Chapter 2: G. Williams-Jones, J. Stix & P.A. Nadeau / Using the COSPEC in the Field
63
Chapter 2
Using the COSPEC in the Field
Glyn Williams-Jones1, John Stix2 and Patricia A. Nadeau1
1
Department of Earth Sciences, Simon Fraser University, Burnaby, British Columbia,
Canada, V5A 1S6. email: glynwj@sfu.ca
2
Department of Earth & Planetary Sciences, McGill University, Montreal, Quebec,
Canada, H3A 2A7. email: stix@eps.mcgill.ca
1. INTRODUCTION
This chapter is aimed at both the first time and experienced user and details the possible
methods by which COSPEC measurements are made in the field. Whether you are a graduate
student going abroad for fieldwork or a Volcano observatory staff member performing
background measurements, it is crucial to understand the strengths and weaknesses of COSPEC
measurements before commencing a survey. It is also important to tailor the most appropriate
method with the activity level of the volcano. For example, what is a sufficient number of
traverses for a background or reconnaissance survey compared to monitoring of an imminent
eruption?
We therefore look first at the importance of meteorological conditions, notably determining
the wind speed, and then discuss the various techniques (e.g., ground-based mobile, airborne,
stationary, etc.), along with their advantages and disadvantages, used for COSPEC
measurements. The different methods of recording the COSPEC data (e.g., chart-recorder,
datalogger, computer, etc.) are then examined, followed by a discussion of potential problems
that may arise while making measurements and the best way to remedy the situation. Data
reduction is also examined, with a discussion of the formulae used to calculate SO2 fluxes and
the various methods of treating analogue and digital data. Once the measurements have been
made and the data analysed, one must then decide upon the most appropriate and useful way to
present the data in order to facilitate interpretation.
2. METEOROLOGICAL CONDITIONS
2.1.
Wind
SO2 flux calculations are dependent upon knowledge of the plume direction and speed (i.e.,
velocity), however, direct measurement of these components is often exceedingly difficult. We
therefore measure wind speed and wind direction, which we assume to represent the plume
velocity. Ideally, this information is obtained from instruments in the downwind plume of the
volcano (e.g., using a radiosonde). When making airborne measurements, these data can be
obtained by flying at the level of the plume (Doukas, 2002). However, for ground measurements,
Williams-Jones, G., Stix, J. & Nadeau, P.A. (2008) Using the COSPEC in the Field. In: Williams-Jones, G., Stix, J.
& Hickson, C. (eds.) The COSPEC Cookbook: Making SO2 Measurements at Active Volcanoes. IAVCEI, Methods
in Volcanology, 1, 63-119.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
64
this data is not always available, and thus other methods are required. Should the volcano be near
an airport or large city, one can obtain the wind speed (generally measured at ~10 m elevation)
and wind direction by contacting the local meteorological stations or airports. For example, at
Popocatépetl volcano, Mexico, wind data is received from the ground approach radar of aircraft
landing in Mexico City (see also Chapter 5).
If wind data is unavailable from the sources discussed above, measurements must be
attempted downwind of the volcano. Importantly, any wind measurement not taken within the
plume must be considered as a minimum value. In some instances, the wind speed may be too
high or too low for COSPEC measurements. For example, at very low speeds, the plume may
become diffuse and unfocused and subsequently unmeasurable; the user should wait until the
winds increase sufficiently before attempting measurements. On the other hand, excessively high
wind speeds that threaten the aircraft would make airborne measurements impossible.
2.1.1. Theodolite, compass or video measurements
One of the most accurate
Point 1
Point 2
Point 3
methods of measuring the speed
a
a
of the plume is by use of a
plume
theodolite.
An
orienteering
compass
or
rangefinding
binocular also may be used. The
theodolite is placed beneath or
adjacent to the plume in order to
Slant
track optically a packet of gas
Slant
triangle 2
triangle 3
from the summit to a position
downwind. By measuring the
time necessary for this “packet”
Angle t
Angle f
of gas to travel from the source,
gas plume
and knowing the distance Vertical triangle 2
Observer
Point 2
travelled (measured from a map),
c
a
the plume speed may be
angle
A
calculated (Fig. 1). This can also
b
Observer
be accomplished with continuous
video recording of the plume Fig. 1. Determining the speed of a horizontal plume.
(Doukas, 2002; Williams-Jones
et al., 2006). For measurement of
a vertically rising plume, see section 3.10.3.
At least three measurements of a gas parcel should be made with the horizontal angle and
time noted for each point (e.g., Points 1, 2, 3; Fig. 1). One measurement should be made at an
angle that is exactly perpendicular to the plume. The shortest slant path from the observer to the
plume (c) is then calculated, either by knowing the perpendicular horizontal distance between the
observer and the plume (b):
2
1
(1)
c = b/cosA
or by knowing the height of the plume above the observer (a1):
(2)
c = a1/sinA
3
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
65
Once the value of c is known, the distances between Points 1 and 2 (a2) and between Points 2
and 3 (a3) can be calculated:
(3)
a2 = c tanφ
(4)
a3 = c tanτ
From this information and the times recorded for each point, the plume speed is calculated.
As an example, a parcel of gas is traced at Points 1, 2, and 3, with the angles and times noted
(Fig. 1, Table 1). Assume b is 1000
m and angle A is 20°. Then c is Table 1. Angular measurements from a parcel of horizontallymoving gas
1064 m, φ and τ are both 30°, and
Point
Time (hrs)
Angle A (°)
Distance (m)
a2 and a3 are 614 m each. From
1
11:16:38
60
0
the data in Table 1, the parcel has
2
11:17:38
90
614
travelled 1228 m over 120 seconds,
-1
or at a speed of 10.23 m s .
3
11:18:38
120
1228
2.1.2. Anemometer measurements
Wind speed measurements also may be made using an anemometer (handheld or mounted).
At volcanoes where the plume is close to the ground (e.g., Masaya, Nicaragua; Kilauea, Hawaii),
this technique is extremely useful. A fully mechanical anemometer such as the one shown in
Figure 2 allows for the integration of fluctuations in wind speed due to gusting. Measurements
are typically taken over a one minute period with the result in feet per minute or metres per
second (1 ft min-1. = 0.00508 m s-1). Mechanical anemometers may require the application of a
correction factor (generally supplied by the manufacturer) to the results. For one set of data, five
to ten measurements are commonly made in
Handheld anemometers
order to account for variations in wind
speed, with the average being used in the
SO2
calculations.
Some
electronic
anemometers will automatically average
wind speeds (e.g., Kestral 1000 pocket
anemometer) for a given interval, reducing
the uncertainty due to wind gusts. Over a
day of COSPEC measurements, at least
three sets of wind speed measurements
should be made at the beginning, middle,
Stationary anemometers
and end of the day, in order to characterise
the fluctuations in wind speed. A more
detailed study of wind speed variation can
be made using a continuously recording
automated anemometer. In March 1998,
February 1999 and March-April 2000,
continuous wind speed measurements were
made downwind of Masaya volcano at an
Instituto
Nicaragüense
de
Estudio
Fig. 2. Examples of handheld and stationary anemometers.
Territoriales (INETER) seismic station.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
66
Located on the Llano Pacaya ridge (930 m a.s.l.), the station was at least 300 m above the active
vent with the result that wind speed measurements are essentially made almost at the column
height.
Care must be taken when deciding upon the placement of ground-based wind speed
instruments. According to the World Meteorological Organisation (WMO), “The standard
exposure of wind instruments over level, open terrain is 33 feet (10 metres) above the ground”
(WMO, 1996). If there are obstacles present, the anemometer is located at a distance downwind
from the obstacle that is at least ten times the height of the obstacle (Fig. 3a). If the instrument
must be near a tall building, it should be placed upwind at a distance from the building equal to
the building’s height, to avoid turbulence caused by the building (Fig. 3a,b). However, if one
were to strictly apply this rule to volcanoes, the wind station would have to be quite far from the
volcano to avoid the effects of turbulence; for example, one should be five to ten kilometres
away from a 1000 m high volcano. Towers such as those used for TV, radio and cellular
transmissions are ideal for siting an anemometer. In practice, however, the installation of such
stations in proximity to volcanoes is not always possible, leaving handheld anemometers or
overflying aircraft as the most likely sources of wind speed data.
When using a handheld anemometer, the effects of ground layer shear may be reduced
somewhat by mounting the instrument on a long pole. Measurements are greatly facilitated if the
anemometer has a timer and recording capability, such that it can be placed aloft and set to
record for a given period of time. With a handheld digital or mechanical anemometer, one can
reduce the uncertainties by starting the anemometer at ground level and quickly raising the pole
(Fig. 4).
a
10 m
T
H
10x H
10x T
b
1x Height
above
1x Height before
Wind
5x - 10x Height beyond
Fig. 3. (a) Standard positioning for ground wind instrumentation recommended by the World Meteorological
Organisation; (b) suggested siting of anemometers near ground level, downwind of an obstacle (WMO, 1996).
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
Fig. 4. Placement of an
anemometer upon a 1-2 mlong
pole
to
reduce
boundary-layer effects near
the ground. If installing a
continuously
recording
stationary anemometer, it
should be placed as high as
possible (see Fig. 3).
67
1-2m
long
Continuously
recording
anemometer
2.1.3. Dual spectrometer method
With the introduction of new small and low-cost spectrometers (e.g., FLYSPEC, see Chapter
6), new techniques have been developed to determine a more accurate representation of volcanic
plume velocity. With only two instruments it is possible to derive the apparent speed of a plume.
To do so, the location of the most concentrated part of the plume is located by completing a
preliminary traverse beneath the plume (see below). If local geography and infrastructure
beneath the concentrated segment permits, each of the spectrometers is placed on a tripod and set
to collect SO2 path-length concentrations. By placing the time-synchronized instruments along a
line parallel to the plume’s propagation (Fig. 5), similar signals are recorded on each of the
instruments. The combination of the known distance between the spectrometers (from an
integrated GPS units or a tape measure) and the time lag between corresponding points on the
data plots allows for the calculation of plume speed by comparing the data signals recorded by
each spectrometer, and noting the time lag between them (Fig. 6; Williams-Jones et al., 2006).
Beyond a simple two-spectrometer approach to determining plume speed, a rotating mirror
may be integrated into a single spectrometer’s field housing, allowing for scanning of two
sectors of the sky during a single traverse. Doing so allows for determination of plume height,
and subsequently plume transport speed (McGonigle et al., 2005a). Such a method does become
complicated, however, in locales lacking a flat road network on which to traverse, as is the case
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
68
Fig. 5. Field deployment for determination of plume speed at
Masaya, Nicaragua. The FLYSPECs were mounted
downwind of the gas source on small, lightweight camera
tripods separated by 20 m. Measurements were made for 30
minutes, with a GPS antenna in each configuration to ensure
time-synchronization of the datasets.
FlySPEC
FlySPEC
GPS
antenna
Spirit
level
Notebook
computer
SO2 pathlength concentration (ppm-m)
a)
140
120
100
80
Upwind
Downwind
100
60
16 s
13:51
13:52
13:53
20
60
40
20
0
-20
-40
13:39
13:44
13:49
13:54
13:59
14:04
Local time
Fig. 6. a) SO2 pathlength concentration for two FLYSPECs at Masaya, Nicaragua on March 25, 2003. Instrument
separation is 40.5 m, determined by tape measure. Inset is a 4-min. window showing an apparent time separation
(16 s) between the 2 signals. b) The SO2 signals for the entire 30-min. sampling period are compared to each other
for timeshifts between -60 and 60 s at 0.1 s iterations. The maximum correlation coefficient (r2 = 0.959) for the
signals occurs at a time difference of 13.1 s, which for a 40.5 m separation, results in a plume speed of 3.1 m s-1.
(modified from Williams-Jones et al., 2006)
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
69
at many volcanoes. In deriving plume height, the pitch of the instrument at the time of each
spectral acquisition must be known. Corrections for pitch compensate for undulating topography
and resultant deviations of the instrument’s field of view from the intended azimuth and
inclination, but such continuous measurements of pitch may not always be practical or feasible.
As an alternative to the determination of plume speed, additional spectrometers may be
added to the two-spectrometer configuration in order to determine plume velocity (McGonigle et
al., 2005b; Williams-Jones et al., 2006), much in the manner of the two-spectrometer approach.
2.1.4. Wind direction
The direction of the wind and therefore the gas plume is also important for the calculation of
SO2 flux. The wind direction can be obtained from meteorological stations or local airports.
However, for mobile COSPEC measurements, the most accurate and reliable method entails
determining the direction from the chart record (or digital record) and map of the volcano. The
plume azimuth (the angle that the plume direction makes with respect to geographic north/south)
is required to correct for the lack of perpendicularity of the traverse segments with respect to the
plume direction (see below). Specifically, one determines the azimuth of the plume by locating
the ground position on a topographic map at which the maximum SO2 peak on the chart record
was detected, and drawing a line on the map from this point to the volcano’s summit (Fig. 7).
600
400
600
1633
1400
1200
1000
800
600
1000
Fig. 7. Determination of the plume azimuth using maximum or median SO2 concentrations from the ground
based chart record (inset). Numbered dashes on the road refer to segment markers. Example from Arenal
volcano, Costa Rica. (Williams-Jones et al., 2001)
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
70
One also may use the midpoint of the plume from the chart record, rather than the maximum
point, in order to determine the plume direction. The midpoint may represent the average plume
direction, whereas the maximum SO2 peak may be due, in part, to heterogeneities in the plume.
The user is urged to adopt a consistent method, using either the midpoint or the highest peak on
the chart record. For stationary measurements, the wind direction can be determined using a
theodolite or compass as above (Fig. 1).
2.2.
Clouds and condensed water
vapour
Condensed water vapour in the
plume may increase the path of
ultraviolet rays in the plume, resulting
in a scattering effect that will
overestimate the observed SO2 burden
by as much as 10% (Millán, 1980;
Stoiber et al., 1983). The position of
the plume with respect to any cloud,
fog, or haze can also have an affect on
measured SO2 concentrations. If the
plume is above the cloud, scattering of
the ultraviolet rays passing through
both the plume and cloud may cause a
secondary plume to appear on the chart
record, increasing the total measured
SO2 emission (Fig. 8). It is therefore
preferable that, where possible,
measurements be made on clear days
or when the cloud is above the plume.
2.3.
Sun
(UV radiation)
Plume
Scattering
Cloud/haze/fog
Primary
plume
Secondary
plume
Fig. 8. Development of a secondary plume when cloud, haze or
fog are between the COSPEC and the plume because of
increased scattering of UV rays.
Online Wind and Meteorological information
Meteorological stations typically are run by federal, provincial/state, and municipal
governments as well as by regional airports. These government agencies can therefore be
excellent sources of wind speed, pressure, humidity, and temperature data. The principle
international meteorological agency is the World Meteorological Organisation (WMO), which
has a webpage at www.wmo.ch and maintains a list of International Weather Monitoring stations
(and their reference codes) which can be found through at www.wmo.ch/web/www/ois/oishome.htm. By far the best source for meteorological data and information is through the U.S.
National Oceanic and Atmospheric Administration (NOAA) webpages (e.g., Environmental
Information Services, www.eis.noaa.gov). A real-time archive of radiosonde data (civilian and
some military sites) for most of North, Central America and the Caribbean is available from the
US National Weather Service (raob.fsl.noaa.gov), while US and International meteorological
and surface wind data can be found at weather.noaa.gov/weather/ccworld.html. The Weather
Underground page (www.wunderground.com) is extremely useful as it maintains current and
archived surface wind data.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
2.4.
71
Sun angle and Plume dynamics
2.4.1. Sun angle
16:30 A GC: 3.7
19:30 AGC: 4.5
20:00
AGC: 4.3
17:00
AGC: 3.7
19:55
AGC: 4.4
16:36
(local time)
The COSPEC is an ultraviolet spectrometer that is dependent on the amount of ultraviolet
radiation entering the instrument. In the early morning and late afternoon at mid-latitudes, the
lower sun angle results in low amounts of ultraviolet radiation entering the vertically oriented
telescope. This leads to higher background noise and reduced precision. Background tests
performed in Montreal, Canada (45.52°N, 73.57°W) on April 29, May 2 and 5, 1997, show a
gradual increase in the Automatic Gain Control (AGC - represents the amount of ultraviolet
radiation reaching the optics), towards nightfall (Fig. 9). More significantly, on May 2, the
“noise” of the background SO2 signal increased by up to a factor of four from 14:30 to 20:00
local time. Interestingly, the relative absorption initially decreases from 14:30 to ~17:30. It then
increases dramatically, peaking at ~19:20
Background COSPEC tests
before decreasing towards the end of
a
measurements at 20:00 (Fig. 9a, b). This
increase and subsequent decrease near
AGC
sunset is due to the rapidly changing sun
angle at this time, resulting in varying
levels of ultraviolet radiation entering the
COSPEC. Measurements should therefore
be made during high sun angle so that
sufficient ultraviolet radiation enters the
UV absorption
instrument. In general, the measurement
b
period
should
be
approximately
symmetrical around solar noon (e.g., 1st
July in Montreal: 11:56 and Anchorage:
13:03 local time). However, during solar
High
noon, i.e., when the sun is at its highest
calibration
“zenith angle”, the ultraviolet radiation
entering the COSPEC may cause spikes in
the data. This problem is reduced in the
COSPEC V by baffles in the spectrometer
c
and the Cassegrain telescope, which
buffer some of the ultraviolet radiation
entering the instrument. Earlier versions
of the COSPEC have baffles only in the
telescope and are thus more vulnerable to
the ultraviolet radiation of solar noon (R.
Dick, Barringer Research, personal
communication, 1997). If an older
COSPEC is being used or if spikes
Fig. 9. Background COSPEC tests on (a) April 29, (b)
continue to occur, it is advisable to stop
May 2, and (c) May 5, 1997 performed in Montreal,
measurements and turn off the COSPEC
Canada. The diagrams show a steady increase in AGC
for approximately half an hour or until the
and signal “noise” towards sunset. Note an approximately
sun passes its apex. During airborne
four-fold increase in background noise from ~16:30 to
~20:00 on May 2, 1997. Spikes represent calibrations
measurements, banking turns may
taken every 30 min.
sometimes pick up solar spikes when
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
travelling east west. An excellent
online resource for calculating
solar noon, as well as apparent
sunrise and sunset can be found at
the NOAA Surface Radiation
Research
Branch
webpage
(www.srrb.noaa.gov/highlights/su
nrise/sunrise.html).
2.4.2. Plume dynamics
72
UV radiation
Increased
scattering
Increased
scattering
Plumes are rarely vertically or
laterally homogenous, and thus
heterogeneities in the SO2
concentrations of the plume may
effect the calculated burden. Fig. 10. Scattering of ultraviolet radiation at the thin edges of a plume.
However, if mobile measurements
are being made directly below the
plume, the ultraviolet radiation will pass through the entire plume thickness and be absorbed by
SO2, minimising scattering effects and allowing for complete measurement of SO2 in the vertical
axis. Only at the thin edges of the plume will there be an increased scattering effect (Fig. 10).
Spatial heterogeneities in the plume then can be integrated into the measurement by completely
transecting the plume during a traverse.
The opacity of the plume also may have an effect on the accuracy of the COSPEC
measurements. If the plume is heavily laden with ash, ultraviolet radiation is partially blocked by
the ash (rather than being absorbed by the SO2) increasing the apparent measured SO2 by raising
the effective background levels. An increase in the AGC would suggest that this was the case.
The AGC will tend to increase gradually towards the end of the day as the sun angle decreases
and the amount of UV radiation entering the instrument decreases. Aerosols between the plume
and the ground also will scatter ultraviolet radiation, further reducing the ground measured SO2
(Millán, 1980). However, laboratory experiments by Andres and Schmid (2001) indicate that the
COSPEC reliably measures SO2 burdens within a 10% accuracy for plumes that are up to 50%
opaque.
3. USING THE COSPEC
In the field, the COSPEC can be connected to any 12 V DC storage battery or other 12-volt
(or converted 115 VAC, e.g., car cigarette lighter) power source. The instrument should be
allowed to warm up for at least half an hour prior to use, in order for the electronics to stabilise at
a constant operating temperature. The COSPEC is connected to a chart recorder (analogue)
and/or a computer or datalogger (digital). The COSPEC generally draws 800 mA (R. Dick,
Barringer Research, personal communication, 1997) while a chart recorder typically draws 400
mA for a total of 1.2 Am. One can therefore nominally run both instruments on a 6 Ah (Amp
hour) gel cell for 5 hours (6/1.2). In practice, we suggest using two 6 Ah gel cells in parallel or
one 12 Ah gel cell to allow for 10 hours of measurement time.
The COSPEC may be used in many different configurations depending upon access to the
volcanic plume. Thus, accessibility to the volcano and monitoring strategy must be determined
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
73
prior to setting up the instrument. We first discuss this strategy and other general considerations,
then examine the different COSPEC configurations.
3.1.
Assessing access to the volcano
One of the most important factors in any COSPEC survey is determining the best access to
the volcano and thus the best possible technique for that situation. Use of an airplane is ideal for
measurements if the plume is not too close to the ground. If, as is often the case, planes are not
available or the plume is too close to the ground, then mobile ground-based measurements
should be attempted. One must then determine whether there are roads running approximately
normal to the plume direction (such as at Masaya, Nicaragua and Arenal, Costa Rica). If there
are no appropriate roads, stationary measurements should then be attempted. If even stationary
access is impossible (e.g., White Island, New Zealand; Mt. Spur, Alaska), than airborne
measurements (fixed wing or helicopter) may be the only viable option.
3.2.
Monitoring strategy - how often to make measurements?
The rate at which repeated surveys are made is greatly influenced by access to the equipment
and by safety, budgetary, and time considerations. The presence or absence of volcanological
observatories in proximity to the volcano also plays an important role in determining the
frequency of surveys. A survey should consist of at least 5-10 traverses or measurements.
In times of relative quiescence or dormancy, monthly surveys are generally sufficient to
maintain a baseline flux. Volcanoes such as Galeras and Nevado del Ruiz, Colombia, and
Kīlauea, Hawaii, are currently surveyed quite frequently, at least once a week, due to their
continued activity. For example, when Popocatépetl volcano (Mexico) was in crisis, ground and
airborne survey are made at least three times a week (see Chapter 5).
The type of activity also will affect the decision. If the volcano is constantly emitting gas,
fewer measurements may be necessary to characterise the plume (e.g., Masaya, Nicaragua).
However, if the volcano is erupting or “puffing” (e.g., Arenal, Costa Rica), more measurements
will be necessary to properly represent daily emissions. In the case of increased seismic activity
and/or deformation, a large number of COSPEC surveys will prove useful in better
understanding the state of the volcano. Post-eruption decay of SO2 emission can generally be
characterised by a reduced number of surveys. A return to non-crisis levels would then only
necessitate baseline monitoring (e.g., monthly).
3.3.
Optimising the distance for measurements
Ideally, measurements should be made as close to the volcano as possible, in order to reduce
dispersion and dilution of the plume. However, as one approaches the volcano, the plume can
become more heterogeneous. If SO2 concentrations are elevated near the summit to the point
where the levels exceed the high COSPEC calibration cells (and a high concentration kit is
unavailable), it will be necessary to increase the distance from the crater to allow for some
dilution of the plume and a lower SO2 concentration, albeit over a larger width. Thus it is
important to choose a distance from the crater where the plume is relatively homogeneous
without being so far removed that there is significant attenuation and dispersion of the gas. This
can lead to an apparent reduction in SO2 emission making the measurements unrepresentative.
While moving downwind is quite feasible during airborne surveys, stationary or ground-based
surveys may be seriously limited because of access difficulties.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
3.4.
Concentrations and
calibrations
74
a
The COSPEC is equipped with two
quartz-glass
cells
with
fixed
concentrations of SO2. These cells are
used as calibrations before and after
Use high calibration cell height and concentration
measurements are made. A low
in calculations
concentration gas cell is typically on the b
order of 50-150 ppm·m SO2 while the
high concentration gas cell contains
between 300-500 ppm·m SO2. Prior to
and following measurements beneath a
Use low calibration cell height and concentration
in calculations
gas plume, both calibration cells should
be used in order to determine which is
appropriate for the flux calculations.
Ideally, the maximum plume peak c
height should always be less than the
high calibration, with the high
calibration consequently being used in
Move further downwind to allow dispersion and
the flux calculations (Fig. 11a).
dilution of plume or put in high concentration kit
However, if the maximum peak height
of the plume is less than that of the low
calibration cell, then the low calibration d
should be used in calculations (Fig.
11b). If the plume substantially exceeds
the high concentration cell (Fig. 11c),
Rapid high calibrations during measurements will
measurements should be made farther
confirm proper operation of the instrument
from the volcano where the plume is
more
dispersed
and
dilute. Fig. 11. Examples of SO2 emission chart records and the
calibration-cell concentration that should be used in final flux
Alternatively, a high SO2 burden calculations. (a) The high calibration heights and concentrations
correlator disc assembly and high should be used when the maximum anomaly peak height is
concentration calibration cells may be equal to or lower than the high calibration height. (b) When the
installed, allowing for the measurement maximum anomaly peak height is equal to or lower than the
of concentrations ranging from ~1,000- low calibration height, the low calibration should be used. (c)
When the plume peak height exceeds that of the high
10,000 ppm·m. However, changing a concentration calibration, COSPEC measurements should be
correlator assembly requires great care made further downwind if possible or the high concentration kit
and should be performed in a clean should be used. (d) During long traverses, proper operation of
the COSPEC may be verified by performing rapid high
environment (e.g., laboratory).
In the field, it is advisable to make a calibrations during the measurement.
note of the typical heights and the peak
height ratio of the two calibration peaks as they appear on the chart recorder paper for a given
voltage (e.g., 0.5 V). If the ratio of the calibration peaks is different from that normally observed,
this might indicate problems with the COSPEC (e.g., leaking calibration cell). Similarly, it is
also useful to make a few rapid high-cell calibrations while in the plume to verify that the
relative calibration peak height remains more or less constant (Fig. 11d).
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
3.5.
75
Mobile ground-based measurements by vehicle
3.5.1. Setting up the instrument
In the case of ground-based mobile COSPEC measurements, the instrument is mounted in a
vehicle (car, van, truck, etc.). In some cases where roads were unavailable, the COSPEC has
been mounted on horseback or even onto a backpack frame (Stoiber et al., 1983). A minivan is
an ideal vehicle as it allows for the easy installation of equipment through sliding side doors. The
instrument is normally placed in the passenger seat on a box or crate (plastic Coke or beer crates
work well and are often easy to find in the field) and is secured to both the crate and seat with
elastic cords or webbing. To minimise vibration to the instrument, foam padding should be
placed between the COSPEC and the straps and the boxes. This assembly brings the COSPEC to
the level of the window and allows the right angle mirror of the Cassegrain telescope to protrude
from the side window. The extension tube is attached to the right angle mirror so that the
instrument is effectively “looking” vertically (Fig. 12). There must be sufficient clearance
between the doorframe and the Cassegrain telescope so that the telescope is never in danger of
hitting the frame of the vehicle. Foam padding should be attached around the tube as a
precaution. The right angle mirror should be oriented such that the extension tube is vertical (i.e.,
the COSPEC may be angled into the seat) to ensure that the field of view is as close to vertical as
possible during travel. A 12-volt battery or the vehicle’s battery acts as the power supply for the
COSPEC. The COSPEC is connected to the battery by a power cable with alligator clips on one
end, supplied with the instrument. The clips
should be secured with electrical tape on the
battery to prevent accidental slippage which
can result in electrical short-circuits. The
instruments may also be connected to the
vehicle's 12-volt supply via the cigarette lighter
socket using a plug for car accessories. For
data collection, the COSPEC is connected to a
chart recorder, portable computer, and/or
datalogger via data cable (COSPEC end
provided, customised end for the output
device). The clock of the computer/datalogger
should always be synchronised with the
operator's watch before starting measurements.
3.5.2. Making the measurements
Fig. 12. Proper installation of a COSPEC in a
vehicle. Straps or webbing can be used to secure the
instrument. Care should be taken that the telescopes
are “looking” vertically upwards and unable to be
knocked (e.g., against the door) during the
measurements. Plastic cases (coke or beer) can be
used to bring the COSPEC up to the level of the
window.
As roads near volcanoes are rarely straight
or perpendicular to the gas plume, an
individual traverse below the plume is divided
into segments in order to correct for the
deviation from perpendicularity of the traverse
segment with respect to the column (Fig. 13).
The azimuth (angle from geographic
north/south) of each segment must be
measured from a map so that a cosine
correction can be made later (Fig. 14). The
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
76
segment positions are marked on the chart recorder with the event recorder or noted for the
computer by recording the time when a segment boundary is reached in the vehicle. This can be
greatly facilitated by using a handheld GPS connected via a datalogger or directly to a laptop
computer. It is therefore possible, with the aide of fairly simple software, to accurately determine
the vehicle’s position every second and knowing the source location, correct for the deviation
from perpendicularity of the road.
The instrument is driven along the plume segment at an approximately constant speed. The
vehicle speed is by necessity highly dependent upon the width of the plume. On a volcano such
as Popocatépetl where the plume is typically very wide (20-50 km, as seen from the road), the
vehicle is driven at ~50-60 km hr-1; speeds must be greatly reduced when passing through towns
or villages. On the other hand, at volcanoes such as Arenal or Soufrière Hills with narrow plumes
(2-3 km), 10-20 km hr-1 is the typical vehicle speed. The quality of the roads where the
measurements are made also will affect the optimum vehicle speed; one can drive at a more
constant speed on a fully tarred road than on a gravel road full of potholes (where one is
swerving from side to side to miss the holes). The time constant of the COSPEC should be set to
either 1 or 2 in order to reduce the effect of overhanging trees. However, if distal measurements
(e.g., tens of km from the volcano) are being made and/or if the plume has low concentrations of
SO2, one should use a higher time constant (4 or 8) in order to improve the signal to noise ratio.
Gas cell calibrations are made before and after each traverse in order to calculate the plume
Fig. 13. An example of the segmentation of a complete COSPEC traverse around Masaya volcano, Nicaragua. The
distance and azimuth for each segment must be determined as well as the position of the plume. Field descriptions of
the segment markers greatly facilitate measurements. This becomes unnecessary if GPS data is integrated with SO2
data. A map should accompany each survey and volcanological events (e.g., ash fall) should be noted. Inset is an
example of how to make the cosine correction for perpendicularity to the plume (also see section 6).
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
77
burden later. Both gas cells should be used (see section 3.4). It is also wise to perform quick
calibrations while in the gas column if conditions (e.g., ultraviolet light level, ash, etc.) are
changing rapidly (see section 3.4).
3.5.3. Advantages/disadvantages
Ground-based mobile surveys are often the most practical method for the majority of
volcanoes with some road access, as they are relatively easy and inexpensive to perform. In some
cases where the plume is hugging the ground, this is the only feasible technique. However, this
technique is clearly impossible in areas where there are no roads. When there are roads, they are
rarely normal to the plume direction, necessitating the segmentation of the roads and increasing
the number of calculations. In many cases, the only accessible roads are far downwind, resulting
in possible dilution and dispersion of the plume and thus increased uncertainty. Roads are also
frequently bordered by trees which overhang the road and interfere with the measurements by
blocking ultraviolet radiation. This effect may be partially overcome by temporarily raising the
pen of the chart recorder or briefly pausing the computer when passing under trees. Stopping the
chart recorder/computer when passing through towns (with interference from telephone/power
poles) is also an option. Although this results in loss of data, one can easily interpolate the
missing data if the plume is significantly wider than the portions of road which contain
trees/towns.
3.6.
Mobile ground-based measurements by boat
3.6.1. Setting up the instrument
During the eruption of Soufrière Hills, Montserrat, ground-based COSPEC measurements
were often made by boat. As with mobile land-based measurements, the COSPEC is set up so
that the COSPEC is looking vertically. The instrument should be installed in a protected area
(e.g., behind driver’s windshield, etc.), in order to prevent contact with salt water, while also
maintaining a clear field of view for the COSPEC. To further protect the COSPEC, plastic bags
or sheeting can be placed around the instrument, telescope, cables, and power supply. The
instrument should be secured in a seat or on a crate with foam padding to absorb any vibrations.
If the sea is calm, the COSPEC simply can be placed on the deck of the boat. A 12-volt battery
(gel cell or car battery) or the boat’s battery acts as the power supply for the COSPEC. The
COSPEC is connected to the battery by a power cable with alligator clips on one end and secured
with electrical tape. The battery should be secured to limit any movement (from tossing of the
boat) during the survey. For data collection, the COSPEC is connected to a chart recorder,
portable computer, and/or datalogger via data cable (COSPEC end provided, customised end for
the output device). These should all be protected from water/spray (e.g., plastic covering) and
synchronised with the operator's watch prior to starting measurements.
3.6.2. Making the measurements
The boat is steered in a straight course as normal to the plume as possible. The position of the
boat at any given time may be determined by taking back sights with a compass on two
landmarks and tracing lines on a map back from those points. The intersection of the two bearing
lines will be the position of the boat (Fig. 14). With points on either side of the plume, the
azimuth or bearing of the traverse can be determined; the traverse should be drawn on a map to
facilitate calculations (Fig. 14). The azimuth of the traverse is required to correct for the lack of
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
78
perpendicularity of the traverse with respect to the plume direction (Fig. 13). The plume azimuth
can be determined in the same way as the ground-based vehicular method; assuming that the
point of maximum SO2 on the chart record/computer represents the centre of the plume and
knowing the speed of the boat, the position can be plotted on the map and the azimuth
determined (Fig. 14). The determination of the boat’s position, and consequently the traverse and
plume bearings, can be greatly facilitated through the use of onboard or handheld navigation
systems (e.g., GPS, Loran, Navstar, etc.). The time constant should be set to 1 or 2. The boat
should travel at a speed fast enough to allow for rapid traverses (at approximately constant
speed) below the plume and good resolution. High and low calibrations must be made on both
sides of the plume.
353 o
66o
A
9o
327 o
C
B
93o
C
A
B
Fig. 14. Determination of the traverse (navigation positions A and B) and plume azimuth for mobile boatmounted COSPEC measurements. Bearings should be taken for A and B using back sighting on landmarks
prior to entering and on leaving the plume area.
3.6.3. Advantages / disadvantages
In this method, flux calculations are simplified as the traverses are independent of road
systems and thus generally consist of one straight traverse (Fig. 14). Boats also are generally
easily available. One must be particularly careful in protecting the instruments from the spray of
water thrown up by the boat. If the wind is very strong, roughness of the water may make
measurements difficult and increase the uncertainty of the measurements because of excessive
rocking of the boat. The traverse speed should therefore be reduced to minimise this rocking.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
79
Care must also be taken to note the possible effects of an onshore breeze that may blow the
plume back upon itself resulting in unrepresentative measurements.
3.7.
Mobile ground-based measurements by foot.
3.7.1. Setting up the instrument
In rare cases where airborne, ground-based vehicular or stationary measurements are not
possible, COSPEC surveys may be made on foot. This was necessary at Poás volcano (Costa
Rica – February, 2001), because of extremely low fluxes from fumaroles near the crater lake and
a poor road network. The COSPEC was attached to a backpack frame (Fig. 15a) and carried
beneath the plume. A 12-volt battery (gel cell or car battery) acts as the power supply for the
COSPEC. The COSPEC is connected to the battery by a power cable with alligator clips on one
end and secured with electrical tape. The small size and light weight of gel cell batteries makes
them preferable. For data collection, the COSPEC is connected to a chart recorder and/or
datalogger/computer via data cable (COSPEC end provided, customised end for the output
device). The clock of the computer/datalogger must be synchronised with the operator's watch
(or vise versa) prior to starting measurements. At least three people were required in order to
carry the COSPEC, chart recorder, battery and GPS (Fig. 15b).
a
Backpack frame
(b) When making a traverse beneath the plume, at least
three people are required in order to carry the
COSPEC, chart recorder, battery and GPS.
Measurements were made in February, 2001 on the
crater floor of Poás volcano, Costa Rica.
COSPEC
chart
recorder
b
COSPEC
Fig. 15. (a) Proper installation for ground-based
COSPEC surveys on foot. Note the COSPEC is
attached to a backpack frame and the telescope is
oriented upwards.
chart
recorder
GPS
3.7.2. Making the measurements
As with any COSPEC measurement, the orientation and location of the plume must first be
determined in order to define the optimum traverse. Calibrations are made before and after
traversing the plume. Care must be taken to walk at as constant speed and as straight as possible.
Furthermore, because of the limited length of the power/data cables, at least two of the three
people (those carrying the COSPEC, battery and chart recorder/datalogger) walk close together
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
80
and at the same speed to avoid straining the cables. In the case of the Poás survey, a handheld
GPS was used to determine the position and direction of the traverse; it also allowed one to
determine the speed of the traverse. The direction and azimuth of the plume was estimated
visually. The time constant should be set to 1 or 2. Wind speed measurements were made using a
handheld anemometer on a “Dome” located immediately above the fumaroles, i.e., the gas
source.
3.7.3. Advantages/disadvantages
This method is generally only advantageous when all other mobile survey options are
impossible (e.g., ground-based vehicular, airborne). Nevertheless it is still a viable option in a
number of cases (e.g., Poás, Costa Rica, Fuego, Guatemala, Kawah Ijen, Indonesia). In the case
of Poás, being so close to the plume resulted in very high concentrations while the traverse over
rocky ground resulted in a “noisy” signal (due to rocking of the COSPEC). The fact that surveys
are made on foot means that only a small number of measurements can be made in a given
period of time; however the small plume width at this proximity to the source meant that a
traverse could be made in ~15 minutes. Access into the crater was difficult and time consuming
and extra care was necessary to get the instrument safely into the crater. This type of
measurement is now significantly easier to perform (by only one person) with the new
lightweight UV spectrometer systems (see also Chapter 6).
3.8.
Mobile measurements by fixed wing aircraft
Light single-engine planes such as Cessnas, Pipers and Senecas (Fig. 14d) are ideal for
airborne measurements, as they are relatively inexpensive, generally quite easy to find,
commonly have a removable hatch just behind the wing, and travel slowly enough to allow for a
reasonable amount of measurement time beneath the plume (see also Chapter 4). However,
virtually any plane can be modified/adjusted to carry a COSPEC (Fig. 16). For some highaltitude volcanoes, some single engine planes are incapable of reaching the higher altitudes,
making a larger pressurised plane necessary.
3.8.1. Setting up the instrument
The COSPEC is usually set up with the telescope extension tube oriented vertically upward
(Fig. 17) and protruding at right angles out a window or door (i.e., need to remove a
window/door) or through an open hatch (Fig. 17). The wing of the aircraft must not obstruct the
vertical view of the telescope.
If the plane must be pressurised or open windows/doors are not acceptable, a fused quartz
window can be installed in the roof in order to allow ultraviolet radiation to pass through to the
COSPEC (Fig. 16e). In some instances, it is possible to purchase windows or doors from crashed
aircraft in order to specifically modify them to allow the COSPEC to protrude. The opening
around the extension tube may also be sealed with a plastic barrier in order to reduce noise. This
plastic “door” should be taped down with duct tape in a crosshatch pattern made up of multiple
layers of overlapping duct tape (Fig. 17b). One must ensure that the vertical layering of the tape
starts from the door outwards for the windward side, i.e., first strip nearest to opening, second
strip overlapping half of first strip, third strip covering half the second, etc. In other words, it is a
process of always moving from the back forward to the front of the aircraft. Thus, for the
downwind side of the opening, the first strip of tape should start away from the opening, with the
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
81
Twin Engine aircraft
a
b
Window
Pass throug h
hatch
c
Emergency
window
Rep lace
plex iglass
Low Wing
Cargo
door
Rem ov e or cut
ho le in door
Door
Extension
Doo r
(e.g., A erocommander)
(e.g., Piper N avajo)
(e.g., Twin O tter)
Multi-engine aircraft
Single Engine aircraft
d
e
Upward facing
window
High Win g
M ad e of quartz
glas s
Cargo
door
Rem ov e or cut
ho le in door
(e.g., Cessna Skyhawk)
(e.g., C-130)
Fig. 16. Examples of various single and multi-engine aircraft and the appropriate COSPEC installation method.
Inset photos modified after (a) Dexter Francis, (b) NASA DFRC, c) Mike Reyno, d) Cessna Inc., e) Lockheed
Martin Inc.
a
c
b
Fig. 17. Proper installation of a
COSPEC in a single engine aircraft.
(a) The COSPECs must be securely
strapped within the plane. Note that
the COSPEC field of view is always
unobstructed. (b) The correct way to
secure a plastic cover using duct
tape. Note that the tape is in a
crosshatched pattern. In some
instances of extreme cold (e.g.,
Alaska), this tape can become brittle
and crack. (c) A sealed Perspex
d window or (d) reinforced card can
also be used. Photos c and d courtesy
of the USGS Volcano Hazards
Program.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
82
last strip oriented vertically next to the opening (Fig. 17b). This ensures that wind friction on the
leading edge of tape will minimise effects on the taped sheet of plastic when the wind passes
over the taped area.
As with ground-based measurements, the COSPEC is connected either to the aircraft power
supply (with a 24 to 12 V converter) or to an independent battery (e.g., a 12-volt gel cell battery).
It is important that liquid lead-acid not be used. The COSPEC is also connected to a chart
recorder and/or portable computer/datalogger, the clock of which must be synchronised with the
operator's watch before starting measurements.
3.8.2. Making the measurements
Huejotzingo
A
Before the survey can start, one must first
determine the location and direction of the
plume. If the wind direction is not known (see
Popocatepétl
section 2.1.3), a complete traverse around the
volcano should be made using the COSPEC
θ
as a reconnaissance tool. Once the lateral
extent of the plume is determined by
0
5
10 km
COSPEC (several orbits may be necessary),
Atlixco
the lower limit of the plume must be found.
Commonly, the bottom of the visible plume B
does not coincide with that of the SO2 plume.
Xm
Multiple passes at levels below and in the
plume using your nose to detect the odour or
SO2 will allow you to make measurements
Plume
hugging the bottom of the plume. The
COSPEC is flown normal to the plume and as
d = 5.8 cm
d = 8.1 cm
close to the bottom of the plume as possible to
Huejotzingo
Atlixco
23, 930 m
ensure that the plane is in fact below the
plume. As with any COSPEC measurement,
Distance measured from map
calibrations are made before and after
Therefore, 8.1 cm
23,930 m
X = 21,831 m
=
5.8 cm
Xm
traversing the plume.
The amount of time in which the Fig. 18. Determination of traverse segment width and
instrument and plane are beneath the plume is azimuth during airborne measurements, using (a) two
landmarks and (b) the proportion of actual distance on a
short, due to the high speed of the aircraft. map to the width of the plume on the chart record.
Note that an altitude correction to the relative
airspeed is required: 2% added to airspeed for every 1000 ft (~300 m) of altitude. Thus one
should reduce the airspeed in order to maintain a constant ground speed. The chart recorder
should be set at a high chart rate (e.g., 60 mm min-1) in order to obtain sufficient chart resolution
while the plane is beneath the plume. The time constant of the COSPEC should be set to one
second. The traverse segment width and azimuth may be measured by a number of methods:
1
2
(a) The aircraft ground speed can be determined using two timed points on the ground (or
estimated by a 2% increase per 1000 ft of altitude as above), marked on the chart record,
with the distance between the same points measured from a map. This will give the
average ground speed of the aircraft. Knowing the chart speed, one can then determine
the chart time under the gas plume and multiply this value by the ground speed to
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
83
determine the width of the plume (Fig. 18a). The traverse azimuth can be determined by
measuring the angle between the traverse line and geographic north/south on a map.
(b) The plume width is also determined by taking the proportion of (1) the chart length of the
plume between the two landmarks noted on the chart record and using the (2) distance
between the landmarks, measured from a map (Fig. 18b).
(c) Determine the position of the aircraft at a given point in time using a GPS; this is done
repeatedly to obtain the aircraft’s ground speed and then doing the calculations as in
methods (a) or (b) above. The traverse azimuth also can be determined using the GPS and
method (a).
The wind speed can also be determined by flying with and against the wind (along the axis of
the plume) and comparing the true air speed with the true ground speed. For example, if it takes
3 minutes per km flying upwind and 5 minutes per km downwind, then the wind speed is 2 min
km-1 divided by 2, i.e., 1 min km-1 or 16.7 m s-1. New methods using GPS technology are now
available and greatly facilitate these measurements (Doukas, 2002).
3.8.3. Advantages/disadvantages
Aircraft are by far the most effective method of making COSPEC measurements, since many
measurements can be made over a very short period of time. Traverses are independent of road
systems and thus generally consist of
Traverse under plume
a
one straight traverse, which greatly
facilitates flux calculations. The velocity
of the plume also may be measured by
flying along the axis of the plume at
plume height. However, aircraft rental is
Helicopter flying
expensive, and access to a plane is
towards you
sometimes
difficult
(although
COSPEC
government support is sometimes
possible). If the plume hugs the surface
of the volcano, airborne measurements
are not possible.
3.9.
Mobile measurements by
helicopter
b
Traverse down plume
Plume coming
towards you
3.9.1. Setting up the instrument
While helicopter-borne COSPEC
measurements
are
not
normally
performed on a regular basis, trials at
Mount Etna have resulted in the
development
of
two
techniques
(Caltabiano et al., 1992). In the first
technique, the COSPEC was mounted
with the telescope protruding out the
doorway of a single rotor (2 blades)
Augusta-Bell AB-212 helicopter. The
Helicopter
descending
COSPEC
Fig. 19. Installation of a COSPEC in a helicopter for (a)
horizontal (below the plume) and (b) vertical traverses of the
plume. The field of view of the COSPEC in (b) lies outside the
circumference of the rotor blades. Modified after Caltabiano et
al. (1992). Inset photo courtesy of Michael Doukas.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
84
extension tube was placed so that it was “looking” vertically up through the rotor blades above it
(Fig. 19a). The second method entailed positioning the COSPEC such that the telescope’s field
of view is oblique to the plane of the rotor blades without being intersected by the blades (Fig.
19b). The COSPEC was connected to an independent power supply (e.g., 12-volt battery) as well
as to a chart recorder and portable computer.
3.9.2. Making the measurements
(5)
AGC (V)
SO 2 Absorption (V)
The first measurement technique is similar to that for fixed-wing aircraft, with the helicopter
being flown along a traverse beneath the plume. Calibrations must be made before and after
traversing the plume. At Mount Etna, the time constant was set to 8 seconds in order to reduce a
component of the high-frequency interference pattern from the rotor blades intersecting the field
of view of the COSPEC. The optimum time constant setting may vary depending on the type of
helicopter, as the rotation frequencies of the rotor blades and the number of blades will differ
from one model to another. If the time constant is high, the helicopter should be flown slowly
across/beneath the plume in order to allow time for instrumental response. The resulting traverse
record (Fig. 20), although maintaining the repetitive interference patterns, may nevertheless
allow for reasonable measurements of a given plume. This assumes that the concentration and
size of the plume is such that the rotor blade interference patterns are minor thus resulting in a
low signal/noise ratio (Fig. 20). As with airplane COSPEC measurements, the wind speed can be
determined using a GPS or by plotting the position of two landmarks on a map and measuring
the time it takes to fly between both points. Similarly, the plume azimuth is determined by
tracing a line on a map
from the volcano summit
10
10
High calibrations
to the position of
maximum SO2 on a map,
8
8
AGC signal
taken from the chart
record, and measuring
6
6
the line’s angle with
respect to geographic
4
4
north/south. The traverse
SO signal
segment
azimuth
is
2
2
estimated in a similar
fashion; plot the position
0
0
of two landmarks on a
Fig. 20. Example of a chart record for horizontal traverse helicopter-mounted map and measure the
COSPEC measurements. Note the repetitive interference pattern caused by the angle of the line from
rotor blades. The time constant is set to 8. Modified after Caltabiano et al. (1992). geographic north (Fig. 7).
The second COSPEC technique, which attempts to eliminate the rotor interference, entails a
vertical traverse of the plume. Starting at a point above or below the plume, the helicopter is
allowed to descend or ascend vertically at a constant velocity. The time that the helicopter is
traversing the plume can be determined by:
2
T = ΔH/Vd
where ΔH is the difference in altitude from the top to bottom of the plume, i.e., the virtual plume
thickness, and Vd is the descent velocity of the helicopter (Fig. 21). By noting the altitude on the
chart record every 100 m and knowing the chart speed, one may determine the descent velocity
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
85
of the helicopter. The horizontal distance (Dh) that the helicopter is offset by the force of the
prevailing wind may be determined by:
(6)
Dh = (Vw + Vh)⋅ΔH/Vd
where Vh is the horizontal speed of the helicopter needed to counter the wind speed, Vw. The
angle of deviation (θ) of the flight path from normal is then calculated from:
(7)
θ = arctan (Dh/ ΔH)
10
10
8
8
6
6
4
4
2
2
0
0
VW
AGC (V)
SO 2 Absorption (V)
Once the angle of deviation is known, the cosine of this angle is used in the flux calculations (see
section 6). A typical chart record for this technique is shown in Fig. 21.
In both cases, determining the position and speed of the helicopter, and plume speed is
greatly facilitated with a GPS. As with fixed-wing aircraft, the first technique also can take
advantage of geographical points of reference to determine the position and ground speed of the
aircraft. Both techniques require a horizontal traverse along the axis of the plume in order to
determine the plume speed and azimuth (see section 3.8). However, the second technique is more
problematic with respect to determination of the aircraft’s position at any given time, unless a
GPS is used.
θ
VW
VD ΔH
Dh
Fig. 21. Trigonometry of a helicopter-mounted vertical COSPEC traverse and example of the resulting chart
record. Note the lack of interference patterns on the chart record because the COSPEC field of view is outside
the area affected by the blades (see Fig. 17). Modified from Caltabiano et al. (1992).
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
86
3.9.3. Comparison of helicopter rotor blade RPM with COSPEC time constants & orientations
Setup
To examine the effects of the helicopter's rotor blade and position of the COSPEC, we
performed some preliminary tests in August 1997 by varying the orientation and the time
constant of a COSPEC V. We used a Bell 206 Jet Ranger helicopter which had a double-blade
rotor turning at 330 rpm during normal operation. The tests were made while the helicopter was
powered up on the ground. The COSPEC was mounted on the back seat of the helicopter, and
the right angle mirror and extension tube were attached to the COSPEC in order to look
vertically upward through the rotor blade. The COSPEC was mounted in two orientations. It was
initially placed on its side on the helicopter seat so that the entrance slits were perpendicular to
the rotor blade (Fig. 22a). The instrument was then placed in a normal position on the seat so that
the slits were parallel to the rotor blade (Fig. 22b). For each COSPEC orientation, tests were
made at time constants of 1, 4, and 8. The backgrounds, high calibrations, and low calibrations
were measured on a chart recorder.
Noise levels
The chart records from the test are shown in Fig. 23. The data reveal at least two frequencies
of noise. First, there is high frequency noise on the order of seconds. A lower frequency
component also is observed on the order of tens of seconds. This lower frequency component
appears to lengthen as the time constant decreases. For a given COSPEC orientation, the high
frequency noise amplitude of the signal decreases as the time constant is increased (Fig. 23). For
a given time constant, the high frequency noise amplitude is higher when the COSPEC is
oriented vertically. This is likely due to the fact that the entrance slits of the COSPEC are
perpendicular to the rotor blade, leading to decreased resolution. The signal appears to be least
noisy when the COSPEC is oriented horizontally with a time constant of 4. This can be seen both
from differences between maximum and minimum values for a given time interval (Table 2) and
from average standard deviations for a series of intervals (Table 3). Using the differences data,
noise levels are between 192-196 ppm·m for a time constant of 1, and 56-78 ppm·m for time
constants of 4 and 8 (Table 2). The difference values should be regarded as maximum noise
levels, and they are probably not the best measure of the true noise level. A better way is to
examine the average of the standard deviations for the different time intervals (Table 3). Using
the 2σ average standard deviation, noise levels for time constants of 4 and 8 are 15-27 ppm·m.
While it is difficult to estimate the noise for a time constant of 1 using this method, the noise is
probably double that for time constants of 4 and 8, i.e., 30-60 ppm·m.
Accuracy
Backgrounds are generally higher by about three vertical chart units when the COSPEC is in
a vertical orientation, compared to a horizontal position. The chart unit values are roughly similar
for the high and low calibrations for the different COSPEC orientations and time constants.
However, the calibration values are attenuated compared to reference calibration values without
the helicopter's rotor turning (background 21.5 units, high calibration 47 units, low calibration
14.5 units, high/low = 47/14.5 = 3.2) (Table 3). The high/low ratios closest to the reference are
those for a horizontal COSPEC position with time constants of 1 and 4 (high/low = 3.3),
followed closely by the vertical position at time constant of 8 (high/low = 3.4). The elevated
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
a
87
b
Fig. 22. Examples of a
COSPEC (a) verticallyoriented on the helicopter
seat with entrance slits
normal to the rotor blades
and
(b)
horizontallyoriented on the seat with
the entrance slits parallel to
the rotor blades.
Horizontal
a
Vertical
AGC
High Cal
AGC
High Cal
Low Cal
Low Cal
SO 2
SO2
Time Constant = 1
b
AGC
AGC
High Cal
High Cal
Low Cal
Low Cal
SO2
SO2
Time Constant = 4
c
AGC
AGC
High Cal
Low Cal
High Cal
Low Cal
SO2
SO2
Time Constant = 8
Fig. 23. Examples of chart records from COSPEC experiments in a Bell 206 helicopter, Vancouver Airport,
Canada, 25 August 1997. Experimental conditions: helicopter rotor ~330 rpm (single blade); chart speed 60
mm min-1; SO2 0.5 V; AGC 5 V; high calibration cell 326 ppm·m; low calibration cell 103 ppm·m. Each
vertical division on the chart paper equals 1 cm. (a) Time constant = 1. The vertically-oriented COSPEC
produces a signal which is significantly noisier than the horizontally-mounted COSPEC. The horizontal
position appears to show both short-period and longer-period noise. (b) Time constant = 4. For both
orientations, the noise is substantially lower compared to a time constant of 1. Two timescales of noise are
again observed. Short-period noise is lower for the horizontally-oriented COSPEC compared to verticallymounted. (c) Time constant = 8. For both orientations, short-period noise is lower than time constants of
either 1 or 4. However, longer-period noise remains a problem.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
88
Table 2. Comparison of COSPEC slit orientation and noise levels due to helicopter blade interference.
COSPEC Orientation
Time Constant
Background 1*
High Calibration
Background 2
Low Calibration
Background 3
Average
Noise Level (ppm⋅m)
†
1
25
30
30
23
25
Vertical
4
12
10
9
9
9
8
10
7
5
10
7
26.6 ± 3.2
196
9.8 ± 1.3
78
7.8 ± 2.2
59
1
21
21
30
22
29
Horizontal
4
8
6
7
10
7
8
9
5
9
8
7
24.6 ± 4.5
192
7.6 ± 1.5
56
7.6 ± 1.7
57
*For each background or calibration, the numbers represent the difference between minimum and maximum
values from the chart record over a given time interval.
†
The noise levels are determined by comparing the difference between above values and average peak values of
the high calibration (326 ppm⋅m) (see Table 3)
Table 3. Averaged chart record values for backgrounds and calibrations during helicopter tests.
COSPEC Orientation
Time Constant
Background 1*
High Calibration
Background 2
Low Calibration
Background 3
Average background
Average std. dev. of
5 values above
1
~18
~60
~15
~25
~14
Vertical
4
12.3 ± 2.4
53.0 ± 2.2
10.9 ± 1.2
27.8 ± 1.4
13.2 ± 1.1
Horizontal
4
8
8.4 ± 1.0 14.2 ± 1.8
53.3 ± 1.0 54.8 ± 1.3
9.9 ± 0.8
9.8 ± 1.7
22.4 ± 1.0 22.3 ± 1.7
8.5 ± 1.3 10.9 ± 1.3
8
13.3 ± 2.2
58.4 ± 0.9
15.6 ± 1.3
28.1 ± 1.7
17.0 ± 1.5
1
~10
~54
~12
~25
~15
15.7 ± 2.1
-
12.1 ± 1.2
1.7
15.3 ± 1.9
1.5
12.3 ± 2.5
-
8.9 ± 0.8
1.0
11.6 ± 2.3
1.6
-
27
23
-
15
24
44.3
9.3
4.8
40.9
15.7
2.6
43.1
12.8
3.4
41.7
12.7
3.3
44.4
13.5
3.3
43.2
10.7
4.0
Noise level (ppm⋅m)†
from 2 sigma avg. std.
dev.
High cal. - Avg. bkd.
Low cal. - Avg. bkd.
High / Low
*Values for backgrounds and calibrations are in arbitrary units from the chart record
†
The noise levels are determined by comparing the 2 sigma average standard deviation of the backgrounds and
calibrations to the average peak height of the high calibration (326 ppm⋅m)
The reference calibration values, when the helicopter was not operating, are 47 chart units for the high calibration
(326 ppm⋅m) and 14.5 units for the low calibration (103 ppm⋅m). Thus, the high/low ratio is 47/14.5 = 3.2, which
is the same as the concentration ratio (326/103 = 3.2).
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
89
value for a horizontal position at time constant of 8 (high/low = 4.0) is due mainly to an
increased average background value (11.6) which decreases the low calibration value more than
the high calibration. The opposite is true for a vertical orientation at time constant of 4; the
high/low ratio of 2.6 results from a comparatively low background which raises the low
calibration value more than the high calibration.
Overall, it is recommended that the best choice for this particular type of helicopter is a
horizontal COSPEC orientation with time constant of 4. The horizontal position provides better
resolution of the SO2 signal, while the time constant of 4 decreases the noise level. This selection
appears to improve accuracy, precision, and detection limits when using this helicopter.
However, until more experience and data are gathered, these test results should be regarded as
preliminary on this helicopter.
3.9.4. Advantages/disadvantages
Helicopters have the same advantages as airplanes: rapid measurement periods at variable
elevations, straight traverses, and the ability to accurately determine the plume speed.
Helicopters also have the ability to access areas that a fast fixed-wing aircraft can not. However,
the principle disadvantage of helicopter-borne COSPEC measurements is the interference pattern
created by the rotor blades (when using the “through the blades” technique). In some instances,
the general plume signal may be visible, as the interference patterns caused by the rotor-blades
should be constant. However, unless one is using a computer for data collection and has access to
software capable of removing or reducing the interference patterns, calculation from chart
recorder data may be difficult. Helicopters can also unstable in the often turbulent and variable
winds that are common around many volcanoes. In some cases, helicopters may be incapable of
obtaining sufficient lift near high altitude volcanoes. Helicopters are also generally far more
expensive than fixed wing aircraft to rent and operate.
3.10.
Ground-based stationary measurements
Making stationary measurements with a COSPEC is more difficult than mobile
measurements. Since the instrument is rarely looking vertically, atmospheric scattering effects
are more pronounced, and the geometric relationship of the instrument to the gas plume requires
more consideration. In particular, the main difficulties are in establishing the true width of the
plume and its velocity (speed and direction). On the other hand, a greater number of
measurements can be made in a stationary mode compared to mobile techniques. On balance, it
is our opinion that stationary measurements are inferior to mobile measurements. However, there
are many situations when stationary measurements are the only possibility and therefore should
be attempted. Such situations include difficult access to the volcano due to lack of roads, lack of
vehicles, danger due to heightened activity, semi-continuous monitoring (see Chapter 6), etc.
3.10.1. Setting up the instrument - the tripod
To make stationary measurements, the COSPEC should be mounted on a sturdy, solid tripod.
The tripod should be set up in a position that can be easily located on a map, and this position
should be known in relation to the gas plume.
The tripod should be able to scan both vertically and horizontally over a large angular range
(360° horizontally and 180° vertically). The tripod must have some means to measure the
vertical and horizontal angular position of the instrument. Ideally, this consists of angular scales
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
90
on the tripod which are adjusted by means of
a
goniometers which can be smoothly turned (a)
by a hand crank (Fig. 24a). The goniometers
allow the angular position of the COSPEC
to be altered at a steady rate. The angular
position also can be varied by means of
motorised
goniometers.
While
this
arrangement permits a constant rate of
movement, the motor must be powered by
batteries which can fail; this can be
overcome through the use of a generator.
Many tripods do not have such
arrangements, and alternatives can be used.
To measure the vertical angle of the
COSPEC, a clinometer or plumb bob can be
(b)
used (Fig. 24b). To measure the horizontal b
angle, a transparent 360° scale can be
mounted where the tripod rotates, and a
pointing device can be taped near the scale
to serve as an angular marker (Fig. 24c).
3.10.2. Making the measurements
To measure a gas plume, the objective is
to scan the instrument through the plume in
a direction that is normal to the plume's
c
movement. Thus, one scans horizontally for (c)
a vertical plume, and vertically for a
horizontal plume. It is essential to scan at a
rate which is as constant as possible, either
by means of hand-cranked goniometers, by
motor, or by simple grips on the tripod. The
scan of the plume should also be made at
high angles in order to reduce, as much as
possible, ultraviolet scattering (Fig. 25).
Generally, measurements require two
people, one person to move the angular Fig. 24. Examples of tripod-mounted COSPECs for
position of the COSPEC and the other to stationary measurements. Photos b and c from
record data on the chart paper. One person USGS/Cascades Volcano Observatory.
can accomplish the measurement if a
motorised goniometer is used. The chart speed should be set fast (60-80 mm min-1) to record
sufficient detail on the paper. As the first person adjusts the tripod at a slow and constant rate, he
or she should call out each 5° or 10° increment to the second person. The second person then
records the increment using the event marker on the chart recorder, noting the time to the nearest
second. In this manner, the scan rate can be established, and the number of degrees traversed
through the plume can therefore be calculated (Fig. 26). Motorised goniometers can be set at a
given scan rate. In this example with a chart speed of 60 mm min-1, the scan rate is 10° vertically
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
91
b
a
UV scattering
COSPEC
COSPEC
Fig. 25. COSPEC orientation for stationary plumes. The plume should be scanned at high angles in order to
reduce ultraviolet scattering.
Firsto calibration
at 0 from vertical
11:15 hrs
Second
calibration
o
at 60 from vertical,
11:18 hrs
Gas plume
o
0
11:15:20
o
10
o
20
30
11:15:40 11:16:00
Scanning vertically
downward
o
11:16:20
40
o
50
o
o
60
11:16:40 11:17:00 11:17:20
Chart speed
60 mm/min
Scan
rate
o
0.5 /s
Fig. 26. Example of a chart record for a traverse through a plume using the stationary ground technique. Note
that the degree of inclination and time should be recorded on the chart record during and after the
measurement. In this example, 0º refers to an arbitrary starting point with the field of view of the COSPEC
outside the plume and is thus independent of the actual angle between the COSPEC and the ground.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
92
every 20 seconds, or 0.5° s-1. Although recording the time is good practice, it is not strictly
necessary as long as the chart speed is known. For example, if a traverse covers 20° over 40 mm
of chart paper with a chart speed of 60 mm min-1, then the duration is 40 seconds for a scan rate
of 20°/40 s or 0.5° s-1.
Before and after scanning, both high and low calibrations should be made, with the time and
the angular position (both vertical and horizontal) of the COSPEC noted on the chart paper (Fig.
26).
At the same time that the chart recorder is being used, the data also can be recorded digitally
on a computer or datalogger. Normally a recording interval of 0.5 s or 1 s is used, with the time
recorded on the chart paper. The clock of the computer/datalogger must be synchronised with the
operator's watch (or vice versa) prior to starting measurements.
3.10.3. Vertical plumes
In the case of a vertical plume, the COSPEC is generally angled upward and scanned
horizontally from side to side. The COSPEC must be looking sufficiently above the crater to
avoid any interference, but at the same time
low enough that the plume is not too a
dispersed and wide (Fig. 27a). The entrance
slits of the COSPEC should be oriented
vertically, parallel to the plume, in order to
achieve the best possible spatial resolution
(Fig. 27b). This is particularly important in
the case of a thin or distant plume. In this
case, only the Cassegrain telescope is used
(without the corner mirror), with the quartz
plate cover as protection.
As mentioned above, the main
b
difficulties are the plume width and the
ascent speed of the plume. We will address
each of these parameters in turn. To a first
approximation, the width of the plume as it
leaves the crater will be no larger than the
width of the crater itself. However, the
Entrance slit(s)
plume could be significantly thinner if the
oriented vertically
source of the gas is from individual
fumaroles. As the plume rises above the
crater, it widens or coalesces. Therefore, it is
COSPEC scanned
from side to side
important to calculate the visual width of the
plume directly from the COSPEC
measurements.
A schematic diagram of a COSPEC
measuring a vertical plume is shown in Fig. Fig. 27. COSPEC orientation for vertically rising
28. Since the COSPEC is frequently plumes. Note that the entrance slits are oriented
positioned below the crater, the instrument vertically and that the COSPEC is scanned horizontally
is tilted at an angle A on its tripod. across the plume. The COSPEC position should be
Normally, angle A is measured, and the cross wind to the volcano such that the plume travels.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
93
horizontal distance between the COSPEC and the vent (b) is measured from a map. The height of
the gas column above the COSPEC (a) and the width of the plume (d) are unknown. Our
objective is to solve for d. First we calculate c (Fig. 28a):
(8)
c = b/cosA
To solve for d, we need to know the angle β of the oblique slice through the gas plume. β is
the number of degrees through which the COSPEC intersects the gas plume during a horizontal
scan (Fig. 28a, b). This value can be determined from the start and end of the plume on the chart
record. Once β is known, we use its half value to solve for d (Fig. 28c):
(9)
tan(β/2) = (d/2)/c
(10)
d/2 = c tan(β/2)
(11)
d = 2c tan(β/2)
Since the COSPEC cuts through the plume at an oblique angle, a correction must be made for
perpendicularity. This is simply the cosine of angle A (Fig. 28d).
Fig.28. Determining the
width of a vertically
rising plume.
A
B
d
β
β/2
angle β
c
a
c
COSPEC
d/2
d/2
β/2
COSPEC
b
angle A
D
C
COSPEC
look
direction
Vertical
plume
β/2
angle A
d/2
COSPEC
c
Vent
Rise velocity of a vertical plume
The remaining unknown parameter is the rise speed of the plume. To determine the speed, an
observer tracks an individual parcel of gas during its upward movement. Recording the parcel on
videotape and scaling it with a time stamp may also facilitate rise speed determination. For the
parcel, periodically record its angle (A) with a clinometer and the time (Fig. 29). By knowing this
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
94
angle and the horizontal distance between the observer and the vent (b), the height of the parcel
(a) can be determined:
a = b tanA
(12)
An example is given in Table 4 with b equal to 300 m. The height difference between
10:15:20 and 10:15:40 hrs is 163 m, and between 10:15:40 and 10:16:00 hrs 162 m. Thus, the
rise rate is 162.5 m every 20 seconds, or 8.1 m s-1.
Point
1
2
3
Time (hrs)
10:15:20
10:15:40
10:16:00
Angle A (°)
33
50
60
Height (m)
195
358
520
Table 4. Angular measurements from a
parcel of upward-moving gas
Horizontal distance b = 300 m
Fig. 29. Determining the rise rate
of a vertical plume. Points 1, 2,
and 3 are arbitrarily chosen.
Point 3
Point 2
a3
Point 1
a2
a1
A2
Observer
A3
A1
b
The availability of new low-cost hand-held digital cameras with video recording features
(~30 s) gives a measurement option that does not require repeated angular measurements (Fig.
29). Rather, it is possible to determine the vertical “Field of View” (vFOV) for a given focal
length lens and distance, b, from the volcano. As such, it becomes possible to determine the area
of plume seen in the camera image (Fig. 30) and deduce the image resolution (pixel per metre) of
the vertically moving puffs of gas.
In order to determine the vFOV, one must relate the frame or sensor size of the camera (e.g.,
35 mm) to the focal length. As digitial cameras do not use film, the frame size is directly related
to the size of the CCD (Charge-Coupled Device) and can be determined from the specifications
of the camera (e.g., Image Format, vertical/horizontal, of 22.2 x 14.8 mm or 3456 x 2304 pixels).
Most digital cameras also have a focal length multiplier or conversion factor to relate the lens
focal length to that of 35 mm film cameras. It should be noted that with “point and click” digital
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
95
cameras where there is minimal control on the focal length used, the minimum or maximum
(e.g., 6.0 - 72.0 mm for 0 to 12x optical zoom) should be used such that the plume immediately
above the crater can be seen at sufficient resolution to observe puffs.
Thus, for a tripod-mounted digital SLR camera with a 55 mm lens (with multiplier of 1.6x)
focused at infinity, the angular vFOV can be determined as follows:
(13)
angular vFOV = 2 * arctan (frame size / (focal length * multiplier * 2))
= 2 * arctan (22.2 mm / (55 mm * 1.6 *2))
= 14.4º
Using Equation 12, for a horizontal distance, b, of 300 m and angular vFOV, A, of 14.4º, the
vFOV in metres, a, equals 77 m, which for a 3456 x 2304 pixel image. This results in a
resolution of 45 pixels per metre. Thus, using simple video processing software, it becomes
possible to track a puff of gas from the first video frame (or time-stamped photo) as it moves
vertically through the image (and vFOV) until it passes out of the image or last video frame.
Thus, if a puff was observed to pass from pixel 50 through to pixel 3000 in 20 s of video, the
vertical plume rise speed, V, would be:
(14)
V = (2950 pixels / 45 pixel m-1) / 20 s
= 3.3 m s-1
Digital image / video
Fig. 30. Determining the rise
rate of a vertical plume with
a digital camera or video.
Time
5s
pixel / m
10 s
a vFOV
(m)
0s
pixel / m
Angular
vFOV A
Tripod-mounted
digital camera
b
3.10.4. Horizontal plumes
For horizontal plumes, two general aspects are important. First, the operator should orient the
COSPEC approximately perpendicular to the plume. Second, since the entrance slits of the
COSPEC need to be parallel with the direction of the plume for greatest spatial resolution, the
COSPEC is oriented to scan with the slits horizontal (Fig. 31). For this orientation, the rightangle mirror and extension tubes are required, in addition to the Cassegrain telescope. To scan
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
96
plume
COSPEC rotated
clockwise or counterclockwise
about its long axis
Entrance slit(s)
oriented horizontally
Fig. 31. COSPEC orientation for horizontal plumes. Note that the entrance slits are oriented horizontally and
that the COSPEC is scanned vertically across the plume.
Cross-section
of cylindrical
plume
d/2
d/2
c
angle β
a
angle β/2
angle A
COSPEC
b
Fig. 32. Determining the width of a horizontal cylindrical plume.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
97
across the plume, the COSPEC is rotated back and forth on the tripod keeping the COSPEC axis
parallel to the plume (Fig. 31).
Once again, the two parameters that are difficult to measure are the plume width and plume
velocity. Plume dynamics modify the shape of the plume constantly, and this factor complicates
the width measurement. For simplicity, we assume two models, (1) a cylindrical plume and (2) a
diffuse plume. The plume spreads out, diffusing laterally with distance from the volcano and also
under low-wind conditions. The true shape of the plume probably falls somewhere between these
two cases.
We first examine a cylindrical plume, since this configuration is geometrically the most
simple. A schematic diagram is shown in Fig. 32. As with vertical plumes, we assume that the
horizontal distance between the COSPEC and the plume centre (b) is known from maps. The
angle from horizontal to the plume centre (A) is determined by summing (1) the angle from
horizontal to the plume base, determined from the chart record, and (2) half the number of
degrees subtended by the plume (angle β/2), also determined from the chart record. Using b and
A, we calculate c, which is the slant distance from the COSPEC to the plume centre:
(15)
c = b/cosA
Normally, the diameter of the plume is small relative to b. However, if the plume diameter is
large, then d is large relative to b, and the determination of b will be subject to some uncertainty
(Fig. 32). As an alternative for calculating c, b is not known but the height of the plume (a) is
from Equation 12. Thus:
(16)
c = a/sinA
Once c is known, we can calculate the plume diameter (d):
(17)
d/2 = c sin(β/2)
(18)
d = 2c sin(β/2)
a
b
d/2
d/2
c
d/2
d/2
c
a
angle β
angle β
angle β /2
angle A
COSPEC
b
COSPEC
Fig. 33. Determining the width of a horizontal cylindrical plume: (a) nearly overhead; (b) directly overhead.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
98
In the case of a plume nearly overhead (Fig. 33a), the calculations are the same. In the case of
a plume directly overhead (Fig. 33b), we need to know the distance from the ground to the plume
centre (c). Here the top of the volcano could be used as a proxy for the centre of the plume. As
before, the plume diameter is:
d = 2c sin(β/2)
(19)
Again, there is a potentially large error measuring c if the plume is either thick or close to the
ground. In both cases, it will be difficult to distinguish the centre of the plume from its base.
Under certain conditions, plumes can spread, becoming wider in the process (Fig. 34).
Dispersed plumes can be distinguished from more cylindrical plumes by their greater angular
spread, longer chart records, lower concentrations, and asymmetric traces on the chart record
(Fig. 34). A dispersed plume also should be suspected if the COSPEC detects its far edge near
the horizon.
B
A
COSPEC
COSPEC
Chart Record
Scanning vertically down
narrow plume
lateral spreading
of plume
Fig. 34. Distinguishing (a) cylindrical and (b) flattened horizontal plumes.
The geometry of dispersed plumes is difficult to treat, and we initially adopt an approximate
technique by measuring the width of the plume from one side (Fig. 35). We require prior
knowledge of the height of the plume above the COSPEC (a), which can be difficult to estimate.
One approach is to assume that the height is equal to the height of the volcano above the
COSPEC. Another way to determine a is by estimating, if possible, the distance between the
COSPEC and the near edge of the plume (b1):
(20)
a = b1 tanθ
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
Plume
a
θ
a
α
COSPEC
b1
b2
Fig. 35. Determining the width of a horizontal flattened plume. plumes: (a) stubby; (b) tapered.
A
Stubby plume
b 2 overestimated
Plume
a
COSPEC
a
b1
b2
B
Tapered plume
Plume
a
COSPEC
a
b1
b2
Fig. 36. Configuration of horizontal flattened
b 2 not overestimated
99
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
100
Once a is established, the width of the plume is then calculated by subtracting the distance
between the COSPEC and the near edge of the plume from the distance between the COSPEC
and the far edge (Fig. 35):
(21)
width = (b2-b1) = (a/tanα) - (a/tanθ)
This equation is very sensitive to the
value of α at angles less than 10°.
Correspondingly, the COSPEC
Plume
should be placed as near to the
plume as possible, in order to
increase α. Another problem is that
the value of b2 is overestimated if
a
a
the plume edges are stubby (Fig.
36a). The problem is lessened if the
α
θ
edges are thinned and tapered (Fig.
COSPEC
36b).
b
b
a
a
These complications can be
(b + b ) =
= plume width
+
tan θ
tan α
overcome if the measurements are
made beneath the plume, rather than Fig. 37. Determining the width of a horizontal flattened plume from
beneath the plume.
from the side (Fig. 37). This is our
preferred orientation for a flattened
horizontal plume. In this case, the two component lengths, b1 and b2, are summed:
1
1
(22)
2
2
width = (b1 + b2) = (a/tanθ) + (a/tanα)
Once the width is determined, the plume speed can be measured as shown in section 2.1.
3.10.5. Advantages/disadvantages
This is the least reliable of techniques, as significant error is incurred due to the large and
highly variable path-lengths from the plume, resulting in signal attenuation from scattering.
Varying plume geometries also add to the error in the flux measurements. However, in many
cases this is the only possible technique due to the volcano’s inaccessibility either by air or
ground.
4. RECORDING THE DATA
It is important to record as much information as possible (e.g., date, time, wind speed and
direction, volcanic activity, etc.) before and during a measurement; at the end of the day, the data
processing and interpretation will be significantly easier. Data recording from the COSPEC
usually consists of an analogue paper chart recorder, a portable computer, and/or a datalogger.
We now discuss these three options in turn.
4.1.
Chart recorder
4.1.1. Useful chart recorder characteristics
1) 1-2 channels
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
2)
3)
4)
5)
6)
7)
101
internal rechargeable battery if possible
flatbed so easier to write on – paper lies flat
millimetre paper with squares on it
spring loaded pens and quick drying ink
small enough to hold in your lap
variable paper speeds and attenuation voltages
4.1.2. How to do it
Data from the COSPEC is in analogue form (i.e., a voltage), and thus a chart recorder has
been the traditional method of data collection. The chart recorder is connected to the COSPEC
data outlet and may be powered either by internal batteries or a 12-volt power source such as a
storage battery, gel-cell battery, or the vehicle power supply. Prior to commencing
measurements, recharge the batteries and keep spares on hand. The chart recorder should be set
to an appropriate chart speed in order to give good spatial resolution. For example, over a
medium duration measurement on the ground (e.g., 30 min), the chart may be set to 20 mm min1
, while for longer measurements (e.g., 1-2 hrs), the speed may be set to 160 or 180 mm hr-1 (2.73.0 mm min-1). By contrast, aircraft measurements should use fast chart speeds (40 mm min-1 or
higher), since the time spent beneath the plume is short. Airborne measurements at Colima
volcano, Mexico, in January 1997 (traverse time ~5 minutes) used chart speeds of 50-60 mm
min-1, while car-based measurements at Arenal volcano in March 1996 (traverse time ~20
minutes), had chart speeds of 20-30 mm min-1. Chart speeds as slow as 180 mm hr-1 were used at
Popocatépetl in April 1997, where a typical ground traverse would take approximately one hour.
The attenuation voltage of the recorder can be changed in order to enhance the signal
amplitude coming from the COSPEC. For example, if the concentration in the plume is low and
the resulting peak heights are small, a decrease in the voltage of the chart recorder (e.g., from 1
V to 0.5 V) will increase the apparent peak and calibration height, increasing resolution and
A
1V
0.5 V
High ppm-m cell
calibration
Low ppm-m cell
calibration
B
0.5 V
1V
Fig. 38. The results of (a) reducing the chart recorder voltage from 1 to 0.5 V to increase resolution and (b)
increasing from 0.5 to 1 V to fit the plume trace to the chart record. At least one reconnaissance orbit or
traverse is needed to properly set the voltage.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
102
facilitating subsequent calculations
High ppm-m cell
and reducing potential reading error A
calibration
Pen up
(Fig. 38a). The inverse is also true;
Pen up
should the incoming signal be too high
Low ppm-m cell
calibration
36
37
for the chart paper, an increase in the
Segment
chart recorder voltage will decrease
event mark
38
35
the apparent peak height (Fig. 38b).
As a rough guideline, one wants to
maximise the height of the high (or B
Interference
Interference
low) calibration peak on the chart
paper, assuming that the maximum
peak height of the plume fits. In
Segment
practice, at least one complete
event mark
reconnaissance
measurement
is
normally required to determine
optimum chart recorder settings.
Interpolate line
When passing under UV obstacles
to facilitate
measurements
or trees (ground-based mobile), the
pens of the chart recorder can be lifted
off the paper in order to reduce the
interference. Once the signal has Fig. 39. (a) Example of a chart record with the pen lifted off the
returned to the previous level, the pen paper at intervals in order to avoid spikes caused by interference
may be placed back on the paper (Fig. of overhanging trees. (b) If the interference patterns are left on
39a). Should stretches of road have the chart record, an interpolated line may be drawn through the
significant amounts of tree coverage, interference in order to facilitate calculations.
it is advisable to stop briefly in clear
treeless areas every so often in order to get at least some measurements for that section of the
traverse. One also may leave the pen recording continuously, despite the interference (as is
always the case when using a datalogger/computer), and attempt a best fit to the curve in order to
aid in calculations (Fig. 39b). In this case, a time constant of 1 on the COSPEC is essential so
that the recording pen will quickly return to its correct position once the trees have been passed.
The operator should also decide how he/she prefers to read the chart record, i.e., from left to
right or right to left. This can be easily adjusted by changing the polarity (i.e., switch the wires at
the back of the chart recorder) of the input signal. A copy of the chart record should always be
made as a backup.
4.1.3. What needs to be recorded?
The more information that is recorded before and during a given measurement, the easier it is
to interpret when in the lab. Prior to making any measurements, the following details should be
written on the chart record and/or notebook:
Date
Volcano
Survey type (e.g., air, stationary, car)
Chart speed (e.g., 20 mm min-1)
Time Constant on COSPEC
Voltage attenuation (e.g., 0.5 V)
Pen colour defined (e.g., red = SO2, blue = AGC)
COSPEC model
AGC (on ground and at takeoff if doing airborne
measurements)
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
103
During measurements:
Note time at event or segment marker
Direction of measurement (e.g., Eastwards =
traverse #1)
Use codes for events (e.g., T = tree, St. = stop)
A two-channel chart recorder is often used, allowing for the simultaneous measurement of
SO2 and AGC. Monitoring both the AGC and SO2 signal is useful but not essential; in some
instances SO2 and CO2 (via LIDAR) are measured simultaneously. The traces made by the chart
recorder pen represent the variation in UV absorption by SO2, with the actual SO2 concentration
(ppm·m) being determined by comparison to the calibration peaks. When passing segment
boundaries on the road or landmarks when airborne, the position and local time are marked on
the chart paper using the event marker (Fig. 39). Noting the position and/or time of the segments
on the chart record and in a notebook is necessary for the subsequent flux calculations of the
individual segments, depending on the method used.
4.1.4. Advantages/disadvantages
Chart recorders have the advantage of being very simple to use. They allow for immediate
visual determination of the plume position and give the operator a permanent backup paper copy
of the data. The chart recorder is also ideal for areas where there is intermittent interference (e.g.,
when passing below overhanging trees during ground measurements), as by lifting the pen from
the paper, one effectively removes the interference patterns that otherwise make interpretation of
the data difficult (Fig. 39b). It has the disadvantage of requiring time consuming manual
calculation after measurements have been made. The chart recorder pens will tend to dry up and
on uneven roads or in turbulent conditions (e.g., aircraft, boat), it is often difficult to write clearly
on the record. Furthermore, in contrast to a computer or digital datalogger, a person is required to
watch it.
4.2.
Computer and analogue/digital converter (ADC)
4.2.1. How to do it
The data coming from the COSPEC are initially in analogue (voltage) format, and thus an
analogue-digital converter (ADC) of some sort is required to digitise the information. The digital
data (Fig. 40a) are collected on a portable computer or datalogger, with measurements being
taken every second. For traverses lasting up to an hour, the measurement interval is usually one
second (e.g., 3600 data points over an hour). For longer traverses, an interval of 2, 5, or 10
seconds can be used, depending on the length of the traverse. The data can then be processed and
graphed using commercial spreadsheet software. From the graph, the beginning and end of the
plume traverse may be located from event marker times or notes taken, consequently the flux
may be calculated (Fig. 40b).
Many different ADCs exist on the market today, and they are continually evolving. For this
reason, we do not want to recommend a specific make or model. Rather, we would like the user
to consider the following points:
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
•
•
•
•
•
the ADC ideally should not require a
separate power supply, instead being
powered by the computer or
independently;
the resolution should be sufficient to
record short plumes; usually 8-bit
resolution (1 part in 256) suffices;
the sampling rate should be variable,
from at least 1 sample s-1 to 1 sample
per 10 s;
multi-channel ADCs allow one to
record SO2 and the AGC (and
perhaps temperature, air pressure,
CO2, GPS altitude & position, etc.);
the computer software should be
simple to use in the field;
use of an ADC requires a portable
laptop computer which needs to be
powered for a sufficient length of
time (typically several hours).
4.2.2. What needs to be recorded?
A
Time stamp
(1 second interval)
B
13 :52:5 4
13 :52:5 5
13 :52:5 6
13 :52:5 7
13 :52:5 8
13 :52:5 9
13 :53:0 0
13 :53:0 1
13 :53:0 2
13 :53:0 3
13 :53:0 4
13 :53:0 5
13 :53:0 6
13 :53:0 7
13 :53:0 8
13 :53:0 9
13 :53:1 0
13 :53:11
13 :53:1 2
13 :53:1 3
13 :53:1 4
13 :53:1 5
13 :53:1 6
13 :53:1 7
13 :53:1 8
13 :53:1 9
13 :53:2 0
13 :53:2 1
13 :53:2 2
13 :53:2 3
0.4 5024
0.4 5204
0.4 541
0.4 5337
0.4 5393
0.4 5662
0.4 5432
0.4 4809
0.4 508
0.4 588
0.4 5296
0.5 2775
0.7 1733
0.7 676
0.7 857
0.7 8363
0.7 9042
0.79 527
0.7 9949
0.7 993 9
0.7 977
0.7 9656
0.7 9666
0.7 9425
0.7 9715
0.7 9513
0.7 8847
0.5 5686
0.4 8311
0.4 6448
Relative background SO2
absorption signal = "0"
High calibration
peak
0.85
0.8
0.75
0.7
Digital units
•
104
0.65
0.6
0.55
0.5
0.45
0.4
13:52:08 13:53:10 13:54:13 13:55:16 13:56:18 13:57:21 13:58:23 13:59:26 14:00:29 14:01:31 14:02:34 14:03:36
Time
Fig. 40. Example of (a) digital data received from the
As with the chart recorder, the UV COSPEC and (b) the resulting digital graph. Note the
absorption signal should be measured. It is grey area which represents the digital data for the high
also necessary to record the exact time at calibration peak and that the background is for a clear
each segment marker in order to determine sky background.
the position of each segment in the digital
record. Thus, one must synchronise the operator’s watch with the internal computer clock (or
vice versa) prior to start of the measurements. This should be done at the beginning of each day’s
measurement, as computer clocks are often not very accurate. It should also be recorded in field
notes and on the chart paper at the start, at several places during the traverse and at the end of the
measurement.
4.2.3. Advantages/disadvantages
Computer recording has the advantage of allowing rapid and precise data collection. Recent
advances in palm top computing greatly enhance the portability of these instruments. However,
in order for extended use, the portable computer should be capable of being connected to a
constant power supply (e.g., 12 V car battery using a cigarette lighter adapter), as computer
batteries are notorious for their short life spans. Software that graphs the data in real time is
desired. A chart recorder may also be connected in series to record hard copy data
simultaneously. Spikes in the data, caused by interference from overhanging trees, radio
interference, solar spikes and unknown transgressions, must be removed from the data set before
making calculations; unless specialised software is used, this can be very time consuming.
Because the COSPEC output is in analogue form, an analogue/digital converter is required. This
small converter is often battery powered, and thus data will be lost if the battery fails.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
4.3.
105
Dataloggers
Instruments exist that allow for the digital collection of data from the COSPEC without using
a portable computer. Dataloggers are typically small enough to be handheld, while some of the
newer varieties are sufficiently small to be attached directly to the COSPEC and/or chart
recorder (Fig. 41). Before purchasing a datalogger, one must determine the features required for
efficient use with a COSPEC:
• is the memory size sufficient to store a day’s
worth of points (e.g., >20,000 points) or merely
an hour (e.g., 3,600 points) at a sampling rate of
1 sample s-1?
COSPECs
• is the power supply sufficient for long periods
IV & V
of measurement, or must one change batteries
frequently?
• how easy is data transmission and conversion;
can one easily transfer the data to a computer
after the measurements?
• what is the range of sampling rates (e.g., 1
sample s-1, 1 sample hr-1, etc.)?
Laptop/
• is there an internal clock to synchronise with
ADC
the computer, operator’s watch or GPS
receiver?
• what is the resolution of the instrument, i.e., 8- Chart
bit (1 part in 256), 12-bit (1 part in 4096) or 16- recorder
bit ?
• does the datalogger have more than one
channel, i.e., can one record both SO2, AGC,
CO2, GPS, temperatures, etc. at the same time?
• is the instrument sufficiently small to make it Fig. 41. Example of a computer / ADC
connected to a chart recorder and COSPEC.
more practical than a portable computer?
4.3.1. Advantages/disadvantages
The main advantage over portable notebook computers for digital recording is the relatively
small size and sealed electronics of the datalogger. Dataloggers often have shock-resistant,
sealed cases for protection from the elements. However, the need to transfer and convert the data
from the datalogger to the computer and the limited memory capacity of some dataloggers can
make this method less attractive for data collection.
Of all the recording techniques mentioned above, the computer/datalogger is by far the most
efficient means of recording data, and subsequently calculating SO2 flux. The paper chart
recorder and field notebook can nevertheless be used as redundant, backup systems.
5. TROUBLESHOOTING
A properly tuned and aligned COSPEC can be expected to perform reliably for extended
periods of time. However, it is nevertheless advisable that one test the instrument (turn on the
COSPEC and record a section of chart with low and high calibrations) at regular intervals
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
106
depending on the usage (monthly to daily during a crisis). Keeping a record of all maintenance in
a logbook is also advisable, especially if the COSPEC eventually has to be returned to the
manufacturer for repair.
5.1.
Tuning the micrometer
9.4
8.65
AGC
Increasing voltage reading (decreasing light)
2
8.0
Positive Signal
8.50
**
Once the micrometer has initially been set by the manufacturer, it should not be changed.
However, if for some reason the micrometer position changes, it will be necessary to readjust it
to the original position (see the manufacture’s guide for the specific grating angle for your
instrument). This is done by first connecting the COSPEC to a power supply and turning it on.
The telescope is connected to an artificial light source (the “telescope lamp”), supplied by the
manufacturer. One must remember to turn on the lamp. The COSPEC is then connected to a twochannel chart recorder to measure SO2 and AGC so that the variation of the grating angle can be
determined (Fig. 42). Note that the chart speed should be slow and with an appropriate
attenuation voltage set (see section 4.1). The outer protective casing of the micrometer should be
removed (if one exists) and the locking screw, at the base of the micrometer, should be released.
With the high calibration SO2 cell in the light path of the instrument, the grating angle is
decreased by manually turning the micrometer screw at a constant and smooth rate. Starting at
short wavelengths (higher AGC readings), turn the micrometer towards longer wavelengths
(decreasing AGC values). On the chart record, the pen will draw a series of positive and negative
peaks (from “zero”) for the SO2 signal (Fig. 42). Continue slowly turning the micrometer screw
until these peaks lose amplitude. From the chart record, the second peak to the left of the highest
peak corresponds to the proper grating
angle for the instrument (Fig. 42). One
should then return to the shorter
wavelengths and again start turning the
micrometer screw slowly towards longer
wavelengths while watching the chart
record. One continues to turn the
micrometer screw until the chart recorder
pen reaches the top of second peak to the
left of the highest peak. The locking screw
should then be tightened and the protective
micrometer case should be replaced. This
process should be repeated a number of
AGC
SO Signal
times until the operator becomes familiar
with it.
5.2.
Changing the printed circuit
boards
Fig. 42. Typical chart record of the variation of the SO2
and AGC signals as the micrometer is tuned. The optimal
micrometer position (wavelength) is at the top of the
peak which is located two peaks to the left of the highest
peak (in this case, 8.50). Data from Barringer Research
Ltd.
Dust and ash can get inside the
COSPEC and environmental conditions
can corrode the contacts on the boards.
Contacts can be cleaned with a pencil
eraser. A spare set of circuit boards always
should accompany the COSPEC when in
the field. The circuit boards come as part
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
107
of a matched set and thus should always be replaced together. One must never mix boards from
different versions of the COSPEC, e.g., COSPEC IV to V.
5.3.
Electrical problems/power sources
5.3.1. Power supply
If there are problems with the COSPEC power, the first thing to verify is that everything
plugged in and that any batteries being used are fully charged.
5.3.2. Wiring problems
If the COSPEC is not functioning properly, verify that the connections are good and that
there are no loose wires. If a contact has broken, it should be resoldered or screwed back in
place. Spare power and data cables, fuses, and duct tape should always accompany the
instrument.
5.3.3. Electrical fields
Electrical fields such as those generated by ground-based radio antennae and pilot
transmission/communication may affect COSPEC measurements by causing spikes or noise on
the chart record. If possible, radio communication should be limited to periods before or after the
traverse. These electrical interferences are often difficult to avoid and must be eliminated from
the chart record or digital record before doing the flux calculations. With a digital record
(computer or datalogger), the spikes may be removed using commercial spreadsheet software. In
the case of analogue chart recorders, these points may be ignored by simply plotting a line
through the interference pattern (Fig. 39) or by digitising the chart.
5.3.4. What happens when you hook things up backwards?
If the analogue wires (the SO2 and
ground) are connected to the chart recorder
backwards, the relative SO2 signal and
calibrations will appear inverted (Fig. 43).
If the COSPEC output is switched to the
analogue/digital converter (ADC) and to
the computer, the ADC may stop or the
computer program recording the data may
crash. If the program is relatively flexible,
the data simply may be recorded as
inverted values, with the resulting data plot
resembling that of the inverted signal on a
correct chart record.
Inverted
SO 2 signal
Low calibration
High calibration
Fig. 43. Example of an inverted SO2 signal, which is the
result of connecting the SO2 and ground wires backwards.
5.3.5. Chart recorder/computer problems
Dust and ash may cause serious problems with the instruments (COSPEC, chart recorder,
computer, datalogger, etc.) if allowed to accumulate. One should therefore clean the instruments
after an ashy day or after having worked in a dusty environment (dirt/gravel roads, etc.).
Cleaning the COSPEC glass and mirrors with high purity alcohol and swabs will help maintain
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
108
high signal quality. By making COSPEC measurements with both chart recorder and
computer/datalogger connected, one reduces the possibility of total data loss should one of the
data recorders fail.
6. REDUCING THE DATA
In all COSPEC measurements, the SO2 flux is determined by calculating the flux for each
segment, which is then summed to determine the total SO2 flux for a given traverse. In the case
of airborne or boat-mounted measurements, only one flux calculation per traverse is generally
performed, as it is rarely necessary to divide the traverse into multiple segments. The SO2 flux
(F) in metric tonnes per day is calculated using the following equation:
(23)
F = [SO2]pl cosθ dseg νwind CF
where [SO2]pl is the average path-length concentration of SO2 (ppm·m) in the plume for the
segment, θ (º) is the deviation from perpendicularity of the segment of road with respect to the
gas plume (Fig. 13) and dseg is the length (m) of a particular segment. For ground-based mobile
measurements, the segment width is determined from a map or GPS. The term νwind is the
average wind speed (m s-1) and Cf is a conversion factor changing ppm·m3 s-1 into metric tonnes
per day (t d-1). One ppm·m of SO2 is one cubic centimetre of SO2 gas uniformly mixed in one
million cubic centimetres of air and viewed by the COSPEC over an optical path of one metre at
a pressure of 101.325 kPa and a temperature of 20°C. Thus, the conversion factor (CF) is derived
as follows:
(24)
CF = ρSO2 CfSTP 0.001 10-6 86,400 = 0.00023
where ρSO2 is the density of SO2 gas at standard temperature and pressure STP (2.8579691 kg mat 0°C and 101.325 kPa), CfSTP is a correction factor (273.15/293.15 = 0.9317755) to change
the SO2 gas density from 0°C to 20°C (273.15 K to 293.15 K), 0.001 converts the kg term in the
density to metric tonnes, 10-6 converts ppm to mass units, and 86,400 are the number of seconds
in one day. Thus, CF is equal to 0.00023008194, with units of t s m-3 d-1 ppm-1.
The average concentration of SO2 for the segment ([SO2]seg) is calculated from:
3
(25)
[SO 2 ]seg =
Pseg
Pcal
[SO 2 ]cal
where [SO2]cal is the concentration of the appropriate calibration gas cell in ppm·m, Pcal is the
peak height of the appropriate calibration gas cell in arbitrary units and Pseg is the average peak
height for the segment calculated from:
(26)
Pseg =
segment area
segment width
If the average segment peak height is lower than the low calibration peak height, then the low
calibration cell should be used in the above calculation. If the average segment height falls
between the high and low calibration, then use the average of the two. If average segment height
is greater than the high calibration peak height, use the high calibration.
When making stationary COSPEC measurements, the flux is calculated from:
(27)
E = cosA d νrise CF [SO2]
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
109
where A is the angle is the COSPEC inclination (see section 3.10; Fig. 28d), d is the plume width
(m), νrise is the plume rise speed (m s-1); CF is the conversion factor (t s m-3 d-1 ppm-1; Eqn. 24)
and [SO2] is the average concentration (ppm·m; Eqn. 25).
An example of a typical COSPEC flux calculation follows (Eqn 23; Table 5). For a traverse
between segment markers L2 and L3 at Masaya volcano, Nicaragua, the segment width (dcol) and
average peak height (Pseg) are 3,400 m and 2.8 cm, respectively. With a high calibration peak
height (Pcal) of 5.85 cm and concentration ([SO2]cal) of 326 ppm·m, the segment concentration
([SO2]seg) is calculated to be 135.9 ppm·m (Eqn. 25). The deviation from perpendicularity of the
segment of road with respect to the gas plume (θ) is 81º and the measured wind speed (νwind) is
10 m s-1. Thus, with a conversion factor (CF) of 0.00023008194 (Eqn. 24), the SO2 flux E for
segment L2-L3 is calculated to be 1050 tonnes per day. (Eqn. 23). The total SO2 flux (1270 t d-1)
for the traverse is thus the sum of segments L1.5-L2, L2-L3 and L3-L4 (Table 5).
Table 5. Example of a spreadsheet for COSPEC calculations of ground-based mobile measurements at Masaya
volcano, Nicaragua
Instrument:
Route:
Segment
No.
L-2-L-1
L-1-L0
L0-L1
L1-L1.5
L1.5-L2
L2-L3
L3-L4
L4-L4.5
L4.5-L5
L5-L6
L6-L7
L7-L8
L8-L9
COSPEC V
Las Palmas to El
Crucero
High cal (ppm-m):
326
Low cal (ppm-m):
103
Difference of
Length of Azimuth of Azimuth of segment & Abs. value
segment
segment
plume
of Δθ
plume: Δθ
(m)
(degrees)
(degrees)
(degrees)
(degrees)
3350
285
256
29
29
650
5000
4000
3300
3400
1400
3600
1700
2800
5500
3000
1800
248
350
309
315
337
0
325
28
9
343
15
14
256
256
256
256
256
256
256
256
256
256
256
256
-8
94
53
59
81
-256
69
-228
-247
87
-241
-242
8
94
53
59
81
256
69
228
247
87
241
242
Sin
of Δθ
Abs.
Cal cell Cal cell Segment Segment SO2 flux
value of Wind speed conc.
Peak
Peak
conc.
(metric
(m/s)
(ppm-m) height
height (ppm-m) tons/day)
Sin Δθ
0.4399
0.4399
10
284
5.85
0.1253
0.9955
0.7396
0.7996
0.9557
-0.7705
0.8838
-0.4258
-0.6730
0.9792
-0.6004
-0.6129
0.1253
0.9955
0.7396
0.7996
0.9557
0.7705
0.8838
0.4258
0.6730
0.9792
0.6004
0.6129
10
10
10
10
10
10
10
10
10
10
10
10
284
284
284
284
284
284
284
284
284
284
284
284
5.85
5.85
5.85
5.85
5.85
5.85
5.85
5.85
5.85
5.85
5.85
5.85
0.00
0.52
2.8
0.36
0.00
0.00
0.00
25.24
135.93
17.48
0.00
0.00
0.00
0.00
0.00
0.00
Total:
0.00
0.00
0.00
0.00
164.24
1049.90
54.60
0.00
0.00
0.00
0.00
0.00
0.00
1268.74
↓
Total SO2 Flux: 1270 t d-1
6.1.
Chart recorder
6.1.1. Area calculations – determining Pseg
While there are a number of different ways of determining Pseg, it is advisable to have several
methods available as backups in case of technical problems.
Box counting
There are a number of different methods for calculating chart recorder data. The most basic
method is the “box counting” method. The chart record first should be broken into segments (if
necessary, e.g., a ground survey) so that individual calculations and corrections can be made (a)
Box counting method of calculating the average peak height for a given segment; (b) simple
triangle method - half base x height = area; (c) reduction of the segment width to exclude the part
outside the plume (Fig. 44). The width (of each segment) of the anomaly is divided into equally
spaced (e.g., 1 mm) divisions. One then determines the baseline or background signal, which
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
Simple Triangle
A
26
Plume signal values
should be a nearly horizontal line
above which is the plume signal.
The baseline is then set to zero and
values assigned (relative to this
baseline) to the mid-point of each
division (Fig. 44a). For example, if
the chart recorder uses metric paper,
one divides the plot into millimetrewidth columns and then assigns a
value to each division relative to the
zero baseline. This is repeated
throughout the plot. Each data
point, with its assigned y-value, is
then summed for the segment in
question, and then an average value
determined. This average value thus
represents the average concentration
of the plume for that segment in
units of millimetres. The same is
done with respect to the calibration
peaks. The average segment height
in mm is then related to the
calibration peak height in mm so
that
the
plume-segment
concentration in ppm·m may be
determined using Equations 25 and
26. In cases where the segment is
not entirely in the plume, reduce the
segment width to exclude the part
outside the plume (Fig. 44c).
110
Average of all boxes
= 9.9
24 .9
24
22
24 .3
23 .5
20 .3
18
19
17 .3
14
12
12 .3
10
7. 3
7. 3
6
2
zero
baseline
5. 5
3. 5
0
0
1
2. 3
2
Segments 2
1
1 0. 4 0 0 0
3
4
B
Height
Base
1
3
2
Segments
4
C
Segments
1
2
3
3a 4
partia l dista nc e
distance 3- 4
Fig. 44. (a) Box counting method of calculating the average peak
height for a given segment; (b) simple triangle method - half base x
height = area; (c) reduction of the segment width to exclude the
part outside the plume.
A very simple and quick method of determining the area beneath a chart record is to calculate
the best fit triangle to the plume anomaly; the area is equal to ½ Base times Height (Fig. 44b).
While this is a very quick method, it is much less accurate than box counting and should only be
used for rapid first order estimates.
Planimeter
The second method for calculating chart data uses a mechanical or digital planimeter to
measure the area beneath the segment of the plume. The “bull’s-eye” or pointer of the planimeter
is placed on the chart recorder strip and zeroed. Moving in a clockwise direction, one then uses
the “bull’s-eye” to trace the SO2 curve for the segment in question. One must start and end at the
same segment point, i.e., follow the curve to the end of the segment and return along the baseline
to the starting position. Typically, two to three planimeter measurements are made for each
segment to check precision and accuracy. One then divides the area by the width of the chart
segment to obtain the average anomaly height for the segment (Eqn. 26). The average anomaly
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
111
height is then compared to the height of the calibration peak(s) to obtain the concentration of SO2
for the segment (Eqn. 25). A drawn area of 1 square cm or 20 mm on a side should be used to
practice using the planimeter prior to starting measurements of the anomaly. This method has the
advantage of being somewhat faster than the box counting method for multiple iterations on the
same segment and has a higher precision and accuracy in the case of very steep slopes on the
segment curve. These mechanical techniques (box counting, simple triangle and planimeter) are
simple and relatively fast ways of making area measurements.
6.2.
Digital data
A
10:03:08
10:03:10
10:03:12
10:03:14
10:03:16
10:03:18
10:03:20
10:03:22
10:03:24
10:03:26
10:03:28
10:03:30
10:03:32
10:03:34
10:03:36
10:03:38
10:03:40
10:03:42
10:03:44
10:03:46
10:03:49
10:03:51
10:03:53
6.2.1. Area calculations
0.16494
0.19397
0.14692
0.12441
0.05447
0.03007
0.04937
0.0805
-0.45645
-0.88753
0.06222
-0.69804
-0.72422
0.02795
-0.18312
-0.1787
-1.01576
0.09837
0.00166
-0.80589
-0.54895
-0.54831
-0.52391
Relative SO2
absorption
signal
The basic digital method is exactly the
Time stamp
same as the box counting method, except there
Interference from trees,
is no need to give each data point a value.
street lamps, power lines
Using any commercial spreadsheet program,
one separates the data into segments (if
necessary), based on the times for each B
segment division noted on the chart paper
during measurement. One then takes the
average of the data points for the desired
segment and the average of the data points for
the appropriate calibration, in order to
calculate concentrations in ppm·m, as shown
above (Eqn. 25). As the software will average
all data points for a given segment, one must
remove all interference spikes due to trees,
radio transmission, etc., prior to averaging
Time
(Fig. 45). These data spikes may be removed
manually by using a spreadsheet program and Fig. 45. Example of (a) digital data received from the
and (b) the resulting digital graph, showing
selectively erasing data, or by creating a COSPEC,
noise caused by interference from overhanging trees
macro to remove all data points greater than a by the road.
given limit (i.e., smoothing). A backup must
be made before beginning in order to avoid erasing useful data.
A more efficient method graphically analyses the data and determines the area beneath the
curve for the desired segment. This may be done with scanned (auto-digitised) chart recorder
data and digital data recorded by computer or datalogger. Many of these software also have the
ability to filter out repetitive interference patterns (e.g., trees, helicopter blades, etc.). However,
in order to make this technique truly efficient, a software package written specifically for
COSPECs is required, i.e., one that automatically records the data digitally, presents it
graphically, and analyses the area beneath the graphed curve.
1
Digital units
0.5
0
9:50:24
9:53:17
9:56:10
9:59:02
10:01:55
10:04:48 10:07:41
-0.5
-1
6.3.
The use of spreadsheets in data reduction
Even if one is manually calculating concentrations from chart recorder strips, the use of
computer spreadsheets greatly facilitates data reduction and flux calculations. As discussed
above, spreadsheets can treat the initial raw digital data, letting the operator selectively remove
interference spikes. Once the constant values (e.g., segment width, segment azimuth) and flux
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
112
calculation are entered, one need only enter the variable data (e.g., wind speed, column azimuth,
segment concentration) to determine the total SO2 flux for a given traverse (Table 5).
6.4.
Uncertainty calculations
Constraining the uncertainties in COSPEC flux calculations can be problematical. The most
important uncertainty is clearly the accurate determination of plume height wind speeds (see
section 2.1), however, calibration, environmental and instrumental uncertainties must also be
accounted for. A total “uncertainty” for a give flux can be calculated using the square root of the
sum of the squares (Table 6):
(28)
Total Uncertainty = U 12 + U 22 + U 32 + K + U n2
It is up to the user to decide whether to present these uncertainties as “error bars” or merely state
the average uncertainty for a measurement.
6.4.1. Calibration uncertainties
If there appears to be
significant variations between
calibrations before and after a
given measurement (e.g., due to
variable cloud cover), the most
“representative” calibration or
calibration closest to the plume
should be used in the
calculations. Multiple traverses
on a given day will allow the
user to determine what is a truly
representative calibration peak.
The user may also consider
performing two or three flux
calculations, (i.e., a low-, highand
average-calibration
calculation) and thus presenting
minimum,
average
and
maximum values.
Table 6. “Calculated” uncertainties for SO2 measurements, Arenal
volcano, 1995-1996.
Calibration cell concentrations:
COSPEC IV:
300 ppm-m
COSPEC V:
339.2 ppm-m
Digital record reading uncertainty:
Variation in car speed:
Windspeed determination:
1995: 0-60%
1996: 6-26%
30%
14%
Total uncertainty (square root of the sum of the squares):
1995
Minimum:
Maximum:
Average:
Minimum:
1996
Maximum:
Average:
6%
60%
31%
9%
27%
15%
2%
2%
2%
5%
6.4.2. Background uncertainties
Variations in calculated SO2 fluxes between individual traverses may be due, in part, to
fluctuations in wind speed and direction, changes in cloud cover, and change in sun angle,
resulting in variable amounts of solar ultraviolet radiation. The opacity of an eruptive plume also
varies due to changes in ash content that will increase the absorption of ultraviolet radiation
(Andres and Schmid, 2001). Instrumental uncertainties include instrument calibration (± 2%),
digital/analogue chart reading uncertainties (± 2%), varying vehicle/aircraft speed (± 5%), and
wind speed measurement (± 0-60% or greater) (see below, Table 7; Casadevall et al., 1981;
Stoiber et al., 1983). Chart and vehicle speed uncertainties can be reduced somewhat by digital
recording the SO2 absorption and vehicle speed (see sections 4.2 and 4.3).
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
113
6.4.3. Wind speed uncertainties
At Arenal volcano in 1995 and 1996, the uncertainty in wind speed measurement was due, in
large part, to the fact that measurements were made at the base of the volcano (elev.: ~550 m),
which did not represent wind speeds at the summit of the volcano (1690 m) where degassing is
taking place (Williams-Jones et al., 2001). As there is approximately a 1-1.5 km difference in
altitude between the ground and the plume, the plume velocity may be up to 10 times greater
than that measured on the ground (Willett and Sanders, 1959). In an extreme example, a
radiosonde profile of wind speeds recorded from the Mexico International Airport (2230 m a.s.l)
shows that wind speeds were almost 20 times greater at the summit height of nearby
Popocatépetl volcano (5430 m a.s.l) than ground level winds at the airport (Williams-Jones et al.,
2006).
Wind speed measurements are also affected by instrumentation uncertainties. Wind
measurements made in 1995 have an average standard deviation of ~30%, as the digital
anemometer that was used gave readings that varied continually with local wind gusts (Table 7).
The operator thus was forced to estimate the average range of speeds at the time of measurement.
The 1996 wind measurements used a fully mechanical anemometer that allowed for the
integration of wind speed over an interval of one minute. Consequently, the standard deviation
was ~14%, or about half that of the previous year (Table 6). Total average uncertainties of ~31%
and ~15% for the SO2 flux were calculated for 1995 and 1996, respectively (Table 6). However,
these uncertainties are probably significantly underestimated and thus, where possible, plume
height wind speed measurements should be made (e.g., with the dual spectrometer method, see
section 2.1.3).
Table 7. Wind speed measurements 4 km west (elev. ~550 m) of the summit of Arenal volcano, Costa Rica, 1995-1996.
1995
Date
-1
Speed (m s )
2.5
1996
Avg. StdDev % Dev
(m s-1)
2.5
Avg.
(m s-1)
28/02/96 2.23 2.13 2.39
2.3
Date
Speed (m s )
StdDev % Dev.
28/02/95
2.5
04/03/95
4.1
01/03/96 3.29 3.84 2.25
3.1
05/03/95
3.9
3.6
3.8
0.21
5.7
03/03/96 2.63
2.6
08/03/95
2
1
1.5
0.71
47.1
05/03/96 2.77 4.09 4.09
3.7
0.76
20.9
22/03/95
2.6
06/03/96 2.73 3.83
3.4
0.58
17.1
23/03/95
2
1.8
3.1
0.36
11.7
24/03/95
1.2
1.2
1
Average:
13.6
27/03/95
2.2
1.3
3.5
30/03/95
4
3.5
4.1
2.6
31/03/95 2.25
01/04/95
4
2.5
-1
3
1.9
0.18
9.4
1.6
0.95
59.6
2.3
1.13
48.8
3.8
0.35
9.4
1.59
55.3
2.3
1.8
2.9
Average 33.6
08/03/96 2.82 3.33
3.6
0.13
5.8
0.81
25.8
0.00
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
114
6.4.4. Wind / plume direction uncertainties
Uncertainties in plume azimuth have a comparatively minor impact on the final SO2 flux
calculation. A difference of one degree, for example, gives a deviation of only 0.3%, while a
difference of ten degrees results in deviation of 4.6%. Thus, while the user should attempt to
determine the plume direction as accurately as possible, small uncertainties in direction can be
neglected.
7. PRESENTING THE DATA
Once the SO2 flux data have been collected and reduced, the data must be presented in a
manner which facilitates interpretation. Firstly, because of the difficulties in accurately
measuring wind speed as well as other uncertainties, the actual flux for a survey (e.g., of 10
measurements) should be rounded to the nearest 10 t d-1, i.e., 2640 ± 90 t d-1 instead of 2639 ± 93
t d-1. Critically, the wind speed and direction should always be presented in conjunction with SO2
flux data and distance from the volcano. The averaged SO2 fluxes for a day, month or year must
also be presented along with values for individual traverses. In order that other users may
2500
2000
16 July 1992 eruption
SO2 data with
measured windspeed
1500
1000
500
0
1500
1000
16 July 1992 eruption
SO2 data with standard
windspeed of 1 m s-1
eruption
-1
SO2 emission rate (t d )
Fig. 46. Diagrams for Galeras volcano,
Colombia, showing (a) SO2 flux with standard
wind speed of 1 m s-1 and SO2 flux with
measured wind speed, modified after Zapata et
al. (1997); (b) SO2 flux and duration of long
period seismic events versus time. Dots and
connecting lines refer to SO2 flux, bars are the
durations of seismic events. Note the low SO2
flux prior to the eruptions. Modified after
Fischer et al. (1994).
eruption
A
(a)
500
May
Durations of LP events (s)
200
June
July
August
Jan. 11
Phase 1
Phase 2
Mar. 12
Phase 3
SO2
800
Eruption
150
1,000
600
100
50
Eruption
400
Eruption
duration
200
Feb. 1
0
0
December 92
January 93
February 93
March 93
-1
(b)
B
SO2 emission rate (t d )
0
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
115
-1
ln [SO2] (t d )
properly compare data from a given
6.0
volcano taken at different times, all of
+
++
++
the above data should be routinely
+
+ + +
+ +
+
+ +
++
+
+
+
presented.
+
+ +
+ +
+ ++
5.5 + ++ ++ +
+
+
+
+
+
+
+
+
+
+ ++
A plot of SO2 fluxes calculated using
+++ + + ++
+ + + ++++ + + + +
+
+
+
+
+
+
+
++ ++ ++ +++ +++
+ + + +++ + +
both the measured wind speed and a
+ + + + ++
++ +
+ + + +++ +
++ ++ + +
+ + ++ +
-1
++
+
++
+ ++ ++
+
+
+
+
+
+
+ +
+
standard wind speed of 1 m s is a
++ ++ ++++ + ++
5.0 + +++++++ ++++++++ ++ +++++ + ++++ ++
++ + + +
+ + + + ++ + + ++ +
+ ++ + + +
useful method to eliminate some of the
++ ++
+
+ + +++ ++
+ ++
+
++ + +
+ + ++
+ +
wind speed uncertainties (Fig. 46a,
++
+
+ ++ +
+ +
+
+ +
+
+
++
Zapata et al., 1997). By standardising to
+
+
+
4.5
+
-1
1 m s , one eliminates the wind
+
contribution, facilitating comparison
+
with other flux data standardised in the
4.0
same manner. The measured wind speed
0
20
40
60
80
is nevertheless needed in order to
Time (days)
determine the calculated fluxes.
Another common method of Fig. 47. Diagram for Halema’uma’u crater, Kīlauea
presenting the data is by plotting SO2 volcano, Hawaii, showing ln (SO2) versus time with moving
fluxes and other measurements versus statistics applied to reduce noise. Top curve is the upper
time. A line graph is used to show the semimidmean, middle curve is the midmean, and the lower
variation in the SO2 flux with time, curve is the lower semimidmean. From Connor et al. (1988).
while seismicity is shown as a bar chart
(Fig. 46b, Fischer et al., 1994). One also can plot SO2 flux versus time as a log normal graph to
reduce some of the scatter resulting from large variations in flux (Fig. 47, Connor et al., 1988),
although one risks obscuring some detail. Moving statistics (Cleveland and Kleiner, 1975) can be
used with high noise to signal ratios in order to search for structure in the data sets. These
moving statistics can enhance the scatter plot and bring out structures in otherwise noisy data.
With large datasets, filtering methods also may be applied to the data in order to reduce noise. At
Etna, for example, a low-pass filter (threshold of 28 days) and a high-pass filter (threshold of 60
days) were applied to the data to minimise noise and allow identification of three distinct
degassing intervals (Fig. 48, Caltabiano et al., 1994). Another method for seeing changes in a
data set is to plot cumulative SO2 flux versus time (Fig. 49, Caltabiano et al., 1994). This plot
also reduces noise and clarifies changes in the degassing behaviour of the volcano, which will be
evident from changes in the slope of the cumulative curve (Fig. 49). Cumulative plots also can be
used for presenting seismic and deformation data, and thus this approach permits comparison
between different datasets.
One also should decide whether all of the data (SO2, seismicity, deformation, etc.) for a given
period is presented or whether only the average measurements of a day, month, year, etc., are
presented. Using only the average SO2 flux is yet another method of reducing the “noise” of a
given data set to bring out any possible correlations. For example, comparison of average
monthly SO2 with average monthly micro-gravity change at Masaya volcano, Nicaragua, shows
a distinct correlation (Fig. 50; Williams-Jones et al., 2003).
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
(a)
Unfiltered
-1
SO2 Flux (t d )
25,000
20,000
15,000
10,000
5,000
116
Fig. 48. SO2 flux versus time at Mount Etna
(a) before filtering and (b) after filtering with
cutoff threshold of 28 days and (c) 60 days.
Note the smoothing effect of increasing the
cutoff threshold. Zones A, B, and C represent
three different phases of eruptive activity.
From Caltabiano et al. (1994).
0
Jan. 89
Jan. 88
(b)
Jan. 90
Jan. 91
Time
SO2 Flux (t d -1)
25,000
Jan. 92
Filtered - 28D
B
A
20,000
C
15,000
10,000
5,000
0
27-Oct-87
(c)
22-Aug-88
18-Jun-89
14-Apr-90
9-Feb-91
5-Dec-91
Time
SO2 Flux (t d -1)
25,000
Filtered - 60D
A
20,000
B
C
15,000
10,000
5,000
0
27-Oct-87
22-Aug-88
18-Jun-89
14-Apr-90
9-Feb-91
5-Dec-91
Time
Fig. 49. Cumulative SO2 flux versus time at
Mount Etna. Note the change in the slope of the
curve for the three intervals of eruptive activity.
From Caltabiano et al. (1994).
8
SO2 Flu x (Mt)
A
B
C
6
4
2
0
27- Oct-87
22- Aug -88
18- Jun- 89
14- Apr- 90
Time
9-F eb-91
5-D ec-91
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
117
SO 2 flux
Gravity
change
-20
1 Standard deviation
of repeat measurements
-40
-1
SO 2 emission rate (t d )
2500
Error (20 μGal) on
1 measurement
2000
-60
-80
1500
500
0
1992
-100
?
1000
?
-120
?
?
Gravity change (μ Gal)
0
3000
-140
-160
1993
1994
1995
1996
1997
1998
1999
2000
2001
Fig. 50. Average monthly SO2 fluxes (circles) compared with average monthly gravity change (diamonds) at
Masaya volcano, Nicaragua. Grey circles represent average COSPEC measurements while shaded region represents
one standard deviation of flux values (~30%). Black diamonds denote repeat gravity measurements; white diamonds
denote single measurements. From Williams-Jones et al. (2003).
Acknowledgements
We thank Robert Dick of Barringer Research Ltd. for discussion regarding the inner working
of the COSPEC and Robert Andres of the Department of Space Studies, University of North
Dakota, for information on the influence of plume opacity on COSPEC measurements. We also
thank John MacGregor of Transport Canada for time in the Bell helicopter and Gary Grass, also
of Transport Canada, for assistance with the helicopter measurements.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
118
REFERENCES
Andres R. J., and Schmid J. W. (2001) The effects of volcanic ash on COSPEC measurements;
Journal of Volcanology and Geothermal Research, v. 108, p. 237-244.
Caltabiano, T., Romano, R., and Budetta, G. (1994) SO2 flux measurements at Mount Etna
(Sicily); Journal of Geophysical Research, v. 99, p. 12809-12819.
Caltabiano, T., Guiduzzi, G., Leuzzi, S., and Romano, R. (1992) Helicopter borne COSPEC SO2
flux measurement; Acta Vulcanologica, v. 2, p. 95-98.
Casadevall, T. J., Johnson, D. A., Harris, D. A., Rose Jr., W. I., Malinconico Jr., L. L., Stoiber,
R. E., Bornhorst, T. J., Williams, S. N., Woodruff, L., and Thompson, J. M. (1981) SO2
fluxes at Mount St. Helens from March 29 through December, 1980; in The 1980 Eruptions
of Mount St. Helens, ed. P. W. Lipman and D. R. Mullineaux, U.S. Geological Survey,
Professional Paper, p. 193-200.
Cleveland, W. S., and Kleiner, B. (1975) Enhancing scatter plots with moving statistics;
Technometrics, v. 17, p. 447-454.
Connor, C. B., Stoiber, R. E., and Malinconico Jr., L. L. (1988) Variation in sulfur dioxide
emissions related to Earth tides, Halemaumau crater, Kilauea volcano, Hawaii; Journal of
Geophysical Research, v. 93, p. 14867-14871.
Doukas, M.P. (2002) A new method from GPS-based wind speed determinations during airborne
volcanic plume measurements. U.S. Geological Survey Open-File Report 02-395, pp. 1-13.
Edmonds, M., Herd, R.A., Galle, B. and Oppenheimer, C.M.(2003) Automated, high timeresolution measurements of SO2 flux at Soufrière Hills Volcano, Montserrat. Bulletin of
Volcanology, v. 65(8), p. 578-586.
Fischer, T. P., Morrissey, M. M., Calvache V., M. L., Gomez, D. M., Torres, R. C., Stix, J., and
Williams, S. N. (1994) Correlations between SO2 flux and long-period seismicity at Galeras
volcano; Nature, v. 368, p. 135-137.
Stoiber, R. E., Malinconico Jr., L. L., and Williams, S. N. (1983) Use of the correlation
spectrometer at volcanoes; in Forecasting Volcanic Events, ed. H. Tazieff and J. C. Sabroux,
Developments in Volcanology, 1, Elsevier, Amsterdam, p. 425-444.
Millán, M. M. (1980) Remote sensing of air pollutants. A study of some atmospheric scattering
effects; Atmospheric Environment, v. 14, p. 1241-1253.
McGonigle, A.J.S., Inguaggiato, S., Aiuppa, A., Hayes, A.R., and Oppenheimer, C. (2005a)
Accurate measurement of volcanic SO2 flux: Determination of plume transport speed and
integrated SO2 concentration with a single device. Geochemistry Geophysics Geosystems, v.
6, Q02003, doi:10.1029/2004GC000845.
McGonigle, A.J.S., Hilton, D.R., Fischer, T.P., and Oppenheimer, C. (2005b) Plume velocity
determination for volcanic SO2 flux measurements. Geophysical Research Letters, v. 32,
L11302, doi:10.1029/2005GL022470.
Willett, H. C., and Sanders, F. (1959) Descriptive Meteorology; Academic Press Inc., New York,
355 p.
Chapter 3: J. Stix, G. Williams-Jones & C. Hickson / Applying the COSPEC at Active Volcanoes
119
Williams-Jones, G., Stix, J., Heiligmann, M., Barquero, J., Fernandez, E., and Duarte Gonzalez,
E. (2001) A model of degassing and seismicity at Arenal Volcano, Costa Rica. Journal of
Volcanology and Geothermal Research, v. 108, p. 121-141.
Williams-Jones, G., Rymer, H., and Rothery, D.A. (2003) Gravity changes and passive
degassing at the Masaya caldera complex, Nicaragua. Journal of Volcanology and
Geothermal Research, v. 123, p. 137-160.
Williams-Jones, G., Horton, K.A., Elias, T., Garbeil, H., Mouginis-Mark, P.J., Sutton, A.J., and
Harris, A.J.L. (2006). Accurately measuring volcanic plume velocity with multiple UV
spectrometers. Bulletin of Volcanology, v. 68, p. 328-332.
World Meteorological Organization (1996) Guide to meteorological instruments and methods of
observation, WMO, Switzerland.
Zapata, J. A., Calvache V., M. L., Cortés J., G. P., Fischer, T. P., Garzón V., G., Gómez M., D.,
Narváez M., L., Ordóñez V., M., Ortega E., A., Stix, J., Torres C., R., and Williams, S. N.
(1997) SO2 fluxes from Galeras volcano, Colombia, 1989-1995: progressive degassing and
conduit obstruction of a Decade Volcano, Journal of Volcanology and Geothermal Research,
v. 77, p. 195-208.