Sulfur dioxide emissions from Papandayan and Bromo , two Indonesian volcanoes

Indonesia hosts 79 active volcanoes, representing 14 % of all active volcanoes worldwide. However, little is known about their SO 2 contribution into the atmosphere, due to isolation and access difficulties. Existing SO 2 emission budgets for the Indonesian archipelago are based on extrapolations and inferences as there is a considerable lack of field assessments of degassing. Here, we present the first SO 2 flux measurements using differential optical absorption spectroscopy (DOAS) for Papandayan and Bromo, two of the most active volcanoes in Indonesia. Results indicate mean SO2 emission rates of 1.4 t d −1 from the fumarolic activity of Papandayan and more than 22–32 t d −1 of SO2 released by Bromo during a declining eruptive phase. These DOAS results are very encouraging and pave the way for a better evaluation of Indonesian volcanic emissions.


Introduction
Volcanic degassing into the atmosphere constitutes one of the external expressions of subsurface magmatic and hydrothermal manifestations.Reciprocally, any changes in the chemical and physical properties of the plume are generally symptomatic of modifications in the magmatic reservoir and/or conduits.Among the volcanic volatile components released into the atmosphere, sulfur dioxide (SO 2 ) enhances considerable research interest due to its non-negligible roles in the atmospheric chemistry, atmospheric radiation, hydrological cycle and climate, as well as acidic precipitation and air quality (Charlson et al., 1992;Jones et al., 2001;Penner et al., 2001;Stevenson et al., 2003).It is a relatively abundant species in the volcanic plume, typically in third place behind H 2 O and CO 2 with around 5 mol% of gas content along with H 2 S (Shinohara, 2008).SO 2 has a very low background level in the atmosphere and strong identifiable optical absorption features in the ultraviolet (UV) skylight region that offer various options for spectroscopic detection in the atmosphere (McGonigle et al., 2003).SO 2 is thus a readily measurable species, widely recognized as an important and highly desirable component of multidisciplinary volcano monitoring.Many observatories routinely measure SO 2 emission rates in support of their monitoring networks.The global SO 2 emission budget estimates over the last four decades range from 1.5 to 40 Tg yr −1 (Table 1) -a very large range due to the diversity of methodology used, the extending number of studied volcanoes and the increasing development of measurement techniques.
According to the IPCC report (2001), the global volcanic SO 2 emission budget is highly uncertain, because only very few of the potential sources have been measured and the variability between sources and between different stages of activity is considerable.Over the last ten years, the increasing development in the field of remote sensing has improved our knowledge on the distribution of volcanic volatile sources across the earth and even on the remote and less accessible edifices (McGonigle et al., 2004;Mather et al., 2006;Bani et al., 2012;McCormick et al., 2012).But still many volcanoes on earth have never had their degassing rates evaluated.This is the case in Indonesia where, despite the high number of active volcanoes, the past SO 2 emission estimates were based on extrapolation and inference (Nho et al., 1996;Halmer et al., 2002;Hilton et al., 2002), whilst SO 2 flux measurements were carried out only on 4 volcanoes (Table 1).This present work aims to point out this misrepresentation and further highlight a possibility to constrain the SO 2 emission better from Indonesian volcanoes using differential optical absorption spectroscopy (DOAS).
We present SO 2 flux measurements, obtained in June 2011 from two volcanoes in Indonesia -including Papandayan and Bromo (Fig. 1), two volcanoes among the most active in Indonesia, and they represent two end-members of volcanic degassing types: a fumarolic emission on Papandayan and an open vent degassing from Bromo

Papandayan and Bromo volcanoes
Papandayan is a complex stratovolcano culminating 2665 m above sea level with a base diameter of ∼ 8 km (Fig. 2) located 45 km south-southeast of Bandung city, West Java.The edifice became well known after its 1772 eruption that caused the collapse of the northeast flank, leading to a devastating debris avalanche over 250 square kilometers that destroyed about 40 villages and killed nearly 3000 people (Abidin et al., 2006).Papandayan ranks 11th out of the 13 deadliest eruptions on earth (Blong, 1984).Other eruptions have been reported for this volcano in 1882, 1923-1927, 1942, 1993and 2002(Abidin et al., 2006)).The latest eruption in 2002 is well detailed in Abidin et al. (2006) -about 6000 people were evacuated.Magmatic degassing on Papandayan occurs mainly in two fumarolic zones, aligned in a northwestsoutheasterly direction and located in a deformed, horseshoeshaped eastern crater (Fig. 2) (Mazot et al., 2007(Mazot et al., , 2008)).
Bromo is located 75 km south of Surabaya, East Java.It occupies the central part of the Tengger caldera -a well-defined and a roughly square structure around 7 km wide (Fig. 2).The caldera rim culminates at more than 2600 m above sea level, and the inner caldera floor is around 2100 m a.s.l.More than 60 explosive eruptions (mainly with VEI = 2) have been reported to have occurred on Bromo over the past four centuries (source: GVP).However, in contrast to Papandayan, no causalities were reported except for the two tourists killed during the 2004 eruption after venturing too close to the volcano.The present-day active crater, through which magmatic degassing occurs, is Bromo's smallest (500 m in diameter) and northernmost crater (Fig. 2) (Andres and Kasgnoc, 1998;Nho et al., 1996;GVP Bromo -03/1995 (BGVN 20:03); GVP Bromo -05/2004 (BGVN 29:05)).

Methods
SO 2 fluxes were measured using a USB2000 ultraviolet spectrometer, and SO 2 column amounts were retrieved by following standard DOAS calibration and analysis procedures (Kraus, 2006;Platt and Stutz, 2008).The spectral range of the spectrometer is 280-400 nm with a spectral resolution of 0.5 nm FWHM (full width at half maximum).Light entered the spectrometer through a telescope (8 mrad FOV) and via a fiber optic bundle.The Vispec program (http:// vispect.sourceforge.net/)was used to field-track the volcanic plume.Total integration times of 3 s (exposure time 300 ms, 10 co-added spectra) and 1.6 s (exposure time 200 ms, 8 coadded spectra) were applied to Papandayan and Bromo respectively.Reference spectra included in the non-linear fit were obtained by convolving high-resolution SO 2 (Bogumil et al., 2003) and O 3 (Voigt et al., 2001) cross sections with the instrument line shape.A Fraunhofer reference spectrum and ring spectrum, calculated in DOASIS, were also included in the fit.The optimum fitting windows of 302-325 nm and 300-320 nm for Papandayan and Bromo respectively were evaluated by obtaining a near random fit residual with minimum deviation.Figures 3 and 5 show examples of the SO 2 fit.Each spectrum position was determined from a continuously recording GPS unit.Wind speeds were obtained using a handheld anemometer at high points, to the east of the Tengger caldera rim for Bromo and about 200 m above the northern fumarole zone on Papandayan.On this latter volcano, DOAS SO 2 flux measurements were performed in walking-traverse mode (McGonigle et al., 2002).The spectrometer was carried with the telescope pointing to the zenith while walking across the northern part of the eastern crater (Fig. 2).The plume was drifting to the northwest at the time of measurement.A complementary USB4000 spectrometer was positioned on a fixed mode, operating within the 292-446 nm spectral range and with 0.3 FWHM spectral resolution.SO 2 column amounts were retrieved using the same procedures as for the USB2000.On Bromo, access into the caldera was possible using a 4WD vehicle, so DOAS traverses were done on a vehicle (Fig. 2).During measurement, the wind was from the east, forcing the plume partially above the inaccessible relief zone to the west (Fig. 2).To ensure measurements across the entire plume, the telescope was positioned with an inclination of around 30 • from the zenith.

Errors in the SO 2 flux measurements
Error in the SO 2 flux measurements is derived from four different factors, including the retrieved column amount, the distance perpendicular to the plume transport direction, the angle between the assumed wind direction and the traverse path and the plume transport speed (Mather et al., 2006).Error in the retrieved SO 2 column amount depends on many factors (Stutz and Platt, 1996;Hausmann et al., 1999;Kern et al., 2010), but we assume that the dominant error is induced by variable cloudiness that we compensate using artificial constant dark, calculated from each recorded spectrum, in the range of blind pixel (pixel below 290 nm) (Tsanev, 2008).Such corrections account for dark spectrum, offset and stray light.We estimate that the error in the column amount contributes 0.006-0.014to the squared variation coefficient of the total flux, whilst error contributions from the distance traversed perpendicular to the plume and from the assumed wind direction following the approached detailed in Mather et al. (2006) are 0.001-0.006and ∼ 0.003 respectively.Note however that all these errors are negligible in comparison to uncertainties in the plume speed (e.g., Stoiber et al., 1983).We assumed that the plume transport speed is conservative throughout our measurements period with a relative error of ∼ 30-35 %, consistent with Stoiber et al. (1983).

Results and discussion
The DOAS measurements obtained in this work are summarized in Table 2, while Figs. 4 and 5 display plots of all traverses and static measurements.Non-linear fits of recorded spectra under Bromo and Papandayan highlight strong SO 2 signals in the plume (Figs. 3 and 5) with maximum concentrations largely exceeding 100 ppm.m above the background level.It is therefore evident at this stage that Papandayan and Bromo release SO 2 into the atmosphere.

Papandayan's SO 2 emission rate
Results indicate that SO 2 emission rate on Papandayan fluctuates between 0.4±0.1 and 2.8±0.8t d −1 with a mean value of 1.4 ± 0.5 t d −1 .This fluctuation is consistent with DOAS static measurements (Fig. 4) where the SO 2 column amount  The mean emission rate is deduced from traverses 2 and 4 (in bold), whose profiles are closer to the real degassing of Bromo.Traverses 1, 3 and 5 account for a small portion of the plume (see text for further detail).
increased progressively from ∼ 40 ppm.m to ∼ 140 ppm.m over a period of 30 min before dropping to the background level in the following 15 min.The SO 2 flux increased accordingly in traverses 1 to 5 and then decreased in traverse 6.Further measurements are required to delimit Papandayan emissions better.However, it is likely that changes in Papandayan's SO 2 emission rate come from subsurface magmatic-hydrothermal processes with regular magmatic gas discharges pumping up to 0.03 kg SO 2 s −1 into the atmosphere.In any case, Papandayan's SO 2 contribution to the atmosphere is relatively small compared to other volcanic sources (Andres and Kasgnoc, 1998).Assuming that the DOAS results are representative, this volcano releases only about 500 tons of sulfur dioxide into the atmosphere annually.This low SO 2 release into the atmosphere is expected for fumarolic-type activity.Mazot et al. (2008) provide a compilation of Papandayan gas chemistry, and mean SO 2 concentration (0.11 mol%) is significantly lower than H 2 S concentration (0.51 mol%).Assuming that these concentrations are representative, and using the H 2 S / SO 2 molar ratio of 4.6, the H 2 S emission rate from Papandayan can be estimated at around 3.4 t d −1 , more than twice the SO 2 emission rate.According to experimental studies and thermochemical modeling of volatile partitioning between vapor and liquid in two-phase hydrothermal systems, CO 2 is the most abundant hydrothermal gas followed by H 2 Sother sulfurous gases are negligible (Symonds et al., 2001;Drummond and Ohmoto, 1985;Giggenbach, 1980;Reed andSpycher, 1984, 1985;Spycher and Reed, 1989).The strong availability of H 2 S suggests the existence of active hydrothermal processes beneath Papandayan's fumarolic activity, and that a portion of the SO 2 released from the magmatic source probably sinks out by hydrolysis (4SO 2 + 4H 2 O = H 2 S + 3H 2 SO 4 and 3SO 2 + 2H 2 O = S + 2H 2 SO 4 ) (Holland, 1965) or is trapped by other hydrothermal processes.Consequently, when considering SO 2 flux measurements for monitoring purposes, the existence of hydrothermal processes should be taken into account.Alternatively, H 2 S may be a good candidate for monitoring as suggested by Symonds et al. (2001) and Aiuppa et al. (2005).In any case, the DOAS measurement results highlight the potential of SO 2 monitoring of this fumarolically active volcano, and the 1.4 t d −1 of SO 2 released into the atmosphere can henceforth be used as a baseline for future SO 2 flux measurements.

Bromo's SO 2 emission rate
The SO 2 flux measurements from Bromo vary roughly between 0.7 ± 0.2 t d −1 and 32 ± 11.2 t d −1 , but unlike the Papandayan survey, there were no static measurements to support this investigation.Furthermore, despite the strong signal obtained in the SO 2 fit procedure (Fig. 5), all the traverses were not completed (Fig. 5), and SO 2 fluxes for traverses 1, 3 and 5 were dramatically reduced in comparison to traverses 2 and 4. The reason for this disparity was the inclination of the In May 2012, a second DOAS survey was organized on Bromo, but, surprisingly, the results showed no SO 2 emission from the active crater (Fig. 6).Bromo is therefore not a persistent source of SO 2 in the atmosphere, as widely thought.However, this volcano has a high frequency of eruptive activity -about one eruption every 6-7 yr since 1804 (http://www.volcano.si.edu/index.cfm),which indicates that it is nevertheless a major contributor of SO 2 to the atmosphere.Regular measurements over a period of 6-7 yr are necessary to determine the SO 2 emission rate of this volcano better.In any case, this work highlights the potential for DOAS traverses on Bromo and encourages systematic DOAS deployment for monitoring and degassing studies given the short periodicity of eruptive events.

Conclusions
We present the first DOAS SO 2 flux estimates for Papandayan and Bromo, two of the most active volcanoes in Indonesia.Results indicate mean SO 2 emission rates of 1.4 t d −1 from Papandayan's fumarolic activity and more than 22-32 t d −1 of SO 2 released by Bromo during a declining eruptive phase.Results further indicate that Papandayan's SO 2 release is sustained by the regular discharge of gas, although much of the SO 2 is likely trapped by subsurface hydrothermal processes, leading to significant release of H 2 S into the atmosphere.Bromo's SO 2 releases appear not to be persistent over time.This volcano is nevertheless a major source of volcanic degassing into the atmosphere given its 6-7 yr cycle of periodic eruptive activity.In contrast, the permanent degassing on Papandayan represents a negligible contribution of SO 2 to the atmosphere outside eruptive periods.Finally, the DOAS measurements obtained on Papandayan and Bromo are very encouraging given the numerous volcanoes in Indonesia whose degassing has never been evaluated.In addition, this work establishes benchmarks for SO 2 flux monitoring on both Bromo and Papandayan.

Fig. 1 .
Fig. 1.Java island with its 18 active volcanoes.The target volcanoes, Papandayan and Bromo, are indicated.The location of Java and the studied volcanoes within the Indonesian archipelago is provided below.

Fig. 2 .
Fig. 2. Maps of the Papandayan summit and the Bromo caldera.The DOAS traverse zones are shaded in gray.The main degassing points are shown: fumarole zones on Papandayan and the active crater on Bromo.Pictures provide a synoptic view of degassing during the field measurements.

Fig. 3 .
Fig. 3. Example of DOAS SO 2 fit on Papandayan.Blue lines are recorded spectra.The background spectra were acquired by pointing outside the plume.

Fig. 4 .
Fig. 4. Plots of traverse (above) and static (below) measurement results obtained on Papandayan.Time axes were aligned highlighting the increase of SO 2 column amounts in both static and traverse measurements.

Fig. 5 .
Fig. 5. Example of DOAS SO 2 fit on Bromo (left).Measurement spectra are blue.Traverse measurement profiles are shown (right).Traverses 1, 3 and 5 did not catch the bulk concentration while traverses 2 and 4 commenced in the plume (see text).

Fig. 6 .
Fig. 6.Bromo degassing observed during two different periods.In June 2011 (left), during measurement, the degassing was clearly visible 2 km from the volcano.In May 2012, there was no plume on Bromo and no SO 2 detected by DOAS.Only a stagnate white vapor was observed in the active crater (right).

Table 1 .
SO 2 flux contributions from Indonesian volcanoes in the global volcanic emission budgets.

Table 2 .
Estimated SO 2 emission rates for Papandayan and Bromo.