Search

co-auteur étranger; International audience; The parallel Halmahera and Sangihe arcs in eastern Indonesia are sites of active arc-arc collision of considerable interest in developing understanding of the geodynamics and geochemistry of subduction zones. Owing to the comparative remoteness of the region, few ground-based studies of the volcanoes have been undertaken. Here, we report and integrate gas measurements and (isotope) geochemical analyses of lava samples for Dukono, Ibu, Gamkonora, Gamalama, and Makian volcanoes of the Halmahera arc. Summing gas fluxes for all five volcanoes indicates arcscale emission budgets for H 2 O, CO 2 , SO 2 , H 2 S, and H 2 of 96300±27000, 2093±450, 944±400, 79±20, and 15±4 Mg/d, respectively. Dukono is the greatest source of SO 2 and H 2 , while Ibu emits the most H 2 O and H 2 S. Both Gamalama and Ibu are significant CO 2 sources. Dukono (farthest from the trench) releases the most CO 2-poor gas. Geochemical and isotopic analyses of recent ejecta emphasize the role of high fluid fluxes in the mantle wedge, necessary for partial melting of depleted mantle. Pb, Nd, and Sr isotope ratios, combined with Ba/Nd, Zr/Nd, Ba/Th, and Zr/Nb ratios, provide evidence for compositional variability along the Halmahera arc, and indicate decreasing subducted sediment contribution from south (Makian, Gamalama) to north (Gamkonora, Ibu, Dukono). Additionally, fluids formed by dehydration of altered crust become prominent at the northern volcanoes. Isotopic and Ba/Nb ratios from the Neogene and Quaternary sources compared to the current magmas further emphasize the evolution of magma genesis since the Neogene.

International audience; With at least four eruptions over the last 20 years, Sirung is currently one of the more active volcanoes in Indonesia. However, due to its remoteness, very little is known about the volcano and its hyperacid crater lake. We report here on the first measurements of gas and heat emissions from the volcano. Notable is the substantial heat loss from the crater lake surface, amounting to 220 MW. In addition, 17 Gg of SO2, representing 0.8% of Indonesian volcanic SO2 contribution into the atmosphere, 11 Gg of H2S, 17 Gg of CO2, and 550 Gg of H2O are discharged into the atmosphere from the volcano annually. The volatiles degassed from Sirung magmas are subjected to hydrothermal fluid-rock interactions and sulfide depositions, initiated by the disproportionation of SO2. These processes lead to distinct gas compositions and changing lake water chemistry (in the sub-craters and the main crater lake). However, the occurrence of SO2-rich fluids and strong gas flux appear to highlight a rapid fluid transfer to surface, avoiding re-equilibration with lower temperature rocks/fluids in the conduits.

Volcanic plumes are sites of dynamic chemistry involving halogen gases. Here we present new data on the relative abundances of SO2, BrO and OClO gases emitted from Soufrière Hills Volcano [SHV). They were collected during an eruptive hiatus but during sustained degassing at this halogen-rich volcano. By comparison with data from a previous study during an eruptive phase and application of the data and modeling of Villemant et al. (2008), we suggest that, after consideration of errors, either the rate of HBr conversion to BrO is variable, ranging from ∼30% to ∼15%, and/or the relative partitioning of Cl and Br into the gas phase from the melt changes according to eruptive activity. We examine the potential implications of this for fluid-melt partitioning, and compare our results with data from the experimental literature. Our work contributes toward understanding the controls on the BrO/SO2 ratio for volcano monitoring purposes; the changes in plume chemistry with regard to bromine at the onset of lava extrusion may be large and rapid. OClO was detected in the plume at SHV for the first time. This species has only previously been detected in emissions from Mount Etna (using ground-based methods) and from Puyehue Cordon Caulle (using satellite-based methods). No HCHO or NOy species were detected in the spectra.

[1] Direct-sampling and remote-sensing measurements were made at the crater rim of Masaya volcano (Nicaragua) to sample the aerosol plume emanating from the active vent. We report the first measurements of the size distribution of fine silicate particles (d < 10 μm) in Masaya's plume, by automated scanning electron microscopy (QEMSCAN) analysis of a particle filter. The particle size distribution was approximately lognormal with modal d ∼ 1.15 μm. The majority of these particles were found to be spherical. These particles are interpreted to be droplets of quenched magma produced by a spattering process. Compositional analyses confirm earlier reports that the fine silicate particles show a range of compositions between that of the degassing magma and nearly pure silica and that the extent of compositional variability decreases with increasing particle size. These results indicate that fine silicate particles are altered owing to reactions with acidic droplets in the plume. The emission flux of fine silicate particles was estimated as ∼1011 s-1, equivalent to ∼55 kg d-1. Sun photometry, aerosol spectrometry, and thermal precipitation were used to determine the overall particle size distribution of the plume (0.01 < d(μm) < 10). Sun photometry and aerosol spectrometry measurements indicate the presence of a large number of particles (assumed to be aqueous) with d ∼ 1 μm. Aerosol spectrometry measurements further show an increase in particle size as the nighttime approached. The emission flux of particles from Masaya was estimated as ∼1017 s-1, equivalent to ∼5.5 Mg d-1 where d < 4 μm. Copyright 2009 by the American Geophysical Union.

We report measurements of Hg, SO2, and halogens (HCl, HBr, HI) in volcanic gases from Masaya volcano, Nicaragua, and gaseous SO2 and halogens from Telica volcano, Nicaragua. Mercury measurements were made with a Lumex 915+ portable mercury vapor analyzer and gold traps, while halogens, CO2 and S species were monitored with a portable multi gas sensor and filter packs. Lumex Hg concentrations in the plume were consistently above background and ranged up to 350 ng m-3. Hg/SO2 mass ratios measured with the real-time instruments ranged from 1.1 × 10-7 to 3.5 × 10-5 (mean 2 × 10-5). Total gaseous mercury (TGM) concentrations measured by gold trap ranged from 100 to 225 ng m-3. Reactive gaseous mercury accounted for 1% of TGM, while particulate mercury was 5% of the TGM. Field measurements of Masaya's SO2 flux, combined with the Hg/SO2 ratio, indicate a Hg flux from Masaya of 7.2 Mg a-1. At Masaya's low temperature famaroles, Hg/CO2 mass ratios were consistently around 2 × 10-8, lower than observed in the main vent (Hg/C02 ∼ 10-7). Low-temperature fumarole Hg fluxes from Masaya are insignificant (∼150 g a-1). Ratios of S, C and halogen species were also measured at Masaya and Telica volcanoes. CO2/ SO2 ratios at Masaya ranged from 2.8 to 3.9, comparable to previously published values. At Masaya molar Br/SO2 was 3 × 10-4 and I/SO2 was 2 × 10-5, suggesting fluxes of 0.2-0.5 Mg HBr d-1 and 0.02-0.05 Mg HI d-1. At Telica the Br/SO2 ratio was also 3 × 10-4 and the I/SO2 ratio was 5.8 × 10-5, with corresponding fluxes of 0.2 Mg HBr d-1 and 0.06 Mg HI d-1. Gases at both volcanoes are enriched in I relative to Br and Cl, compared to gases from volcanoes elsewhere. Copyright 2008 by the American Geophysical Union.

Here we report results from a multidisciplinary field campaign at Villarrica volcano, Chile, in March 2009. A range of direct sampling and remote sensing techniques was employed to assess gas and aerosol emissions from the volcano, and extend the time series of measurements that have been made during recent years. Airborne traverses beneath the plume with an ultraviolet spectrometer yielded an average SO2 flux of 3.7kg s-1. This value is similar to previous measurements made at Villarrica during periods of quiescent activity. The composition of the plume was measured at the crater rim using electrochemical sensors and, for the first time, open-path Fourier transform infrared spectroscopy, yielding a composition of 90.5mol% H2O, 5.7% CO2, 2.6% SO2, 0.9% HCl, 0.3% HF and

Near-source observations of aerosol size distributions ( 1 [mu]m) ash particles. The number ratio of fine and coarse particles fluctuated strongly within short intervals of time, indicative of changes in the fragmentation energy. The near-source ash fallout was poorly sorted (interquartile range 16-80 [mu]m); ash grains sized

International audience; Cascade impactors were used to sample volcanic aerosol from Masaya (Nicaragua) in 2007, 2009 and 2010. Differences were found in the size distributions of volcanic aerosol between these recent campaigns and with a campaign in 2001: (1) SO42− showed modes in both the fine (1 μm; with high Na+/K+) fractions in all of the recent campaigns despite being unimodal in 2001 (

International audience; The measurement of SO2 flux from volcanoes is of major importance for monitoring and hazard assessment purposes, and for evaluation of the environmental impact of volcanic emissions. We propose here a novel technique for accurate and high time resolution estimations of the gas flux. We use two wide field of view UV spectrometers capable of collecting, instantaneously, light from thin parallel cross sections of the whole gas plume, obviating the need for either traversing, scanning or imaging. It enables tracking of inho-mogeneities in the gas cloud from which accurate evaluation of the plume velocity can be made by correlation analysis. The method has been successfully applied on Mt. Erebus volcano (Antarctica). It yields estimations of the plume velocity and gas flux at unprecedented time resolution (1 Hz) and high accuracy (uncertainty of 33%). During a ∼2 h experiment on 26 December 2006, SO2 flux varied between 0.17 and 0.89 ± 0.2 kg s −1 with a vertical plume velocity varying between 1 and 2.5 ± 0.1 m s −1. These measurements provide insight into the short-term variations of the passive degassing of this volcano renowned for its active lava lake. A cyclicity in flux, ranging from about 11-24 min, is evident. We propose two physical mechanisms to explain this degassing pattern, associated to periodic supply of either gas-rich magma or gas alone into the lake. The dual-wide field of view DOAS technique promises better integration of geochemical and geophysical observations and new insights into gas and magma dynamics, as well as processes of magma storage and gas segregation at active volcanoes.

Volcanoes provide important contributions to atmospheric budgets of SO2 and reactive halogens, which play significant roles in atmospheric oxidative capacity and radiation. However, the global source strengths of volcanic emissions remain poorly constrained. These uncertainties are highlighted here by the first measurements of gas emission rates from Ambrym volcano, Vanuatu. Our initial airborne ultraviolet spectroscopic measurements made in January 2005 indicate fluxes of 18–270 kg s-1 of SO2, and 62–110 g s-1 of BrO, into the atmosphere, placing Ambrym amongst the largest known contemporary point sources of both these species on Earth. We also estimate high Cl and F fluxes of ~8–14 and ~27–50 kg s-1, respectively, for this period. Further observations using both airborne and spaceborne remote sensing reveal a fluctuating SO2 output between 2004 and 2008, with a surge in the first half of 2005, and underline the substantial contribution that a single passively degassing volcano can make to the atmospheric budget of sulfur and halogens.

Here we report measurements of the chemical composition and flux of gas emitted from the central lava lake at Erta 'Ale volcano (Ethiopia) made on 15 October 2005. We determined an average SO 2 flux of ∼ 0.69 ± 0.17 kg s − 1 using zenith sky ultraviolet spectroscopy of the plume, and molar proportions of magmatic H 2 O, CO 2 , SO 2 , CO, HCl and HF gases to be 93.58, 3.66, 2.47, 0.06, 0.19 and 0.04%, respectively, by open-path Fourier transform infrared (FTIR) spectrometry. Together, these data imply fluxes of 7.3, 0.7, 0.008, 0.03 and 0.004 kg s − 1 for H 2 O, CO 2 , CO, HCl and HF, respectively. These are the first FTIR spectroscopic observations at Erta 'Ale, and are also some of the very few gas measurements made at the volcano since the early 1970s (Gerlach, T.M., 1980b. Investigation of volcanic gas analyses and magma outgassing from Erta 'Ale lava lake, Afar, Ethiopia. Journal of Volcanology and Geothermal Research, 7(3–4): 415–441). We identify significant increases in the proportion of H 2 O in the plume with respect to both CO 2 and SO 2 across this 30-year interval, which we attribute to the depletion of volatiles in magma that sourced effusive eruptions during the early 1970s and/or to fractional magma degassing between the two active pit craters located in the summit caldera.

We report here the first measurements of SO 2 emission rates for three volcanoes in Hokkaido. Observations were made by scattered light ultraviolet (UV) spectroscopy using miniature spectrometers carried in vehicles or on foot, or attached to scanners, with data processing following variants on differential optical absorption spectroscopy. The following mean fluxes were calculated for measurements in 2003 (and 2004): 0.032 (0.14), 0.24 (0.28) and 2.4 kg s − 1 for Meakan-dake, Tarumae, and Tokachi-dake, respectively. These are thought to be the only sites of high temperature volcanic degassing on Hokkaido at present, and thus the modest combined flux of 2.7 kg s − 1 represents, to first order, the total volcanic emission rate of SO 2 for the island. This represents only a small fraction of the SO 2 output of about 80 kg s − 1 for Miyakejima volcano in 2003. The apparently low sulfur emission from Hokkaido volcanoes may reflect the scrubbing efficiency of their hydrothermal systems though H 2 S emissions remain unquantified. The ability to measure comparatively weak fumarole emissions (e.g., at Meakan-dake and Tarumae) emphasizes the value of walking traverse UV spectroscopy.

The correlation spectrometer (COSPEC), the principal tool for remote measurements of volcanic SO2, is rapidly being replaced by low-cost, miniature, ultraviolet (UV) spectrometers. We compared two of these new systems with a COSPEC by measuring SO2 column amounts at Kīlauea Volcano, Hawaii. The two systems, one calibrated using in-situ SO2 cells, and the other using a calibrated laboratory reference spectrum, employ similar spectrometer hardware, but different foreoptics and spectral retrieval algorithms. Accuracy, signal-to-noise, retrieval parameters, and precision were investigated for the two configurations of new miniature spectrometer. Measurements included traverses beneath the plumes from the summit and east rift zone of Kīlauea, and testing with calibration cells of known SO2 concentration. The results obtained from the different methods were consistent with each other, with

We report here the first observations of NO2 emission from Mt. Erebus, a volcano with an active lava lake located on Ross Island, Antarctica. Erebus generates a persistent plume, which is entrained at an altitude of about 4 km above sea level. Its NO2 flux, measured by scattered light ultraviolet spectroscopy in December 2003, was equivalent to ∼0.6 Gg (N) yr-1. The total reactive nitrogen supply may be significantly higher than this since other NOy species are likely to have been present in the plume. We believe the NO2 is generated by thermal fixation of atmospheric nitrogen at the hot lava surface, forming NO, which then reacts rapidly with oxidants including ozone to yield NO2. Erebus volcano has displayed lava lake activity for many decades and may, therefore, play a significant long-term role in Antarctic tropospheric chemistry, and represent an important source of nitrogen deposited to the ice surface. © 2005 Elsevier Ltd. All rights reserved.

Anak Krakatau is the active cone that has built up in the caldera of Krakatau volcano after the 1883 cataclysmic eruption, in the Sunda Strait. Initially submarine, this new cone definitely emerged from the sea in 1930 and since then has progressively grown up through both explosive and effusive eruptions (similar to one eruption every 3 years). Here we report on the first quantification of volcanic gas output from Anak Krakatau, based on airbom UV measurements of the SO2 flux in 2014, and then discuss its implication in terms of magma degassing budget since 1930. We find that during non-eruptive activity Anak Krakatau passively emits 190 +/- 40 tons per day of SO2, which is comparable to the emission rate during lava dome extrusion at Merapi, central Java, but substantially more than those measured on few other Indonesian volcanoes (Tangkubanparahu, Slamet, Bromo and Papandayan). Anak Krakatau thus appears to be an important persistent emitter of volcanic volatiles in the Indonesian arc, even though this very active region still remains weakly documented on that aspect. Combining with available data for the composition of its high-temperature (similar to 700 degrees C) crater gases, Anak Krakatau may release annually 0.07 Tg of SO2, 3 Tg of H2O, and 0.13 moles of He-3. Using published data for the sulfur content of its feeding magma, we estimate that about 1.3 km(3) of magma may have been degassed during its subaerial growth over the past 85 years. The subaerial cone represents only 14% of this volume. Thus, a substantial fraction (1.1 km(3)) of the degassed magma did not extrude and may have accumulated in the plumbing system. This inference is consistent with geophysical and petrologic evidence of the presence of dense magma bodies in the shallow crustal basement of Krakatau volcano.

International audience; The spatial and temporal coverage of measurements of previous termvolcanicnext term gas emissions remains patchy. However, over the last decade, emissions inventories have improved thanks to new measurements of some of the lesser-known previous termvolcanicnext term areas. We report on one such region - the Vanuatu island arc, in the Southwest Pacific - for which we now have sufficient systematic observations to offer a systematic emissions inventory. Our new estimate is based on SO2 flux measurements made in the period 2004-2009 with ultraviolet spectroscopy techniques for the following volcanoes: Yasur, Lopevi, Ambrym, Ambae, Gaua and Vanua Lava (from south to north). These are the first ever measurements for Lopevi, Gaua and Vanua Lava. The results reveal the Vanuatu arc as one of Earth's prominent sources of previous termvolcanicnext term degassing with a characteristic annual emission to the atmosphere of ~ 3 Tg of SO2 (representing about 20% of hitherto published global estimates). Our new dataset highlights the sustained prodigious degassing of Ambrym volcano, whose 5 Gg d-1 mean flux of SO2 represents nearly two-thirds of the total budget for the Vanuatu arc. This confirms Ambrym as one of the largest previous termvolcanic sources worldwide comparable to Etna, often considered as the most vigorous source of previous termvolcanic emission on Earth. We also report a high degassing for Ambae of ~ 2 Gg d-1 SO2, representing more than 28 % of the Vanuatu arc budget. Thus, 90 % of the SO2 output from Vanuatu is focused in the central part of the arc (from Ambrym and Ambae) where magmas originate from enriched Indian-type mantle and where peculiar tectonic conditions could favour high magma production rates.

Since the first detection of bromine monoxide in volcanic plumes attention has focused on the atmospheric synthesis and impact of volcanogenic reactive halogens. We report here new measurements of BrO in the volcanic plume emitted from Kīlauea volcano – the first time reactive halogens have been observed in emissions from a hotspot volcano. Observations were carried out by ground-based Differential Optical Absorption Spectroscopy in 2007 and 2008 at Pu'u' O ' o crater, and at the 2008 magmatic vent that opened within Halema'uma'u crater. BrO was readily detected in the Halema'uma'u plume (average column amount of 3×10 15 molec cm −2 ) and its abundance was strongly correlated with that of SO 2 . However, anticorrelation between NO 2 and SO 2 (and BrO) abundances in the same plume strongly suggest an active role of NO x in reactive halogen chemistry. The calculated SO 2 /BrO molar ratio of ~1600 is comparable to observations at other volcanoes, although the BrO mixing ratio is roughly double that observed elsewhere. While BrO was not observed in the Pu'u' O ' o plume this was probably merely a result of the detection limit of our measurements and based on understanding of the Summit and East Rift magmatic system we expect reactive halogens to be formed also in the Pu'u' O ' o emissions. If this is correct then based on the long term SO 2 flux from Pu'u' O ' o we calculate that Kīlauea emits ~480 Mg yr −1 of reactive bromine and may thus represent an important source to the tropical Pacific troposphere.

[1] We report the first combined measurements of the composition and flux of gas emitted from Nyiragongo volcano by ground-based remote-sensing techniques. Ultraviolet spectroscopic measurements made in May/June 2005 and January 2006 indicate average SO(2) emission rates of 38 kg s(-1) and 23 kg s(-1), respectively. Open-path Fourier transform infrared spectroscopic measurements obtained in May/June 2005, January 2006, and June 2007 indicate average molar proportions of 70, 24, 4.6, 0.87, 0.26, 0.11, and 0.0016% for H(2)O, CO(2), SO(2), CO, HCl, HF, and OCS, respectively. The composition of the plume was remarkably similar in 2005, 2006, and 2007, with little temporal variation in proportions of CO(2), SO(2), and CO, in particular, on the scale of seconds or days or even between the three field campaigns that span a period of 24 months. This stability persisted despite a wide range of degassing behaviors on the surface of the summit crater's lava lake ( including discrete strombolian bursts and lava fountains) and variations in the SO(2) emission rate. We explain these observations by a regime of steady state degassing in which bubbles nucleate and ascend in chemical equilibrium with the convecting magma. Short-term ( seconds to minutes) temporal fluctuations in the SO(2)-HCl-HF composition were observed, and these are attributed to shallow degassing processes.

The explosions and subsequent fire at the Buncefield oil depot in December 2005 afforded a rare opportunity to study the atmospheric consequences of a major oil fire at close range, using ground-based remote-sensing instruments. Near-source measurements (less than 10 km) suggest that plume particles were approximately 50% black carbon (BC) with refractive index 1.73−0.42i, effective radius ( R eff ) 0.45–0.85 μm and mass loading approximately 2000 μg m −3 . About 50 km downwind, particles were approximately 60–75% BC with refractive index between 1.80−0.52i and 1.89−0.69i, R eff ∼1.0 μm and mass loadings 320–780 μg m −3 . Number distributions were almost all monomodal with peak at r 2 (70 ppbv), NO 2 (140 ppbv), HONO (20 ppbv), HCHO (160 ppbv) and CS 2 (40 ppbv). Our measurements are consistent with others of the Buncefield plume, and with studies of the 1991 Kuwaiti oil-fire plumes; differences from the latter reflecting in part contrasts in combustion efficiency and source composition (refined fuels versus crude oils) leading to important potential differences in atmospheric impacts. Other measurements made as the plume passed overhead approximately 50 km downwind showed a reduced solar flux reaching the surface, but little effect on the atmospheric potential gradient (electric field). The wind speed data from the day of the explosion hint at a possible explosion signature.

The Central American volcanic arc supplies a significant proportion of the persistent annual global sulphur dioxide emissions from volcanoes. In November/December 2003, we completed a survey of the arc section from Mombacho to San Cristobal in Nicaragua recording individual mean fluxes of 800, 530 and 220 Mg day � 1 in the plumes from San Cristobal, Telica and Masaya, respectively. An assessment of fluxes published since 1997 along the entire Central America arc yields a mean total arc flux of SO2 of 4360 Mg day � 1 or 8-16% of the annual estimated global volcanic SO2 flux to the troposphere. New field data shows that Masaya volcano continues to show stable HCl/SO2 and HF/SO2 ratios, suggesting a sustained flux of these components of ~220 and 30 Mg day � 1 , respectively (1997 to 2004). Masaya's plume composition also appears to have been stable, between 2001 and 2003, with respect to all the particulate species measured, with significant fluxes of SO4 2� (4 Mg day � 1 ), Na + (0.9-1.3 Mg day � 1 ) and K + (0.7 Mg day � 1 ). Extrapolating the Masaya plume species ratios to the entire Central American arc gives mean HCl and HF fluxes of 1300 and 170 Mg day � 1 and a particulate sulphate flux of 40 Mg day � 1 for 1997 to 2004, although without further understanding of the degassing processes and sources at depth of these different volatiles, these arc-scale estimates should be treated with caution. Combining our arc scale mean SO2 flux with published measurements of volcanic gas compositions with respect to CO2 and H2O allows us to estimate mean CO2 fluxes of 4400-9600 Mg day � 1 and H2O fluxes of 70,000-78,000 Mg day � 1 for the arc. Preliminary comparisons of these estimates of outgassing rates with published volatile input fluxes into the Central American subduction zone, suggest that Cl is more efficiently recycled through the subduction zone than CO2. The results for H2O are inconclusive. D 2005 Elsevier B.V. All rights reserved.

Volcanoes have only recently been recognized as a potentially major source of reactive bromine species to the atmosphere, following from the detection of bromine monoxide (BrO) in the plume emitted by Soufriere Hills Volcano, Montserrat. However, BrO is not expected to be emitted in significant quantity from magma, presenting a puzzle regarding its formation. We report here new field measurements of the tropospheric plume emitted by Mt. Etna, Italy, which provide the first direct evidence of fast oxidation of halogen species in a volcanic plume, and lead to an explanation of how BrO is generated from magmatic HBr emissions. We show that the timescale of BrO formation (a few minutes after emission into the atmosphere) is consistent with rapid heterogeneous halogen chemistry involving sulphate aerosol in the plume. The model highlights considerable complexity to the oxidative chemistry of volcanic plumes.

[1] Papua New Guinea (PNG) hosts some of the most prolific degassing volcanoes in the world. Collectively, they are thought to be responsible for a significant contribution of sulfur dioxide to the regional and global atmosphere. Despite their importance, very few measurements of SO2 fluxes from PNG volcanoes have been made in the past, leading to uncertainty in estimates of total global volcanic sulfur budgets. We report here sulphur dioxide flux measurements obtained by ultraviolet spectroscopy during July and August 2003 for the following PNG volcanoes: Tavurvur, Ulawun, Pago, Langila, Bagana and Manam. We found fluxes of 1.3, 7.4, 1.4, 2.9, 23 and 2.1 kg s−1, respectively, for these sources. This amounts to a total of 38 kg s−1 of SO2, constituting some 6–9% and 7–12% of the estimated global time-integrated volcanic emissions to the atmosphere, and troposphere, respectively.

We report here the first measurements of the SO 2 flux from Erta 'Ale volcano (Ethiopia); the measurements were obtained in March 2003 with a portable ultraviolet spectrometer. Emission rates averaged 0.7 kg.s - 1 from the active lava lake and 1.3 kg.s - 1 from the whole volcano (including fumarolic emissions in the north part of the caldera). This modest output combined with an estimate of the melt sulfur content indicates a magma-supply rate of 350-650 kg.s - 1 (∼0.13-0.25 m 3 .s - 1 ). Radiant heat power from the lava lake, measured by infrared thermography, was found to vary between ∼5 and 30 MW according to activity of the lake surface and time elapsed since resurfacing events. We show that 8%-14% crystallization and/or 30-60 °C cooling of the upwelling magma reaching the lake, as well as degassing, are sufficient to increase magma density in the upper conduit and drive convection between the surface and a feeding reservoir. Fluctuations in the system, such as degree of vesiculation and magma supply rate, can be buffered by ascent or descent of the magma level within a flared vent region whose geometry directly controls lake surface area and hence heat loss.