allows the total CO2 flux from 33 measured volcanic gas plumes to be calculated as 59.7 Mt/yr. We note that this measured flux by itself is higher than the maximum estimated for global passive degassing from Williams et al. (1992), highlighting the fundamental importance of direct measurements of volcanic CO2 fluxes in quantifying the volcanic CO2 inventory.
In Table 3, we report diffuse CO2 fluxes from historically active volcanoes, which have been verified through isotopic analysis to be of magmatic origin. This list is not an exhaustive collation of all diffuse CO2 degassing measurements, but relfects the most updated or complete diffuse CO2 flux measured at each volcano. The total CO2 flux from the 30 measured volcanoes is 6.4 Mt/yr, including emissions from diffuse soil degassing and those measured in groundwater.
CO2 fluxes from tectonic structures, hydrothermal systems or inactive volcanic areas are reported in Table 4, distinguishing between measured (or estimated) fluxes from soils from those dissolved in groundwaters. The two major measured contributors to the total tectonic, hydrothermal and inactive volcano CO2 flux of 66 Mt/yr are tectonic degassing in Italy (10 Mt/yr, Chiodini et al. 2004) and hydrothermal emissions from Yellowstone (8.6 Mt/yr, Werner and Brantley 2003). We also include in this list an estimate of the total CO2 flux produced by hydrothermal activity in Indonesia-Philippines (1.8 Mt/yr, Seward and Kerrick 1996) and the subaerial Pacific rim (44 Mt/yr, Seward and Kerrick 1996). These estimates are based on extrapolations from the CO2 emissions observed from the 150 km long Taupo Volcanic Zone (New Zealand) to the 18,000 km long Pacific Rim.
The global emissions of CO2 from volcanic lakes were recently assessed by Pérez et al. (2011), who pointed out that volcanic lakes had not been included in previous estimates of global geological carbon efflux (e.g., Kerrick et al. 2001; Mörner and Etiope 2002). They found that CO2 emissions increased with increasing acidity in volcanic lakes, reflecting the acidity of the volcanic gas discharge. Measurements were conducted on 32 volcanic lakes which were divided into three types of water based on pH, alkali, neutral and acid. Average flux per unit area for each type was then used to calculate a global volcanic lake estimate, extrapolating to an estimated number of volcanic lakes in the world (769). This number of lakes is greater than the number of lakes reported in the literature (138) by a factor which reflects the regional under-sampling between actual lakes and lakes reported in the scientific literature. This methodology assumes that the average acidity-emission rate relationship in the 32 measured lakes is a faithful average representation of the global lake population. A further estimate was produced by defining 4 populations in the measured data set based on frequency, emission rate and lake size and extrapolating to all 769 volcanic lakes. The combination of these two approaches yielded a global volcanic lake CO2 emission of 117 ± 19 Mt/yr, of which 94 Mt/yr is attributed to magmatic degassing.
There are three main submarine sources of CO2, MOR, arc volcanoes and fore-arc degassing (which appears to be dominated by CH4 emissions (Füri et al. 2010)). Global emissions from MOR have been determined by various authors, as reported in Table 5. The large spread of MOR fluxes, from 4.4 to 792, reflects uncertainties in the dissolved contents of C and global 3He fluxes. Marty and Tolstikhin (1998) performed a careful examination of the CO2/3He ratios used and determined a median value of 2.2 × 109 with standard deviation of 0.7 × 109. Using a 3He flux of 1000 ± 250 mol/yr (Farley et al. 1995) they derived a MOR CO2 flux of 97 ± 40 Mt/yr CO2. The more recent determination of MOR CO2 flux from Resing et al. (2004) who measured CO2/3He in MOR hydrothermal plumes of 55 ± 33 Mt/yr is in reasonable agreement with that estimate.
While measurements of CO2 release from the cooler flanks of MORs and submarine arc volcanoes increase in number each year, global estimates of submarine CO2 emissions are extremely difficult to make. The large areal extent and our relatively poor knowledge of the submarine surface suggests that there is ample opportunity for unknown or unrecognized active volcanism (e.g., cold liquid CO2 emissions, Lupton et al. 2008), but at the current time it is not possible to make quantitative estimates of the global CO2 emissions from such sources.
INVENTORIES OF GLOBAL VOLCANIC DEEP CARBON FLUX: IMPLICATIONS FOR THE GEOLOGICAL CARBON CYCLE
Estimates of global deep carbon emission rates
In Table 6, we summarize the measured fluxes from the subaerial sources and MOR, and attempt to extrapolate from these measurements to global estimates of the CO2 flux for each source. In the case of volcanic plume passive degassing the GVN catalogue (Siebert and Simkin 2002) indicates that there are ~150 such actively degassing volcanoes on Earth. While our catalogue of 33 CO2 flux plume measurements (Table 2: total flux 59.7 Mt/yr) is significantly larger than previously collated, it reflects only 22% of the total number of active volcanoes. While our current compilation includes some large emitters, suggesting that the major sources have been already identified, we highlight how this total flux has increased due to previously unrecognized large emissions from e.g., Ambrym (Vanuatu). Further large, but as yet unquantified, sources of CO2 emissions may be present in Papua New Guinea, the Banda Sunda arc and the Vanuatu island chain. We therefore conclude that the clearest and probably most accurate way to extrapolate from the current catalogue of plume emissions to a global estimate is through a linear extrapolation. Extrapolating from the measured 33 to an estimated 150 plume-creating, passively degassing volcanoes we estimate that the global plume CO2 flux is ~271 Mt/yr (see Table 6).
The total number of historically active volcanoes reported by GVN is ~550, and 30 (5.5%) of these have had diffuse CO2 soil degassing fluxes quantified, as reported in Table 3, for a total of 6.4 Mt/yr. Extrapolating to a global flux, assuming a similar distribution of fluxes in the unmeasured fluxes as seen in those measured, produces a total of 117 Mt/yr from diffuse degassing from the flanks of historically active volcanoes (Table 6).
The total diffuse CO2 flux from inactive volcanoes, hydrothermal and tectonic structures reported in Table 4 is more challenging to extrapolate to a global scale. Our current constraints on the tectonic CO2 flux comes almost entirely from the work of Chiodini et al. (2004) who examined actively degassing tectonic structures in Italy. The abundance of such structures on Earth is unknown, and this therefore represents a source of great uncertainty in estimates of total deep carbon flux. This uncertainty makes it challenging to sensibly extrapolate to a global estimate of tectonic CO2 fluxes, and therefore we use only reported fluxes, and highlight the possibility that the true total may be significantly larger. Emissions from hydrothermal systems estimated by Seward and Kerrick (1996) are already extrapolated to cover a significant proportion of the volcanically active surface of the Earth. We therefore use the total presented in Table 4 for the CO2 emissions from tectonic, hydrothermal and inactive volcanoes as a lower limit for the global emission of CO2 from these sources.
Summing the extrapolated passive plume, diffuse degassing, lake degassing global estimates and emissions from inactive, hydrothermal and tectonic structures produces a total subaerial volcanic flux of 540 Mt/yr, and a global emission (including MOR emissions) of 637 Mt/yr (Table 6). Thus global volcanic CO2 fluxes are only ~1.8% of the anthropogenic CO2 emission of 35,000 Mt per year (Friedlingstein et al. 2010).
Comparison with previous estimates of subaerial volcanic CO2 flux
There have been several papers which estimate the global CO2 flux, as shown in Table 7. Our update of the global volcanic CO2 flux, 637 Mt/yr, is larger than the maximum suggested by Marty and Tolstikhin (1998) of 440 Mt/yr. This is in part because CO2 emissions from volcanic lakes were not addressed in that work. The total subaerial flux we calculate of 540 Mt/year is also higher than that proposed by Mörner and Etiope (2002), due primarily to the improvement in measurements of persistently degassing volcanoes. We note that Mörner and Etiope (2002) included the fluxes from single eruptive events from Pinatubo (1991) and Mt. St. Helens (1980) in their inventory of volcanoes contributing to the annual global CO2 flux. Other papers cited in Table 7 appear to have significantly underestimated the global subaerial CO2 flux, primarily due to a lack of field measurements.
Balancing CO2 emission rates with weathering and subduction rates
In the absence of a continual supply of CO2 from volcanic and tectonic degassing the CO2 content of the atmospheres and oceans would be gradually depleted through CO2 removal by weathering (Gerlach 1991). The fact that, instead, pre-industrial CO2 concentrations are relatively stable suggest a balance between CO2 removal by weathering and CO2 supply by Earth degassing on timescales of ~0.5 Ma (Walker et al. 1987; Berner 1991). Over such timescales weathering of carbonates has no impact on removal of atmospheric CO2, because HCO3− supplied to the ocean by carbonate weathering releases the CO2 it captured from the atmosphere during calcite precipitation (Berner 1991). Therefore in the long timescale of the geological carbon cycle CO2 emissions from geological sources should balance consumption from silicate weathering and oceanic crust alteration. Gaillardet et al. (1999) found that CO2 consumption from continental silicate weathering was 515 Mt/yr, which matches well with our estimates of subaerial volcanic CO2 degassing (540 Mt/yr). However, inclusion of 300 Mt/yr CO2 released by metamorphism (Morner and Etiope 2002) produces a total lithospheric subaerial CO2 emission of 840 Mt/yr. This is larger than the current estimates of silicate weathering, suggesting that, assuming steady-state, weathering rates might be slightly higher to absorb all the emitted CO2.
Dasgupta and Hirschmann (2010) calculated the total ingassing of CO2 into subduction zones from the combination of three lithologies in the subducting slab, altered oceanic crust, sediments and mantle, producing an estimated CO2 consumption rate of 403 Mt/yr. We note that this is lower than the CO2 consumption rate due to silicate weathering (515 Mt/yr, Gaillerdet et al. 1999), which is slightly inconsistent (but within uncertainties of such estimates), as the eventual destiny of CO2 consumed by silicate weathering will be deposition on the seafloor or precipitation within the oceanic crust. In order to maintain steady-state quantities of CO2 in the exosphere this consumption should be balanced by the total emission from MORs, subaerial degassing and metamorphism (calculated here to be 937 Mt/yr). Given the likely underestimate in our total lithospheric CO2 emissions arising from lack of knowledge of tectonic degassing, it appears reasonable to conclude that the ingassing rate may be an underestimate. However, it is clear that there are large uncertainties in both sums.
THE ROLE OF DEEP CARBON IN VOLCANIC ACTIVITY
Original CO2 contents of magma
The volatile content of magmas, together with their evolving viscosity and degassing behavior during ascent, helps to determine whether a volcano will be quietly degassing or violently erupting for a given magma input rate. Models of magma dynamics require knowledge of original volatile content in order to reproduce physically accurate processes occurring during an eruption and in quiescent phases. Furthermore, knowledge of the original volatile contents of magmas allows calculation of the magma mass required to produce an observed gas flux, permitting quantitative comparison of fluxes with geophysical and volcanological observations. Measurements of original volatile contents are therefore of great interest.
Melt inclusions (MIs) provide records of original volatile contents, through analysis of pockets of melt trapped inside growing crystals during magma ascent or storage, and such studies have been carried out on many eruption products. However, the presence of a separate fluid phase at the moment of inclusion entrapment will produce an underestimate in the concentrations of dissolved volatiles. Wallace (2005) concluded that no melt inclusions sample arc magmas undegassed with respect to CO2. Blundy et al. (2010) used MI measurements from Mt. St. Helens to show that the dissolved volatile contents of shallow magmas were strongly affected by CO2-rich fluids rising from magmas at greater depth, concluding that the original CO2 contents of arc magmas was likely to be significantly higher than that recorded in MIs. Using inferred CO2 contents in arc andesites and dacites of 1.5 wt% they calculate that the CO2 contents of parental, mantle-derived basalts would contain 0.3 wt%. This relatively high CO2 content is in agreement with previous estimates of volatile contents of arc magmas (Wallace 2005).
Original CO2 contents of magmas can also be estimated by assuming a steady-state condition for a persistently degassing volcano, and comparing the observed CO2 flux together with the flux of a more soluble gas species whose original volatile content has been well-characterized, such as SO2. Gerlach et al. (2002) performed such a calculation for Kīlauea volcano, Hawai’i, concluding that the bulk CO2 concentration required to match magma input rates and CO2 output rates was 0.70 wt%. Dixon and Clague (2001) measured a dissolved CO2 content of at a depth of 1500–2000 m at the Loihi seamount in the Hawai’i chain. While the CO2 concentrations were very low, the samples contained fluid-filled vesicles with high CO2 contents, which allowed a bulk CO2 concentration of up to 0.63 wt% to be determined, in fair agreement with the estimate from Gerlach et al. (2002). A recent study (Barsanti et al. 2009) introduced a more complex note to the examination of original CO2 contents at Hawai’i, with a statistical analysis of MI CO2 and H2O concentrations which revealed distinct magma batches some of which could contain 2–6 wt% CO2. High CO2 contents of magmas feeding Etna and Stromboli have also been proposed based on degassing mass balance calculations, with amounts ranging between 1.6 and 2.2 wt% (Spilliaert et al. 2006; Burton et al. 2007a). These few available estimates of original CO2 contents from mass balance determinations open the possibility that CO2 contents of magmas feeding active volcanoes are in general higher than is expected based on CO2 contents of melt inclusions.
Importance of a deep exsolved volatile phase on magma dynamics and eruptive style
The presence of a CO2-rich volatile phase at great pressure can strongly affect the dynamics of magma ascent and eruption, because the style and intensity of eruptive activity is controlled in part by the distribution of gas phase in a magma during eruption (Eichelberger et al. 1986; Jaupart and Vergniolle 1988). Persistently degassing volcanoes can release vast amounts of gas at the surface non-explosively, implying storage within the crust of large volumes of degassed magma (Crisp 1984, Francis et al. 1993) via magma convection (Kazahaya et al. 1994).
Exsolved volatiles can ascend from depth, accumulating in foams that can produce Strombolian activity (e.g., Menand and Phillips 2007; Jaupart and Vergniolle 1988). Gas can stream through magma from depth to the surface (Wallace et al. 2005), as surmised to occur at Soufrière Hills volcano, Montserrat (Edmonds et al. 2010) and Stromboli volcano (Aiuppa et al. 2010). Perhaps most importantly of all however, exsolved gas accumulation can produce powerful explosive eruptions. The eruption of Pinatubo in 1991 (Pallister et al. 1992) was one of the most violent in recorded history. It produced a much greater mass of S than was to be expected from dissolved S contents and the volume of erupted material, suggesting the presence of a voluminous pre-eruptive gas phase (Wallace and Gerlach 1994), likely produced from basaltic underplating crystallizing as anhydrite (Matthews et al. 1992) which triggered the eruption (Pallister et al. 1992).
MAGNITUDE OF ERUPTIVE DEEP CARBON EMISSIONS
It is useful to compare the CO2 emission rates for subaerial volcanism of 540 Mt/yr reported in Table 7 with a single large eruption such as the ~5 km3 eruption of Mt. Pinatubo in 1991, producing ~50 Mt of CO2 (Gerlach et al. 2011), equivalent to merely ~5 weeks of global subaerial volcanic emissions. The Pinatubo 1991 syn-eruptive emission is therefore dwarfed by the time-averaged continuous CO2 emissions from global volcanism. Indeed, the present day CO2 emission rate from the lake filling the crater formed during the eruption of Pinatubo is 884 t/d (Perez et al. 2011), suggesting that in the 31 years since that eruption ~10 Mt of CO2 has been produced, ~20% of that emitted during the eruption.
Using the volumes of erupted material produced by the three largest eruptions of the last 200 years (Self et al. 2006) we may estimate their CO2 emissions, assuming a similar erupted volume to CO2 emission amount to that estimated for Pinatubo (10 Mt CO2 per km3 erupted, equivalent to ~1 wt% CO2 content). The eruption of Tambora (Indonesia) in 1815 is estimated to have produced 30 km3 of products (Self et al. 2006), with an inferred output of 300 Mt of CO2. Krakatua in 1883 (Indonesia) and Katmai-Novarupta in 1912 (Alaska) each produced 12 km3, and ~120 Mt of CO2. The total CO2 output of the four largest eruptions in the last 200 years is therefore ~600 Mt of CO2, slightly less than we estimate for subaerial volcanic degassing in a single year, and therefore only 0.6% of the amount of gas released through continuous volcanic activity in the same time period. It therefore appears that the continuous degassing of active and inactive volcanoes dominates the short-lived paroxysmal emissions produced in large eruptions.
Crisp (1984) calculated that the average eruption rate from volcanoes over the last 300 years was 0.1 km3 magma per year, which with ~1 wt% CO2 content suggests an annual output of ~1 Mt CO2, only 0.2% of the estimated annual subaerial CO2 emissions. This demonstrates that degassing of unerupted magma dominates degassing of erupted lava on the planet, and emphasizes the fundamental role that unerupted magmatic intrusions must have in contributing to the global volcanic CO2 flux. Such intrusions may produce unexpectedly high CO2 emissions if they interact with crustal carbonates (Troll et al. 2012). This important process could be assessed quantitatively if a method could be developed for measuring volcanic 12C/13C ratios in the field.
In recent years, measurements of CO2 flux from volcanoes and volcanic areas have greatly increased, particularly on persistently degassing volcanoes, of which ~22% have had their CO2 flux quantified. Notwithstanding this progress, it is clear that the CO2 emissions from the majority of volcanic sources are still unknown. Using the available data from plume measurements from 33 degassing volcanoes we determine a total CO2 flux of 59.7 Mt/yr. Extrapolating this to ~150 active volcanoes produces a total of 271 Mt/yr CO2. Extrapolation of the measured 6.4 Mt/yr of CO2 emitted from the flanks of 30 historically active volcanoes to all 550 historically active volcanoes produces a global emission rate of 117 Mt/yr. Perez et al. (2011) calculated the global emission from volcanic lakes to be 94 Mt/yr CO2. The sum of these fluxes produces an updated estimate of the global subaerial volcanic CO2 flux of 474 Mt/yr. Emissions from tectonic, hydrothermal and inactive volcanic areas contribute a further 66 Mt/yr to this total (Table 6), producing a total subaerial volcanic emission of 540 Mt/yr. An extrapolation to a global estimate is not straightforward for tectonic-related degassing, as the number of areas which produce such emissions is not known. Given the fact that ~10 Mt/yr is produced by Italy alone it is possible that the global total is significant, and this merits further investigation. We highlight also that the magnitude of CO2 emissions from both cold and hot non-MOR submarine volcanic sources are currently effectively unknown.
Our subaerial volcanic CO2 flux matches well with estimates of CO2 removal rates of 515 Mt/yr due to silicate weathering, which, over timescales of 0.5 Ma, should balance lithospheric CO2 emissions. However, inclusion of the metamorphic CO2 flux of 300 Mt/yr calculated by Morner and Etiope (2002) produces a total subaerial lithospheric flux of 840 Mt/yr, suggesting that, assuming steady-state, weathering rates might be slightly higher in order to absorb all the CO2 emitted from the lithosphere.
The global subaerial CO2 flux we report is higher than previous estimates, but remains insignificant relative to anthropogenic emissions, which are two orders of magnitude greater at 35,000 Mt/yr (Friedlingstein et al. 2010). Nevertheless, it is clear that uncertainties in volcanic CO2 emission rates remain high and significant upward revisions of the lithospheric CO2 flux cannot be ruled out. This uncertainty also limits our understanding of global volcanic carbon budgets and the evolution of the distribution of CO2 between the crust and the mantle. Furthermore, with the notable exception of continuous CO2 flux monitoring at a handful of volcanoes we have very little data with which to assess CO2 flux variations across different timescales. It is clear that there is much further work to be done surveying CO2 emissions from both active and inactive volcanoes.
Continuous global CO2 emissions from passively degassing volcanoes over timescales longer than a few months dominate CO2 emissions produced by relatively short-lived eruptions. Nevertheless, we highlight that dramatic CO2 emissions may occur during magmatic intrusion events, and that the sporadic and short-term nature of field measurements to date may lead to such events being missed. To this end, robust field-portable instruments capable of measuring 12C/13C ratios in volcanic CO2 emissions would be of great utility in order to distinguish CO2 produced during metamorphism of crustal carbonates from magmatic CO2.
We thank Adrian Jones and Bob Hazen for the opportunity to contribute to this volume. Adrian Jones’ and Tamsin Mather’s constructive reviews of an earlier version of this work are greatly appreciated. Sara Barsotti is thanked for her simulations of CO2 concentrations in the Etna plume with VOLCALPUFF. MB acknowledges support from ERC project CO2Volc 279802.