2 Newconceptformonitoring SO emissionsfromTavurvurvolcanoinPapuaNewGuinea D G B T 15

B

T

15

D

G

New concept for monitoring SO

2

emissions

from Tavurvur volcano in Papua New Guinea

Author:

Julia WALLIUS

Supervisor:

Prof. Olafur GUDMUNDSSON

A thesis submitted in fulfillment of the requirements for the degree of Bachelor

in

Physics

Sammanfattning

Nytt koncept för att övervaka SO2 utsläpp från vulkanen Tavurvur i Papua Nya

Guinea

Julia Wallius

Under en period på 29 dagar under oktober månad 2016 kördes ett dubbelstrålat DOAS instrument med brett synfält på ön Matupit i Papua Nya Guinea. Detta för att mäta svaveldioxidutsläppen från vulkanen Tavurvur, som ligger på Gazelle halvön i provinsen East New Britain. På grund av otillförlitliga inställningar under de första 17 dagarna så utgick de slutgiltliga resultaten från mätningar tagna mellan den 19:e och 30:e oktober med ett två dagars uppehåll under den 25:e och 26:e oktober då instru-mentet användes för att istället erhålla vindhastighetdata. Det genomsnittliga flödet av SO2 bestämmdes för denna period till att ligga på 0.27 kg_s med en vindhastighet på 3.9 m

s. Användandet av ett dubbelstrålat DOAS instrument med brett synfält rekom-menderas vid Tavurvur när gasutsläppen är låga och vindriktningen är övervägande sydvästlig. En vinkel på 13 grader över horisonten användes för att erhålla refer-ensspektra av himlen för varje skanningsuppsättning men på grund av SO2spår i dessa spektra så anses denna vinkel vara för liten och att höja denna mellan 5-10 grader för referensspektra rekommenderas för framtida mätningar.

Nyckelord: Optisk fjärranalys, DOAS, Vulkanövervakning, SO2-utsläpp Examensarbete C i geofysik, 1GE037, 15 hp, 2017

Handledare: Olafur Gudmundsson

Institutionen för geovetenskaper, Uppsala universitet, Villavägen 16, 752 36 Uppsala (www.geo.uu.se)

Hela publikationen finns tillgänglig på www.diva-portal.org

Abstract

New concept for monitoring SO2 emissions from Tavurvur volcano in Papua New

Guinea

Julia Wallius

flux of SO2 was determined for the period to be 0.27 kg_s with a wind speed estimation of 3.9 m_s. The use of a dual-beam wide-field-of-view DOAS instrument is recommended at Tavurvur when the degassing levels are low and the wind is predominantly southwest-erly. An angle of 13 degrees above the horizon was used to obtain the sky reference spectrum for each set of scans, but on account of SO2 traces in these spectra this an-gle is considered too low and lifting the sky reference 5-10 degrees is recommended for future measurements.

Keywords: Optic remote sensing, DOAS, Volcano monitoring, SO2 emissions Independent Degree Project C in Geophysics, 1GE037, 15 credits, 2017 Supervisor: Olafur Gudmundsson

Department of Earth Sciences, Uppsala University, Villavägen 16, SE-752 36 Uppsala (www.geo.uu.se)

Acknowledgements

A note on the text

Contents

Acknowledgements iii

List of Abbreviations vii

1 Introduction 1

1.1 Rabaul caldera . . . 1

1.2 Method and purpose of study . . . 2

2 Background 5 2.1 Differential Optical Absorption Spectroscopy. . . 5

2.1.1 Lambert-Beer’s Law . . . 6

2.2 DOAS Measurements and Corrections . . . 7

2.2.1 Fraunhofer spectrum . . . 8

2.2.2 Ring effect . . . 8

2.2.3 High-pass filtering . . . 9

2.2.4 Dark current . . . 9

2.2.5 Sky reference . . . 9

2.3 Dual-beam wide-field-of-view DOAS . . . 9

2.4 Elevation angle . . . 11

2.5 Wind speed calculation . . . 12

2.6 Previous DOAS measurements on Tavurvur . . . 12

3 Method 13 3.1 Materials and instrument assembling . . . 13

3.2 Configuration and testing of instrument. . . 15

3.3 Installation . . . 15

3.4 Scanner settings and properties . . . 16

5 Discussion 25

5.1 Wind direction and speed . . . 25

5.2 Recommendation. . . 25

6 Conclusions 26

List of Abbreviations

COSPEC Correlation Spectroscopy

D Day

DB-WFOV Dual-beam wide-field-of-view

DCO-DECADE Deep Carbon Observatory, Deep Earth Carbon Degassing DECADE Deep Earth Carbon Degassing

DOAS Differential Optical Absorption Spectroscopy

GPS Global Positioning System

KG Kilogram

M Meter

NOVAC Network for Observation of Volcanic and Atmospheric Change

O3 Ozone

PPM Parts Per Million

RVO Rabaul Volcanological Observatory

S Second

SO2 Sulphur dioxide

T Ton

Chapter 1

Introduction

For many years, volcanological monitoring has been an important tool used in order to further our understanding of the workings of volcanoes and the degree to which we can accurately predict a coming volcanic eruption (United States Geological Survey 2016). To this end, scientists have long recognized that the surveillance of volcanic gases play an important part, as gases dissolved in magma provide the driving force of volcanic eruptions. The observation of any changes in the release of certain gases from a vol-cano can therefore help forecast volcanic activity and provide insight into the processes behind eruptions (United States Geological Survey 2015). More specifically, the be-havior of magma bodies as they ascent and descent in relation to eruptions can be better understood by conducting high temporal SO2emission measurements over long time periods (Galle et al.,2010). Together with other geophysical data it is also possi-ble to interpret the relationship between magma ascent and conduit and hydrothermal processes (Olmos et al., 2007) (Sparks, 2003). In fact, the relationship between gas signals and other geophysical measurement methods is conceivably strong, and it is possible that the process of volcanic degassing is responsible for, or at least closely related to, observed ground deformation and seismicity (Oppenheimer and McGonigle,

2004).

Volcanoes also contribute to the release of many toxic gases into the atmosphere, which makes them a relevant study in context with the Earth’s climate. During recent years this specific topic has been given more and more attention, as the emission of a wide variety of volatile species, some of which are converted into aerosols during transport in the atmosphere, makes volcanism one of the main climate-forcing agents and a natural volatile source with important impact upon ecosystems at both local and regional scales (Allard et al., 2016). By measuring the gas emissions, evaluating the data and observing any underlying trends, the potential impact of volcanic emissions on the atmosphere and climate can begin to be understood.

1.1

Rabaul caldera

Chapter 1. Introduction

At the northeastern end of the province East New Britain lies Rabaul, a small town inside a large caldera known as Rabaul Volcano. For many years Rabaul was consid-ered the most important settlement of the province and served as a popular boating destination for both commercial and recreational purposes (Wikipedia 2017), before the eruptions of the sub-vents Tavurvur and Vulcan in 1994 devastated the town.

The present-day shape of the caldera, with a wide opening where the sea forms Blanche Bay to the east, is reckoned to have been formed by a major eruption around 1400 years ago (Volcano Discovery 2017). Besides Tavurvur and Vulcan, other vents along its 14 km length include Turanguna, Rabalanakia, Sulphur Creek, Kombiu and the Beehives.

The Rabual Volcanological Observatory (RVO) is a branch of the Geological Survey of Papua New Guinea and monitors 37 volcanoes (World Organization of Volcano Ob-servatories 2015) spread out through Papua New Guinea. It was formed in 1937, after Vulcan and Tavurvur erupted and killed 507 people (Wikipedia 2017). When Vulcan and Tavurvur had their next eruptions in 1994, the people of Rabaul were better pre-pared by having planned for such an event, and only five people were killed. However, the huge amounts of ash sent into the air during the eruptions caused 80 percent of the buildings of the town to collapse, and Rabual was completely devastated. This caused the province’s capital, along with a major part of the population, to move from Rabaul to the nearby town Kokopo.

Papua New Guinea is considered one of the most significant sources of volcanic degassing in the world (UN World Conference on Disaster Risk and Reduction 2015). After the eruptions in 1994, the most closely monitored volcano, Tavurvur, was identi-fied by Andreas and Kasgnoc (1998) as a sporadic emitter and capable of an output of 301 kg_s of SO2 during eruptive periods (Andres and Kasgnoc, 1998) for the years 1973-1998. More recently, RVO conducted measurements of the volcano for the pe-riod 2009-2014 and found an estimated 3 kg_s of SO2, using a measurement method of traverses with a portable FLYSPEC system (Horton et al., 2006). From the experi-ence gained during the eruptions of Tavurvur between 1994 and 2014, the emission of SO2has been regarded as an important indicator of precursory eruptive activity for this volcano (Mulina,2015)

1.2

Method and purpose of study

Chapter 1. Introduction

Chapter 1. Introduction

FIGURE 1.1: Overview of Rabaul Volcano (Wunderman, 1994). Active vents are marked with black dots and extinct cones with white dots. RVO

Chapter 2

Background

The sensing method used in this study is based on spectroscopic observation of molec-ular species, in this case SO2. By observing their specific electronic, vibrational and rotational transitions as recognized in the process of absorption, specific gases are detectable(Oppenheimer and McGonigle, 2004). In the context of volcano monitoring, a combination of traverses or scans of volcanic clouds using a spectroscopy based sensing method and the measurement of plume transport speed can determine fluxes of different gases (Galle,2003)(Oppenheimer and McGonigle,2004).

DOAS is a recognized technique used for the monitoring of SO2 emissions from volcanoes in order to obtain gas fluxes from active volcanoes around the world, as verified by the project NOVAC (Network for Observation of Volcanic and Atmospheric Change) which utilizes DOAS instruments in a global network of stations for the quan-titative measurement of volcanic gas emissions (Galle et al., 2010)(NOVAC Project 2017). The DOAS instrument, in particular the mini-DOAS, is thanks to its size and weight relatively easy to carry up to a volcanic crater or to carry around the base of a volcano making it an attractive alternative to other remote sensing techniques such as for example COSPEC (Correlation Spectroscopy) (Humaida,2010).

2.1

Differential Optical Absorption Spectroscopy

The following section on Differential Optical Absorption Spectroscopy is largely based on Stefan Kraus Ph.D dissertation on DOAS (Kraus, 2006) and Samuel Brohedes pa-per on the same subject (Brohede,2002).

Chapter 2. Background

introduced the method (Platt and Perner, 1983b). To accurately obtain the desired information, for example the concentration of gas species within a certain volume of gas, it is necessary to understand the absorption behavior, i.e. the impacts of matter to electro-magnetical rays and how the species change the light that goes through it. To understand this, it is necessary to take a look at Lambert-Beer’s Law.

2.1.1 Lambert-Beer’s Law

Lambert-Beer’s Law defines how a layer of different substances will diminish the in-tensity of light at a specific wavelength. Equation 2.1 describes how the absorption structure looks like:

I(λ) = I0(λ)e−σ(λ)·L·n (2.1)

I(λ) is the measured intensity while I0(λ) represents the source intensity, both at given wavelengths. I0 is absorbed according to the substance’s number density n (molecules per cm3_{) along a path of light L. The so called cross-section σ(λ) gives the} wavelength dependent absorption strength of the substance and describes the light absorption at a specific wavelength λ.

For later evaluations it is important to know the cross-section as precisely as possi-ble; different substances have different specific cross-sections. These can, for exam-ple, be obtained from lab measurements or literature data.

The dimensionless quantity σ(λ) · L · n is often referred to as the Optical Depth and here represented by τ .

When measuring light observed through the atmosphere different scattering pro-cesses also contribute to the radiation extinction. Though not technically absorption processes, Rayleigh and Mie scattering behaves in a similar way by scattering light away from the line of vision. Adding this scattering to equation 1 results in the follow-ing:

τ (λ) = lnI0(λ)

I(λ) = L(σ(λ) · n +  R

(λ) + M(λ)) (2.2)

The product of the Rayleigh cross-section and the number density nair for air cre-ates the Rayleigh extinction coefficient R_{(λ). The Mie extinction coefficient }M_{(λ) is} similarly made up of the Mie cross-section multiplied by nair.

To include all atmospheric species with significant absorption cross-sections in the specified wavelength the Lambert-Beer’s law is extended. If the atmosphere is optically thin, i.e. if τ is much smaller than 1, then the additional species i can simply be added:

τ (λ) = L(X i

Chapter 2. Background

In the equation above an additional function has also been added, A(λ), which describes the attenuation of the instrument optics.

The cross-sections σican be separated into two parts, as seen in equation 2.4, de-pending on their variation: σs

i(λ)varies slowly with wavelength and σ 0

i(λ)varies rapidly with wavelength. This latter, rapidly varying, part is also known as the differential cross-section.

σi(λ) = σis(λ) + σ 0

i(λ) (2.4)

Rayleigh and Mie scattering, and the attenuation factor A(λ) are all slowly varying with wavelength and can therefore be expressed as a slowly varying part of the optical depth τ . Combining equation 2.4 with 2.3 gives:

τ (λ) = LX i σ_i0(λ) · ni | {z } rapid, τ0 + L(X i (σs_i(λ) · ai+ R(λ) + M(λ)))A(λ) | {z } slow, τs (2.5)

By pairing the rapidly varying part τ0(λ) of the measured optical depth to the ab-sorption resulting from the differential cross-sections, a separation of the slowly and rapidly varying optical depth is possible. Rayleigh and Mie scattering can together with the attenuation factor now be disregarded by only studying differential cross-sections and the rapidly varying part of the optical depth:

τ0(λ) = lnI 0 0(λ) I(λ) = L X i σ_i0(λ) · ni (2.6)

Equation 2.1 is dependent on the knowledge of the original light source intensity I0 which, in real-life measurements using the sun as light source, can be very difficult to establish as it includes simulating an atmosphere free of absorption. When using DOAS instruments only the differential optical depth is important which includes the much simpler task of instead simulating an atmosphere free of differential absorption.

Figure 2.1 shows the relationship between Optical Depth, Differential Optical Depth, light source intensity, and measured light intensity.

2.2

DOAS Measurements and Corrections

Chapter 2. Background

FIGURE 2.1: The logarithm of the fraction between the light source

inten-sity I0(λ) and the measured light intensity I(λ) describes Optical Depth

τ (λ). The fraction between the interpolated source intensity I ∗ (λ) and the measured intensity I(λ) describes the Differential Optical Depth τ0(λ).

electromagnetic radiation, then Mie scattering occurs. Rayleigh scattering is what pro-duces the blue light of the sky, while Mie scattering gives the sun a white glare, and fog and mist a white light.

In order for the final spectra to be reliable, a number of different corrections must be applied to the light collected with the telescope and spectrometer. The ones used in this study are listed and briefly explained below.

2.2.1 Fraunhofer spectrum

When using the sun as a light source a correction known as Fraunhofer must be applied to the spectrum. This is due to the fact that the sun’s photosphere absorbs a part of the light emitted by itself, so absorption structures will already exist even if there is no atmosphere in the light path. The Fraunhofer correction is evaluated while recording a spectrum looking directly into the sun. This is then applied as a cross-section reference during evaluation of subsequent spectral measurements.

2.2.2 Ring effect

Chapter 2. Background

Ring-corrected Fraunhofer correction is used as yet another reference spectrum during evaluation.

2.2.3 High-pass filtering

To remove all the low frequency parts of the spectrum that are not required for evalua-tion a high-pass filter it applied. This removal is desirable since the procedure behind the DOAS method depends on the separation of the measured data into two parts: high frequency and low frequency data (see section 2.1).

2.2.4 Dark current

A current, known as the dark current, is generated within the instrument even when the spectrometer is not illuminated. This current changes based on the temperature of the instrument, but as the power budget for most stations is limited, regulating this temperature is not possible and all spectra must be corrected for this offset in the pro-cessing stage. To do this, no light is permitted to enter the spectrometer during one single spectral measurement and taken to represent the dark current. All the subse-quent spectra collected within a limited time span will be assumed to have the same dark current, which is then subtracted from every spectrum before further processing.

2.2.5 Sky reference

Another necessary correction to enhance the sensitivity to volcanic SO2emissions con-sists of removing the effect of the background atmosphere. This is done by measuring a spectrum outside of the plume, called the "sky spectra" or "sky reference", which all the subsequent plume spectra are divided by to reduce this background "noise" made up mostly of spectral lines originating from H2O and CO2 and molecular scattering of shorter wavelengths.

2.3

Dual-beam wide-field-of-view DOAS

Chapter 2. Background

When having this setup with a wide-field-of-view, the telescope will "see" in the same scene both regions with high and low concentrations of gas. The radiation the instrument receives will be the sum of radiation from these low and high concentration regions. Since each of these regions are subject to the absorption law, which is ex-ponentially dependent on the concentration, the total signal will come from the sum of exponential terms, which are non-linear. This is not equivalent to having a single expo-nential of the concentrations of the different regions. If all regions instead have a low gas concentration, the absorption law exponential is expanded in a polynomial and only the first term is important. This means that the absorption of different sectors depend linearly on the concentration. The sum of the signals is then equivalent to the sum of concentrations. If the emission of SO2 is large, the concentration will normally be high in some regions and low in others, which would make this WFOV method inaccurate. But for the low emission situation at Tavurvur, it may prove ideal.

Chapter 2. Background

2.4

Elevation angle

It is important to determine a correct elevation angle for the scanner. This is to ensure that the scanner measures the plume against a clear-sky background. An elevation angle that is too high can result in the scanner only observing clear sky, not sampling the plume at all, while an elevation angle that is too low might result in the scanner being blocked by a mountain or other obstacle and not receiving enough light from behind the plume.

To optimize the elevation angle of the scanner the intensity of radiation in the two spectrometer channels needs to be observed as a function of elevation and azimuth angles. The geometry involved in calculating the angle is shown in figure 2.3 below.

FIGURE 2.3: Sketch of elevation angle calculation. The box on the left represents the scanner while the box on the right represents the highest

obstacle in the viewing direction of the telescope.

The box on the left represents the scanner and the box on the right represents the position of the highest obstacle in the viewing direction of the scanner telescope. ∆L is the distance between the two while ∆h is the difference in elevation of the boxes. Θ denotes the elevation angle above the horizontal and ∆ω the field of view of the telescope. These parameters give an equation as follows:

Θ = atan(∆h + ( 1 2) √ ∆h2_{+ ∆L}2_∆ω ∆L ) (2.7)

Chapter 2. Background

2.5

Wind speed calculation

By doing two spectral measurements simultaneously, via two separate spectrometer channels, the dual-beam aspect of the modified DOAS make wind speed measure-ments possible (Johansson, 2009). The fields of view of the two channels are sepa-rated by a small angle. The view fields are directed toward the middle of the gas plume at different distances, creating intersections of the instrument fields of view with the plume. The distance between these two intersections can be calculated if the distance to the plume is known.

The total column variations that have been measured in both directions is registered in a time series, which shows the time delay in variations in the total column (Galle et al.,2010). This delay, together with the distance between the two intersecting fields of view, is then used to derive the wind speed.

2.6

Previous DOAS measurements on Tavurvur

Chapter 3

Method

The DOAS scans were conducted over a period of 29 consecutive days during the month of October 2016. The work process was divided up into three stages: assem-bling the instrument box that was to be installed; the installation process itself; and finally the evaluation of the data. Before the work could begin a suitable measurement location was chosen, where the instrument could be installed. When choosing this lo-cation four things had to be considered: the proximity to the volcano; the predominant wind direction; potential topographical obstacles existing behind the plume; and the safety of the instrument at the particular chosen site. To obtain quantitative results of gas-emission rate the sensor should ideally observe the plume from the side (Arellano, Galle, and Wallius,2017a). Taking this into account with the rest of the considerations, RVO suggested a site in a village on Matupit island, where an RVO monitoring and telemetry site was already in place. This site is southwest of the Tavurvur plume and perpendicular to it in southeasterly wind. It also provides a clear backdrop. It is at about a 3,2 km distance from the volcano, with low risk of instrument theft or vandal-ism. After a preliminary inspection of the site, it was concluded that the mast used for the telemetry transmission antennas offered a good viewpoint, and it was decided to install the scanner around three meters up on the mast to provide it with a field-of-view above the surrounding treetops.

A team of RVO employees and Swedish scientists worked on the preparation and installation of the instrument between the dates of September 28th and October 2nd in 2016. During this time general planning, building an instrument platform and case, configuration and testing of the instrument, test of telemetry, mounting the instrument and telemetry, tests of azimuth and elevation angles of the scanner, and final tests of the system and data transmission to RVO were conducted.

3.1

Materials and instrument assembling

Chapter 3. Method

FIGURE3.1: Instrument in the lab at RVO

Chapter 3. Method

3.2

Configuration and testing of instrument

Before mounting the instrument and platform on the antenna mast, testing had to be done to make sure that the instrument components were all working properly after assembling them in the small box. This was done on site, where RVO had a previously established telemetry station inside an old Japanese bunker. The tests were done by connecting the instrument’s power and data cables to the existing radio and power installations inside the station and it was concluded that every component was working properly.

The scanner configuration was also changed to match the conditions that applied; i.e. the compass direction, the elevation of the instrument, the latitude and longitude coordinates of the site, the tilt angle and cone angle etcetera (see Table 3.1). This was not given extremely high priority however, since these configurations would also be changeable from RVO after the installation process thanks to the successful telemetry connection.

3.3

Installation

The second stage of the work process was to install the instrument at the chosen Matupit site. This was done with assistance from RVO, more specifically from RVO employee Ezequiel and Chalmers scientist Santiago Arellano who did the most of the mounting and installing of the instrument and platform on the mast.

The telescope was mounted so that the small window on its side had an unob-structed view of the volcanic gas plume as seen in figure 3.3. When activated, the stepper motor rotated the telescope head between the angles required to take a sky reference, a dark current reference, and the plume scans. These angles were decided based on trial and error, with the accurate settings not finalized until the 19th of Octo-ber, 16 days into the study. A table listing the final parameters for the spectrometer as well as their simple definitions can be found in figure 3.4.

It was noted during this installation that two components of the system had failed: the timer used to control the operation time of the instrument and the electronic distribu-tion unit. These were therefore replaced, solving the problems, but RVO staff informed that similar failures had occurred after the installation of a stationary DOAS scanner system at the Rabaul Hotel (Arellano, Galle, and Wallius,2017a).

The elevation angle of the instrument was determined according to equation 2.12, using known values for the distance, heights and field of view:

Chapter 3. Method

Together with equation 2.12 the elevation above the horizontal became:

Θ = atan(110 + 1 2 √ 1102_{+ 3200}2_0.133 3200 = 5.77deg (3.2)

Each step of the stepper motor is 1.8 degrees. The elevation angle therefore corre-sponded to three or four steps of the scanner’s stepper motor.

3.4

Scanner settings and properties

The coordinates and characteristics of the scanner are shown in Table 3.1 and Figure 3.2 shows the location and pictures of the station. Topographical obstacles in the viewing direction are at a distance of 3,2 kilometers away while the distance to a plume drifting towards the northwest is estimated to be around 2 kilometers.

3.4.1 Measurement modes

During the study, two measurement strategies were employed: a ’sensitive detection’ mode and an ’average flux’ mode. The sensitive detection mode was used for the greater part of the study, with the average flux mode running only once or twice a week in order to gain wind speed data.

Sensitive detection

In this mode one channel of the spectrometer measured at full spectral resolution and an averaging time matching the fluctuation rate of the emission (typically 30 seconds), allowing frequent and sensitive detection of the total column density of SO2 (Arellano, Galle, and Wallius, 2017a). To obtain gas fluxes from this mode wind-speed data had to be acquired independently.

Average flux

Chapter 3. Method

Though the common wind patterns observed at Tavurvur make this atypical, it must be noted that large column densities can result from gas emissions being transported towards the measurement station, since the optical path through the gas would in this case be large. In such situations the estimate of the flux must therefore be done with care as the extent and speed of the plume would be difficult to determine (Arellano, Galle, and Wallius,2017a).

TABLE3.1: Coordinates and characteristics of the scanner

Name Matupit

Latitude -4.243760

Longitude 152.190128

Altitude 32 m asl

Tilt angle 0 deg

Cone angle 90 deg

Spectrometer serial number D2J2355

Compass direction 328 deg

MotorStepComp 119

IP address 192.168.1.33

Chapter 3. Method

FIGURE 3.2: A) Location of the station, viewing direction (red), typical

di-rection of the plume (black) and topographic profile in the viewing didi-rection. B) Photograph of the viewing direction with a sketch of an emission and

Chapter 3. Method

FIGURE 3.3: The general and specific parameters of the scanning DOAS

instrument.

3.5

Data evaluation

Evaluation of the measured spectra was done using a software called NovacProgram, developed at Chalmers. When calculating the fluxes in this program, a wavelength range of 308-317 nanometers was used. Below this range there was a strange spike in the measurements while wavelengths above it showed little sensitivity to SO2. A high-pass filter was applied to remove unwanted low frequency parts from the data, and the Fraunhofer and Ring effect corrections discussed in section 2.2 were now applied in the form of reference spectra. The SO2 fluxes were derived from the spectra using differential cross-sections of SO2 and O3. The fitting procedure for this is implemented using a non-linear least squares fit. A more detailed description of the NovacProgram can be found in the article of Galle et al. about NOVAC and volcanic gas monitoring (Galle et al.,2010).

Chapter 3. Method

FIGURE 3.4: The final settings used in NovacProgram to evaluate the spectra.

calculating the SO2 fluxes. The final two parameters needed to acquire the flux was an estimate of the distance to the gas plume and the plume width. The former was judged from a map of the area to be about 2000 m, while the latter was calculated by multiplying the distance to the plume by the field-of-view of the telescope (₁₅2 radians).

Chapter 4

Results

4.1

Offsets

After calculating the SO2 fluxes using Novac there were a lot of negative values in the results. Negative values will show up if the "sky reference" spectrum before each set of measurements contain traces of SO2.

On the 23rd of October the instrument was run in "sensitive mode". Figure 4.1 shows the measured SO2 emission during the nine hour operating period of the scan-ner. It consists of 34 individual measurements, each containing a clean air sky refer-ence spectrum and 20 individual plume measurement spectra with a duration of 30-45 seconds. When evaluating these columns of the plume spectra, each one is divided by the preceding sky reference to cancel out any background light not relevant to the measurements. This sky reference should not contain any SO2. However, since there are negative values for the SO2 emission, the figure below shows this not to be the case. In each data-set there is a varying degree of offset caused by the sky reference containing SO2 traces. These negative numbers can be compensated by including an offset-level into the flux calculation. This level tells NovacProgram where the "true" zero-level is, and can be calculated automatically or be explicitly specified.

Chapter 4. Results

FIGURE 4.1: Measurement of the emission from Tavurvur on the 23rd of October done in "sensitive mode".

FIGURE 4.2: The emission measurements from the previous figure after correcting for different offset. The offset correction values are shown in

Chapter 4. Results

4.2

SO2 fluxes

The average SO2 flux during the period 19th to 30th of October 2016 was found to be 0.27 kg_s . Two days, the 25th and 26th, were excluded from this period as they were used to take wind measurements in average flux mode. The variation of the averages may be a result of different amount of SO2 in the plume, or because of a change in wind direction or wind speed which would also create an increase or decrease in the SO2 fluxes.

FIGURE 4.3: Each red dot represents the average SO2 flux for that day.

The total average over the entire period was found to be 0.27 kg_s.

Chapter 4. Results

FIGURE 4.4: Wind measurement data for the 25th of October, 2016. The average is shown as a red line.

Chapter 5

Discussion

5.1

Wind direction and speed

If the wind direction changes, and instead moves towards or away from the instrument, there will be an increase or decrease in the resulting calculated flux. This is because the instruments calculation of the SO2 flux is based on the cross-sections of the plume. If the wind changes and travels towards the instrument then th optical path through the plume, which the instrument integrates along, will be larger then the cross-section and the resulting flux will be larger as well. If the wind travels away from the instrument, the opposite will instead occur with a resulting flux that is smaller then the actual value.

If there is a significant change in wind speed compared to the values acquired during October 25th and October 26th then the SO2 fluxes will be affected since the wind speed is used when calculating the fluxes.

5.2

Recommendation

Chapter 6

Conclusions

Bibliography

Allard, P. et al. (2016). “Prodigious emission rates and magma degassing budget of major, trace and radioactive volatile species from Ambrym basaltic volcano, Van-uatu island Arc”. In: Journal of Volcanology and Geothermal Research 322. Un-derstanding volcanoes in the Vanuatu arc, pp. 119 –143. ISSN: 0377-0273. URL:

http://www.sciencedirect.com/science/article/pii/S0377027315003273. Andres, R.J. and A.D. Kasgnoc (1998). “A time-averaged inventory of subaerial

vol-canic sulfur emissions”. In: Journal of Geophysical Research: Atmospheres 103.D19, pp. 25251–25261. ISSN: 2156-2202. DOI: 10.1029/98JD02091. URL: http://dx. doi.org/10.1029/98JD02091.

Arellano, S., B. Galle, and J. Wallius (2017a). “New concept for gas monitoring of Tarvurvur volcano, PNG”. Installation report, unpublished internal document. — (2017b). “Report on volcanic plume measurements on volcanoes in Papua New

Guinea”. Fieldwork report, unpublished internal document.

Boichu, M. et al. (2004). “High temporal resolution SO2 flux measurements at Erebus volcano, Antarctica”. In: Journal of Volcanology and Geothermal Research 190, pp. 1455–1470.

Brohede, S. (2002). “Differential Optical Absorption Spectroscopy -How does it work and what can it measure?” In: URL: http://home.elka.pw.edu.pl/~rgraczyk/ DOAS.pdf.

Galle, B. et al. (2010). “Network for Observation of Volcanic and Atmospheric Change (NOVAC) - A global network for volcanic gas monitoring: Network layout and instru-ment description”. In: Journal of Geophysical Research 115.

Galle B., et al. (2003). “A miniaturised ultraviolet spectrometer for remote sensing of SO2 fluxes: a new tool for volcano surveillance”. In: Journal of Volcanology and Geothermal Research 119.1-4, pp. 241–254.

Horton, K.A. et al. (2006). “Real-time measurement of volcanic SO2 emissions: valida-tion of a new UV correlavalida-tion spectrometer (FLYSPEC)”. In: Bulletin of Volcanology 68.4, pp. 323–327.DOI: 10.1007/s00445-005-0014-9. URL:http://dx.doi.org/ 10.1007/s00445-005-0014-9.

Humaida, H. (2010). “SO2 emission measurement by DOAS (Differential Optical Ab-sorption Spectroscopy) and COSPEC (Correlation Spectroscopy) at Merapi volcano (Indonesia)”. In: Indonesian Journal of Chemistry 8.2.ISSN: 2460-1578.URL:http: //pdm-mipa.ugm.ac.id/ojs/index.php/ijc/article/view/358.

Johansson M., et al. (2009). “A miniaturised ultraviolet spectrometer for remote sensing of SO2 fluxes: a new tool for volcano surveillance”. In: Bulletin of Volcanology.DOI:

BIBLIOGRAPHY

Kraus, S. (2006). “DOASIS A Framework Design for DOAS”. PhD thesis. Mannheim University.

Lampel, J., U. Friess, and U. Platt (2015). “The impact of vibrational Raman scatter-ing of air on DOAS measurements of atmospheric trace gases”. In: Atmospheric Measurement Techniques 8.9, pp. 3767–3787.DOI:10.5194/amt-8-3767-2015. Mulina, K. (2015). “Monitoring SO2 emissions from Tavurvur volcano, Rabaul, PNG”. NOVAC Project (2017). URL:http://www.novac-project.eu/indexProjectSum.htm. Olmos, R et al. (2007). “Anomalous Emissions of SO2 During the Recent Eruption

of Santa Ana Volcano, El Salvador, Central America”. In: Pure and Applied Geo-physics.

Oppenheimer, C. and JS A. McGonigle (2004). “Exploiting ground-based optical sens-ing technologies for volcanic gas surveillance”. In: Annals of Geophysics 47.4, pp. 1455–1470.

Platt, U. (1994). “Differential optical absorption spectroscopy (DOAS)”. In: Air monitor-ing by spectroscopic techniques, Chem. Anal. Ser. Vol. 127, pp. 27–84.

Platt, U. and D. Perner (1983a). “Measurements of Atmospheric Trace Gases by Long Path Differential UV/Visible Absorption Spectroscopy”. In: Optical and Laser Re-mote Sensing. Ed. by Dennis K. Killinger and Aram Mooradian. Berlin, Heidelberg: Springer Berlin Heidelberg, pp. 97–105. ISBN: 978-3-540-39552-2. DOI: 10.1007/ 978-3-540-39552-2_13. URL:http://dx.doi.org/10.1007/978- 3- 540-39552-2_13.

— (1983b). “Measurements of Atmospheric Trace Gases by Long Path Differential UV/Visible Absorption Spectroscopy”. In: Optical and Laser Remote Sensing. Ed. by Dennis K. Killinger and Aram Mooradian. Berlin, Heidelberg: Springer Berlin Hei-delberg, pp. 97–105. ISBN: 978-3-540-39552-2. DOI: 10.1007/978-3-540-39552-2_13.URL:http://dx.doi.org/10.1007/978-3-540-39552-2_13.

Sparks, R. S. J. (2003). “Dynamics of magma degassing”. In: Volcanic Degassing. UN World Conference on Disaster Risk and Reduction (2015). URL: http : / / www .

wcdrr.org/sasakawa/nominees/3907.

United States Geological Survey (2015).URL:http://volcanoes.usgs.gov/activity/ methods/gas.php.

United States Geological Survey (2016). URL: https : / / volcanoes . usgs . gov / vhp / monitoring.html.

Volcano Discovery (2017). URL: https : / / www . volcanodiscovery . com / rabaul -tavurvur.html.

Wikipedia (2017).URL:https://en.wikipedia.org/wiki/Rabaul.

World Organization of Volcano Observatories (2015). URL: http : / / www . wovo . org / 0500_0504.html.

Wunderman, R. (ed) (1994). “Global Volcanism Program. Report on Rabaul (Papua New Guinea)”. In: Bulletin of the Global Volcanism Network. Smithsonian Institution.