DECADE campaign to Papua New Guinea (report) Contributors : Brendan McCormick Kilbride, Lois Salem, Roberto D’Aleo, Santiago Arellano, Bo Galle, Julia Wallius, Peter Barry, Kila Mulina Overview In September 2016, seven researchers from European universities undertook a month-long campaign of volcanological fieldwork in Papua New Guinea in collaboration with the Rabaul Volcanological Observatory (RVO). The research chiefly focussed on the gas emissions of active volcanoes in three regions of the country (Figure 1). In the following report, we summarise the fieldwork and describe immediate future work drawing on our samples and measurements.

Figure 1 Overview map of the New Britain and Bougainville regions of Papua New Guinea, with major field sites marked. Fieldwork East New Britain (Rabaul) Overview of logistics The active cone of Tavurvur (Figure 2) is readily accessed by car from Rabaul, and a climb of only 40 minutes is needed to reach the crater rim. The terrain is very challenging however, particularly on the crater rim and inside: the entire ground comprises piles of bombs and boulders, often separated by steep and wide crevasses. A walk around the crater rim was attempted but resulted too difficult and dangerous, and travelling across the crater to the central degassing area took approximately one hour from the rim. The walls of the crater are very steep and only accessible from tracks on the northern and eastern flanks.

Figure 2 Partial view of Rabaul caldera, with active cone Tavurvur on the left of the image (and inactive cone Mt Turanguna, or “South Daughter” on the extreme left). The second most-recently active centre is Vulcan, across the caldera on the right of the image. In the foreground and at other locations towards Tavurvur are hot springs of around 60-80°C. Remote sensing of gas emissions During four days, a combination of scanning and traverses with DOAS systems were conducted to obtain the flux of SO2 from Tarvurvur. The strategy was to co-locate 2 scanning DOAS in a configuration that would allow to get the plume height and direction, and a third instrument below the plume to measure plume speed by the dual-beam correlation method. Simultaneously, walking traverses with a mobile-DOAS unit were performed along the northern base of the volcano. Tarvurvur presently has passive, weak, ash-free emissions that tend to accumulate inside the crater and then release sporadically, without forming a continuous plume, which made challenging to estimate the gas flux. However, by prolonged measurement sessions some emissions were captured and the background flux could certainly be characterized. The weather conditions were good for most of the time, but on several occasions we experienced quick changes from clear sky to mild or heavy rain, that did not last long. The temperature was above 30°C most of the time. At the end of the campaign, we installed a wide-field-of-view fixed system looking to the emissions close to the crater from a distance of about 2 km, in the premises of a station operated by RVO on the Matupit village. The instrument will operate continuously until at least the end of October 2016, under the supervision of Swedish MSc. student, Julia Wallius, and will complement the existing SO2 flux monitoring station at Rabaul Hotel, in operation since 2015. In tandem with the DOAS we deployed a dual UV camera mobile system. Measurements of SO2 emissions were performed at different distances, initially at a distance of 1400 meters, but the flux was too weak so on subsequent days we moved closer, up to 300 meters (crater rim). The goal was to characterize the different fumaroles on top of the volcano and inside the crater.

Figure 3 Top. Overview of Tavurvur crater, looking south from crater rim. Highest temperature emissions (~150°C are from the site with bluish plume on the central crater floor). Bottom. Photo of Giggenbach sampling at this central site obtained from UAV. In-situ measurements of gas emissions A Multigas instrument was deployed in the crater of Tavurvur volcano. Due to the rough terrain on the crater rim, no rim traverse walk could be completed. For each sampling point, temperatures were also measured with a thermocouple. To support this thermal data IR photography was obtained of the entire crater with a FLIR camera during the evening, to avoid the sun-heat interference. Three fumaroles were also measured by Multigas. In parallel to the MultiGAS measurements, a compact system dubbed ‘Sunkist’, developed by Heidelberg University, was used to obtain measurements of the CO2 and SO2

concentrations (and ancillary temperature, pressure and relative humidity data) on two fumaroles and the central vent of Tarvurvur. The goal here was to get to independent sets of measurements for comparison and to test the performance of the compact system, later used in the campaign for plume sampling onboard an unmanned hexa-copter. Sampling of gas emissions Gas emissions were sampled in Giggenbach bottles for full chemical analysis, and copper tubes for noble gas analysis. The ideal samples for characterisation of magmatic gases are high temperature (>400 C) emissions but these were not present. In Rabaul, the highest temperature emissions encountered were ~150 C, at two locations in the crater of the dormant volcano Tarvurvur. We obtained both Giggenbach and copper tube samples here. Unfortunately conditions in the central crater were not optimum for effective sampling: the terrain comprised rubbly piles of boulders and deep crevasses, and consequently it was difficult to access any location where the gas emissions were directly emanating from the substrate; rather they were mixing extensively with air in the pore space beneath our feet. Analyses of two of the Tavurvur samples by Tobias Fischer confirm our fears of air contamination. Around Rabaul, hot spring emissions were also sampled: gas bubbles were observed around the coastline of the caldera, and two locations were sampled by Giggenbach and copper tube. Airborne photogrammetric surveys Rabaul is an ideal location for airborne surveys using the Phantom 4 drone as the maximum flight altitude of 500 m exceeds the summit elevation of all but the largest of the various volcanoes around the caldera. Tavurvur’s northern flank is covered with volcanic bombs and bomb craters, and there is the possibility that we can use our airborne footage to undertake studies of the dynamics of the eruption which ejected these bombs. Petrological sampling Petrological sampling at Rabaul focussed on the joint eruption of Vulcan and Tavurvur in 1994 and subsequent eruptions of Tavurvur. Sampling was undertaken under the guidance of RVO geologist Mikhail Sindang. Representative samples of both cones’ 1994 tephra were obtained, and a detailed log through deposits of Tavurvur’s explosive products from 2006 to the present day was made.

Bougainville (Bagana) Overview of logistics Bagana is among the most remote volcanoes in Papua New Guinea, and is not presently monitored by any instruments. An observer, Thomas, employed by RVO provides daily reports by radio from a village on the SW side of the volcano. In recent times, activity has focussed on the NW side, and there have been some delays in news of fresh unrest reaching RVO. Access is by flight from Rabaul to Buka, and then by boat to Torokina (Figure 4). There is one car in the Torokina region which can be used to transport people and gear inland to Gotana, where there is accommodation sufficient for a large group. After periods of high rainfall the Torokina river can become impossible to cross safely, and at these times the villages inland of Torokina become cut off. The return trip followed the same route, and it was delayed for some hours due to heavy rain. Gotana is approximately 2-3 hours walk from the base of the volcano through thick forest, involving numerous river crossings, and local guides are essential to reach the volcano. On the NW side, the village of Wakovi sits high on a cliff above the Torokina river and so is safe from possible lahar or PDC originating from the volcano, which sits at the head of the river. Wakovi is roughly one hour’s walk from a location suitable for UV remote sensing of the gas plume, and around 2 hour’s walk from the base of the edifice proper. Our opinion is that Wakovi is a better base for working on Bagana than Gotana, but Gotana was chosen because it has more available accommodation and the current RVO observer has relatives there. In future campaigns, we will work from Wakovi but only after sensitive negotiation with RVO. An alternative route to access the volcano is on the SW side, via the village of Patsikopa (Figure 5). Beyond Patsikopa the trails through the forest are substantially less well-travelled, making guides even more essential (in some cases to cut a trail). This route also has the additional hazard of geothermally-heated rivers which must be crossed: care needs to be taken to avoid scalding.

Figure 4 Environs of Bagana volcano, showing volcanoes in white and settlements in yellow.

Figure 5 View of Bagana volcano, around 5:30pm, from Patsikopa village. Clear views of the summit are very rare.

Remote sensing of gas emissions Scanning-DOAS was conducted during four days from two locations at about 30 and 60 minutes walk from Wakovi, on the NW of Bagana. A walking-traverse with a mobile-DOAS was attempted once across a huge lahar deposit on the foot of the volcano, but it was not possible to find a suitable path beyond the deposit to complete the traverse under a wide and meandering plume. On one day, measurements were done to obtain plume height and direction with 2 scanners, whereas on the rest of the days, only one scanner was used for flux measurements and one for plume-speed measurements under the plume. The emission was continuous and strong, with moderate amounts of ash and high amount of condensed particles. The weather conditions were good early in the morning, when most of the valid measurements could be taken, and progressively changed to cloudy and heavy rain in the afternoon. The heavy rain may produce secondary lahars and river overflows, making the return hike to the camping site in Gotana dangerous and extenuating; therefore, the last two days were spent entirely in Wakovi, which made possible to take good-quality early-day measurements. In-situ measurements of gas emissions Direct sampling of the gas plume was achieved by a compact ‘Sunkist’ CO2/SO2 instrument onboard a 6-propellers ‘SkyEye’ drone. Two successful measurements were conducted on Bagana, reaching about 1600 m above ground level at the same site of simultaneous scanning-DOAS measurements. The available batteries limited the flight time to about 15 minutes (which is ca. 3 times less than possible for this system), making necessary to find a location below the plume to launch the drone. The ‘SkyEye’ has been tested for payloads of

up to 1.5 kg (the ‘Sunkist’ weights 0.5 kg) up to 2 km altitude, beyond which radio signal may be lost. Sampling of gas emissions At Bagana no sites of magmatic gas emission could be reached due to the steepness and instability of the volcanic edifice, and ash venting eruptions from the closest degassing site. Hot springs around the base of the edifice were investigated but manifested as heated (50-60 C) rivers, which flowed too fast for locations with bubbling gas to be identified. Airborne photogrammetric surveys Flights over Bagana’s lower slopes were made from both the NW and SW sides (Figure 6). The elevation of these two locations limited the height of the edifice that could be mapped, but nonetheless a large volume of footage was shot and will be useful in developing a new DEM for the volcano. In particular, we obtained footage of the 2014 and 2005 lava flows, which were also sampled. The airborne footage also proved useful in assessing the potential of an ascent of the volcano. Currently we deem this unsafe due to the recent occurrence of large debris avalanches and moreover the presence of an active flank vent (around 750-800m elevation on the NW side) exhibiting active degassing and ash venting eruptions. We also noted large rockfalls from the summit crater down the upper slopes and through the fumarole field around the flank vent.

Figure 6 Example view of Bagana field site, looking NE from the Phantom 4 UAV. The 2014 and 2005 lava flows were mapped from the air, and also sampled. Petrological sampling We sampled blocky lava flows erupted in 2005 and 2014, from locations on the SW and NW flanks of the volcano respectively (Figure 6). As the drone footage testifies, these lava flows are huge features, and it is likely that there are better samples higher up on the edifice. Our

samples are taken from the rubbly tip of each lava flows, given that these locations were the most readily sampled from our basecamp location. To sample other locations, we would need to explore the possibility of camping at the base of the edifice or even some distance up the lower slopes. Given the recent (one month old) lahar deposits around the base of the volcano we did not judge this a prudent approach at the present time. West New Britain (Kimbe, Ulamona) Overview of logistics Travel to Kimbe was done by flight from Rabaul to the Hoskins airport, it is also possible to drive for at least 8 h under a partially built road. Work in the Kimbe area began with a visit to the Kimbe Volcano Observatory, an outpost of RVO, including staff from the Provincial Disaster Office. We were informed that the northern end of the Willaumez Peninsula (the location of Dakataua volcano) was affected by a local land dispute (including inter-tribal violence) and was therefore off-limits. For the remainder of our time around Kimbe we were accompanied by an armed police escort. It may be that this is always necessary in the Willaumez/Talasea area. Around Kimbe, an ideal basecamp is the conservation centre Mahonia Na Dari, located very close to Garbuna volcano, and within easy drive of other geothermal fields in Talasea and in the future Dakataua volcano too. Pago volcano lies east of Kimbe, close to Hoskins airport but is difficult to access (requiring up to a day of jungle hiking to reach the base of the most recently active cone) and was not visited this time. Ulawun volcano lies further east (3-4 hours drive from Hoskins/Kimbe) and the basecamp for work on this volcano was Ulamona Mission. It is possible to reach Ulamona by boat from Rabaul and in future campaigns this may be a better option that flying to Hoskins and then driving back east. Roads from Rabaul to Ulamona are not always passable during intervals of heavy rain. Remote sensing of gas emissions With the support from the Disaster Provincial Office, who provided a volcano guide and a driver with a car, two members of the group concentrated on gas flux observations in Ulawun volcano (Figure 7), while the other members conducted gas and petrological sampling in the Kimbe area. During three days it was possible to measure the flux of SO2 with a combination of car-traverses with a mobile-DOAS and stationary measurements with one scanning-DOAS and one dual-beam instrument below the plume. From the traverses, a suitable location below the plume was found for temporary deployment of the scanners. Traverses were done on the northern and western sides of a gravel road around Ulawun, at a typical distance of 10 km from the volcano and an elevation of 60 m a.s.l. Between this road and the volcano flanks there are dense oil palm plantations, and several partially passable roads. The weather conditions were good in the early hours of the day and turned to cloudy by mid-morning, with occasional showers. Ulawun is presently degassing passive but continuously, forming an ash-free plume that typically reaches a few tens of m above the 2334 m high summit.

Figure 7 Ulawun volcano at dawn. Like Bagana, summit views are very rare.

In-situ measurements of gas emissions The drone+CO2/SO2 instrument was used on two occasions at Ulawun. It was necessary to make incursions through some roads in the oil-palm plantations to find a location below the plume at a higher elevation than the ring road. The drone was lifted about 1900 m above ground, which was enough to reach the dispersed plume. Due to higher cloud cover and elevation than in Bagana, the system was out of sight during these measurements but performed according to the expectations. Sampling of gas emissions In the Kimbe area, two geothermal sites were sampled, at Pangalu (Figure 8) and Cilanga. In each place, there were bubbling pools to sample (60-100C), and at Pangalu a limited number of low temperature fumaroles (~100C). A sample was obtained of low temperature fumarolic emissions (~98 C) from the solfatara field of Garbuna volcano (Figure 9). The sampling work at these locations was mostly of a reconnaissance nature. Cilanga aside, the geothermal areas studied are very large and offer opportunities for extensive sampling in order to fully characterise their emissions. Our initial sampling, aerial footage, and access on foot all lay important groundwork for future sampling campaigns. Airborne photogrammetric surveys The purpose of airborne surveys where undertaken in Kimbe was mostly to aid reconnaissance of each location, in order to better plan more detailed field campaigns in the future. For example, airborne imagery obtained at Garbuna revealed the existence of numerous sites of degassing across the edifice which were not evident from a ground-based perspective.

Figure 8 Aerial photographs of the main geothermal areas of the Talasea coast, centred on Garua Harbour volcanic system (no identified eruptions in Quaternary or Holocene) close to Pangalu village.

Figure 9 Two views from the Phantom 4 UAV of the summit area of Garbuna volcano, with its extensive fumarole fields. Top view is looking ~NE, and south view approximately N. Petrological sampling No samples were taken during this leg of the campaign. Exposure in the geothermal areas is of poor quality, with rocks being extensively altered and encrusted. No eruptions of the Talasea volcanic system (beneath Pangalu hot springs) are reported in either the GVP Holocene or LaMEVE Quaternary databases. When visiting the Garbuna edifice we were without a local geologist who could advise to the age/location of specific deposits; sampling recent eruptions could be a goal of future trips. The Talasea coast is mostly covered by

explosive deposits from large eruptions of either Dakataua in the north or Witori in the south, which have had numerous historical eruptions; careful sampling is necessary to distinguish these deposits in the field. Future Work Briefly, we outline the imminent work we will undertaken from the basis of our field campaign. Petrological analyses Petrological analyses at Rabaul require discussion with a research group at the Earth Observatory of Singapore, who have also recently sampled the recent eruptive products of Tavurvur. McCormick and Salem intend to meet with them at AGU Fall Meeting and ensure that complementary work is undertaken. Analyses of Bagana lavas will characterise the chemistry of the 2005 and 2014 effusive eruptions and place it in the context of earlier (pre-1990s) eruptions, for which geochemical exists. We seek to understand the variance in composition through time and between eruptions. We are particularly interested in the possible differences in volatile budget of distinct eruptions. Our samples of effusive products will be compared to samples of 2016 ash we collected, again with a focus on exploring the differences in volatile content of products erupted under different styles of activity. Gas geochemistry Giggenbach flask samples are being analysed by Tobias Fischer at University of New Mexico, for bulk chemistry (major, minor, trace elements) and isotopic analyses. Copper tube samples will be analysed by Peter Barry at the University of Oxford, with a focus on isotopic analyses of noble gas species. Gas emission rates Chalmers is currently working on the evaluation of SO2 gas emission rate measurements conducted at Tarvurvur, Bagana and Ulawun. Three main datasets exist: scanning-DOAS measurements, plume-speed dual-beam measurements, and traverses with mobile-DOAS. We are also working on the evaluation of one year of data collected by a RVO stationary scanning-DOAS system installed by VDAP at Rabaul Hotel; as well as on the evaluation of one month of data from a temporary installation at Matupit, with a more sensitive wide-field-of-view remote sensor. This system may stay longer if found useful and if funding can be secured for its permanent transfer to RVO. During the next weeks, we will report the results of CO2 fluxes from the three volcanoes, based on the measurements done with the Sunkist instrument on the crater of Tarvurvur and from the drone inside the plumes of Bagana and Ulawun. This work will involve laboratory calibrations and analysis with colleagues from Heidelberg and Mainz (contact person is Nicole Bobrowski). Drone imagery

In Cambridge, we are using imagery obtained from drone flights at Bagana to reconstruct the topography and potentially volume of recent lava flows. These will be compared to satellite radar measurements led by PhD student Amy Sharp, under the supervision of Prof Geoff Wadge at the University of Reading, and provide a means of ground-truthing these relatively novel satellite observations. The imagery obtained from Rabaul will allow a DEM of the Tavurvur cone to be constructed and potentially a study of ballistic ejecta during the 2014 eruption. Satellite observations Also in Cambridge, we are undertaking analyses of longterm satellite records of SO2 and thermal emissions from Bagana, seeking to build an understanding of differences in gas output during different styles of activity (ash venting, effusive eruption, quiescence). Thermal emissions provide a good proxy for intervals of effusive activity and are measured by MODIS and ASTER instruments in the infrared. SO2 emissions are monitored in the IR and UV by numerous instruments, but particularly the UV spectrometer OMI. The measurements made by the Chalmers and Palermo members of the group will provide important ground truthing for these satellite SO2 observations. Funding We acknowledge funding from the Deep Carbon Observatory DECADE project; from the NERC Centre for Observations and Modelling of Earthquakes, Volcanoes and Tectonics; from the SIDA Minor Field Studies program; from Chalmers University of Technology; from the University of Cambridge; from the University of Palermo; and from Rabaul Volcano Observatory.