R S A V: Emote Ensing of Ctive Olcanoes

Annu. Rev. Earth Planet. Sci. 2000.
28:81106
Copyright q 2000 by Annual Reviews. All rights reserved
REMOTE SENSING OF ACTIVE VOLCANOES

Peter Francis and David Rothery
Earth Sciences, Open University, Milton Keynes MK7 6AA UK, e-mail:
D.A.Rothery@open.ac.uk
deceased
Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

by Universita degli Studi di Napoli Frederico II on 06/19/13. For personal use only.
Key Words volcanoes, satellite remote sensing, thermal anomalies, eruption

plumes, radar interferometry, natural hazards
Abstract The synoptic coverage offered by satellites provides unparalleled
opportunities for monitoring active volcanoes, and opens new avenues of scientific
inquiry. Thermal infrared radiation can be used to monitor levels of activity, which is
useful for automated eruption detection and for studying the emplacement of lava
flows. Satellite radars can observe volcanoes through clouds or at night, and provide
high-resolution topographic data. In favorable conditions, radar inteferometery can
be used to measure ground deformation associated with eruptive activity on a centimetric scale. Clouds from explosive eruptions present a pressing hazard to aviation;
therefore, techniques are being developed to assess eruption cloud height and to
discriminate between ash and meterological clouds. The multitude of sensors to be
launched on future generations of space platforms promises to greatly enhance
volcanological studies, but a satellite dedicated to volcanology is needed to meet
requirements of aviation safety and volcano monitoring.
INTRODUCTION
In 1972, the United States sent the Earth Resources Technology Satellite (ERTS)
into orbit. Later renamed Landsat 1, this was the first of a series of polar orbiting
satellites to provide routine, high-resolution, multispectral images of the Earths
surface for civilian applications. Its launch could be said to initiate the modern
era of remote sensing of active volcanoes. A further major step was the October
1974 launch of the first Geostationary Operational Environmental Satellite
(GOES), which provided continuous monitoring of North American weather.
Since then, a plethora of polar orbiting and geostationary satellites has been
launched, carrying sensors working in all parts of the electromagnetic spectrum
from ultraviolet to microwave. Some of these satellites were spectacularly shortlived, whereas others worked for many years; however, none had volcanology as
a prime objective. Meteorological observations were the dominant routine application, followed by agricultural and related environmental issues and resource
exploration. Thus, none of the sensors employed was designed or intended for
volcanological applications. Despite this, data from remote sensing satellites have
00846597/00/05150081$14.00
81

82
FRANCIS n ROTHERY
been successfully applied to a wide range of volcanological problems, furnishing

good examples of the unexpected spin-offs that characterize the history of science.
Launch of the Earth Observing System EOS AM-1 platformnow named
Terrain late 1999 will mark a new era of Earth remote sensing (Mouginis-Mark
et al 1991). Part of NASAs broader Mission to Planet Earth, this ambitious project
is mostly concerned with exploring global climate change and environmental
issues. From its initiation, however, volcanological observations have been
included among the wide range of mission objectives.
Here, we review some of the significant advances achieved in volcanology by
the generation of satellites that led up to the EOS era. Earlier reviews of volcano
remote sensing (e.g. Francis 1989, Rothery et al 1995, Francis et al 1996) tended
to focus on individual sensors, wavelength regions, or techniques. Given the novelty of the observations, it was natural for such publications to celebrate the
acquisition of unprecedented types of data and thus to focus on the methodology
of obtaining results, rather than on underlying science issues. In this review,
therefore, we focus on science rather than sensors or techniques, and attempt to
evaluate the contribution of remotely sensed data to addressing broader volcanological questions. Satellite remote sensing can yield an improved understanding
of volcanic processes simply by providing an enormously enhanced monitoring
capability. It can also provide qualitatively different kinds of scientific insights
from anything previously possible: literally, new ways of looking at volcanoes.
We try to distinguish between these two perspectives. Rather than reviewing the
literature exhaustively, we highlight some key papers. We concentrate on satellite
remote sensing that can be used routinely, rather than aircraft campaigns, which
are usually experimental and brief. We do not comment on many important studies
that have simply proved the use of new technologies or achieved technique development, but we conclude by looking forward briefly to volcanological objectives
that may be gained in the future.
RADIANT THERMAL ENERGY BUDGETS OF

VOLCANOES AND LAVA FLOWS
Physical Background
The Earths atmosphere is highly transparent to infrared radiation in windows
at 35 lm and 814 lm. These lie in what is conventionally referred to as the
thermal infrared part of the spectrum because it is at these wavelengths that matter
radiates most strongly at normal temperatures. Using systems sensitive to these
wavelengths, it is commonplace to detect differences in sea surface temperatures
of less than 0.18C in satellite images (e.g. Njoku 1990). This precision is possible
because the sea surface is a locally homogenous emitter of thermal infrared radiation. By contrast, active volcanoes represent massive heterogeneous anomalies
at much higher temperatures. They are easily detectable in the thermal infrared

VOLCANO REMOTE SENSING
83
even when their dimensions are less than the spatial resolution of the imaging
system, and they frequently saturate sensors originally intended for much lower
temperature applications. Volcanoes also emit detectable thermal radiation in the
short wavelength infrared or even the visible parts of the spectrum. Lava extrusions at temperatures of ;10008Cwhich may be visibly radiant even in daylightare the most obvious volcanic thermal phenomena, but there are many
others. These include fumaroles emitting steam and other gases that may heat the
ground to several hundred degrees; pyroclastic flow deposits consisting of hot,
solid, fragmented material; columns of ash and gas blasted into the atmosphere
by explosive eruptions, or rising convectively; and lakes of hot water ponded in
the craters of active volcanoes.
Volcanic thermal anomalies were noted on images shortly after routine weather
satellite observations began. Early examples were the 1978 and 1981 eruptions
of Etna detected by the NOAA 5 satellite (Archambault et al 1980, Bonneville et
al 1985). On 19 November 1979, Landsat 3 obtained the first high-resolution,
short-wave infrared images of an active lava flow on Sierra Negra volcano in the
Galapagos (Rothery et al 1988, McClelland et al 1989).
These early observations were necessarily qualitative. They benefited from the
hitherto unprecedented capability of satellites to provide synoptic coverage of the
Earth, and thus to obtain images of the most remote volcanoes as easily as the
most accessible. Extraction of meaningful quantitative thermal data is complicated by problems of scale. Large lava flows may have widths of many tens of
meters and extend many kilometers in length, but their thermal structure is invariably complex, typically consisting of small, incandescent areas of newly exposed
high temperature material, surrounded by much larger areas of cooler, chilled
material. This means that the hot components of volcanic surfaces imaged by
satellite sensors are almost invariably smaller than the pixel (picture element) size
of the sensor. This problem is obvious with meteorological satellites, which typically have pixel sizes of 14 km, but it is also serious with the ;30 m pixels of
Landsat and similar satellites used for detailed investigations. Because the radiant
thermal energy detected by a satellite sensor is integrated over the entire area of
each pixel, estimation of the temperatures of hot volcanic surfaces requires simultaneous determination of the relative surface areas covered by components at
different temperatures.
For remotely sensed data at a single wavelength, no solution is possible. For
multispectral data (i.e. data recorded at more than one wavelength), however,
Dozier (1981) and Matson & Dozier (1981) proposed an elegant solution based
on the Planck distribution law, which relates the spectral radiance (L) at wavelength k to its thermodynamic temperature T:
L(k,T ) 4 ekc1k15/[exp(c2/kT) 1 1]
(1)
where ek is the spectral emissivity and c1 and c2 are constants. Working with forest
fires, and assuming there are two components at different temperaturesone
hot and the other backgroundMatson and Dozier recognized that the radi-
FRANCIS n ROTHERY
84
ance measured by the satellite sensor at two different wavelengths could be used
to estimate both the temperature of each component and the fractional area of
each pixel covered by each component. For any one wavelength k (or channel of
a sensor), the total radiance detected by the sensor is the weighted average of the
radiance from each component:
Rk(tot) 4 s[ f (R hot) ` 1 1 f )R back]
(2)

where R hot is the radiance of the hot component at temperature T (K) derived
from Plancks equation; Rback is the radiance of the cooler background component,
also specified by Plancks equation; f is the fractional area of the surface covered
by the hot component; and s is the atmospheric transmittance function for the
wavelength concerned. Where radiance data at two infrared spectral channels with
central wavelengths kx and ky are available, Equation 2 can be written:
RX 4 sX[ f L (kX, Thot) ` (1 1 f )L(kX, Tback)]
(3)
Ry 4 sy[ f L(ky, Thot) ` (1 1 f )L(ky, Tback)]
(4)
and
Because Rx and Ry are recorded by the satellite, two out of the three remaining
parameters f, Thot and Tback, can be determined if the third can be estimated or
assumed, and if sx and sy can be constrained from standard atmospheric radiative
transfer models.
Equations 3 and 4 form the basis of the dual band method, which was first
applied to volcanoes by Rothery et al (1988), after they noted a persistent hot
spot in a summit crater of Lascar volcano (Figure 1, color insert), north Chile
(Francis & Rothery 1987). Subsequently, the same technique has been refined
and applied to volcanoes around the world by many authors, in a continuing string
of papers usefully reviewed by Flynn et al (2000). Rothery et al (1988) assumed
a value for the low-temperature background and used the technique to estimate
the temperature of the high-temperature material; they thus demonstrated that rock
at near magmatic temperatures had been present at the surface 21 months before
Lascar erupted in September 1986, following a prolonged period of repose. In a
significant refinement of the method, Oppenheimer (1991) pointed out that
because the core temperatures of typical basaltic lava flows vary little (they are
typically around 11008C), improved estimates of crust temperatures and their
fractional areas could be obtained by assuming values for core rather than background temperatures. Oppenheimer (1993) also showed that improved estimates
of radiant emittance could be obtained by using data in three rather than two
simultaneously acquired bandsyielding estimates for the core, crust, and cool
background to a lava flow. This has obvious implications for providing input to
models of lava flow behavior, which can be applied either retrospectively (to seek
to understand how a flow has behaved) or used in a real-time monitoring situation
to predict the motion of a flow. Other refinements include the work of Harris et
al (1998), who used hyperspectral data collected in the field to show that a high-

85
temperature lava surface is itself composed of at least two thermal sources, and
incorporated this into their remote sensing models.
We note that developments in the acquisition and interpretation of remotely
sensed thermal data for volcanoes on Earth have been paralleled by similar
advances in the study of the active volcanoes on Jupiters satellite Io (e.g. Davies
1996, McEwen et al 1998). Although the styles of volcanism on Io are very
different from those on Earth, the data sets used are also obtained by remote
sensing (from the Hubble Space telescope and the Galileo orbiter), and identical
problems of pixel size arise.
Shortly after discovering that volcanic thermal anomalies could be detected
on Landsat images, Francis et al (1989) argued that changes in radiated emitted
power with time might provide a valuable means of monitoring the level of activity of volcanoes. This is desirable, because active vents are generally not accessible to conventional methods, even when the volcanoes themselves are
accessible. In a first attempt to explore this approach, Glaze et al (1989) used
Landsat Thematic Mapper data to estimate the radiant thermal energy flux of
three active volcanoes: ErtaAle (Ethiopia), Erebus (Antarctica), and Lascar
(North Chile). Their approach yielded underestimates, since they did not take into
account factors such as the power radiated by the warm background. Radiation
is of course only one of several mechanisms by which volcanoes lose heat. Other
important factors are the convective and conductive fluxes; rainfall (which can
be modeled); and cooling by hydrothermal circulation, which is much more difficult to constrain. Subsequently, successively more sophisticated attempts to use
remote sensing to measure volcanic thermal budgets have been made, leading to
some richer insights into volcanic processes.
Thermal Energy Budgets of Volcanoes: Implications for

How Volcanoes Grow
Volcanoes such as Stromboli (Italy) are puzzling to volcanologists because they
have been prodigious sources of both gases and thermal energy throughout historic times, yet they have erupted negligible quantities of lava at the surface.
Some, notably Kilauea in Hawaii, have sustained lava lakes in their summit craters
for decades and perhaps centuriesthe historic records are often too inadequate
to be certain. Clearly, effective convective links must exist between the active
surface vents and subterranean magma reservoirsbut how large are these, and
what is taking place within the reservoirs? Washington first posed this question
about Stromboli as early as 1917.
Using satellite-derived estimates of radiant thermal fluxes from volcanoes in
tandem with emission rates of sulfur dioxide measured by conventional groundbased techniques, Francis et al (1993) argued that the fluxes of 5 2 103 W m12,
which can be inferred for the Halemaumau lava lake during much of the nineteenth century, imply that a flux of magma into the volcano of ;0.6 3.0 2 103
kg s11 was required to sustain the surface phenomena. This rate is broadly con-

86
FRANCIS n ROTHERY
sistent with the rate at which lava has been erupted at the surface since 1924:
;104 kg s11. During the nineteenth century, when the lava lake existed, lava was
erupted as flows outside the confined lake at an average rate of only ;3 2
102 kg s11. Francis et al (1993) thus concluded that Kilauea and similar persistently active volcanoes must grow endogenously, either by intrusion of dykes or
by cumulate formation deep within the volcano. They also noted that efficient
convective circulation can be driven by density differences between gas-rich
magma and magma that has degassed via a persistently active vent. Thus, the
resulting density increase could inhibit eruption of lavas and promote cumulate
formation at depth.
Petrologic and tectonic studies of Kilauea have demonstrated that it is indeed
underlain by large volumes of cumulate olivine, which may be contributing to
the lateral spreading of the shield volcano by plastic deformation (Clague et al
1994). Volcanologists need to take into account the rapid rates of endogenous
growth implied by thermal budgets when considering the long-term evolution of
volcanic edifices. These may be spreading sideways or sagging downward by
lithospheric loading, as well as growing more conventionally upward by the accumulation of surface lava flows.
Several papers subsequently pursued this approach, demonstrating how basic
measurements done by remote sensing can stimulate new ways of looking at
volcanoes, or lend support to developing models. Harris & Stevenson (1997a,b)
quantified the thermal budgets of Vulcano and Stromboli, and Stevenson & Blake
(1998) built on this, using theoretical and laboratory modeling to investigate the
role of convective overturn of degassed magma in sustaining persistent volcanism.
Harris et al (1999) used three bands of Landsat TM data to constrain the radiated
heat flux and from this derived estimates of the convective and conductive losses
in order to find the total heat budget. They concluded that the persistence of lava
lakes over decades or even centuries may not necessarily imply endogenous or
cryptic growth, but that lakes could be sustained by magma recycling from a
sufficiently large, deep reservoirif it were hotter than the temperature at which
crystallization would commence in the magma chamber. Prolonged activity (with
or without crystallization) must be accounted for by circulation of magma
between the surface conduit and a deep reservoir, driven either by thermal convection or by degassing. In either case, the rate at which magma is required to
solidify permanently is much less than the rate at which magma is circulating
within the conduit. Along similar lines, Delmelle et al (in press) cited the remotely
sensed heat flux of the Masaya volcano (Nicaragua)which has an open, persistently degassing ventas evidence that the volumetric rate of magma degassing is several orders of magnitude greater than the rate at which magma is
solidifying in the conduit system. They concluded that gas-rich magma continually arrives at a shallow, vesiculated layer where it degasses and then sinks (with
minimal cooling), to be replaced by the next influx of gas-rich magma arriving
from depth.
87

THERMAL REMOTE SENSING OF LAVA FLOWS

Understanding the dynamics of lava flow emplacement is of more than academic
interest. Basaltic flows can travel quickly, reaching speeds in excess of 10 km
hr11. This means that they are capable of rapidly engulfing towns and villages
in their path, and although it is rare for people to be killed, they can lose their
homes and livelihoods. Therefore, it is important to develop the capability to
define areas on the flanks of a volcano that are at greatest risk from lava flow
inundation, and how soon they may be affected.
TV documentaries often show dramatic images of volcanologists poking temperature sensors into active lava flows. Although useful in the early days of volcanology (and still good for TV), this approach is of limited value for two
fundamentally different reasons. First, it is physically difficult to obtain meaningful data in a hazardous and rapidly evolving situation. Second, point source
measurements are of dubious value in the absence of a model for the thermomechanical evolution of lava flows. A single lava flow may have an area of many
square kilometers, with a complex distribution of temperature: horizontally, vertically, and through time.
The synoptic view provided by air- and space-borne instruments, coupled with
the multispectral capability of modern infrared sensors, ensures that remote sensing has much to contribute to studies of lava flow emplacement (Figure 2, color
insert). Rothery & Pieri (1993) showed that remote sensing makes it possible to
observe the whole of a lava field simultaneously, and to monitor the evolution of
the field through time in a consistent manner. In practice, spatial and temporal
resolution tends to be incompatible: the higher the spatial resolution of a sensor,
the less often it is likely to view the target. However, infrequent high-resolution
images can provide information that is complementary to that obtained from more
frequent low-resolution coverage and vice versa. Because they are such powerful
sources of thermal energy, lava flows are easily detected even by low spatial
resolution infrared detectors, such as those on meteorological satellites. The eruption of Surtsey, off Iceland, in 1966 was probably the first to be detected by
satellite sensors (Williams & Friedman 1970, Friedman & Williams 1970) and
there have been many subsequent reports of thermal observation of eruptions from
space. Rothery et al (1988) made the first high spatial resolution observations of
lava flows from space on a 1979 Landsat Multispectral Scanner image of Sierra
Negra, Galapagos, but detailed analyses were not made until Pieri et al (1990)
studied a Landsat Thematic Mapper image of a June 1984 flow from the southeast
crater of Etna, Italy. Using an early form of the dual-band model, they concluded
that only 10 to 20% of the total energy was radiated from hot zones or fractures,
which formed less than 1% of the surface area of the flow. They also concluded
that boundary layer and thickness effects dominated over radiant zones in cooling
the flow.
Thermal evidence for passage of basaltic lava through tubes was conclusively
demonstrated by Realmuto et al (1992), using radiance measurements from an

88
FRANCIS n ROTHERY
aircraft deployment of the NASA Thermal Infrared Multispectral Scanner (TIMS)

to Kilauea volcano, Hawaii in 1988. They found that molten lava in a tube in the
active flow field resulted in surface temperatures that were at least 108C above
ambient (when integrated across a whole pixel), and used this temperature contrast
to create maps of concealed tube systems over distances of many kilometers.
Tubes were several meters deep in the upper portion of the flow field, but only
13 m deep in the lower parts.
The consistency of surface temperatures above the tubes provided a convincing
demonstration of how lava flowing in tubes can retain its heat, and thus travel
very long distances. This is an essential requirement in the construction of the
very large, low-relief basaltic shield volcanoes and plateaux found on the Earth
and other planets (cf Greeley & Spudis, 1981). Immense basalt provinces such
as those of the Deccan and Columbia River have long been regarded as the
accumulations of flood basalts, erupted at prodigious effusion rates. Improvements in the understanding of basalt lava field formation have led to the suggestion that they may, by contrast, be formed relatively slowly, by tube-fed systems
(Self et al 1996).
On the basis of field observations, Rowland & Walker (1990) demonstrated
that for Hawaiian basaltic lava flows, effusion rates of .20 m3 s11 produce
rapidly advancing channelized aa flows (characterized by rubbly, irregular surfaces), whereas lower effusion rates (;5 m3 s11) produce slow-moving, tube-fed
pahoehoe flows (characterized by smooth, lobate surfaces). Given the practical
importance of being able to forecast the areas that may be overrun by lavas,
estimates of effusion rate may provide a useful tool for hazard assessment. Making
the assumption (suggested by certain theoretical models) that the instantaneous
radiant thermal energy detected by a satellite is related to the longer-term effusion
rate of a flow, Harris et al (1997a, 1998) used data from Landsat TM and AVHRR
to estimate volumetric effusion rates for Etna and Kilauea. They found a good
correlation between their remotely sensed estimates of effusion rate and estimates
made on the ground. For example, for Kilauea, using TM data for 23 July and
11 October 1991, they estimated effusion rates of 1.8 5 0.5 m3 s11 and 0.8 5
0.3 m3 s11 respectively, consistent with rates of ;1.4 and 0.9 m3 s11 respectively,
measured on the ground. Using lower resolution meteorological satellite data for
26 additional dates between those of the TM acquisitions, they obtained broadly
similar results. This is significant, because low-resolution data are always going
to be available on a more frequent basis than high-resolution data, so it is important to have confidence that useful information can be extracted from the lower
resolution images. In another paper, Harris et al (1997b) published graphs of
effusion rate against time derived from AVHRR images for several eruptions that
had not been closely studied on the ground.
In terms of hazard mitigation, this work represents a significant step forward,
although debate continues as to why remote thermal determination of effusion
rates appears to work so well. For the first time, volcanologists can make quantitative estimates of the magnitude and rates of lava eruptions, and, coupled with
89

local topographic data, assessments of areas likely to be affected. To date, however, only retrospective analyses of existing data have been carried out. To validate the effective operational use of the technique, a demonstration experiment
is required, with real-time data being used to predict the future evolution of a
flow field.
Effusion rate monitoring by thermal remote sensing is not limited solely to
lava flowsit can also be valuable for measuring effusion rates of lava domes.
This is important because pyroclastic flows triggered by dome collapse are much
more deadly than lava flows, and such flows are likely to be especially frequent
when a dome is expanding as a result of injection of lava. In a study of the growth
and collapse of the dacite lava dome of Mt. Unzen (Japan), Wooster & Kaneko
(1998) showed that the 1.6 lm spectral radiance as determined from nighttime
ATSR data correlates with the rate of degassing, which in turn is closely related
to the lava effusion rate into the dome.
Eruption Detection and Monitoring

Apart from volcanoes, only a limited range of hot spots is readily detectable by
satellite. Natural hot spots are almost exclusively forest fires, whereas industrial
sources are mostly gas flares and large smelters. Because fires tend to be short
lived and industrial sources tend to be located in fixed sites, satellite detection of
new volcanic events is practical on a worldwide basis. This detection is clearly
most useful for remote volcanoes, such as those of the Aleutian, Kurile, or Indonesian archipelagos, where eruptions might otherwise go entirely undetected;
however, it is also a valuable adjunct for monitoring volcanoes that have wellestablished volcano observatories. Care must be taken in using satellite data to
distinguish between genuine volcanic hot spots and, for example, forest fires on
or near the volcano. This can usually be done by using an appropriate geographic
mask and thresholding filters.
Eruption monitoring has developed rapidly. In a seminal paper, Harris et al
(1997b) demonstrated the use of low-cost meteorological satellite data (AVHRR)
in monitoring Etna, Krafla, Cerro Negro, Fogo, Lascar, and Erebus. As part of
the lead-up to the EOS generation of space platforms, workers at the University
of Hawaii have used data from GOES geostationary weather satellites (GOESEast at 758W and GOES-West at 1358W) to provide continuous, real-time access
to thermal images of ten selected volcanic sites via the World Wide Web (http://
virtual1.pgd.hawaii.edu/goes). GOES satellites are part of an equatorial girdle of
geostationary weather satellites at an altitude of 35,770 km, from which vantage
point the field of view covers the entire visible disc of the Earth (not quite a full
hemisphere). Images can be recorded every 1530 minutes. Unlike their fellows
at other longitudes (e.g. Meteosat, at 08W), the GOES satellites include a channel
at 3.7 lm (4 km pixel size), which is sensitive to magmatic temperatures. On the
hotspot Web site, images are displayed in a variety of band combinations selected
to emphasize thermal anomalies over background, together with radiance mea-

90
FRANCIS n ROTHERY
surements and cloud cover assessment updated every 1530 minutes. By selecting
a threshold value for the difference in temperature between different channels,
alarm algorithms have been developed to detect new hotspots automatically and
send e-mail alert messages to volcanologists.
On 15 September 1998, an eruption of the volcano Cerro Azul on Isla Isabela
in the Galapagos islands was successfully detected in this way, four hours before
it was seen by ground-based observers. This was the first demonstration of automated eruption detection, and it highlights the future potential of automated systems (Harris et al, in press). Use of the Internet also means that for the first time,
volcano observatories around the world will have access to free, real-time data
for volcanoes that may be inaccessible for physical or financial reasons.
A similar approach has been adopted by the Alaska Volcano Observatory,
which is responsible for monitoring one of the worlds largest volcanic provinces.
Extending over 2500 km, this chain includes about a hundred potentially active
volcanoes. More than 350 eruptions have been recorded from 45 different volcanoes over the last 200 years (McClelland et al 1989). Although the area is thinly
populated, traffic in several major air corridors is at risk from eruptions. The
region is too far north to be well seen from a geostationary orbit; however, the
advantage is that the ground tracks of polar orbiting satellites are so closely spaced
that there are about twenty AVHRR imaging opportunities each day, which provides ample coverage. AVO has developed automated monitoring algorithms
(Dean et al 1998) that successfully detected an eruption of Okmok volcano, well
before local observers confirmed the effusion of new lava.
Ideally, remote monitoring of volcanoes requires long time series of data, so
that patterns of activity can be discerned and used for forecasting. Lascar volcano
in north Chile was the first thermally radiant volcano to be identified (Francis &
Rothery 1987) so it has the longest data coverage. Wooster & Rothery (1997)
used the Along Track Scanning Radiometer (ATSR) aboard the ERS-1 spacecraft
to monitor the volcano for the period 199295, extending the coverage and
improving on the temporal resolution of the previous TM time series (Oppenheimer et al 1993). Designed for routine mapping of sea surface temperatures,
ATSR is broadly similar to the AVHRR in that it has roughly 1 km pixel size,
but it is better suited to monitoring radiance from high-temperature volcanic
sources.
Fortuitously, Wooster & Rotherys time series of data spanned the largest eruption of Lascar for ;9,000 years. They showed that variations in radiance from
Lascar were consistent with an independent model (Matthews et al 1997) for the
growth and subsidence of a lava dome within the summit crater. They suggested
that large explosive eruptions are more likely when the short wavelength infrared
signal has fallen from a high to a low level, because this signifies that the lava
has collapsed back into the conduit, hindering passive degassing. In practice, this
conclusion may be difficult to use in forecasting, because decreases in thermal
radiance also precede decreases in the rate of supply of gas or magma from depth.
91

STUDIES OF VOLCANO TOPOGRAPHY AND

DEFORMATION BY SYNTHETIC APERTURE RADAR
Topographic data are almost the first requirement for any volcano study. On the
one hand, information is needed about the fixed, large-scale shape of the volcano
in order to forecast where lava flows may go. On the other hand, small changes
in topography are common precursors to eruptions. For example, prior to its
catastrophic eruption of May 18, 1980, the north flank of Mt. St. Helens was
swelling by up to one meter per day (Lipman & Mullineaux 1981). Although
precise topographic data were available for Mt. St. Helens, many of the worlds
volcanoes lack basic topographic maps.
Because radars work over narrow, operator-defined frequency ranges (within
the microwave region of the spectrum), returned signals contain information about
travel time, range, and phase. This information can be used not only to create
images, but also to derive precise topographic data and information about surface
properties such as roughness and dielectric constant. Radar has several major
advantages over remote sensing systems working at shorter wavelengths. Radar
will penetrate clouds, and moreover, because it is an active rather than passive
technique, the target can be imaged at any time of day or night. As a result,
volcanoes that are normally cloud-covered, or that are located at high latitudes
so that they experience long winter darkness, can be monitored year round.
Although preceded by a series of classified military missions, routine civilian
observations of large areas of the Earths land surface with radar sensors did not
become a reality until the launch of the first European Remote Sensing satellite
ERS-1 in July 1991. There has also been a series of civilian airborne and shortduration Space Shuttle experiments, so the potential of radar for studying volcanoes has been thoroughly investigated (e.g. Mouginis-Mark 1995, Zebker et al
1996). Orbital radars have already provided the first images of many of the
worlds most remote volcanoes, from the Aleutians to the sub-antarctic (Garvin
et al 1999).
Radar is by no means a panacea for volcano remote sensing, however. Because
it is an active technique, it is power-hungry and generates much excess heat
aboard the spacecraft. Also, though synthetic aperture radar (SAR) systems provide spatial resolutions equivalent to those of antennae many meters across, they
are still bulky. Furthermore, though radar images can be relatively easily interpreted as quasi-photographic descriptions of topography, retrieval of quantitative
data requires sophisticated, computationally intensive data processing and an
understanding of the effects of atmospheric water vapor on microwave
transmission.
Radar Experiments
To investigate the potential of SAR; techniques for volcano studies, the aircraftborne Topographic Synthetic Aperture Radar (TOPSAR) system was deployed
to Hekla (Iceland), Vesuvius (Italy) and the Galapagos islands (Evans et al 1992,
FRANCIS n ROTHERY

92
Mouginis-Mark & Garbeil 1993, Rowland 1996). The aircraft carried a pair of
C-band (5.6 cm) antennae, which provided height measurements to a vertical
accuracy of 15 m and a spatial resolution of about 10 m. In July 1991, TOPSAR
data collected over Hekla were used to create a Digital Elevation Model (DEM),
which demonstrated that small-scale topography and local slopes could be derived
to high precision. Pressure ridges less than 2 m high were detectable. Such
detailed topography of lava flows is difficult to obtain by conventional means,
even for easily accessible volcanoes. A similar TOPSAR study of Fernandina
volcano, Galapagos islands, provided a topographic database for the entire island
volcano, which far excels the best previously available photogrammetric
topography.
Many experiments were also carried out with the Space Shuttle, notably the
Shuttle Imaging Radar (SIR-A, -B and -C) and the 1999 Shuttle Radar Topography Missions (SRTM). Gaddis (1992) investigated the effects of surface texture
and roughness on the intensity and polarization of backscattered radar signals,
and showed that differences could be used to discriminate features such as aa
and pahoehoe lava flows: aa and blocky lava flows are radar bright, whereas
smoother pahoehoe flows are radar dark. During the 1994 SIR-C missions, experiments were conducted to investigate the optimal combinations of radar wavelength, pulse bandwidth, polarization, look direction, and incidence angle for a
variety of scientific applications. Studies of the Virunga volcanic chain (Central
Africa) by MacKay & Mouginis-Mark (1997) provided valuable insights into the
potential applications of radar remote sensing to volcanoes in general. Data at 3.1
cm (X band), 5.7 cm (C band), and 24 cm (L band) were collected during the
experiment. As an example of the complexity of radar data sets, Table 1 summarizes the broad range of data options available during the Virunga experiment.
Although complex, multi-parameter radar experiments such as SIR-C and
SRTM are instructive, they are not suitable for routine observations: Space Shuttle
missions are costly, generally last only about ten days, and do not provide global
TABLE 1 Radar parameters used during the 1994 SIR-C experiment over the Virunga
volcanoes. Data from MacKay & Mouginis-Mark (1997). SRL-1: Shuttle Radar Lab-1, April
1994; SRL-2: Shuttle Radar Lab-2, October 1994.
Data
take
Mission
58.60
58.61
154.90
170.90
170.10
58.61
SRL-1
SRL-2
SRL-2
SRL-2
SRL-1
X-SAR
Polarization
Bandwidth
(MHz)
Incidence
angle
Look
direction
Swath
width
(km)
Range
resolution
(m)
HH,HV
HH,HV
VV
VV
HH
X-VV
20
10
40
40
20
10
36.7 42.78
37.7 48.18
20.030.88
20.030.88
34.5 46.48
41.6 47.18
Left
Left
Right
Right
Right
Left
36.5
71.4
46.4
46.4
71.6
60.2
12.511.1
24.520.2
11.07.3
11.07.3
13.210.4
23.821.6
93
coverage. Apart from their contribution to the design of later Earth-orbiting

radars, perhaps the most important scientific outcome of experiments with airand space-borne imaging radars has been their contribution to the technical background needed to interpret volcanic phenomena on Venus, a planet for which
radar provides virtually the only information about surface features. Head et al
(1992) provides an excellent introduction to the volcanology of Venus as deduced
from Magellan Venus orbiter radar data.

Earth Orbital Radars

The first radar satellite (ERS-1) was joined by ERS-2 in 1995. These satellites
were placed in a sun-synchronous orbit at an altitude of 780 km, and carried a C
band (5.6 cm) radar with a spatial resolution of 30 m. Rowland et al (1993) used
ERS-1 images collected between October 1991 and February 1993 to study the
volcanoes of the Alaska peninsula and Aleutian island chain. This was a valuable
contribution, because remoteness and physical difficulties of access have limited
conventional ground-based field studies in this area; also, persistent cloud cover
and long periods of darkness during the arctic winter hamper remote sensing
surveys at short wavelengths.
Rowland et al (1993) used the radar images to map pyroclastic flow deposits
covering some 140 km2 around the Aniakchak caldera. They also showed that
the new lava and debris flows from Westdahl and Mt. Spurr volcanoes (which
both erupted during the period of observation) could be identified and mapped
by means of their high radar backscatter and contrast with the background. They
concluded that radar satellites offer an important method for monitoring volcanoes
such as those of the Aleutians, but noted that the 238 incidence angle at which
the ERS-1 radar beam illuminates the topography causes strong foreshortening
and layover in areas of steep terrain, which complicates interpretations. Furthermore, in the case of ERS satellites, there may be delays of up to 35 days between
successive imaging opportunities, which limits the utility of the method for routine monitoring.
Differential Interferometric Synthetic Aperture Radar

(DINSAR)
Measurements of the displacement field of the 1992 magnitude 7.3 Landers earthquake (California) provided the classic example of the potential of satellite SAR
interferometry to detect ground deformation of the scale that may precede a volcanic eruption (Massonet et al 1993, 1994, Zebker et al 1994). In this case, SAR
images obtained from the ERS-1 satellite before and after the earthquake were
used in conjunction with existing topographic data to construct an interferogram
revealing the displacement of the ground surface. With a pixel size of about
30 m, this yields a far denser spatial sampling than any feasible ground-based
surveying system. Massonets work was a powerful demonstration of the potential

94
FRANCIS n ROTHERY
of radar interferograms to monitor inflation, deflation, and rifting of active

volcanoes.
Each of the SAR images used to construct an interferogram is a map of the
magnitude and phase of the reflected signal. The phase of the radar return from
each pixel is a function of both the range and the phase shift due to reflection of
the wave from the ground surface. If the dielectric properties of the surface (such
as wetness) remain unchanged between the two images, the phase shift can be
eliminated. The remaining path difference is influenced by several factors: (a) the
relative orbital positions of the satellite when each image was acquired; (b) parallax differences when topographic features are observed from slightly different
orbital passes; (c) any changes in position of points on the ground surface between
the acquisition times of the two images; (d) system noise; and (e) atmospheric
noise.
A radar interferogram displays fringes, comparable to Newtons rings in optical
interference, resulting from reinforcement and subtraction of waves. Each successive fringe represents a path difference of half a wavelength. In practice, path
differences due to orbital parameters are minimized by choice of optimal orbits,
and by reducing the number of interference fringes to a minimum at points distant
from areas suspected to be deforming, assuming these show no real displacement.
The path difference arising from the parallax induced by the stereoscopic geometry is eliminated by using a pre-existing standard digital elevation model (which
can itself be constructed from stereoscopic radar images). In areas where a significant change has occurred in the surface characteristics between image acquisitionsfor example because of vegetation, snow, or actual disruption of the
surface such as happens when a volcanic dome collapsesthe signal is incoherent
and fringes cannot be obtained. Where fringes can be defined, however, the path
differences arising from deformation can then be mapped, resulting in a contour
map of the change in range. Each fringe corresponds to one cycle, equivalent to
28 mm (half of the 5.6 cm wavelength of the ERS-1 C-band radar). In the case
of the Landers earthquake, range changes derived from the SAR were comparable
to those calculated from observations of ground surveys of horizontal displacement resulting from the earthquake, with a root mean squared difference of ;34
mm.
Massonet et al (1995) appeared spectacularly to demonstrate the same potential
in a study of Mt. Etna between May 1992 and October 1993. They mapped a
volcano-wide deflation of 21 mm per month, which they attributed to eruption of
;3 2 108 m3 of lava during the 199193 eruption of Etna. However, this deflation did not tally with ground-based surveys. Work by Delacourt et al (1998)
showed that because of the omission of corrections for tropospheric water vapor
absorptions, part of the deformation signal seen by Massonet et al (1995) was
simply an effect of the altitude of the volcano, which resulted in an overestimate
of the subsidence rate by a factor of two. Lanari et al (1998) provided a longer
time-series analysis, which also showed post-eruptive reinflation, and documented
the changing depth of the modeled source of deformation that is presumed to

95
correspond to the magma chamber. Now that the means to correct for tropospheric
effects are in place, radar interferometry on a volcano-wide scale is set to become
a routine operational technique for volcano monitoring (Figure 3, see color insert).
Several studies have already capitalized on this remarkable new approach. For
example, Sigmundsson et al (1997) observed subsidence of ;24 mm per year
over the shallow magma chamber at Krafla, Iceland, superimposed on a ;7 mm
per year subsidence along the axis of the spreading segment. They attributed this
to cooling contraction at a depth of about 3 km, coupled with ductile flow of
material away from the spreading axis. Lu et al (1998) documented 140 cm of
subsidence of the caldera floor of Okmok volcano (Alaska) as a result of its
FebruaryApril 1997 eruption.
Although satellite DINSAR holds the promise of exceptionally detailed topographic and deformation studies of volcanoes, it is not as widely applicable as
first hoped. Observations of the ongoing eruption of the Soufriere Hills volcano,
Montserrat, show that it can be difficult to obtain suitable coherent radar images
for many reasons. Rapid variations in the vegetation canopy, differences in surface
dielectric constant due to variations in moisture, steep, irregular, and rapidly
changing topography on the growing and collapsing andesitic lava dome, and
variations in atmospheric transmission caused by variations in water vapour all
conspire to hinder the retrieval of topography. At Montserrat, useful topographic
data were obtained from fans of relatively flat, smooth, unvegetated pyroclastic
flow deposits surrounding the dome, but little coherent data could be derived
from the active dome itself (Wadge et al 2000).
ERUPTION CLOUDS
Explosive eruptions inject clouds containing huge quantities of ash, gas and water
vapour into the troposphere and stratosphere (Figure 4, see color insert). Sparks
et al (1997) provides a useful general volcanological overview. Scientific issues
apart, a crucial practical problem for the remote sensing community is the discrimination of ash clouds from ordinary clouds, because of the threat they pose
to aviation. Disasters were narrowly averted when aircraft unwittingly encountered ash clouds from eruptions of Galunggung in Indonesia, 1982, and of
Redoubt in Alaska, 1989 (Casadevall 1994a), which caused the aircraft engines
to stall. Even when, as in these instances, the engines are eventually restarted and
the aircraft lands safely, the cost of repairing the damage caused to the engines,
avionics, and fabric of an aircraft can reach several tens of millions of dollars in
the case of a Boeing 747 (Casadevall 1994b). An additional challenge is that
discrimination of clouds must be done in near real time, because injection of
eruption clouds to heights of many kilometers may take only minutes, while
aircraft may be approaching at speeds up to 1000 km hr11. Furthermore, aircraft
on long-haul polar or trans-Pacific routes have limited options for diversion to
alternative airports. Timely warnings of eruption clouds can help avoid the sig-

96
FRANCIS n ROTHERY
nificant economic losses associated with landing planeloads of passengers at unintended destinations, and leaving aircraft out of position. Satellite monitoring is
the only way that synoptic eruption cloud data can be gathered rapidly enough
to preempt such problems. To this end, the International Civil Aviation Organization sponsored in 1997 the establishment of nine Volcanic Ash Advisory Centers responsible for issuing Volcanic Ash Advisory Statements to aviators. These
rely heavily on images from meteorological satellites and are available on the
World Wide Web (Figure 5, see color insert). Solid ash particles from an eruption
cloud usually fall to the ground within a few tens of kilometers of the volcano,
except in the very largest cases where fine ash may fall hundreds of kilometers
away. Gases behave differently. If the eruption cloud is confined to the troposphere, it is relatively rapidly dispersed; if injected into the stratosphere ($18 km
at low latitudes; $10 km at high latitudes), however, the SO2 is oxidized to
sulphuric acid, which subsequently forms aerosols through heteromolecular
nucleation of H2SO4 and H2O vapors, at a rate depending on the amount of
available water vapor. The residence time of acid aerosols in the stratosphere is
long (many months), and the scattering to space and absorption of incident solar
radiation by aerosol particles can cause significant surface cooling for periods of
12 years. Aerosols are also important because they provide surfaces upon which
chemical reactions can take place, and may therefore play a significant role in
stratospheric ozone depletion.
Satellite monitoring is thus called on to track separately the silicate, gas, and
aerosol components of eruption plumes.
Tracking the Ascent and Dispersal of Eruption Clouds

Large eruption clouds have been noted on weather satellite images since the
earliest days of the technology, and were used, for example, to study the dynamics
of the eruption clouds of Mt. St. Helens in 1980 (Sparks et al 1986), Colo in 1983
(Malingreau & Kaswanda, 1986), Lascar in 1986 (Glaze et al 1989), and Pinatubo
in 1991 (Holasek et al 1996). Essentially, these studies took advantage of the
unique synoptic and repeat coverage provided by satellites. They concentrated on
determining the height of the plume and tracking its horizontal motion, and did
not draw on the multispectral information contained in the image data.
Early uses of satellite images were empirical; volcanic eruption clouds were
distinguished from ordinary clouds on the basis of their shapes and movement
(e.g. Sawada 1987). However, Prata (1989) and later Holasek & Rose (1991)
showed that the spectral characteristics of andesitic ash provide a more objective
means of discriminating between ash and ordinary clouds, and ultimately of estimating ash amounts in eruption plumes. Images constructed by subtracting
AVHRR channel 4 (10.311.3 lm) from channel 5 (11.512.5 lm) were used to
estimate ash amounts in the eruption cloud from the 1986 eruption of Mt. St.
Augustine. The ability to distinguish ash-laden clouds from others is clearly vital
97
for mitigating aircraft hazards, but AVHRR images by themselves are not always
adequate.

Plume Heights
An important challenge in tracking eruption clouds is determining the height they
reach, because this information can be used to infer the underlying mass eruption
rate. Several approaches have been used, and are usefully summarized in Sparks
et al (1997). Stereo satellite images represent a straightforward method, but currently only the ATSR on the ERS satellites is capable of acquiring stereoscopic
views of transient features. This provides forward viewing and nadir viewing
scans, recorded about two minutes apart. Prata & Turner (1997) showed that these
data could determine cloud-top height to an accuracy of 51 km. We know of
only one study that has yet used this capability to investigate eruption plume
heights (Shocker et al 2000), but similar stereoscopic work is likely to form an
important part of future monitoring.
A crude measure of plume top height can be obtained from radiometric temperature measurements (e.g. Glaze et al 1989). These assume that the emissivity
of the plume is close to unity, and that the plume top temperature is in equilibrium
with the ambient air. Uncertainties arise because emissivities of ash clouds are
poorly known, and because the buoyant momentum of rapidly ascending convective clouds may carry them well above the level of thermal equilibrium
(Woods & Self 1992). For accurate work, it is also necessary to get an atmospheric
temperature profile for the region from radiosonde balloons. These are difficult
to obtain for many parts of the world. Measurements of plume shadow lengths
provide a trigonometric means of estimating height (e.g. Glaze et al 1989), but
only on daytime images. For large umbrella clouds, the value of these measurements is limited because they give heights of the margins of clouds, rather than
their highest central points.
Glaze et al (1999) combined shadow length and temperature determinations
with a photoclinometric study of the December 1989 eruption of Redoubt, Alaska.
They were able to show that the plume rose to an altitude of 3 km while drifting
150 km downwind from the vent. Using the shadow technique alone, they would
have systematically underestimated the plume height by up to 30%. This is a
significant finding, because an underestimate of the plume height by 30% could
lead to an underestimate of the eruption rate by a factor of three. In turn, this
could lead to major errors in predictions made from satellite data of the accumulation rate on the ground of ash falling from the plume, for example.
Plume Gas Measurements

Volcanic gases are overwhelmingly dominated by steam (;95%), with smaller
quantities of carbon dioxide and sulfur dioxide, and traces of gases such as HCl,
HF, and H2. Ideally, remote sensing of volcanic eruption plumes should include
monitoring of gas chemistry during long periods of quiescent, steady-state degas-

FRANCIS
ROTHERY C-1
Figure 1 Detail from a pair of Landsat TM images of the summit crater area of Lascar
volcano, Chile, recorded by day on 27 October (left) and by night on 28 November
(right) 1989. Three bands (or channels) of TM data have been used to compile these
views: band 7, 2.082.35 m, is in red; band 5, 1.551.75 m, is in green; and band 4,
0.760.90 m, is in blue. The area shown is about 2.5 km across in each case. The radiant anomaly in bands 5 and 7 is readily apparent by night, and represents a lava dome
surrounded by hot fumaroles. The anomaly is less apparent by day because the signal
is diluted by reflected sunlight.

C-2 FRANCIS
ROTHERY
Figure 2 Daytime Landsat TM images using the same band combination as in Figure
1, draped over a digital elevation model (viewed looking toward the west) and showing the development of an active lava flow field on Mt. Etna. Left: 2 January 1992;
right: 9 May 1992. The area shown is about 4.3 km across and 9.0 km from front to
back, and covers a scarp-bounded valley on the volcanos eastern flank (the summit is
out of view, beyond the back right corner of each image). The source of the lava flow
is where a fissure intersects the back wall of the valley, and the development of the
flow field was largely controlled by the form of this valley. Only the bright red-orange
areas are thermally radiant; all other areas show up because of reflected sunlight.
Snow appears dark blue, and receded to higher altitudes between the two dates. On 2
January, the lava flow field was at its most vigorous, with large expanses dominated
by radiance from incandescent surfaces. On the 9 May image, radiance is apparent
from open channels and skylights over lava tubes near the source area, and also where
the front of the flow field was still advancing toward the front left. Between these two
areas, the surface slope is much less and the flow field is covered by a relatively
unbroken crust. (Images courtesy of Rob Wright, Open University.)

FRANCIS
ROTHERY C-3
Figure 3 An example of volcanic radar interferometry in practice. On the left is a

radar image of the whole of Mt. Etna (Sicily) with interference fringes superimposed,
representing ground surface displacement between 21 November 1993 and 28
January 1998. On the right is an enlarged view of the area within the green box. The
coast is at the right-hand edge of each image. Elevation changes around the summit
area are indicative of subsidence at a rate of at least 4 cm per year, but of greater
interest are the fringes on the southeastern flank of the volcano. These indicate subsidence across previously unknown NW-SE trending faults, which would be a mirror
image of faults and subsidence known from ground surveys on the northeastern flank
(but not detectable on the interferogram because vegetation and agriculture make the
signal locally incoherent). By combining information from these two sources, a picture emerges of the unbuttressed eastern side of Etna sliding seaward. (Radar inter-

C-4 FRANCIS
ROTHERY
Figure 4 The early stages of the 19 September 1994 eruption of Rabual (Papua, New
Guinea). Top: Photographed looking obliquely westward from the Space Shuttle,
showing the eruption cloud fanning out downwind. Inset: Image from a geostationary
weather satellite (GMS) at 07.32 GMT. Land appears darker than the sea, enabling the
northern coastline of Australia to be made out.

FRANCIS
ROTHERY C-5
Figure 5 A Volcanic Ash Advisory Statement, in this example relying on geostationary weather satellite images to identify and track the plume.

C-6 FRANCIS
ROTHERY
Figure 6 A compilation of TOMS data sets for several successive days revealing
the circumpolar transport of the sulfur dioxide plume from the 1991 eruption of
Mt. Hudson, Chile (courtesy of WI Rose). The values on the color-coded scale are
Dobson units (the standard measure of trace gas column density in the atmosphere).

98
FRANCIS n ROTHERY
sing, because changes in gas composition may presage eruptions. Current satellite
technology, however, is restricted to detection and mapping of sulfur dioxide
when injected in large quantities into the atmosphereusually, but not always,
by large explosive eruptions. Water vapor plays a crucial role in the dynamics of
eruption columns, a role that is complicated by changes in phase that provide
much of the energy driving convection. Unfortunately, estimates of volcanic water
vapor are exceedingly difficult to obtain by remote sensing, because of water
vapors high ambient concentrations. Similarly, volcanic contributions of carbon
dioxide are interesting because of their involvement in the global carbon cycle,
but again they are very difficult to distinguish against normal background levels.
Considerable success has been achieved in using the Total Ozone Mapping
Spectrometer (TOMS) on the Nimbus-7 and Meteor 3 satellites to map the atmospheric distribution of volcanic SO2 (Figure 6, see color insert). Although it was
primarily designed to determine ozone concentrations by measuring backscattered solar ultraviolet irradiance in six narrow wavelength bands in the region
0.306 to 0.360 lm, TOMS is able to take advantage of strong absorptions by SO2
in the same part of the spectrum to measure SO2 concentrations. TOMS provides
a spatial resolution of 47 km 2 47 km at nadir, but averages 62 km. Thus, while
it is well adapted to obtaining estimates of bulk SO2 from large dispersed plumes,
its footprint is too large to permit observations of small-scale volcanic features
for example, the steady-state degassing of SO2 by volcanoes such as Etna or
Stromboli, whose average fluxes are only 4 and 0.8 ktonne/day, respectively. In
a review, Bluth et al (1993) estimated that TOMS successfully detected SO2 from
55 out of 350 known eruptions between 1979 and 1992, and also from several
eruptions not known from ground studies. They also suggested that explosive
volcanism contributes about 4 Mtonne of SO2 annually to the atmosphere, mostly
from a few large eruptions, out of a total of approximately 13 Mtonne (from
explosive and non-explosive eruptions). (This is about 510% of the current
anthropogenic flux.) We list some significant volcanic sulfur dioxide plumes in
Table 2.
Limb-sounding instruments (which look obliquely through the atmosphere,
rather than straight down at the ground) are capable of detecting trace gases at
much lower concentrations than the nadir-pointing TOMS, but the footprints of
these sensors are inevitably even larger. An unanticipated capability of the Upper
Atmosphere Research Satellite (UARS) was to obtain vertically resolved (2.5
5.0 km) daily global maps of volcanic SO2 and aerosols. The Microwave Limb
Sounder (MLS) aboard UARS began operating a hundred days after the June
1991 eruption of Mt. Pinatubo, and immediately detected SO2 from the eruption;
this showed that the gas was distributed in a band around the tropics, with peak
mixing ratios of 15 ppbv (parts per billion by volume) near 26 km altitude. An
initial input of 17 Mtonne was estimated, compared with the 20 Mtonne estimated
from TOMS (Read et al 1993). Although superior to TOMS in resolution, the
MLS, which can detect SO2 down to mixing ratios of 15 ppbv, is also sensitive
only to gas at altitudes greater than 15 km, so tropospheric gas is inaccessible.
99
TABLE 2. Sulfur dioxide plumes for the period 19791985 detected by the TOMS
instrument on Nimbus 7 (from Mouginis-Mark et al 1991). The mystery eruption is now
thought to be that of Nyamuragira in Zaire. Note that most eruptions were explosive, but those
of Mauna Loa and Nyamuragira were large, effusive events.

Volcano
Sierra Negra
Mt. St. Helens
Alaid
Ambrym
Pagan
Mystery
El Chichon
Galunggung
Soputan
Colo, Una Una
Mauna Loa
Fernandina
Soputan
Krafla
Ruiz
Ruiz
Eruption date
Number of days
plume tracked
Nov 13 1979
May 19 1980
April 27 1981
May 8 1981
May 15 1981
Dec 26 1981
March 28 1981
April 5, June 25, July 14 1982
Aug 27 1982
July 24 1983
March 25 1984
March 31 1984
May 25 1984
Sept 5,10, 18 1984
Sept 12 1984
Nov 13 1985
3
4
22
2
2
`12
`64
1, 1, 2
1
1
1
11
2
3,1,1
7
7
Total SO2/kton
340
1,000
5,000
275
190
35
660
Although eruption clouds from large explosive eruptions contain massive

amounts of both gas and ash, the two components have rarely been studied
together. In an important contribution, Schneider et al (1999) used ultraviolet data
from TOMS together with infrared data from AVHRR to study the cloud erupted
by El Chichon, Mexico in 1982. Orbital constraints dictated that observations
from the two satellites were never exactly simultaneous, but useful data were
obtained for the first four days of the life of the cloud. The eruption cloud reached
a maximum altitude of 32 km, where strong wind shear led to separation of the
sulfur dioxide and ash components. The uppermost, sulfur dioxide-rich part was
transported westward in the stratosphere, while the fine silicate ash mostly drifted
southward in the troposphere. Schneider et al (1999) estimated that 7.1 2 109 kg
of gas and 6.5 2 109 kg of fine ash were erupted. The ash measured in the cloud
formed only 0.7% of the mass of ash that fell to the ground near the volcano, and
98% of this fine ash was removed from the cloud over the following four days,
whereas the mass of sulfur dioxide showed only a modest decrease.
100
FRANCIS n ROTHERY

Aerosol Measurements
Wen and Rose (1994) refined the data extraction by showing that AVHRR bands
4 and 5 could also be used to estimate particle sizes and masses in eruption clouds,
and Krotkov et al (1999) showed the power of using AVHRR and TOMS (or
equivalent) data in parallel to track the history of total particle mass and size
distribution during the evolution of an eruption cloud. In some eruptions, such as
that of Rabaul 1994 (see Figure 4), large quantities of ice crystals are thought to
form in rising eruption clouds. Further work is required to ascertain the effects
of ice on the retrieved quantities of ash in such clouds.
Measurements of stratospheric aerosols from volcanic eruptions are perhaps
of more relevance to atmospheric and climate scientists than to volcanologists.
AVHRR satellite data were used to map aerosols from the El Chichon and Pinatubo eruptions, by determining the errors introduced into apparent changes in
sea-surface temperature caused by the aerosols (Strong & Stowe 1993). The technique requires that the plume be transparent enough so that the sea surface is
visible. Thus, very dense, ash-laden plumes cannot be studied. A similar technique
may also be valuable in monitoring the steady state emissions from volcanoes
such as Etna, which rarely emits dense ash clouds.
The Stratospheric Aerosol and Gas Experiment (SAGE II) carried by UARS
scans the Earths sunrise and sunset limbs, with a vertical resolution of 1 km in
order to provide information on the distribution of aerosols. It tracked the dispersal of 1991 Pinatubo aerosol cloud into both northern and southern hemispheres, but was unable to discriminate between aerosol types (McCormick &
Viega 1992).
FUTURE PROSPECTS
There are remarkably bright prospects for volcano remote sensing. Future years
will bring the launch of ever-increasing numbers of remote sensing satellites from
several different countries. In the short term, Landsat 7, launched in 1999, will
extend the capability of this familiar system for many years. NASAs Earth
Observing System (EOS) will greatly enhance monitoring capabilities with both
high- and low-resolution multispectral sensors. Launch of the first platform in the
series (previously known as AM-1, but renamed Terra) is scheduled for December
1999 or later; launch of others in the series is due to continue in 2000 (EOS PM1) and 2002 (EOS CHEM-1). Volcanological investigations to be carried out
under the auspices of EOS will make use of a bewildering array of sensors:
ASTER, CERES, EOSP, ETM`, GLAS, HIRDLS, MISR, MODIS, OMI, SAGE
III, TES, TOMS and VCL. Many other satellites are in an advanced design stage,
such as ENVISAT and RADARSAT-2. Although the string of acronyms gives a
flavor of the range of investigations to be undertaken, it is impractical to review
them all. It is evident, however, that unprecedented amounts of satellite data will
101

become available, accompanied by simultaneous developments in computer data

storage and networking, which will make the data accessible to volcanologists
worldwide at a negligible marginal cost.
These welcome developments are offset by some difficulties. As the preceding
alphabet soup of acronyms hints, the volume of data threatens to become an
unmanageable flood, especially because the number of volcanologists with remote
sensing skills equipped to handle the data remains small and concentrated in a
few countries. Similarly, the calibration, validation, and ground truth studies
required to build up confidence in the results obtained from remotely sensed data
are costly and time-consuming, and thus often lag far behind the publication of
results from spacecraft.
Nonetheless, many volcanically useful tools, which build on the range of techniques described earlier, seem certain to become routinely available in the near
term. These include:
1. Stereoscopic views of eruption cloud altitude and topography, permitting tests
of models of eruption cloud ascent and dispersal.
2. A global system of automated volcanic hot spot detection and monitoring,
using GOES-type geostationary satellites, polar orbiting instruments such as
ATSR (and successors), and Terras MODIS instrument. Detailed follow-up
studies of selected targets will be possible using Terras ASTER instrument,
which has infrared capability. For the first time, it should be possible to inventorize the location, styles, and durations of eruptions worldwide.
3. A global digital topographic map of the Earth between 608N and 608S to
be collected by the Shuttle Radar Topography Mission in 1999. Data from
this mission will have a spatial resolution of 30 m and a vertical accuracy of
15 m, and will provide valuable input for a wide range of volcanic topography
studies.
4. Ground deformation monitoring by means of differential radar interferometry
and possibly lidar (e.g. Garvin 1996).
5. Studies of a range of volcanic cloud microphysical and chemical processes,
especially in the wake of the EOS CHEM-1 platform.
Some conceptually new approaches also seem probable. Hitherto, the TOMS
instrument has only been able to measure SO2 gas emissions from rather large
explosive eruptions. In an encouraging development, Realmuto et al (1994) demonstrated that the airborne Thermal Infra-Red Mapping Spectrometer (TIMS)
could be used to map SO2 concentrations in the gas plume from Mt. Etna, which
emits 40005000 tonne of SO2 per day. This holds the promise that steady state
degassing may be detectable using sensors such as ASTER, which has infrared
channels at the appropriate wavelength. Although this offers the opportunity to
measure quiescent SO2 degassing rates for many currently inaccessible volcanoes,
and thus contribute to understanding the global sulfur cycle, ASTERs 16-day
repeat cycle will preclude its use for frequent monitoring. However, when mea-

102
FRANCIS n ROTHERY
surements of SO2 gas emissions can be combined with data from sensors designed
to measure aerosols (such as SAGE III), it may become possible to investigate
the rates at which sulfate aerosols are formed from volcanic gases in different
contexts, with important implications for understanding the radiative effects of
volcanic aerosols on climate.
Although volcanology has been included in the objectives for EOS since its
inception, volcanological science has necessarily formed a relatively minor component of the overall mission strategy. The competing requirements of different
scientific disciplines dictate that the sensors used for volcanology represent a
compromise with a range of other applications. Eventually, a satellite dedicated
to volcanology will be launched. Such a platform (VOLCAM) was conceived in
1998, and although not ultimately approved for funding, its successors are likely
to embody similar concepts.
VOLCAM was designed primarily with aviation hazards in mind; therefore,
its UV and IR sensors were intended to optimize the ash and sulfur dioxide
capabilities of existing AVHRR and TOMS for tracking volcanic clouds and for
discriminating between volcanic and meteorological clouds. Intended as a geostationary satellite located over the Americas, it would have had the advantage
of 15-minute time resolution. Through optimization of the wavelength bands in
the UV camera, the sensitivity to sulfur dioxide was improved by a factor of ten
over TOMS. The VOLCAM footprint is also smaller by a factor of 16, so much
smaller clouds could be detected. Thus, in addition to determining whether clouds
were safe to fly through, VOLCAM also offered the potential to monitor preeruptive outgassing from all the volcanoes viewed from the satellite.
A satellite such as VOLCAM would permit investigations of some key science
issues, such as the fate of SO2 in eruption clouds. In fresh clouds, SO2 may be
lost by absorption onto ash or ice particles (Rose et al 1995), but few direct data
bear on this theory. In older clouds (more than a day), sulfur dioxide is probably
converted to sulfate in gas phase reactions. Measurements of the lifetime of SO2
in clouds of different ages are required to resolve the relative rates of these
processes.
Also required is a high-resolution thermal system in polar orbit, with spectral
bands and sensor gains tailored for volcanologic applications. Such a satellite
could be rapidly targeted on volcanic hotspots to investigate hot spot alarms
triggered by the low-resolution thermal alert system that is already coming into
place.
POSTSCRIPT
Peter Francis died suddenly on 30 October 1999 at the age of 54, just three weeks
after acting as convenor for a highly successful international meeting on eruptions
of andesite volcanoes. He was a pioneer in satellite remote sensing of volcanoes
and in the use of Fourier Transform Infrared spectroscopy for volcanic gas analysis. He will be greatly missed.
Visit the Annual Reviews home page at www.AnnualReviews.org.
103

LITERATURE CITED
Archambault C, Stoschek J, Tanguy JC. 1980.
Analyse et surveillance thermique de volcans actifs (197879) Bull. PERPSEV No.
11, Paris
Bluth GJS, Schnetzler CC, Kreuger AJ, Walter
LS, et al. 1993. The contribution of explosive volcanism to global atmospheric sulphur dioxide concentrations. Nature
366:32729
Bonneville A, Vasseur G, Kerr Y. 1985. Satellite thermal infrared observations of
Mt. Etna after the 17th March 1981 eruption. J. Volcanol. Geotherm. Res. 24:293
313
Casadevall TJ. 1994b. The 19891990 eruption of Redoubt Volcano, Alaska: implications on aircraft operations. J. Volcanol.
Geotherm. Res. 62:30116
Casadevall TJ, ed. 1994a. Volcanic hazards and
aviation safety. In Proc. First Internat.
Symp. on Volcanic Ash and Aviation Safety.
US Geol. Surv. Bull. No. 2047. 450 pp.
Washington, DC: US Govt. Print. Off.
Clague DA, Denlinger RP, McGee KA. 1994.
Role of olivine cumulates in destabilizing
the flanks of Hawaiian volcanoes. Bull. Volcanol. 56:42534
Davies AG. 1996. Ios volcanism: thermophysical models of silicate lava compared
with observations of thermal emission. Icarus 124:4561
Dean K, Servilla M, Roach A, Foster B, Engle
K, et al. 1998. Satellite monitoring of
remote volcanoes improves study efforts in
Alaska. EOS Trans. Am. Geophys. Union.
79:41318
Delacourt C, Briole P, Achache J. 1998. Tropospheric corrections of SAR inteferograms with strong topography. Application
to Etna. Geophys. Res. Lett. 25:284952
Delmelle P, Baxter P, Beaulieu A, Boardman
L, Burton M, et al. 1999. Interdisciplinary
attention focuses on the restless Masaya
caldera, Nicaragua. EOS Trans. Am. Geophys. Union. In press
Dozier JA. 1981. A method for satellite identification of surface temperature fields of
sub-pixel resolution. Rem. Sens. Environ.
11:22129
Evans DL, Farr T, Zebker HA, van Zyl JJ,
Mouginis-Mark PJ, et al. 1992. Radar interferometry studies of the Earths topography.
EOS Trans. Am. Geophys. Union 73:553
58
Flynn LP, Harris AJL, Rothery DA, Oppenheimer C, et al. 2000. High spatial resolution remote sensing of active volcanic
features using Landsat and hyperspectral
data. In AGU monograph, Remote Sensing of Volcanoes, ed. P Mouginis-Mark, J
Fink, L Glaze. Washington, DC: AGU. In
press
Francis PW, Glaze LS, Rothery DA. 1989. Lascar volcano set to erupt. Nature 339: 434
Francis PW, Oppenheimer C, Stevenson D.
1993. Endogenous growth of persistently
active volcanoes. Nature 366:55457
Francis PW. 1989. Remote sensing of volcanoes. Adv. Space Res. 9:8992
Francis PW, Rothery DA. 1987. Using the
Landsat Thematic Mapper to detect and
monitor active volcanoes: an example from
the Lascar volcano, northern Chile. Geology 15:61417
Francis PW, Wadge G, Mouginis-Mark PJ.
1996. Satellite monitoring of volcanoes in
IAVCEI-UNESCO Manual of Volcano
Monitoring, ed. R Scarpa, R Tilling. pp.
257298. Berlin: Springer-Verlag
Friedman JD, Williams RS. 1970. Changing
pattern of thermal emission from Surtsey,
Iceland, between 1966 and 1969. USGS
Prof. Pap. 700-D:D11624
Gaddis LR. 1992. Lava flow characterization
at Pisgah volcanic field, California, with
multiparameter imagining radar. Geol. Soc.
Am. Bull. 104:695703
Garvin JB. 1996. Topographic characterization
and monitoring of volcanoes via airborne
laser altimetry. In Volcano Instability on the

104
FRANCIS n ROTHERY
Earth and Other Planets, ed. WJ McGuire,

AP Jones, J Neuberg, pp. 13752. London:
Geological Society Special Publication
110. 389 pp.
Garvin JB, Mahmood A, Giguere C. 1999.
Taking a clear look at cloud-covered islands
on a seasonal basis. EOS Trans. AGU No.
80 (5) 48:5354
Glaze LS, Francis PW, Self S, Rothery DA, et
al. 1989. The Lascar September 16 1986
eruption: satellite investigations. Bull. Volcanol. 51:14960
Glaze LS, Wilson L, Mouginis-Mark PJ. 1999.
Volcanic eruption plume top topography
and heights determined from photoclinometric analysis of satellite data. J. Geophys.
Res. 104:29893001
Greeley R, Spudis PD. 1981. Volcanism on
Mars. Rev. Geophys. Space Phys. 19:1341
Harris AJL, Blake S, Rothery DA, Stevens NF,
et al. 1997a. A chronology of the 1991 to
1993 Etna eruption using AVHRR data:
implications for real-time thermal volcano
monitoring. J. Geophys. Res. 102:7985
8003
Harris AJL, Butterworth AL, Carlton RW,
Downey I, Miller P, et al. 1997b. Low cost
volcano surveillance from space: case studies from Etna, Krafla Cerro Negro, Fogo,
Lascar and Erebus. Bull. Volcanol. 59:49
64
Harris AJL, Flynn L, Keszthelyi PJ, MouginisMark PJ, Rowland SK, et al. 1998. Calculation of lava effusion rates from Landsat
TM data. Bull. Volcanol. 60:5271
Harris AJL, Flynn LP, Dean K, Pilger E, Wooster M, et al. 2000. Real-time monitoring of
volcanic hot spots with satellites. In AGU
monograph, Remote Sensing of Volcanoes,
ed. P Mouginis-Mark, J Fink, L Glaze.
Washington, DC: AGU. In press
Harris AJL, Flynn LP, Rothery DA, Oppenheimer C, Sherman SB, et al. 1999. Mass
flux measurements at lava lakes: implications for magma recycling. J. Geophys. Res.
104:711736
Harris AJL, Leszthelyi L, Flynn LP, MouginisMark PJ, Thornber C, et al. 1997a. Chro-
nology of the episode 54 eruption at

Kilauea Volcano, Hawaii, from GOES-9
satellite data. Geophys. Res. Lett. 24:3281
84
Harris AJL, Stevenson DS. 1997a. Thermal
observations of degassing open conduits
and fumaroles at Stromboli and Vulcano
using remotely sensed data. J. Volcanol.
Harris AJL, Stevenson DS. 1997b. Magma
budgets and steady state activity of Vulcano
and Stromboli. Geophys. Res. Lett.
24:104346
Head JW, Crumpler LS, Aubele JC, Guest JE,
Saunders RS, et al. 1992. Venus volcanism:
Classification of volcanic features and
structures, associations and global distributions from Magellan data. J. Geophys.
Res. 97:1315397
Holasek RE, Rose WI. 1991. Anatomy of 1986
Augustine volcano eruptions as recorded by
multispectral image processing of digital
AVHRR weather satellite data. Bull. Volcanol. 53:42035
Holasek RE, Self S, Woods A. 1996. Satellite
observations and interpretations of the 1991
Mount Pinatubo eruption plumes. J. Geophys. Res. 101:2763555
Krotkov NA, Torres O, Seftor C, Kreuger AJ,
et al. 1999. Comparison of TOMS and
AVHRR volcanic ash retrievals from the
August 1992 eruption of Mt. Spurr. Geophys. Res. Lett. 26:45558
Lanari R, Lundgren P, Sansoti E. 1998.
Dynamic deformation of Etna volcano
observed by satellite radar interferometry.
Geophys. Res. Lett. 25:154144
Lipman PW, Mullineaux DR, eds. 1981. The
1980 eruptions of Mt. St. Helens, Washington. USGS Prof. Pap. No. 1250. 843 pp.
Lu Z, Mann D, Freymueller J. 1998. Satellite
radar interferometry measures deformation
at Okmok volcano. EOS Trans. Am. Geophys. Union 79 No. 39:46168
MacKay ME, Mouginis-Mark PJ. 1997. The
effect of varying acquisition parameters on
the interpretation of SIR-C data: the
Virunga volcanic chain. Rem. Sens. Environ. 59:32136


Malingreau JP, Kaswanda. 1986. Monitoring
volcanic eruptions of Indonesia using
weather satellite data: The Colo eruption of
July 28, 1983. J. Volcanol. Geotherm. Res.
27:17994
Massonet D, Briole P, Arnaud A. 1995. Deflation of Mount Etna monitored by spaceborne radar interferometry. Nature
375:56770
Massonet D, Feigl K, Rossi M, Adragna F, et
al. 1994. Radar interferometric mapping of
deformation in the year after the Landers
earthquake mapped. Nature 369:22730
Massonet D, Rossi M, Carmona C, Adragna F,
Peltzer G, et al. 1993. The displacement
field of the Landers earthquake mapped by
radar inteferometry. Nature 364:13842
Matson M, Dozier J. 1981. Identification of
subresolution high temperature sources
using a thermal IR source. Photogramm.
Eng. Rem. Sens. 47:1311
Matthews SJ, Gardeweg MC, Sparks RSJ.
1997. The 1994 to 1996 cyclic activity of
Lascar Volcano, northern Chile: cycles of
dome growth, dome subsidence, degassing
and explosive eruptions. Bull. Volcanol.
59:7282
McClelland L, Simkin T, Summers M, Nielsen
E, Stein TC, et al. 1989. Global Volcanism
197585. Englewood Cliffs, New Jersey:
Prentice Hall. 655 pp.
McCormick MP, Viega RE. 1992. SAGE II
measurements of early Pinatubo aerosols.
Geophys. Res. Lett. 19:15558
McEwen AS, Keszthelyi L, Spencer JR, Schubert G, et al. 1998. High temperature silicate volcanism on Jupiters moon Io.
Science 281:8790
Mouginis-Mark PJ, et al. 1991. Analysis of
active volcanoes from the Earth Observing
System. Rem. Sens. Eviron. 48:5160
Mouginis-Mark PJ, Garbeil H. 1993. Digital
topography of volcanoes from radar interferometry: an example from Mt. Vesuvius,
Italy. Bull. Volcanol. 55:56670
Mouginis-Mark PJ. 1995. Preliminary analysis
of volcanoes with the SIR C/X-SAR radar.
IEEE Trans. Geosci. Remote Sens. 33: 934
41
105
Njoku EG. 1990. Satellite remote sensing of

sea surface temperature. In Surface Waves
and Fluxes, Vol. 2, ed. GL Geernaert, WJ
Plant, pp. 31138. Dordrecht: Kluwer. 352
pp.
Oppenheimer C, Francis PW, Rothery DA,
Carlton RWT. 1993. Infrared image analysis of volcanic thermal features: Lascar volcano, Chile. J. Geophys. Res. 98:426986
Oppenheimer C. 1993. Infrared surveillance of
crater lakes using satellite data. J. Volcanol.
Oppenheimer C. 1991. Lava flow cooling estimated from Landsat Thermatic Mapper
infrared data: the Longquimay eruption
(Chile, 1989). J. Geophys. Res. 21:86578
Pieri DC, Glaze LS, Abrams MJ. 1990. Thermal radiance observations of an active lava
flow during the June 1984 eruption of
Mount Etna. Geology 30:101822
Prata AJ, Turner PJ. 1997. Cloud-top height
determination using ATSR data. Remote
Sens. Environ. 59:113
Prata AJ. 1989. Observations of volcanic ash
clouds in the 1012 lm window using
AVHRR/2 data. Int. J. Rem. Sens. 10:751
61
Read WG, Froidevaux L, Waters JW. 1993.
Microwave limb sounder measurement of
stratospheric SO2 from Mt. Pinatubo volcano. Geophys Res. Lett. 20:1299302
Realmuto VJ, Hon K, Kahle AB, Abbott EA,
Pieri DC, et al. 1992. Multispectral thermal
infrared mapping of the 1 October 1988
Kupaianaha flow field, Kilauea volcano,
Hawaii. Bull. Volcanol. 55:33441
Realmuto VJ, Abrams MJ, Buongiorno MF,
Pieri DC, et al. 1994. The use of multispectral thermal infrared image data to estimate
the sulfur dioxide flux from volcanoes: a
case study from Mt. Etna. Sicily, July 29
1986. J. Geophys. Res. 99:48188
Rose WI, Delene DJ, Schneider DJ, Bluth GJS,
Krueger AJ, et al. 1995. Ice in the 1994
Rabaul eruption cloud: implications for
volcano hazard and atmospheric effects.
Nature 375: 47779
Rothery DA, Francis PW, Wood CA. 1988.
Volcano monitoring using short wavelength

106
FRANCIS n ROTHERY
infrared data from satellites. J. Geophys.

Res. 93:79938008
Rothery DA, Oppenheimer C, Bonneville A.
1995. Infrared thermal monitoring. In
Monitoring Active Volcanoes, ed. B McGuire, CRJ Kilburn, J Murray, pp. 184
216. London: UCL Press. 421 pp.
Rothery DA, Pieri DC. 1993. Remote sensing
of active lava. In Active Lavas, ed. CRJ Kilburn, G Luongo, pp. 203323. London,
UK: University College London Press
Rowland SK, Smith GA, Mouginis-Mark PJ.
1993. Preliminary ERS-1 observations of
Alaskan and Aleutian volcanoes. Rem.
Sens. Environ. 48:125
Rowland SK, Walker GPL. 1990. Pahoehoe
and aa in Hawaii: volumetric flow rate controls the lava structure. Bull. Volcanol.
52:61528
Rowland SK. 1996. Slopes, lava flow volumes,
and vent distributions on Volcan Fernandina, Galapagos islands. J. Geophys. Res.
101:65772
Sawada Y. 1987. Study on analysis of volcanic
eruptions based on eruption cloud image
data obtained by the Geostationery Meteorological Satellite (GMS). Tech. Rep.
Meteorol. Res. Inst. (Japan) 22, Tokyo,
Japan
Schneider DJ, Rose WI, Coke LR, Bluth GJS,
Sprod IE, et al. 1999. Early evolution of a
stratospheric volcanic eruption cloud as
observed with TOMS and AVHRR. J. Geophys. Res. 104, D4:403750
Self S, Thordason TH, Keszthelyi L, Walker
GPL, Hon K, et al. 1996. A new model for
the emplacment of the Columbia River
Basalts as large inflated pahoehoe sheet
lava flow fields. Geophys. Res. Lett.
23:268992
Shocker HL, Rose WI, Bluth GJS, Prata AJ,
Viramonte JG, et al. 2000. Lascar volcanic
clouds of 1993: merging of satellite-based
remote sensings from TOMS, AVHRR and
ATSR during three days of atmospheric residence. In preparation
Sigmundsson F, Vadon H, Massonet D. 1997.
Readjustment of the Krafla spreading segment to crustal rifting measured by satellite
radar interferometry. Geophys. Res. Lett.

24:184346
Sparks RSJ, Bursik MI, Carey SN, Gilbert JS,
Glaze LS, et al. 1997. Volcanic Plumes.
Chichester: Wiley & Sons. 574 pp.
Sparks RSJ, Moore JG, Rice CJ. 1986. The initial giant umbrella cloud of Mt. St. Helens.
J. Volcanol. Geotherm. Res. 28:25774
Stevenson DS, Blake S. 1998. Modelling the
dynamics and thermodynamics of volcanic
degassing. Bull. Volcanol. 60:30717
Wadge G, Scheuchl B, Stevens NF. 2000.
Spaceborne radar measurements of the
eruption of Soufriere Hills volcano, Montserrat during 199699. J. Volcanol. Geotherm. Res. In press
Washington HS. 1917. Geol. Soc. Am. Bull.
28:24978
Wen S, Rose WI. 1994. Retrieval of particle
sizes and masses in volcanic clouds using
AVHRR band 4 and 5. J. Geophys. Res.
99:542131
Williams RS, Friedman JD. 1970. Satellite
observations of effusive volcanism. Br.
Interplanet. Soc. J. 23:44150
Woods AW, Self S. 1992. Thermal disequilibrium at the top of volcanic clouds and its
effects on estimates of the column height.
Nature 355:62830
Wooster MJ, Kaneko T. 1998. Satellite thermal
analysis of lava dome effusion rates at
Unzen Volcano. J. Geophys. Res. 103:
2093547
Wooster MJ, Rothery DA. 1997. Thermal
monitoring of Lascar volcano, Chile, using
infrared data from the Along Track Scanning Radiometer: a 19921995 time series.
Bull. Volcanol. 58:56679
Zebker HA, Rosen P, Goldstein RM, Gabriel
A, Werner CL, et al. 1994. On the derivation of coseismic displacement fields using
differential radar interferometry: the Landers earthquake. J. Geophys. Res. 99:19617
34
Zebker HA, Rosen P, Hensley S, MouginisMark PJ. 1996. Analysis of active lava
flows on Kilauea volcano, Hawaii, using
SIR-C radar correlation measurements.
Geology 102:49598
Annual Review of Earth and Planetary Science

Volume 28, 2000
CONTENTS
Palynology after Y2K--Understanding the Source Area of Pollen in
Sediments, M. B. Davis
Dinosaur Reproduction and Parenting, John R. Horner
Evolution and Structure of the Lachlan Fold Belt (Orogen) of Eastern
Australia, David A. Foster, David R. Gray

Remote Sensing of Active Volcanoes, Peter Francis, David Rothery
1
19
47
81
Dynamics of Volcanic Systems in Iceland: Example of Tectonism and

Volcanism at Juxtaposed Hot Spot and Mid-Ocean Ridge Systems, Agust
Gudmundsson
107
Understanding Oblique Impacts from Experiments, Observations, and

Modeling, E. Pierazzo, H. J. Melosh
141
Synthetic Aperture Radar Interferometry to Measure Earth''s Surface

Topography and Its Deformation, Roland Brgmann, Paul A. Rosen, Eric
J. Fielding
169
Geologic Evolution of the Himalayan-Tibetan Orogen, An Yin, T. Mark

Harrison
211
MARS 2000, Arden L. Albee

Vredefort, Sudbury, Chicxulub: Three of a Kind, Richard Grieve, Ann
Therriault
Climate Reconstruction from Subsurface Temperatures, Henry N.
Pollack, Shaopeng Huang
Asteroid Fragmentation and Evolution of Asteroids, Eileen V. Ryan
Seismic Imaging of Mantle Plumes, Henri-Claude Nataf
New Perspectives on Orbitally Forced Stratigraphy, Linda A. Hinnov
281
305
339
367
391
419
Clathrate Hydrates, Bruce A. Buffett
477
Heterogeneity of the Lowermost Mantle, Edward J. Garnero
509
Spreading Volcanoes, Andrea Borgia, Paul T. Delaney, Roger P.

Denlinger
539
Scaling, Universality, and Geomorphology, Peter Sheridan Dodds,

Daniel H. Rothman
571
Chemical Weathering, Atmospheric CO2, and Climate, Lee R. Kump,

Susan L. Brantley, Michael A. Arthur
611
Self-Ordering and Complexity in Epizonal Mineral Deposits, Richard W.

Henley, Byron R. Berger
669

R S A V: Emote Ensing of Ctive Olcanoes

Uploaded by

Copyright:

Available Formats

R S A V: Emote Ensing of Ctive Olcanoes

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

R S A V: Emote Ensing of Ctive Olcanoes

Uploaded by

Copyright:

Available Formats

Annu. Rev. Earth Planet. Sci. 2000.

REMOTE SENSING OF ACTIVE VOLCANOES

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Key Words volcanoes, satellite remote sensing, thermal anomalies, eruption

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

been successfully applied to a wide range of volcanological problems, furnishing

RADIANT THERMAL ENERGY BUDGETS OF

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

VOLCANO REMOTE SENSING

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Ry 4 sy[ f L(ky, Thot) ` (1 1 f )L(ky, Tback)]

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

VOLCANO REMOTE SENSING

Thermal Energy Budgets of Volcanoes: Implications for

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

VOLCANO REMOTE SENSING

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

THERMAL REMOTE SENSING OF LAVA FLOWS

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

aircraft deployment of the NASA Thermal Infrared Multispectral Scanner (TIMS)

VOLCANO REMOTE SENSING

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Eruption Detection and Monitoring

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

VOLCANO REMOTE SENSING

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

STUDIES OF VOLCANO TOPOGRAPHY AND

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

VOLCANO REMOTE SENSING

coverage. Apart from their contribution to the design of later Earth-orbiting

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Earth Orbital Radars

Differential Interferometric Synthetic Aperture Radar

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

of radar interferograms to monitor inflation, deflation, and rifting of active

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

VOLCANO REMOTE SENSING

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Tracking the Ascent and Dispersal of Eruption Clouds

VOLCANO REMOTE SENSING

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Plume Gas Measurements

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Figure 3 An example of volcanic radar interferometry in practice. On the left is a

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

VOLCANO REMOTE SENSING

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

Although eruption clouds from large explosive eruptions contain massive

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

VOLCANO REMOTE SENSING

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org

become available, accompanied by simultaneous developments in computer data

Annu. Rev. Earth Planet. Sci. 2000.28:81-106. Downloaded from www.annualreviews.org