Eos,Vol. 84, No. 40, 7 October 2003
VOLUME 84
NUMBER 40
7 OCTOBER 2003
PAGES 409–424
EOS,TRANSACTIONS, AMERICAN GEOPHYSICAL UNION
Ground-based Infrared Monitoring
Provides New Tool for Remote
Tracking of Volcanic Activity
PAGES 409, 418
Thermal monitoring of active volcanoes has
long been the domain of satellite and airborne
remote sensing (for reviews of current capabilities, see Harris et al. [2002]). However,
ground-based thermal sensors offer considerable benefits in that (1) they can be located
beneath cloud decks that prohibit aerial views;
(2) they allow small thermal targets to be
resolved; (3) they observe targets with a constant
viewing geometry for long periods of time;
and (4) they provide data at high sample
rates (tens to hundreds of Hz).This latter
capability is extremely attractive when tracking transient or rapidly evolving events, such
as volcanic explosions. In addition, when used
in conjunction with other geophysical data
sets, thermal time series reveal clues as to the
manner in which a volcanic system is erupting.
Consider, for example, an explosion.The
thermal and infrasonic signals will propagate
from the explosion source at different velocities.The delay between the arrival of the two
signals at co-located thermal and infrasonic
sensors can thus be used to constrain source
depth and/or the velocity at which the ejecta
shell is travelling [Ripepe et al., 2002]. However,
permanent, continuously recording, groundbased thermal sensors are a rare sight at
active volcanoes. Reasons for this include
cost, system failure due to the harsh volcanic
environment, vandalism, and/or destruction
during violent volcanic and weather-related
events. However, technology has now advanced
to a point at which it is possible to build inexpensive,robust,expendable thermal monitoring
systems for deployment on active volcanoes.
Over the last 5 years, such a system has been
designed and tested at several active volcanoes.
This has allowed us to use this new volcano
monitoring tool to identify volcanic activity
styles and achieve real-time event tracking.
BY A. HARRIS, J. JOHNSON, K. HORTON, H. GARBEIL,
H. RAMM, E. PILGER, L. FLYNN, P. MOUGINIS-MARK,
D. PIRIE, S. DONEGAN, D. ROTHERY, M. RIPEPE,
AND E. MARCHETTI
Inexpensive Thermal Monitoring Systems
Designed for Volcano Deployment
The prototype system (Figure 1) was designed
to operate in the harsh conditions present at
the rim of Pu’u ‘O’o, the active vent system at
the head of the current eruption of Kilauea
volcano (Hawai’i).At this location, the thermal
system has to cope with an extremely acidic
and wet atmosphere. For example, during
November 2000, the system received 69 cm
of rain in one day alone.
The core of the system is the sensor module.
At a total cost of $1,150, this consists of an
Omega™ thermal infrared (8–14 µm) thermometer contained within a protective Pelican™
case.A selenium-germanium-arsenic window
permits the detection of infrared radiation in the
required wavelength range.The Omega™ thermometer outputs measurements of temperature in the range -18 to 1371°C at a high sample
rate,allowing acquisition of thermal time series.
These time series provide excellent temporal resolution and the ability to distinguish different
types of eruptive events that take place within
the Pu’u ‘O’o crater.The prototype system
employs two 1° field-of-view (FOV) sensors
and one 60° FOV sensor. The 1° instruments
provide extremely detailed thermal time series
for individual vents,whereas the 60° instrument
provides less detail (the signal being damped
as a result of integration over a wider surface
area), but allows thermal detection of any significant activity anywhere on the crater floor
(Figure 1).
The sensors are linked to an acquisition,
power, and transmission hub, built at a cost of
$2,300 and located 30 m from the crater rim.
From here, data are telemetered every 2 seconds, via a repeater site on the flanks of Mauna
Loa,to a reception site at the Hawaiian Volcano
Observatory (HVO). Here, data are archived
and displayed in real time on a dedicated PC.
In addition, thermal time series are updated
every 5 minutes on http://hotspot.higp.hawaii.
edu/puuoo/.The Pu’u ‘O’o system has now been
in continuous operation since March 2001,
with no significant damage or degradation
over a 24-month operational lifetime.
The success of this prototype system led to
the construction and installation of a second
permanent system on Stromboli Volcano (Italy)
and several portable units capable of campaignstyle deployment.The Stromboli system is
based on the Pu’u ‘O’o blueprint, except that
the sensor boxes are housed in a bomb-proof
bunker to protect them from ejecta impact.
The portable system consists of a 1°, 15°, or
60° FOV sensor, where sensors can easily be
switched in and out of the housing cases.This
allows FOVs to be optimally selected,depending
on distance to target and activity style.At
Stromboli,for example,an array of three 15° FOV
sensors located ~250 m from the three active
craters allowed all explosions from each of
the craters to be logged.
Portable modules can be operated in an
entirely self-contained mode with an internal
data logger fixed to the inside lid of the housing
case.A HOBO™ H8 data logger is used that is
capable of recording at a 1-Hz rate, an addition
that adds $170 to the cost of the unit.The
assembled stand-alone device uses standard
AA batteries to power the sensor, weighs 3.1
kg, and measures 28 x 25 x 18 cm.These features make it possible to cheaply respond to
equipment requests using an overnight airborne
carrier.In this way,a device can be in the field
within 2 days of a deployment request. In
addition, a Pelican case-contained logging
module has been constructed for ~$1,000 that
links to the sensor boxes via a hard-wire connection.This option allows logging of eight
data channels at a rate of 200 Hz, and is capable of running for 10 hours on internal battery
power. Connection to a solar panel extends
the operating period indefinitely.
To date,the portable system has been deployed
in campaign-style,short-duration (10–21 day- long)
experiments at Kilauea, Masaya (Nicaragua),
Erta Ale (Ethiopia),Soufrière Hills (Montserrat),
and Villarrica (Chile), as well as Santiaguito,
Fuego, and Pacaya (all in Guatemala),plus
Stromboli,Etna, and Vulcano (all in Italy).The
thermal time series measurements obtained at
each of these active volcanoes allow for precise
timing for the onset of eruptive events.
Furthermore, the style of eruptive activity may
be inferred by interpretation of distinctive
thermal wave forms.
A Library of Thermal Waveforms for
Volcanic Events
By recording the thermal evolution of an
unfolding explosive event, continuously
recorded thermal data provide information
Eos,Vol. 84, No. 40, 7 October 2003
regarding event chronologies and frequencies,
activity style,repose times,and ejection velocities,
as well as precise timing of event onsets.When
provided in real time,such information is potentially an extremely valuable hazard assessment
and monitoring tool.As Figure 1 also shows,
the instruments are capable of resolving individual vents in a spatial context.To date, the
library consists of thermal waveforms for
strombolian eruptions, lava flows, gas jetting,
crater floor collapse, persistent degassing, lava
flow in tubes, rock falls, pyroclastic flow events,
and lava lake activity (Figure 2).
(1) Strombolian eruptions (Figure 2a).
Strombolian eruptions imprint characteristic
waveforms on the thermal time series, where
thermal transients with impulsive onsets correspond to the explosive emission and passage
of gas and ballistics through the thermometer
FOV. The duration of the thermal transients
corresponds to the temporal extent of the
explosion.Time series analysis can reveal fine
detail about the explosion, including multiple
ejection pulses.
(2) Gas jet events (Figure 2b). Gas jet events
have similar waveforms to strombolian eruptions.
Indeed, they may be viewed as ejecta-free,
low-energy versions of their strombolian
counterparts. Here, the duration of the thermal
transient relates to the duration of high temperature gas jetting, and often displays a phase
of waning flow evidenced by exponentially
decaying thermal signals.A characteristic of
gas jet events recorded at Pu’u ‘O’o during
2002–2003 was their tendency to gather into
remarkable clusters of high-frequency, periodic
gas pistoning episodes.
(3) Persistent degassing. Persistent degassing
from an active vent gives a persistently high
thermal signal, in which oscillations relate to
discrete gas puffs emitted from the vent. Each
oscillation builds as the puff rises into the
FOV, and then decays as the puff leaves the
FOV. In this way, the instrument can be used
as a puff-counter, which may provide a proxy
for degassing rates.At Stromboli, degassing
appears to oscillate between periods of highand low-frequency puffing [Ripepe et al., 2002].
(4) Lava flow events (Figure 2c).As with
strombolian eruptions and gas piston events,
the arrival of the lava flow within the instrument
FOV produces an impulsive onset, although the
lower velocity of the lava flow causes the onset
to be slower than in the more explosive cases.
Indeed, the slope of the thermal onset can be
used to extract the velocity of the lava flow front.
Given the time elapsed between the flow front
entering and entirely crossing the ~4 m-wide FOV
(~1 minute), flow front velocity can be calculated; in this case, 4 m per minute.The telltale
trace of a lava flow waveform is the exponentially decaying cooling curve as the flow
surface within the FOV stagnates and cools.
(5) Tube-fed lava flow. Lava flow through
tubes can be targeted through holes or
skylights in the tube roof. Because the flow surface is insulated by the tube roof and is continuously flowing through the thermometer
FOV, the thermal time series maintains a uniformly high temperature level. Subtle oscillations may, however, record slight, periodic
Fig. 1.Two of the thermal sensors on the rim of Pu’u ‘O’o during November 2002, with (below)
single vents targeted by the two 1° FOV instruments (yellow arrows) and the area covered by the
60° FOV (red sector).
variations in lava flux through the tube and/or
emission of gas from the skylight, potentially
revealing flow surges down the tube system.
(6) Crater floor collapse. Crater floor collapse
is marked by catastrophic truncation of the
thermal signal. Effectively, the high-temperature
(vent) source targeted prior to the event falls
out of the bottom of the FOV during floor collapse. In the case of one such collapse event at
Pu’u ‘O’o during 25 August 2002, the collapse
event was timed to an accuracy of 2 seconds.
Insights into Eruptive Processes
Thermal monitoring applications continue
to increase.At Kilauea, Stromboli,Villarrica,
and Santiaguito, multiple 1° FOV instruments
have been employed, aligned along the axis
of explosive emissions, to allow tracking of
thermal pulses up the sensor array. For correlated wave forms and known separation distances between sequential thermometers,
ascent velocity of an ascending pulse can be
obtained from the delay between thermal
pulses when tracked from sensor to sensor.
This information is used to assess the time
evolution of lava tube flow velocities, lava
lake convection velocities, and gas jet velocities
during explosive events.This application provides an effective, low-bandwidth means of
determining muzzle velocity during violent
eruptions, and is somewhat more straightforward than more traditional techniques that
use high-speed film or video to obtain ejecta
travel speeds [e.g., Chouet et al., 1974]
Ground-based thermal monitoring has proven
applications for the study of a variety of volcanic
processes and,through the real-time transmission
Eos,Vol. 84, No. 40, 7 October 2003
a)
of data to nearby volcano observatories, has
potential for the assessment and mitigation of
volcanic hazards.In short,over the 3-year-long
system development and testing period described
here, the thermal sensors have proved to be a
valuable new tool in the analysis of volcanic
phenomena,and in tracking volcanic eruptions.
Data collection and monitoring campaigns in
Hawai’i, Guatemala, Nicaragua,Chile,Ethiopia,
and Italy have shown that the data stream can
be applied to monitor a wide range of volcanic
phenomena,and represents a new,inexpensive
analysis and monitoring tool.
Collaborators who wish to use our sensors
on other active volcanoes are encouraged to
contact A.Harris,the senior author of this article,
at HIGP/SOEST, University of Hawaii, 2525
Correa Road, Honolulu, Hawaii, 96822 USA;
E-mail: [email protected] full wave
form library is still under construction, but will
be available via our Web site (http://hotspot.
higp.hawaii.edu/) to support interpretation
and analysis of future data sets.
Acknowledgments
Funds for equipment construction deployment and data analysis were obtained from
the Hawai’i Institute of Geophysics and Planetology,NASA grants NAG5-9413 and NAG5-10640,
NSF grants EAR-0106349 and EAR-0207734,
The Royal Society, and NERC grant NER/B/S/
2001/00707.
b)
References
Chouet, B., N. Hamisevicz, and T. R. McGetchin,
Photoballistics of volcanic jet activity at Stromboli,
Italy, J. Geophys. Res., 79, 4961–4976, 1974.
Harris,A. J. L., M. J.Wooster, and D.A. Rothery, Monitoring Volcanic Hotspots Using Thermal Remote
Sensing,Advances in Environmental Monitoring
and Modelling,1(3),2002; http://www.kcl.ac.uk/kis/
schools/hums/geog/advemm/vol1no3.html.
Ripepe, M.,A. J. L. Harris, and R. Carniel,Thermal, seismic and infrasonic insights into conduit process at
Stromboli volcano, J.Volcanol. & Geotherm. Res.,
118, 285–297, 2002.
Author Information
c)
Fig. 2. Example waveforms from a suite of volcanic activities including: (a) two consecutive
strombolian events recorded at Stromboli, (b) a phase of periodic gas jetting (pistoning) at Pu’u
‘O’o, and (c) eruption of a sequence of two lava flows onto the floor of Pu’u ‘O’o.Waveforms are
robust, but thermal hetero-geneity within each FOV and different source-instrument distance in
each case means that absolute temperatures are not meaningful.
Andrew Harris, Jeffrey Johnson, Keith Horton,
Harold Garbeil, Hans Ramm, Eric Pilger, Luke Flynn,
and Peter Mouginis-Mark, HIGP/SOEST, University of
Hawai’i, Honolulu; Dawn Pirie, Steve Donegan, and
Dave Rothery,The Open University, Milton Keynes,
U.K.; and Maurizio Ripepe and Emanuele Marchetti,
Università di Firenze, Italy