remote sensing

Article
Uncertainty Analysis of Remotely-Acquired Thermal
Infrared Data to Extract the Thermal Properties of
Active Lava Surfaces
James O. Thompson * and Michael S. Ramsey
Department of Geology and Environmental Science, University of Pittsburgh, 4107 O’Hara Street,
Pittsburgh, PA 15260, USA; mramsey@pitt.edu
* Correspondence: james.thompson@pitt.edu; Tel.: +1-412-624-9234




Received: 28 November 2019; Accepted: 3 January 2020; Published: 5 January 2020


Abstract: Using thermal infrared (TIR) data from multiple instruments and platforms for analysis of
an entire active volcanic system is becoming more common with the increasing availability of new
data. However, the accuracy and uncertainty associated with these combined datasets are poorly
constrained over the full range of eruption temperatures and possible volcanic products. Here, four
TIR datasets acquired over active lava surfaces are compared to quantify the uncertainty, accuracy,
and variability in derived surface radiance, emissivity, and kinetic temperature. These data were
acquired at Kı̄lauea volcano in Hawai’i, USA, in January/February 2017 and 2018. The analysis
reveals that spatial resolution strongly limits the accuracy of the derived surface thermal properties,
resulting in values that are significantly below the expected values for molten basaltic lava at its
liquidus temperature. The surface radiance is ~2400% underestimated in the orbital data compared
to only ~200% in ground-based data. As a result, the surface emissivity is overestimated and the
kinetic temperature is underestimated by at least 30% and 200% in the airborne and orbital datasets,
respectively. A thermal mixed pixel separation analysis is conducted to extract only the molten
fraction within each pixel in an attempt to mitigate this complicating factor. This improved the orbital
and airborne surface radiance values to within 15% of the expected values and the derived emissivity
and kinetic temperature within 8% and 12%, respectively. It is, therefore, possible to use moderate
spatial resolution TIR data to derive accurate and reliable emissivity and kinetic temperatures of a
molten lava surface that are comparable to the higher resolution data from airborne and ground-based
instruments. This approach, resulting in more accurate kinetic temperature and emissivity of the
active surfaces, can improve estimates of flow hazards by greatly improving lava flow propagation
models that rely on these data.

Keywords: thermal infrared remote sensing; volcanic eruption analysis; Kı̄lauea volcano; emissivity;
uncertainty analysis; data synergy

1. Introduction
Using remote sensing data to monitor volcanic eruptions has improved our understanding of
the precursory activity, eruptions dynamics, and eruptive products [1,2]. Thermal infrared (TIR)
data between a spectral range of 8 to 12 µm have been utilized since the early 1960s, with major
developments in orbital, airborne, and ground-based TIR technologies. These systems provide new
data that are important for modeling volcanic activity over a variety of spatial, spectral, and temporal
scales (e.g., [3]). For example, the data have improved the accuracy, reliability, and duration of
precursory evaluations, constrained eruption dynamics, and improved both magma rheological and
thermal models of all volcanic products [1,4,5].

Remote Sens. 2020, 12, 193; doi:10.3390/rs12010193 www.mdpi.com/journal/remotesensing

Remote Sens. 2020, 12, 193 2 of 21

Multi-instrument, multi-platform TIR data of an entire volcanic system must be properly integrated
and cross-calibrated to understand the thermal regime. An orbiting instrument such as the Advanced
Spaceborne Thermal Emission and Reflection Radiometer (ASTER) acquiring moderate spatial (90 m)
and spectral (5 TIR channels) resolution data will provide lower temporal frequency observations
(~16 days) of the larger volcanic system and the ongoing eruption [6]. Typically, these systems miss the
detailed spatial and spectral observations of low-level smaller-scale thermal anomalies and processes
unless specific methods are adapted to directly detect and characterize these, and/or coincident
mid-infrared data are exploited [5,7]. Airborne and ground-based instruments can observe smaller
anomalies with the required spatial, spectral, and radiometric resolutions, but they are unable to
provide data over extended time periods and may miss the synoptic coverage. Hence, there is a need for
instrument and data synergy to improve observations from an individual source as well as the scientific
interpretation of the volcanic system. This synergy requires accuracy and uncertainty quantification
to improve analysis confidence. This issue has been investigated over many scenarios [8–11] but not
for observations of active molten lava surfaces where rapid changes in thermal properties occur both
spatially and temporally at very high surface temperatures (<1450 K), as seen in the recent basaltic
eruptions in Hawaii.
Basaltic volcanism is the most common form of extrusive volcanism on Earth, occurring at every
tectonic setting [12]. Recent significant basaltic eruptions have occurred at Piton de la Fournaise
(Reunion Island); Yasur (Vanuatu); Tolbachik (Russia); and Kı̄lauea (Hawaii), emphasizing the hazards
that these eruptions pose. For example, in 2018 the lava flows from the Lower East Rift Zone of Kı̄lauea
Volcano threatened lives and property in the Puna District of the Island of Hawai’i [13]. During the
4-month long eruption, 23 fissures opened emplacing over 750 m3 of lava that covered the populated
Leilani Estates and destroyed 716 dwellings [14]. Near-daily broadband helicopter-based TIR camera
data were acquired to produce lava inundation maps, but no thermal lava flow propagation forecasts
were calculated, mostly due to the lack of an available robust and reliable modeling approach.
With these recent large flow-producing eruptions, there is an increasing need to develop rapidly
implementable lava flow propagation models to aid in volcanic hazard response. Developing these
models requires improvements in the accuracy of estimating lava flow pathways and velocity to
reduce the hazard risk to proximal communities. Volcanic eruption flow modeling relies on factors
including but not limited to effusion rates, flow dynamics, viscosity, cooling rate, in addition to
continuously-updated topography [15–17]. Typically, these models use data from TIR instruments
to measure cooling and model viscosity. Therefore, improvements in the accuracy and constraint
of uncertainty in the measured surface radiance, kinetic temperature, and emissivity from these
instruments will improve our ability to model downslope lava surface dynamics, rheological changes,
flow propagation, and hazard assessments (e.g., [3,18]).
TIR data for this study were acquired during the 2017 and 2018 NASA Hawaii airborne campaign
and are utilized to determine the spatial and spectral resolution influence on the derived kinetic
temperature and emissivity of the active lava surfaces as they cool. Simultaneous TIR data were
acquired using the ground-based Miniature Multispectral Thermal Camera (MMT-Cam) system [19],
the airborne Hyperspectral Thermal Emission Spectrometer (HyTES) [20], and the MODIS/ASTER
(MASTER) airborne simulator [21], as well as the orbital ASTER instrument [6], with increasingly
lower spatial resolutions, respectively. A primary goal was to constrain the relationship between the
derived emissivity and acquired surface radiance to improve the accuracy of the surface temperatures
derived from the TIR data [1,3].

2. Background
The TIR region is primarily used in volcanology to determine the eruption temperature, and
with later multispectral systems being able to also derive surface emissivity [1]. Previous TIR
spectroscopy studies investigated emissivity, cooling rates, surface textures, and compositions
(e.g., [1,3,22]). However, the extreme thermal heterogeneity of active volcanic surfaces can easily
 [23]) that have investigated the influence of spatialtemperatures andtemperatures
spectralwhere resolution where
a molten ona molten lava calculations
these surface
lava surface coolsover cools
(<1450(<1450 K). Therefore,
K). Therefore, simply simply
deconvo dec
currently temperatures
currently available available low where lowaspatial
spatial molten
resolution lava(>500
resolution surface m)cools
(>500 TIR m)datasets
TIRfraction
(<1450 fraction
K).
datasets
that that
Therefore,
are
that has
that
not areablethe
has
simply
notto maximum
the
able maximum
deconvolving
accurately
to temperature
accurately resolve temperature
the molten
resolve withinwithin each eachpixel pixel will improve
will impro th
the maximum fraction
the maximum that
temperature has
temperature the maximum
or representative
or representative temperature
emissivity emissivity uncertainty
within
spectrum. uncertainty
each
spectrum. ThereThereof
pixel the of
emissivity,
will
are relatively the emissivity,
improve
are relatively kinetic
few studies the kinetic
temperature,
few studiesaccuracy temperature,
(i.e., (i.e., and and radiant
and radiant
flux. flux.
[23]) that[23])uncertainty
have
that have ofinvestigated
investigated the emissivity,
the influencethe kinetic
influence of temperature,
spatial
of spatial and spectral and spectral
and Measuring
radiant Measuring
resolution flux.
resolution accurate
on these accurate
on thermal
these
calculationsthermal properties
calculations properties
over over of a molten
of a molten lava surface
lava surface is also is critic
also
temperatures
temperatures Measuring
where where
Remote a molten accurate
Sens. a molten thermal
lava12,surface
2020, lava
193 surface properties
cools cools (<1450(<1450 of a propagation
molten
K). Therefore, propagation
K). Therefore, lava models
simply surface models
simply [15,24].
is
deconvolving [15,24].
also With
deconvolving With
the
critical
the molten increasing
tothe lava
the molten increasing number
flow numberof spectral
of
3 of 21spectral
bands bands
in more in mo
rec
s.
ote2019,
Sens.11, 2019,
x REVISION
11, x REVISION 3 of 213 of 21
fraction propagation
fractionthat has that the hasmodels the [15,24].
maximum maximum Withtemperature
temperature the increasing withinwithin (e.g.,
number
each each (e.g.,
HyTES
pixel of spectralHyTES
pixel [20]
will improve will and
[20]
bands the
improveand
inthe Mineral
morethe Mineral
recent
accuracy
the and
accuracy TIR Gas
and
andimagers Identifier
andGas Identifier
(MAGI) (MAGI)
[25]), [25]),
the radiative
the radia t
uncertainty (e.g.,ofHyTES
uncertainty the [20] andkinetic the kinetic
Mineraltemperature, and Gas Identifier emissivity emissivity
(MAGI) of an of
object’s
[25]), anthe object’s
surface
radiative surface
can becan
temperatureextracted
be extracted andwith increasing
with increasing accuracy accuracy[26–28]. [26–2A
yrently
available available
low spatial lowof emissivity,
the
spatial emissivity,
resolution resolution (>500temperature,
(>500
m) TIR TIRand
m)datasets datasets radiant
thatand areradiant
that flux.
not areable flux.
notto able accurately
to accurately resolve resolve
emissivity
Measuring Measuring lead
of
accurate an to misclassification
object’s
accurate thermal surface
thermal properties can or be
properties over-simplification
extracted
of a spectrum.
molten
of aThere emissivity
with
molten lava emissivity
increasingof can
the then
true can
accuracy then
be
thermal used be
[26–28]. used
with
fractions A with
approaches
[23].
well-constrained approaches
This such
is made such
as linear
worseas linear
spectral
by spectral
deconvolutideconv
mum
maximum
11, x REVISION
temperature temperature or representative
or representative emissivity emissivity spectrum. aresurface
There lava aresurface
relatively is few
relatively also isstudies
3 of
critical
alsostudies
few 21
critical
to lava
(i.e., to flow
(i.e., lava flow
propagation emissivity
propagation models the can
models then
currently
[15,24]. beavailable
[15,24].
With usedWith
the with low
increasing
the approaches
spatial
increasing number resolution
number ofquantitatively
such as
spectral
of quantitatively
(>500linear
spectralbands determine
m)spectral
TIR
bands
in determine
datasets
more in possible
deconvolution
more
recent that possible
are spectral
not spectral
modeling
able end-member end-member
to
to accurately that defines
that defines
resolve the mineralogic
the miner
t])have
that haveinvestigated
investigated the influence the influence of spatial of spatial and spectraland spectral resolution resolution on these on these calculations calculations over over imagers
recent
TIR imagers
TIR
quantitatively determine possible spectral end-member thermal thermal
that fractions
definesfractions[29–32].
the [29–32]. Additionally,
mineralogical, Additionally, kinetic
textural, kinetic
andtemperaturetemperature (and (and
to a lesser
to a lesser
degre
(e.g.,where
mperatures
tures
ble low
(e.g.,
HyTES aHyTES
where
spatial
[20]
molten a theand
[20]
molten
resolution lavamaximum
the
and
surface
lava
(>500
Mineral
the surface
m)
temperature
Mineral
cools and
TIRcools (<1450 Gas
datasets
and (<1450
K). or
Identifier
Gas representative
Identifier
Therefore,
that K). (MAGI)
areTherefore,
(MAGI)
simply
notrequired
ablerequired
to
emissivity
[25]),
simply [25]),
the
deconvolving
accurately
thespectrum.
radiative
deconvolving radiative
resolve
temperature
the molten There are
the temperature
molten andrelatively and few studies
thermal fractions
(i.e., [23]) [29–32].
that have Additionally,
investigated kinetic
the temperature
influence to determine
of improve to
(and[26–28].
spatial determine
to a[26–28]. the
lesser runout
the
degree,runout
distance distance
emissivity) and hazard
and hazard
potential
is calculations potential
using using
radiant radiant
heat f
ction emissivity
that
emperature hasemissivity
that the
hasof maximum
an
the of
object’s
or representative
an object’s
maximum surface
temperature surface
can
temperature
emissivity
becan extracted
within bewithin
spectrum.
extracted
each with
There each increasing
pixelwith
arepixel
increasing
will
relatively
accuracy
improve
will few the and
accuracy
studies accuracy
the spectral
A
(i.e., accuracy and resolution
well-constrained
A well-constrained
and on these
emissivity required
emissivity can to
then
can
over determine
then
betemperatures
used be used the runout
with with
approaches
where distance
approaches
aand such
molten andsuch as
lava propagation
hazard
linear
as
surface propagation
potential
linear
spectral models
spectral
cools using models
deconvolution
(<1450 [15,24].
radiantK).[15,24].
deconvolution However,
heatmodeling
Therefore, However,
flux in
modeling the lava
to
simply accuracy
the
flow
to accuracy
deconvolving of these of these
derived
the derived
parameters parameo
certainty
nty of the
investigated of
emissivity,
thetheemissivity,
influence kinetic kinetic
oftemperature,
spatial temperature,
and spectral and radiant radiant
resolutionflux. flux.
on these calculations over
quantitatively propagation
quantitatively Remote Sens.
determine models
2019,
determine 11,
possible [15,24].
x REVISION
possible However,
spectral the accuracy temperatureof temperature
these range
derived range
of typicalof
parameters typical
lavas lavas
is
over less is
thewellless constrained
coolingwell 3 constrained
of 21 at the at
various
the various
spatial spatial
resolu
asuring
hereMeasuring accurate
a molten accurate
lava molten
thermal
surface thermal fraction
properties
cools properties
(<1450 that ofK). aspectral
has ofend-member
the
molten a molten
Therefore,
end-member
maximum
lava simply
that
surface
lava defines
that
temperature
surfaceis also
deconvolving
defines
isthe mineralogical,
within
critical
also the
the mineralogical,
critical
to
molten each
lava topixelflowtextural,
lava will
flowtextural,
and andthe accuracy and
improve
thermal temperature
thermalfractions fractions range
[29–32]. [29–32].of typical lavas is less well TIR
constrained instruments.
TIR at instruments.
the variousTherefore, Therefore,
spatial improving improving
resolutions the of accuracy
the
current accuracyof the ofkinetic
the kinetic
temperaturetemperature and e
opagation
tion
as the models models
maximum [15,24]. uncertainty
[15,24].
With
temperature the Additionally,
With increasing
the Additionally,
ofwithin
the emissivity,
increasing number kinetic
number
eachresolution
kinetic
of
pixel
temperature
kinetic of temperature
spectral
will temperature,
spectralbands
improve
(and
bands
in the (and
toinand
more a more
lesser
toradiant
recent
accuracy
a recent
lesser
degree,
TIR and
degree,
flux.
imagers
TIR emissivity)
imagers emissivity) is is
required TIR
required currently
instruments.
to determine to Mineral
determine available
theTherefore,
runout
theGas low
runout spatial
improving
distance distance and the
hazard
and (>500
accuracy
hazardpotential m) of TIR
previously-determined
the
potentialusing datasets
previously-determined
kinetic using
radiant that
temperature
radiant are
heatmolten not able
molten
fraction
and to accurately
fraction
emissivity should should
thenresolve
reduce
then
of the to lava flow reduce
the uncertainty
the uncertainty in flow in mo
flo
g., TES
he HyTES
[20]
emissivity, and
[20] the
and
kinetic the Measuring
Mineral
temperature, and and and accurate
Identifier
Gasradiant thermal
Identifier (MAGI)
flux. (MAGI) properties
[25]), [25]),
the of
radiative
the aradiative
moltentemperature lavaflux
temperature heat
surface
and in
flux lava
andin flow
is lava flow
also critical
propagation the maximum
previously-determined temperaturemolten or
fractionrepresentative
should then emissivity
directly
reduce directly
relyspectrum.
thederivedon rely these
uncertainty There
onparameters
these
thermal are
in overthermal
flow relatively
properties.
theproperties.
model few
analyses studies that (i.e.,
of anpropagation
issivity
yaccurate of
object’s
an models
object’s
thermal surface models [15,24].
propagation
surface
propertiescan be [15,24].
can However,
ofmodels
extracted
be However,
a extracted
molten with the increasing
[15,24]. accuracy
with
lava
the With accuracy
the
increasing
surface
of these
accuracy of
increasing these
accuracyderived
[26–28]. number parameters
[26–28]. ofA spectral
A well-constrained well-constrained over
bands cooling
the
in more cooling recent TIR imagers
temperature directly
temperature [23])
range that
rely range
of have
ontypical
these
of investigated
thermal
typical
lavas lavas
is less the
properties.
is well influence
less constrained
well of is
constrained
also
spatial
at the
critical
and
atvarious
the spectral to
various
lava
spatial
flow
resolution
spatial
resolutions on these
resolutions of calculations
currentof current over
issivity
y can[15,24].
odels then
can then be
With used be
(e.g.,
the used
with HyTES
increasing with
approaches approaches
[20]
number and such
ofthe such
asMineral
spectral linearasbandslinear
spectral
and in spectral
Gas
more deconvolution
Identifier
recent deconvolution
TIR (MAGI) modeling
imagers modeling
[25]), tothe to radiative temperature and
TIR instruments.
TIR temperatures
instruments. Therefore, where
Therefore, improving a molten
improving the lava surface
accuracy
the accuracy of cools
the (<1450
3.ofkinetic
Data
the K).
3. kinetic
Data Therefore,
temperature temperature simply
and emissivity
and deconvolving
emissivity of the of thethe molten
antitatively
tively
0] anddeterminethe determine
Mineral possible possible
emissivity
and spectral
Gas ofspectral
anend-member
object’s end-member surface that defines
canthat bedefines the mineralogical,
extracted thetemperature
mineralogical,
with increasing textural,
andtextural, and and
accuracy [26–28]. A well-constrained
3. Data
previously-determined fraction
previously-determined that moltenhasIdentifier
the
molten
fraction maximum (MAGI)
fraction
should should
[25]),
temperature
then reduce
then
the radiative
reduce
the within
uncertainty each
thea uncertainty pixel
in flow will
in model
flow improve
model
analyses thethat
analyses accuracy that and
nrmal
fractions
object’sfractions[29–32].
surface [29–32].
can Additionally,
emissivity
be Additionally,
extracted can kinetic
with then kinetic
temperature
be
increasing temperature
used with
accuracy (and (and
to
approaches
[26–28]. a
3.1. Study lesser
to A
3.1.radiant lesser
degree,
such
well-constrained
Study
Area Area asdegree,emissivity)
linear emissivity)
spectral is is
deconvolution modeling to
directly
uired
to determine
to relyuncertainty
directly
determine onrely
the theseon
runout
the these of the
thermal
runout thermal
distance
emissivity,
properties.
distance and properties.
hazard
and
kinetic temperature,
hazard
potential potential using
and
using
radiant radiant
heat
flux.
flux
heat influx lava in flow
lava flow
then be used with
3.1. StudyMeasuring Areaapproaches
quantitatively such
determine as linear
accurate thermal properties of a This spectral
possible deconvolution
spectral end-member
molten modeling that to
defines the mineralogical, textural, and
opagation
tion models models[15,24]. [15,24].
However, However, the accuracy
the accuracy of these of these
derived derived parameters Thislava
study
parameters study
was
over
surface
conducted
was
over
the
is also
conducted
cooling
the during
cooling
critical twotofield
during lavacampaigns
two flowcampaigns
field at Kīlaueaat KīlaueaVolcano Volcanin H
determine possible spectralmodels
thermal end-member
fractions [29–32].that
Withdefines Additionally, the mineralogical, kinetic of textural,bands
temperature and(and to a lesser degree, emissivity) is
3. Data 3. Datapropagation
This study
[15,24]. the increasing number
January/February
January/February spectral 2017 and
2017 in
2018,
and more 2018,
a recent
period a periodTIR imagers
when when
both the
both summit
the summit
lava lake
lava and
lake coas
and
mperature
ture rangerange
ns [29–32]. ofAdditionally,
typical
of typical lavaswas lavas
is less conducted
iswell wellduring
lessconstrained constrained two
at the field
at
avarious
the campaigns
variousspatial at
spatial Kīlauea
resolutions resolutions Volcano
ofiscurrent of in Hawai’i, USA, in
current
HyTESkinetic
(e.g., required [20]to determine
and temperature
the Mineral the (and runoutandtoGas lesser
distance Identifier
lava the
degree,
and (MAGI)
flows
lava
hazard
flows
emissivity)
were
potential
were[25]),
active. the
active.
Itand
using
radiative
focusedIt of
radiant
focused temperature
primarily
heat flux andin lava flow
Rermine
uments.
instruments. theJanuary/February
Therefore,
runout Therefore, improving
distance 2017
improving
and the and
hazard 2018,
accuracy
the a of
accuracy
potential period
theusing
of when
kinetic
the radiant both
kinetic
temperature summit
temperature
heat flux and inlavaemissivity
and
lava lake
emissivity
flow of coastal
the the plainprimarilysurface on the onlava the lake
lava in lake theinHalemaʻum
the Halem
3.1. Study3.1. Study
Areaemissivity
Areapropagation of an models
object’s surface [15,24]. can be However,
extracted the
with accuracy
increasing of these
accuracy derived
[26–28]. parameters
A well-constrained over the cooling
ly-determined
viously-determinedlava flows
molten were
molten
fraction active. It focused primarily on propagating
the propagating
lava lake lava in flows
lava
the flows
from
Halemaʻumaʻu from
the Puʻuthe Puʻu
‘Ō’ō
Crater vent
‘Ō’ō and (Figure
vent (Figure
1). Kīlauea
1). Kīlauea
is a basaltic
is a basalt shi
odels [15,24]. emissivity
However, temperature thefraction
can
should
accuracy
then range
should
be used
then
of
of typical
reduce
then derived
thesewith lavas
reduce
the uncertainty
approaches
the uncertainty
isparameters
less such
well
in flow over
asbeen
inmodel
linear
constrained
flow model
the spectral analyses
cooling
at the
analyses that that
deconvolution
various spatial modeling
resolutions to ofyears
current
rely
ectlyof on This
rely propagating
onstudy
these This
these
thermal study
was
thermal lava
conducted
was flows
conducted from
during the
during
two Puʻu field
two ‘Ō’ō vent
campaigns
field campaigns has
(Figure been
has 1).
atresolutions
Kīlauea erupting
Kīlauea
at Kīlauea erupting
Volcano nearly
is a
Volcano nearly
continuously
basaltic
in Hawai’i, continuously
shield
in Hawai’i, for
volcano
USA, USA, in the for past
that
in the 500
past 500 years
[33]. The
[33]. lava
The surfac
lava s
nge typical lavas isproperties.
quantitatively
TIR less properties.
well
instruments. constrained
determine possible
Therefore, at the various
spectral
improving spatial
end-member the accuracy that ofthe
defines
of current the mineralogical,
kinetic temperature textural,
and and(tube-
emissivity of the surface-fed)
has been erupting nearly continuously for the pastduring 500 during
long
years sustained
long
[33]. sustained
The eruptions
lava eruptions
surfaces where where
are pāhoehoe
produced pāhoehoe (tube- and surface-fed)
and and ‘a’ā
and flow
‘a
s.January/February
January/February
Therefore, improving
thermal
2017the and
2017
fractions
previously-determined
2018,
and 2018,
accuracy [29–32].
a period
of the a period whenwhen
kinetic
Additionally,
molten
both the
bothsummit
temperature
fraction kinetic
the and
should
summit
temperature
lava
then
lake
emissivity lavaand
reduce (and
lakeof coastal
tothe
the
and coastal
a thelesser
uncertainty
plain plain surface
degree, in
surface
flowemissivity)
model is during
analyses that
Data lava flows
lava during flows
were long were sustained
active. active.
It eruptions
focusedIt focused primarilywhereprimarily pāhoehoe
on the on lava [33].
(tube-
the lake
lavaThe
[33].
and inlava
lake The
the lake
inlava
surface-fed) inlake
Halemaʻumaʻu
the the in
and Halemaʻumaʻu
Halemaʻumaʻu ‘a’ā Halemaʻumaʻu
flows
Crater are
CraterCrater
and emplaced Crater
and observed observed during
this studythis study was active
was
rmined molten fraction
required should
to determine then reduce
the runout the uncertainty
distance and in flow
hazard model potential analyses using that radiant heat flux in lava flow
directly rely on these thermal properties. 2018 and
2018 varied
and varied
in
propagating
these
[33]. The
propagating
thermal lavalava
properties. flows
lava lakeflows
from in the theHalemaʻumaʻu
from Puʻuthe Puʻu ‘Ō’ō vent ‘Ō’ōCrater (Figure
vent (Figure observed
1). Kīlauea
1). during
Kīlauea is athis basalticasize
is study inwith
basaltic size
was maximum
shield with
active
shield
volcano maximum
from
volcano dimensions
that 2008that dimensions
until of 160ofmeters 160 meters wide wide and 225 andmet 22
propagation models [15,24]. However, the accuracy 1) [34]. 1) of these
During
[34]. Duringthis derived
period,
this parameters
period,
there there
were over
were
fluctuations the cooling
fluctuations in lava in lake
lava activity
lake activity
with continu
with co
yStudy
Area
has been Area2018
has been and
erupting varied
erupting nearlynearlyin size
continuously with
continuously maximum for the dimensions
forpast the 500 pastyears of
500 years 160
[33]. Themeters
[33].lava wide
The surfaces and
lava surfaces 225 meters
are produced long
are produced (Figure
temperature 3. Datarange of typical lavas is less well and constrainedirregular
and atsmall
irregular the varioussmall
explosions, spatial
explosions, resolutions
finally finally
ending ofwith
ending current the
with summit
the summit
collapse collapse
in May in 201
Ma
during 1)
during [34].
longconducted During
sustained
longconducted
sustained this
eruptions period,
eruptions there were fluctuations in lava lake activity with continuous gas plumes
s study
This study was was
TIR instruments. during twowhere
during
Therefore, twowhere
field pāhoehoe
campaigns
field
improving
pāhoehoe
campaigns (tube-
at (tube-
theKīlauea
and
at Kīlauea
accuracy
surface-fed)
and
Volcano surface-fed)
of the Volcano
the in and in‘a’ā
Hawai’i,
kinetic
and flows
Hawai’i, ‘a’ā flows
USA,
temperature
arein emplaced
USA, and
are emplaced
inemissivity of thelevel
[33]. The
[33].andlava
The irregular
lake
lava inlake small
the in explosions,
Halemaʻumaʻu
the Halemaʻumaʻu finally
Crater ending
Crater
observed with
observed time
during time
of
theduringsummit
this of
study
this field
the
collapse
study
was field
campaigns,
in campaigns,
active
was May
active
from 2018the
2008
from lava
the
[13,34].
until
2008 lava
lake
At the
until lake
level was was relativelyrelatively high high but nb
February
uary/February 3.1.
2017previously-determined
and
2017 2018,
and Study
2018,
a period Area
a period when molten when
bothfraction the
bothsummit theshouldsummit lava then lake
lava
reduce and
lakecoastal
theand coastal
plain plain
uncertainty surface
in surface
flow model analyses that
time ofvaried thesize field sizecampaigns, the lava lake approximately
level approximately
was relatively 100–130 100–130
highmeters butmeters
below below
not(Figure the Halemaʻumaʻu
overflowing, the Halemaʻumaʻu CraterCraterrim. Itrim. wasItcontinuo
was con
ws a 2018
flows
were and
2018were varied
and
active. Itinfocused
active.
directly Itinfocused
rely
with
onprimarily
maximum
with
these
maximum
primarily
thermal ondimensions
the onlavadimensions
properties.the lake lava of 160 in
lakeofthe
meters
160 meters
wide
inHalemaʻumaʻu
the wide
and
Halemaʻumaʻu 225and meters
Crater 225Crater meters
andlong andlong (Figure
1) [34]. 1) approximately
During
[34]. Duringthis period,
thisThis
100–130
period,study
there meters
there
werewaswere conducted
below
fluctuations during
the Halemaʻumaʻu
fluctuations in lava with
in laketwo
lava fresh
with field
Crater
activity
lake fresh
lava campaigns
activity upwelling
lavaIt
rim.
with upwelling at
in
wascontinuous
continuous
with Kı̄lauea
the
continuously in
north
gas theplumesVolcano
north
gasthat thatin
migrated
circulating
plumes Hawai’i,
migratedto theto USA,
south,
the south,
cooled cooled
and formedand fo
opagating
ting
waslava flows
conducted lava flows from
during from
thetwo Puʻuthefield Puʻu
‘Ō’ōcampaigns
vent
‘Ō’ō (Figurevent at (Figure1). Kīlauea
Kīlauea 1).Volcano
Kīlaueais a basalticis Hawai’i,
in a basaltic shield shield
USA, volcano in volcano that that
and irregular with
andnearly fresh
irregular in lavaJanuary/February
upwelling in the 2017
north and
that 2018,
migrated solidified
a period
to solidified
the lava.
when
south, The
lava.
both
cooled lava
The the
andthen
lava summit
formedsank
then in
sank the
lava
plates insouth,
the
lake
of south,
cooler, distinguished
and distinguished
coastal by
plain the byoccurrence
the occurrenceof st
n
ary been
erupting
2017erupting
and 2018, 3. small
nearly
Data small
a period
explosions,
continuously explosions,
continuously
when for finally
both the for finally
ending
past
the the 500 ending
past
summit
with
years
500 the
with
lavayears[33]. summit
lakeand
the
The
[33].
and
summit
lavacollapse
The
coastal
collapse
surfaces
lava plain
in May
surfaces arein produced
surface
2018
May [13,34].
2018 [13,34].
are produced At theAt the
time time solidified
ofsustained
the oferuptions
field lava.
surface
theeruptionsfield The
campaigns, lava lava
campaigns, then
flows the sank
were
lava
the in the
active.
lava
lake south,
It focused distinguished degassing
and
primarily degassing
by [34].
on the the The
[34].
occurrence
lava lava
Thelake flows
lava
of
in flows
from
strong
the the
from Puʻu
splattering
Halema the uma‘Ō’ō
Puʻu u vent
‘Ō’ō
Craterwere
vent andactive
were active
for ~30for years
~30 ye
ong
ring
re sustained
long
active. It focused where
primarily where
pāhoehoe
on the pāhoehoe (tube-
lava andlake
(tube-
lake
level level
insurface-fed)
and was was
the surface-fed)
Halemaʻumaʻu
relatively
andrelatively
‘a’ā
andflows high
‘a’ā flows
Crater
high
butemplaced
are and
not
butemplaced
are overflowing,
not overflowing,
approximately and 3.1.
approximately degassing propagating
100–130 [34].
100–130 meters The lavaflows
lava
meters
below flows
the from
below from
Halemaʻumaʻu
the duringthe Pu
the Puʻu
Halemaʻumaʻu u ‘Ō’ō locations
‘Ō’
Craterō ventlocations
vent
Crater
rim.wereover
(Figure
Itwas
rim.was over
numerous
active 1).
It numerous
for ~30
Kı̄lauea
continuously
was eruptive
years
continuously eruptive
erupting
is acirculating
basaltic episodes, episodes,
in many
shield producing producing
volcano that has mostly mostly
pāhoehoe pāhoehoe lava
va].lava
The
flowslake
lava in
from lake
thethe in Study
Halemaʻumaʻu
the
Puʻu Area
Halemaʻumaʻu
‘Ō’ō vent Crater
(Figure Crater
observed 1).observed
Kīlauea is during
this
a basalticstudy
this study
was
shield activevolcano active
from 2008
from
that until
2008 untilcirculating
with fresh
with locations
fresh
lava over
been
upwelling
lava numerous
erupting
upwelling in the innearly
north
theeruptive
north
thatcontinuouslyepisodes,
migrated
that migrated to for
the occasional
producing
the
tosouth,
the occasional
past south,500
cooled ‘a’ācooled
mostly yearslava
and ‘a’ā flow
lava
pāhoehoe
[33].
formed
and [33,35].
flow
The
formed lava
plates [33,35].
lava Thisflows
surfaces
plates
of long
This
cooler,of with eruption
longproduced
are
cooler, the eruption
finally finally
endedended
during in 2018 in with
2018 the wit
ding8varied
andnearlyvaried
in size inwithsize maximum
continuously withfor maximum
the dimensions
past dimensions
500 years of 160 ofmeters
[33]. 160Themeters widesurfaces
lava wide
and 225 andare meters
225 meters
produced long (Figure long (Figure
solidified occasional
solidifiedlava. TheThis
lava. ‘a’ā
lava study
The lavathen
lava was
flowsank
then conducted
[33,35].
in
sank the in This
south,
theduring
long
south, two
eruption
distinguished field
lower
distinguished campaigns
lower
East
finally
by the by Rift
East
ended
occurrence
the atRift
Zone Kīlauea
in
occurrenceZone
(LERZ)
2018 of Volcano
(LERZ)
with
strong oferuption
the
strong in splattering
eruption Hawai’i,
inflows
cessation
splattering thein USA,
Leilani
ofthe theLeilani in Estates
Estates [14]. The
[14].lava Theflows lava flowsobse
During
34]. During this period,
this period, long sustained
therepāhoehoethere
were were eruptions
fluctuations fluctuations where
in surface-fed)
lava in lake pāhoehoe
lavaactivitylake activity (tube-
with and
continuous
with surface-fed) and ‘a’ā are emplaced [33].
stained eruptions where
January/February 2017 (tube- and and
2018, ainperiod whenand ‘a’ā
both flows
the arecontinuous
summit emplaced gas plumes
lava lake
gas plumes
and coastal
dgular and
irregular
lower
degassing
and degassingEast
[34]. The Rift
The
[34]. Zone
lava lava
The lake (LERZ)
flows
lava in flows
from
the eruptionthe
from
Halema Puʻuthe uma the
‘Ō’ō
Puʻu Leilani
u vent
‘Ō’ō
Crater 2017–2018
Estates
were
vent 2017–2018
were
active
observed study
[14].active
for Thestudy
~30
during were
forlava
years
~30were
part
thisflows
years ofpart
erupting
study the of61g
observed
erupting
was inthe many
activeinplain
episode
61g
during manyepisode
from
surface
that
our 2008erupted
that erupted
until from
2018 fromthe east theflankeast flanof P
ake in small small
explosions,
the Halemaʻumaʻu lava
explosions,
flows
finally
Crater
were
finally
ending
observed
active.
ending
with
It
the
withsummit
during
focused
the summit
this study
primarily
collapse was
flowson
collapsein May
active
the
flows
propagated
in
lavafrom 2018
May
lake
propagated 2008 [13,34].
2018
down in until
the
[13,34].
down
the
At the At
Halemaʻumaʻu
Pulama
the
the
Pulama pali andpali Crater
entered
and and
entered
the ocean
the ocean
at Kamokuna
at Kamokuna as a ser
a
locations 2017–2018
locations over campaigns,
over study
numerous
and numerousvaried were eruptivepart
in size of
eruptive withthe
episodes, 61g
episodes,
maximum episode
producing that
producing
dimensions erupted
mostly mostly
of from
pāhoehoe
160 the
pāhoehoe
m wide east
lava flank
and lava
flows of
225flows mPuʻu
with long ‘Ō’ō.
with
the (Figure These
the 1) [34]. During
dethe inofsize
field
the with field
campaigns,
maximum
propagating
the
dimensions
lava
lava
the
flows
lava
lake of
from
lake
level
160 the
level
meters was
Puʻu
was
relatively
wide‘Ō’ō andrelatively
vent
surface225 highmeters
(Figure
surface
lava
high
but
1). long
flow
lava
not
but
Kīlauea and
flow
overflowing,
not
(Figure is
lava
and
overflowing,
a basaltic
tubes
lava tubes shield
(Figure (Figure1)volcano
[35]. 1) that
The
[35]. majority
The majority
of the offlows
the flows
observe ob
occasional flows
occasional ‘a’āpropagated
lava
‘a’ā flow
lava down
[33,35].
flow the
[33,35].
This Pulama long pali and
Thisfluctuations
eruption
long eruption entered
finally thelake
finally
ended ocean atin Kamokuna as a series ofof pāhoehoe
mately
proximately
this period, 100–130 100–130
there meters this
were meters period,
below thethere
below
fluctuations inwere
Halemaʻumaʻu
the Halemaʻumaʻu
lava lake activityCrater Craterin lava
rim.
with wasItended
Itrim.
continuous
inactivity
2018
continuously
was with
2018
continuously
gas with
plumes
the
with
circulating cessation
the
continuous
circulatingcessation gas the of
plumesthe and irregular
lower lowersurface
East hasZone
Rift
East been
lava
Rift erupting
flow
Zone
(LERZ) and
(LERZ) nearly
lava
eruption tubes
eruption continuously
in (Figure
the inLeilani
the 1) for
[35].the
Leilani
Estates The
Estatespast
pāhoehoe,
[14]. 500
pāhoehoe,
majority
The
[14]. years
both
lava
Theof flows
the[33].
both
sheet-like
lava flows Theobserved
sheet-like
flows
observed lava
and
observed surfaces
ropey
andwere
during ropey
in
during
our are
texture.
in
tube-fedproduced
our texture.
Previous Previous
remote remote
sensing sensing
studies stu
h fresh
mall lava upwelling
lava upwelling
explosions, insmall
finally thein explosions,
north
the north
ending that with finally
migrated
that the migrated
summit ending
to thecollapse
to with
south,
the south,the
cooled
in Maysummit
cooled
and 2018 collapse
formed
and formed
[13,34]. plates in plates
At May
of cooler,
the 2018
of cooler,[13,34]. At the time of the field
2017–2018 pāhoehoe,
2017–2018 during
study study
were long
bothwere sustained
sheet-like
part ofpart the of eruptions
and
61g the ropey
episode
61g wherein
episode that pāhoehoe
texture.
erupted
that Kīlauea
Previous
erupted (tube-
from Kīlauea from
the and
eruption
remote
east surface-fed)
eruption
the have
sensing
flank
east have
flank
of and
investigated
studies
Puʻu of ‘a’ā
investigated
Puʻu
‘Ō’ō. flows
oflava
‘Ō’ō.
These are
the discharge
lava
pre-2018
These emplaced
discharge rates rates
[36], lava
[36], flow
lava emplaceme
flow empla
didified
eldlava. The
lava.lava
campaigns, Thethen lava sank
then
lavain
campaigns,
the sank thein
lake south,
the the
levellava south,
distinguished
lakedistinguished
was level was
relatively by the by occurrence
relatively
high thebut occurrence
high
not of strong
but
overflowing,ofnot strong
splattering splatteringapproximately 100–130 m
overflowing,
flows flowsKīlauea
propagated [33].eruption
propagated Thedown lava down
thelake
have in
Pulama
the the
investigated
Pulama Halemaʻumaʻu
pali and
pali lava
entered
and Crater
discharge
enteredthe and
oceanobserved
rates
the lava
and
ocean
at [36],
Kamokunaduring
pathway
lavaat lavapathway
Kamokuna this
mechanisms
flow as astudy
mechanisms
emplacement
series
as awasseries active
of[38].
pāhoehoe The
of[38]. from
tracking
pāhoehoe The2008
areas for until
areas
[37], thisforstudy
this study
were were chosen chosen
for the forhig th
dassing
degassing
100–130 [34].metersThe
[34].lava The
below flows
belowlava the flows
from
the the
from
Halema
Halemaʻumaʻu Puʻuthe uma ‘Ō’ō
Puʻu uvent
Crater ‘Ō’ō
Crater were
vent
rim. Itwere
rim. active
was It wasactive
for ~30 foryears
continuously
continuously ~30 years erupting
circulating erupting
circulatingin many in many
with fresh lava(Figureupwelling in the
surface and 2018
lava and
pathway varied in
mechanisms size with maximum
[38]. The areas dimensions
for observing
this of
observing
study 160molten
were meters moltenlava
chosen wide for and
surfaces
lava the 225
surfaces
high meters
combined combined
probability long
with with
the
of availability
the availability
of a variety
of a variety
of rem o
sationsover surface
overlavanumerous
numerous flow
lava eruptive
and
flowlava
north
andepisodes,
eruptive
that
tubes
lava episodes,
migrated
tubes
(Figure
to to
(Figure
producing
the
1)producing
[35].
south,
1) The[35].majority
mostly
cooled
The
mostly majority
pāhoehoe
and
of theof
pāhoehoe
formed
flows
lava thelava
plates
flows
observed
flows of flows
observed
with
cooler,
were
with
the were
tube-fed
solidified the tube-fed lava.
upwelling
pāhoehoe, pāhoehoe,
in 1)
observing
the[34].
both
north
bothmolten
sheet-like
that
Duringsheet-like
migrated
this
lava
and period,
surfaces
ropey
and
the
ropeythere
in
south,
combined
texture.
in were cooled
texture.fluctuationsand
withPrevious
Previous the
formed
datasets. in remote
availability
remote lava
datasets. plates lake
sensing
of
ofsensing
cooler,
aactivity
variety
studies with
studies of the
of continuous
remote
of thesensing
pre-2018 pre-2018 gasThe plumes
TIR
lava then sank
al
asional
The ‘a’ā
lava lava
‘a’ā
then flow
lavasank [33,35].
flowin [33,35].
the
in the south, This
south, long
This eruption
long
distinguished
distinguished eruption finally
by
by the finally
the ended
occurrenceended
occurrence in 2018 in
of with
2018
strong
of strong the
with cessation
the
splattering
splattering cessation of the of the
datasets. and irregular small explosions, finally ending withlava the summit collapse inand Maydegassing 2018 [13,34]. [34]. The lava flows
[37], At the
wer stKīlauea
34]. Rift
East
The
Kīlauea
Zone eruption
Rift
lavaZone (LERZ)
flows
eruption have
(LERZ)eruption
from
from
have
investigated
the
theeruption
Puʻu
Pu
investigated
in uthe‘Ō’ōin
‘Ō’
lavaLeilani
Leilani
ōthe
vent
ventwere
discharge
lava
Estates
were
discharge
Estates
active[14].
active
rates
forThe
[14].
for
rates
[36],
~30 lava
~30The
years
[36],
flows
lava
years
flow
lava
flows
eruptingobserved
erupting
emplacement
flow observed
in many
emplacement
induring
many duringtracking
our
locations
tracking
our [37],
over numerous eruptive
and lava
and time
pathway
lava pathwayof the
mechanisms field
mechanisms campaigns,
[38]. The
[38]. areas
The the areas
for lava
this lake
forstudy this level
study
were was
were
chosen relatively high but not overflowing,
18 7–2018
study
numerous study
were were
part
eruptive of
part
episodes,theof
episodes,61g the episode
61g
producing episode
producing thatmostly erupted
that
mostly erupted
pāhoehoefrom
pāhoehoe from
the lava eastthe
lava
flowsflank
east flows ofchosen
flank
with
forofthe
Puʻu
with
the
forhigh
Puʻu
‘Ō’ō.
the
occasional
the‘Ō’ō.
Theseprobability
high These
‘a’ā
probability
lava
of
flow
of
[33,35]. This long
observing observing approximately
molten molten
lava surfaces
lava 100–130
surfaces combinedmeterscombined below
with the
with
the ocean Halemaʻumaʻu
availability
the availability of as Crater
a variety
ofseries
a varietyrim.
of It
remote was
of remotecontinuously
sensing sensingTIR TIR circulating
wsopagated
lava propagated
flow down [33,35]. down
the ThisPulama
the
eruption longPulama pali
eruption andpali entered
and
finally enteredthe
ended ocean
the in at
2018 Kamokuna
at
with Kamokuna
the a
cessation as a series
of
of pāhoehoe
the of pāhoehoe
datasets. datasets. with fresh lava finallyupwelling ended in the in north
2018 with the cessation
that migrated to theofsouth, the lower cooledEast andRift formed Zone (LERZ)
plates eruption in the
of cooler,
ava
face
Zoneflow lava
(LERZ) and
floweruption
lava
and tubes lavaintubes (Figure
the (Figure
Leilani 1) [35]. 1) The
Estates [35]. majority
[14]. The The majorityof the offlowsthe flows
observed observed wereour were
tube-fed tube-fed
solidified Leilani lava. The
Estates lava
[14]. then Thesank lava thelava
inflows south,
observedflows observed
distinguished
during our during
by2017–2018
the occurrence studyofwere strong part splattering
of the 61g episode
hoehoe,
e, were
y both part both
sheet-likeofsheet-like and
thedegassing
61g eruptedropey
and ropey
episode in texture. in texture. Previous Previous remote remote sensing sensing studies studiesof the ofpre-2018
the pre-2018
and that [34].that Theerupted
from lava
theflows east from from
flank the ofeast
the Puʻu
Pu flank
u ‘Ō’ō‘Ō’of ō.ventPuʻu
These were ‘Ō’ō. active
flows These for ~30 yearsdown
propagated erupting theinPulama many pali and
auea
eruption eruption
ed down the Pulama have have
investigated investigated
pali and lava entereddischarge
lava discharge
the rates
ocean atrates
[36],
Kamokuna lava
[36], flow
lava as emplacement
flow
a series emplacement
of pāhoehoetracking tracking [37], [37],
locations over numerous eruptive episodes,
entered the ocean at Kamokuna as a series of pāhoehoe surface lava flow and lava tubes (Figure producing mostly pāhoehoe lava flows with the 1) [35].
dw pathway
lava
andpathway
lava mechanisms
tubes mechanisms
(Figure [38].1)The [38].
[35]. areas
The
The areas
for
majoritythisforstudythis
of study
the were flows were
chosen chosen
observed for the forhigh
were thetube-fed
probability
high probability of of
occasional The majority ‘a’ā lava of flow the[33,35].flows observed This long eruption were tube-fed finally pāhoehoe,ended in 2018 both with the cessation
sheet-like and ropey of thein texture.
serving
h g sheet-like
molten molten
lava and surfaces
lavaropey surfacescombined combined with with
the availability
the availability of a variety of a variety of
of remote of remote sensing sensing TIR TIR
lower EastinRift
Previous
texture.
Zone (LERZ)
remote
Previous
sensing
remote
eruption
studies insensing
thethe
of Leilani studies
pre-2018 Estates the
[14].pre-2018
Kı̄lauea The lava flows
eruption have observed
investigated duringlava our discharge
on asets.
have investigated lava discharge rates [36], lava flow emplacement tracking [37],
2017–2018 study were part of the 61g episode
rates [36], lava flow emplacement tracking [37], and lava pathway mechanisms [38]. The areas for that erupted from the east flank of Puʻu ‘Ō’ō. These
way mechanisms flows [38]. The areas down
propagated for thisthe studyPulama were chosen for the high probability of
this study were chosen for thepali high and entered
probability the ofocean
observing at Kamokuna molten lava as a seriessurfaces of pāhoehoe
combined with the
en lava surfaces surface combined lava flow with and thelava availability
tubes (Figure of a variety 1) [35].of The remotemajority sensing of the TIR flows observed were tube-fed
availability of a variety of remote sensing TIR datasets.
pāhoehoe, both sheet-like and ropey in texture. Previous remote sensing studies of the pre-2018
Kīlauea eruption have investigated lava discharge rates [36], lava flow emplacement tracking [37],
and lava pathway mechanisms [38]. The areas for this study were chosen for the high probability of
observing molten lava surfaces combined with the availability of a variety of remote sensing TIR
datasets.
RemoteRemote
Sens. 2019,
Sens.11,
2019,
x REVISION
11, x REVISION
Remote Sens. 2019, 11, x REVISION 3 of 213 of 21 3 of 21
Remote Sens. 2020, 12, 193 4 of 21
currently
currently
available
available
low spatial
Remote low spatial
Sens. currently
resolution
2019, resolution
11, available
(>500(>500
x REVISION m)low TIRm)spatial
datasets
TIR datasets
resolution
that are thatnot
(>500
areable
not
m)to able
TIR
accurately
datasets
to accuratelyresolve
that are
resolve
not able to accurately
4 of 21 resolve
the maximum
the maximum temperature
temperature or representative
theormaximum
representative emissivity
temperature
emissivity
spectrum.
or spectrum.
representative
ThereThereare emissivity
relatively
are relatively
few
spectrum.
studies
few studies
There
(i.e., (i.e.,
are relatively few studies (i.e.,
[23]) [23])
that have
that have
investigated
investigatedthe[23])
influence
thethatinfluence
have
of spatial
investigated
of spatial
and spectral
andthespectral
influence
resolution
resolution
of spatial
on theseonandthese
calculations
spectral
calculations
resolution
over overon these calculations over
temperatures
temperatureswherewherea moltena molten
lava
temperatures
surface
lava surfacecools
wherecools
(<1450a molten
(<1450
K). Therefore,
K).
lava Therefore,
surfacesimplycools
simply
deconvolving
(<1450
deconvolving
K). Therefore,
the molten
the molten
simply deconvolving the molten
fraction
fraction
that that
has thehas maximum
the maximum
fraction
temperature
that
temperature
has within
the within
maximum
each eachpixeltemperature
pixel
will will
improve improve
withinthe each
accuracy
the accuracy
pixeland willandimprove the accuracy and
uncertainty
uncertainty
of theofemissivity,
the emissivity,kinetic
uncertainty
kinetic
temperature,
temperature,
of the emissivity,
and radiant
and radiant
kinetic
flux. flux.
temperature, and radiant flux.
Measuring
Measuringaccurateaccurate
thermal thermal
properties
Measuring
propertiesofaccurate
a ofmolten
a molten
thermal
lava lava
surface
properties
surface
is also
ofis acritical
also
moltencritical
tolava
lava
tosurface
lava
flow flow is also critical to lava flow
propagation
propagation
models models
[15,24].[15,24].
Withpropagation
With
the increasing
the increasing
models number
[15,24].
number
of With
spectral
of spectral
thebands
increasing
bands
in more number
in more
recent of
recent
TIR
spectral
imagers
TIR bands
imagers in more recent TIR imagers
(e.g., (e.g.,
HyTES HyTES
[20] and
[20] the
andMineral
the(e.g.,
Mineral
and
HyTES Gas
and[20]Identifier
Gasand Identifier
the(MAGI)
Mineral
(MAGI)[25]),
and[25]),
the
Gasradiative
the
Identifier
radiative
temperature
(MAGI)
temperature
[25]),
andthe and radiative temperature and
emissivity
emissivity
of an ofobject’s
an object’s
surface surface
emissivity
can be can
extracted
be
of anextracted
object’s
with withincreasing
surface
increasing
canaccuracy
be extracted
accuracy
[26–28].[26–28].
with A increasing
well-constrained
A well-constrained
accuracy [26–28]. A well-constrained
emissivity
emissivity
can thencan then
be used be used
with
emissivity
with
approaches
approaches
can then such besuch
as
usedlinear
aswith
linear
spectral
approaches
spectral
deconvolution
deconvolution
such as linear modelingmodeling
spectral
to deconvolution
to modeling to
quantitatively
quantitatively
determine
determinepossible
quantitatively
possible
spectralspectral
end-member
determine
end-member possible
that defines
thatspectral
defines
the mineralogical,
end-member
the mineralogical, that
textural,
defines
textural,
and theandmineralogical, textural, and
thermalthermal
fractions
fractions
[29–32].[29–32].
Additionally,
thermal
Additionally,
fractions
kinetic kinetic
[29–32].
temperature
temperature
Additionally,
(and (and to a kinetic
lesser
to a lesserdegree,
temperature
degree,
emissivity)
emissivity)
(and tois a lesser
is degree, emissivity) is
required
required
to determine
to determinethe runout
therequired
runout
distancedistance
to determine
and hazard
and hazard
thepotential
runout
potential
distance
usingusing radiant
and radiant
hazard
heat flux
heat
potential
in
flux
lava
inusing
flow
lava flow
radiant heat flux in lava flow
propagation
propagationmodels models
[15,24].[15,24].
However,
propagation
However, the models
accuracy
the accuracy[15,24].
of theseof
However,
these
derived derived
the
parameters
accuracy
parameters over
of theseover
the cooling
derived
the coolingparameters over the cooling
temperature
temperature
rangerangeof typical
of typical
lavas
temperature
lavas
is lessiswell
less
range
constrained
wellofconstrained
typicalatlavas
theatvarious
istheless
various
well
spatial
constrained
spatial
resolutions
resolutions
at of
thecurrent
various
of currentspatial resolutions of current
TIR instruments.
TIR instruments. Therefore,
Therefore,
improving
TIRimproving
instruments.
the accuracy
the Therefore,
accuracyof theofimproving
kinetic
the kinetic
temperature
thetemperature
accuracy andofemissivity
and
the emissivity
kineticof temperature
theof the and emissivity of the
previously-determined
previously-determined molten molten
previously-determined
fraction fraction
should should
then reduce
thenmolten
reduce
thefraction
uncertainty
the uncertainty
should in flow
then
in model
flow
reduce model
analyses
the uncertainty
analyses
that that in flow model analyses that
directly
directly
rely onrelythese
on these
thermal thermal
properties.
directly
properties.
rely on these thermal properties.

3. Data
3. Data 3. Data

3.1. Study
3.1. Study
Area Area 3.1. Study Area
This study
This studywas conducted
was conducted duringThisduring
two
study field
two wasfield
campaigns
conducted
campaigns atduring
Kīlauea
at Kīlauea
twoVolcano
field
Volcanocampaigns
in Hawai’i,
in Hawai’i,
atUSA,
Kīlauea USA,
in Volcanoin in Hawai’i, USA, in
January/February
January/February 2017 2017
and 2018, andJanuary/February
2018,
a period a periodwhenwhen 2017
both both and
the summit
2018,
the summit
a period
lava lake lava
when and
lakeboth coastal
andthe coastal
plain
summit plain
surface
lavasurface
lake and coastal plain surface
lava lava
flowsflowswere wereactive. active.
It focused
lava
It focused flows
primarily primarily
were on active.
theon lavaIt
thefocused
lava
lake lakeinprimarily
thein Halemaʻumaʻu
the on Halemaʻumaʻu
the lavaCrater lakeCraterin
and theand Halemaʻumaʻu Crater and
propagating
propagating lava flows
lava flowsfrom from the
propagating
Puʻu
the Puʻu ‘Ō’ōlava vent
‘Ō’ōflows
vent
(Figure (Figure
from 1). theKīlauea
1).Puʻu
Kīlauea
is‘Ō’ō
a basaltic
isvent
a basaltic
(Figure
shieldshield1).
volcano
Kīlauea
volcanothat is athatbasaltic shield volcano that
has been
has been
erupting erupting
nearly nearly
continuously
has
continuously
been erupting
for the forpast
the
nearly 500
pastcontinuously
years
500 years [33]. [33].
The forlava
The
the surfaces
lava
past surfaces
500 are
yearsproduced
are
[33].produced
The lava surfaces are produced
duringduring
long sustained
long sustained eruptions eruptions
during
where where
longpāhoehoe
sustained
pāhoehoe (tube-eruptions
(tube-
and surface-fed)
and
wheresurface-fed)
pāhoehoe
and ‘a’ā and (tube-
flows
‘a’ā flows
and
are emplaced
surface-fed)
are emplaced and ‘a’ā flows are emplaced
[33]. The
[33].lava
The lake
lavain laketheinHalemaʻumaʻu
the Halemaʻumaʻu
[33]. The lava Crater lake
Crater
observed
in the observed
Halemaʻumaʻu
during during
this studythis
Crater
study
wasobserved
active
was active from
during
from
2008this until
2008 studyuntilwas active from 2008 until
2018 2018
and varied
and varied in size inwith
size withmaximum
2018maximumand varied
dimensions dimensions
in size ofwith160ofmeters
maximum
160 meters widedimensions
wide
and 225 andmeters 225of 160
meters
long
meterslong
(Figurewide
(Figure and 225 meters long (Figure
1) [34].
1) During
[34]. During this period,
this period, there
1) there
[34].
wereDuring were
fluctuations
fluctuations
this period,
in lava inthere
lake
lava wereactivity
lake activity
fluctuations
with with continuous in
continuous
lava gas
lakeplumes
gas
activityplumes with continuous gas plumes
and irregular
and irregular smallsmall
explosions,
explosions,
and finally
irregular
finally
ending ending
smallwithexplosions,
with
the summit the summitfinally
collapse collapse
ending in May with
in May2018
the 2018
summit
[13,34].[13,34].
At
collapse
the At the in May 2018 [13,34]. At the
time time
of the of field
the field
campaigns,
campaigns,
time the oflavathethelava lake
fieldlake
level
campaigns,
level
was was relatively
therelatively
lavahigh lakehigh but level not
butwas overflowing,
notrelatively
overflowing, high but not overflowing,
approximately
approximately 100–130 100–130
meters meters
approximately
belowbelow the Halemaʻumaʻu
the100–130
Halemaʻumaʻu meters Crater below
Crater
rim.the It
rim.was
Halemaʻumaʻu
It continuously
was continuously Crater
circulating
circulating
rim. It was continuously circulating
with with
freshfresh
lava upwelling
lava upwelling Figure
Figure in the 1.north
with
1.in Advanced
the fresh
north
Advanced that lava Spaceborne
migrated
that upwelling
migrated
Spaceborne to Thermal
theintosouth,
the Emission
thenorth
Thermal south,
cooledthat
Emission and
cooled
and Reflection
migratedformed
andandReflection
formed Radiometer
toplates
the south,
plates
of cooler,of(ASTER)
cooled
Radiometer cooler, and visible
formed
(ASTER) false-
plates of cooler,
visible
solidified
solidified
lava. lava.
The lava colorsank
Thefalse-color
then
lava image
then
solidified (RBG: lava.3,2,1)
south,
the The of
south, the
distinguished
lava southeastern
distinguished
then sank by the
in region
by
theoccurrence
the of
south, the
occurrence Island
image (RBG: 3,2,1) of the southeastern region of the Island of Hawai’i, showing the analysis splattering
sank
in the in distinguished
of strong
ofof Hawai’i,
strong
splattering
by showing
splattering
the occurrence the analysis
of strong
and degassing
and degassing [34]. The locations
[34].locations
lava
Theflowslava
and of
offlows
fromtheHalema
degassing
the Halemaʻumaʻu
from
the Puʻu the
[34].Puʻu
‘Ō’ō
uma The Crater
u vent
‘Ō’ō
lava
Crater were
vent lava
flows
lava lakeactive
were
active
from
lake and
the Puʻu
for ~30
and Puʻu
Pu foru ‘Ō’ō
years
~30
‘Ō’ō lava
ventflows
‘Ō’ōyears
erupting
lava inatactive
erupting
were
flows at Kīlauea
manyin many
Kı̄laueafor Volcano
~30 years
Volcano (white
erupting in many
(white
locations
locations
over over numerous boxes).
numerous
boxes). eruptive
TheThe
locations ASTER
eruptive
ASTERepisodes,
over data
episodes,
data were
numerous
were acquired
producing producing
acquired on
eruptive
mostly March
mostly
on March 7, 2017,
episodes,
pāhoehoe pāhoehoe
7, 2017, at 10:06:02
atproducing
lava
10:06:02lava HST.
flows Insert
flows
HST.mostly
with
Insert shows
with
thepāhoehoe
shows the entire
lavastate
thethe entire flows with the
state
occasional
occasional
‘a’ā lava
‘a’ā flow
lavaofof theHawaiian
flow
[33,35].
the Hawaiian
occasional
[33,35].
This long island
This ‘a’ā
island long chain
eruption
lava
chain inthe
eruption
flow
in the[33,35].
finally central
finally
central ended Pacific
This
ended
inlong
Pacific Ocean,
2018 the
ineruption
Ocean, 2018
withthe redwith box
thefinally
red indicating
cessation
the
box cessation
ended
indicatingof thethe
inthe area
theof
of2018
area ofthe
with theASTER
the cessation of the
ASTER
lowerlower
East Rift
EastZoneRift Zone
(LERZ)image
image (LERZ) shown.
eruption
lower
shown. eruption
East
in the RiftinLeilani
Zone
the Leilani
(LERZ)
Estates Estates
eruption
[14]. The[14].in lava
Thetheflows
lava
Leilani flows
observed
Estates
observed
during
[14]. during
The our lava our flows observed during our
2017–2018
2017–2018
studystudy were werepart of part
2017–2018
theof61g theepisode
61gstudy episode
that
wereerupted
that
parterupted
of from
the 61g from
theepisode
east
the flank
east that flank
oferupted
Puʻuof Puʻu
‘Ō’ō.
from ‘Ō’ō.
These
theTheseeast flank of Puʻu ‘Ō’ō. These
flowsflows
propagated
propagated 3.2.
3.2.
downdown Remote
Remote
the Pulama
the Sensing
Sensing
flows Pulama Data
Data
propagated
pali and palientered
and downentered
thetheoceanthe
Pulama ocean
at Kamokuna
paliat Kamokuna
and entered as a series asthe
a series
ocean
of pāhoehoe
ofatpāhoehoe
Kamokuna as a series of pāhoehoe
surface
surface
lava flow
lava flow
and lavaand tubeslavasurface
tubes
(Figure lava
(Figure
1)flow[35].
1)and
[35].
Thelava majority
Thetubes majority
of
(Figure
theofflowsthe
1) [35].
flows
observedTheobserved
majority
were weretube-fed
of thetube-fed
flows observed were tube-fed
3.2.1.
3.2.1.Instruments
Instruments
pāhoehoe,
pāhoehoe,both bothsheet-like
sheet-likeandpāhoehoe,
ropey
and ropey in both
texture.
in texture.
sheet-like
Previous Previous
and remote
ropeyremotesensing
in texture.
sensingstudies Previous
studies
of theofremote
pre-2018
the pre-2018
sensing studies of the pre-2018
Kīlauea
Kīlauea
eruptioneruption
have haveinvestigated
During
During investigated
Kīlauea
the lava
thefield eruption
field discharge
lava discharge
campaigns,have
campaigns, rates
investigated
TIR
TIRrates
[36],data [36],
lavawere
lava
flow
lavadischarge
flow
emplacement
acquired
acquired emplacement
rates
of the
of tracking
the[36], tracking
lava[37],
volcanic
volcanic flow [37],
emplacement
targets
targets onon22 tracking [37],
22separate
separate
and lava
and pathway
lava pathway mechanisms
mechanisms
occasions. and
occasions. Of these, [38].lavaThe
[38]. pathway
areas
The
these,simultaneous areas
for
simultaneous mechanisms
this
for study
this
data data study
fromwere
[38].
from were
chosen
The
all the areas
chosen
all sensorsfor
the sensorsfor
thefor this
high
the
were were study
high
probability
acquired were
probability
acquired chosen
of
three times offor
three(Tablethe high
times1). probability
(Table
These1). of
observing
observing
molten molten
lava surfaces
lava
included
These surfaces
included observing
data combined
from
data combined
molten
from with
ground with
the
lava
ground and availability
the
surfaces
and availability
airborne
airbornecombined
of ainstruments
variety
of awith
instruments variety
ofmade
theremote
of
availability
made remote
sensing
possible
possiblesensing
of
byTIR
byatwovariety
TIR
two of remote sensing TIR
NASA-sponsored
NASA-sponsored
datasets.
datasets. airbornecampaigns
airborne datasets. totoHawaii
campaigns Hawaiiin in support
support of aa proposed proposedorbital orbitalmission missiondata datacollection/analysis
collection/analysis effort.
effort.
Thespaceborne
The spaceborne data
data were
were acquired
acquired bybythethe ASTER
ASTER instrument,
instrument, whichwhich
hashas been
been in orbit
in orbit aboard
aboard the
the Terra
Terra satellite
satellite since December
since December 1999Following
1999 [6]. [6]. Following the failure
the failure of the
of the shortwave
shortwave infrared
infrared (SWIR)system
(SWIR) systemin
in 2008,
2008, ASTER ASTER is now
is now a two-subsystem
a two-subsystem instrument
instrument withwith
eighteight channels
channels between
between 0.520.52
andand 11.65
11.65 µmμmand
and a spatial resolution between 15 and 90 meters [6]. For this study, only the five
a spatial resolution between 15 and 90 m [6]. For this study, only the five TIR channels between 8.125 TIR channels
between
and 8.125
11.65 µm anda 11.65
with μm
spatial with a spatial
resolution resolution
of 90 m are usedof(Table
90 meters areairborne
2). The used (Table
MASTER2). The airborne
and HyTES
MASTER and HyTES instruments were mounted on a NASA ER-2 aircraft that flew at an altitude of
Remote Sens. 2020, 12, 193 5 of 21

instruments were mounted on a NASA ER-2 aircraft that flew at an altitude of ~20 km. MASTER is a
50-channel instrument that detects radiance between 0.4 and 13.0 µm, with a FOV of 85.92◦ resulting in
a ground spatial resolution of ~50 m from the flight altitude [21]. In this study, the seven TIR channels
are used between 8.0 and 13 µm (Table 2). HyTES is a hyperspectral TIR instrument with 256 channels
that detects radiance between 7.5 and 12 µm, with a FOV of 50◦ resulting in a ground spatial resolution
of ~35 m at an altitude of ~20 km [20]. In this study, 128 HyTES channels between 8.3 and 11.6 µm
are used due to an instrument calibration resampling (Table 2). Finally, the MMT-Cam ground-based
system acquired data in six spectral channels plus a broadband temperature channel between 8.0 and
11.5 µm, with a FOV of 45◦ × 37◦ (Table 2) [19].

Table 1. Atmospheric conditions and target locations during each acquisition data.

Acquisition Data Time (HST) No Data Temperature (◦ C) Humidity (%) Target

01/19/2017 20:32 HyTES 27.9 59.8 Lava Lake
01/26/2017 10:59 HyTES 17.3 82.4 Lava Lake
01/30/2018 10:59 - 40.0 25.3 Lava Flows
02/06/2018 11:06 - 28.5 47.9 Lava Lake
02/08/2018 22:42 - 26.4 49.8 Lava Flows

Table 2. Instrument specifications.

MMT-Cam MASTER TIR HyTES ASTER TIR

(Ground) (Airborne) (Airborne) (Orbital)
VOX HgCdTe HgCdTe
Detector QWIP
microbolometer photoconductive photoconductive
Field of View 45◦ × 37◦ 85.92◦ 50◦
Spatial Resolution (m) 0.04/0.3 50 35 90
Spectral Resolution 6 9 (7) 186 5
Daily during Daily during
Temporal Resolution 1 second 5–15 days
campaign campaign
Radiometric Range (K) 233 to 832 245 to 480 240 to 455 200 to 370
Radiometric Accuracy 5% <5% <1% <3%

3.2.2. Data Calibration

Before extracting the surface kinetic temperature and emissivity from the various datasets, the raw
radiance data are calibrated and corrected for instrumentation and atmospheric effects (Figure 2).
The ASTER instrument data are radiometrically calibrated by viewing an internal constant temperature
blackbody and cold space. Occasionally, the internal blackbody is heated and cooled to provide a
multi-temperature radiometric calibration [6]. The MASTER and HyTES data are corrected for aircraft
motion and orthorectified using digital terrain models [20,21]. The MASTER instrument is spectrally
and radiometrically calibrated in the laboratory using two blackbodies pre- and post-campaign, with a
cold blackbody used during the data acquisition [21]. In the laboratory, the HyTES instrument is
spectrally calibrated using narrowband interference filters and radiometrically calibrated using a
blackbody between 4 and 40 ◦ C [20]. Finally, the MMT-Cam instrument is spectrally and radiometrically
calibrated in the laboratory using a blackbody between 20 and 800 ◦ C [19].
The ASTER, MASTER, HyTES, and MMT-Cam radiometrically calibrated at-sensor TIR radiance
data are all atmospherically corrected to derive the at-surface radiance (Figure 2). The MASTER and
ASTER data are corrected using the MODTRAN radiative transfer model with the water vapor scaling
method to optimize the atmospheric correction [39]. The HyTES data are corrected using the in-scene
atmospheric correction method [40]. The MMT-Cam spectral channels are co-registered using a fast
Fourier transform algorithm with centroid matching and then atmospherically corrected using the
SpectralCalc simulator [19].
 [28]. The algorithm first assumes a brightness temperature using a maximum scene emissivity and
the spectral morphology of each pixel is derived. An emissivity calibration curve relating spectral
contrast to the minimum emissivity is then used to constrain the true emissivity values from band
ratios. The calibration curve is empirically determined for each instrument separately using data from
a laboratory
Remote Sens. 2020,spectral
12, 193 library. Finally, the kinetic temperature is then calculated from the maximum
6 of 21
derived emissivity using the inverse Planck function [28].

Figure2.2.A
Figure Aflowchart
flowchartillustrating
illustratingthe
themethods
methodsused
usedin
inthis
thisstudy.
study.

4.4.2.
Methods
Thermally-Mixed Pixel (TMP) Separation Analysis
In most
4.1. Kinetic TIR remote
Temperature andsensing data of active volcanic thermal anomalies, a pixel contains multiple
Emissivity
surface fractions (or end-members) that can include temperature, composition, and texture. The
Kinetic temperature and emissivity are derived from the calibrated surface radiance data using
spectrum of a mixed pixel composed of two or more fractions represents the areal-weighted averages
the temperature emissivity separation (TES) algorithm (Figure 2), first developed for ASTER data [28].
of those end-members rather than being dominated by any one [29]. This problem increases in
The algorithm first assumes a brightness temperature using a maximum scene emissivity and the
complexity with lower spatial resolution data and the increased mixing of potentially more surface
spectral morphology of each pixel is derived. An emissivity calibration curve relating spectral contrast
fractions [41]. However, the end-members within any given pixel can be deciphered with knowledge
to the minimum emissivity is then used to constrain the true emissivity values from band ratios.
of radiance theory, a well-developed spectral deconvolution model, and an understanding of the
The calibration curve is empirically determined for each instrument separately using data from a
spectral signatures of the end-members (e.g., an end-member spectral library).
laboratory spectral library. Finally, the kinetic temperature is then calculated from the maximum
A straightforward solution to the thermally-mixed pixel problem was originally developed by
derived emissivity using the inverse Planck function [28].
Dozier [42] using a dual-band approach to define the two thermal fractions within a pixel (Figure 2).
TheThermally-Mixed
4.2. method uses Pixel
surface radiance
(TMP) values
Separation from two spectral channels to derive the unique
Analysis
combination of each fraction, both the value and proportion:
In most TIR remote sensing data of active volcanic thermal anomalies, a pixel contains
𝑀(𝜆𝑛 ,𝑇𝑖𝑛𝑡) = 𝑝 ∙ that
multiple surface fractions (or end-members) 𝑀(𝜆𝑛can
,𝑇ℎ ) + (1 − 𝑝)temperature,
include ∙ 𝑀(𝜆𝑛 ,𝑇𝑏) , composition, and texture.
(1)
The spectrum of a mixed pixel composed of two or more fractions represents the areal-weighted
averages of those end-members rather than being dominated by any one [29]. This problem increases
in complexity with lower spatial resolution data and the increased mixing of potentially more surface
fractions [41]. However, the end-members within any given pixel can be deciphered with knowledge of
radiance theory, a well-developed spectral deconvolution model, and an understanding of the spectral
signatures of the end-members (e.g., an end-member spectral library).
A straightforward solution to the thermally-mixed pixel problem was originally developed by
Dozier [42] using a dual-band approach to define the two thermal fractions within a pixel (Figure 2).
The method uses surface radiance values from two spectral channels to derive the unique combination
of each fraction, both the value and proportion:

M(λn ,Tint ) = p·M(λn ,Th ) + (1 − p)·M(λn ,Tb ) , (1)

where, M(λn ,Tint ) is the mixed surface radiance in channel n for the mixed temperature (Tint ). M(λn ,Th )
and M(λn ,Tb ) are the surface radiances contributed by the hot temperature fraction and background
temperature fraction, respectively; and p is the proportion of the hot fraction within the pixel area [42,43].
Equation (1) is solved with two simultaneous equations at two different wavelength channels, each
containing two unknown variables. This approach provides a unique solution for the radiance of
one fraction (either molten lava or the background) and its fractional proportion after assuming or
knowing the value for the other radiance value.
In this study, the surface radiance values are unmixed within each pixel for each dataset using a
channel at 8.5 µm and 11.0 µm. The background fraction applied to the analysis is the average value
temperatures where a molten lava surface cools (<1450 K). Therefore, fraction fraction
simplythat deconvolving
has
that the has maximum
the the maximum
moltentemperaturetemperature withinwithin each eachpixel pixel
will improvwill im
fraction that has the maximum temperature within each pixel uncertainty
uncertainty
will of theof
improve emissivity,
the emissivity,
accuracy kinetic kinetic
and temperature,
temperature, and radiant
and radiant
flux. flux.
uncertainty of the emissivity, kinetic temperature, and radiant flux. Measuring Measuring accurate accuratethermal thermal
properties properties of a molten
of a molten lava surface
lava surface is also is
propagation
propagation
Measuring accurate thermal properties of a molten lava surface is also critical to lava flow models models
[15,24]. [15,24].
With With
the increasing
the increasingnumber number
of spectral
of spectral
bands bands
in mor i
propagation models [15,24].
Remote Sens. With the193
2020, 12, increasing number of spectral (e.g.,bands
(e.g.,
HyTES HyTES
in [20]
moreand [20]
recent the
and Mineral
TIR the Mineral
imagers and Gas and Identifier
Gas Identifier (MAGI)
7 of 21(MAGI)
[25]), [25]),
the radiatthe r
(e.g., HyTES [20] and the Mineral and Gas Identifier (MAGI)emissivity [25]),emissivity
the of an ofobject’s
radiative an temperature
object’s
surface surface
canand becan extracted
be extracted with increasing
with increasing accuracy accuracy[26–28[
emissivity of an object’s surface can be extracted with increasingemissivity accuracyemissivity can then
[26–28]. canAthen be used be used
well-constrained with with approaches approaches such suchas linearas linear
spectral spectral
deconv d
emissivity can thenof the non-active
be used regions forsuch
with approaches eachas scene
linearand each
quantitatively
spectral channel
quantitatively
deconvolution indetermine
determine the dataset.
possible
modeling possibleAfter
spectral
to spectralthe thermal
end-memberend-membermixed
that defines
that defines the minera the m
pixel (TMP)
quantitatively determine possibleseparation analysis is applied,
spectral end-member that definesthe thermal
thermal molten
thefractions fraction
fractions
mineralogical, [29–32]. datasets
[29–32]. are
Additionally,
textural, integrated
Additionally,
and kinetic into
kinetic the
temperature TES (and (and
temperature to a lesserto a led
thermal fractions algorithm in order to derive
[29–32]. Additionally, kinetickinetic temperature
temperature required
(and required
toand a to determine
todegree,
emissivity
lesser determine ofthethe runout
the runout
molten
emissivity) distance
is distance
fractions and hazard
only and hazard
(Figurepotential
2).potential
usingusing radiant radh
required to determine
Thesethe runout
values fordistance
each pixelandare
hazard potential
then compared propagation
to propagation
using theradiant
resultsmodels heatmodels
from [15,24].
flux
the in [15,24].
same However,
lava However,
flow
pixels the accuracy
prior tothetheaccuracyof these
unmixing of these
derived derived
paramet pa
propagation models [15,24].Finally,
analysis. However, the accuracy
the variability of these
within each temperature
derived temperature
dataset rangerange
parameters
is quantified of over
typical
of the
to typical
lavas
evaluatecoolinglavas
is the
lessis well
lessconstrained
effect well constrained
of spatial at theat
and various
the various spatialspr
temperature rangespectral
of typical lavas is less
resolution on well constrained at
the discrepancy andtheTIR instruments.
various TIR spatial
uniqueness. instruments. Therefore,
resolutions Therefore, improving
of currentimproving the accuracy
the accuracy of theofkinetic
the kinetic
temperaturetempera
previously-determined
previously-determined
TIR instruments. Therefore, improving the accuracy of the kinetic temperature and emissivity of the molten molten
fraction fraction
should should
then reduce
then reduce
the uncertainty
the uncertainty in flow i
4.3. Accuracy
previously-determined and Uncertainty
molten fraction should thenAssessment directly
reduce the uncertainty directly
relyinon rely
flowthese
onmodelthese
thermal thermal
analyses properties.properties.
that
directly rely on these thermal properties.
The accuracy and degree of variability of the
3. Data
measured surface radiance as well as the derived
3. Data
3. Data
kinetic temperature and emissivity for each TIR dataset are quantified through comparative analysis
(Figure 2). The TMP separation analysis approach 3.1. Study is
3.1. Study held
Area Area as constant as possible to evaluate only the
3.1. Study Area influence of spatial resolution. After spatial resolution, the largest variability between the datasets
This study This study was conducted
was conducted during during
two field two campaigns
field campaigns at Kīlauea
at KīlaueaVolcano Vo
is the spectral resolution, which has less of an effect because the channel locations are commonly
This study was conducted during two field campaigns atJanuary/February Kīlauea January/February
Volcano in 2017 and
20172018,
Hawai’i, andUSA, 2018,
a period
in a period whenwhen both the bothsummit
the summitlava lake lavaand lak
within 0.5 µm of each other. The hyperspectral resolution of the HyTES data increases the level of
January/February 2017 and 2018, a period when both the summit lavalava flows
lava lake flows
were
and wereactive.active.
coastal It focused
plain It focused
surface primarily primarily on the onlavathe lake
lava in lake theinHalem the H
complexity for the comparative analysis to the multispectral resolution datasets. All analyses on the
lava flows were active. It focused primarily on the lava lake propagating
in propagating lava flows
the Halemaʻumaʻu lava flows from from
Craterthe Puʻu the Puʻu
and ‘Ō’ō vent ‘Ō’ō (Figure
vent (Figure 1). Kīlauea
1). Kīlauea
is a basalti
is a b
HyTES data are computed at full resolution, hashowever,
been
has these results are then spectrally degraded to years
propagating lava flows from the Puʻu ‘Ō’ō vent (Figure 1). Kīlauea is abeen
erupting
basaltic erupting
nearly
shield nearly
continuously
volcano continuously
that for the forpastthe 500
pastyears
500 [33]. The [33].lava The su la
perform the later comparison. The HyTES data
during also
during
long
has been erupting nearly continuously for the past 500 years [33]. The lava surfaces are produced provide
sustained
long a
sustained spectral
eruptions resolution
eruptionswhere where comparison
pāhoehoe pāhoehoe (tube- with
(tube-
and surface-fed)
and surface-fed) and ‘a’
a
the MASTER
during long sustained eruptions wheredata acquired
pāhoehoeat(tube-
the same time
[33]. and
and surface-fed) The atThe
[33].lava and a lake
similar
lava
‘a’ā in lake
flowsthespatial
in
Halemaʻumaʻu
are resolution,
theemplaced
Halemaʻumaʻu which
Crater Crater allows
observed observedthe during
during this studythis studywas a
influence of spectral resolution
[33]. The lava lake in the Halemaʻumaʻu Crater observed during to be quantified.
2018
thisand2018 Finally,
study varied
and the
wasvaried sensitivity
in sizein
active with
fromsize2008 of
maximum
withuntil spatial
maximum resolution
dimensions dimensions on the
of 160ofmeters
160 meterswide wide and 225 an
2018 and varied inderived
size with kinetic
maximum temperature
dimensionsandofemissivity
160 meters iswide
1) [34]. determined
1) During
[34].
andDuring225 toperiod,
thismeters quantify
this period,
longthere constraints
there
were were
(Figure on
fluctuations the degree
fluctuations in lava inof
lake
lavaactivity
lake activity with con wi
uncertainty
1) [34]. During this period, there withwerespatial resolution
fluctuations change.
in lava lakeand irregular
activityand with irregular
small
continuoussmall
explosions, explosions,
gas plumes finallyfinallyending ending
with the withsummit
the summitcollapse collapse
in May in
time time
of the
and irregular small explosions, finally ending with the summit collapse in May 2018 [13,34]. At the of field
the field
campaigns, campaigns, the lava
the lava
lake lake
level level
was was
relatively
relatively high hig
bu
time of the field5. campaigns,
Results the lava lake level was relatively approximately
approximately
high 100–130but not 100–130meters meters
overflowing, belowbelow the Halemaʻumaʻu
the Halemaʻumaʻu CraterCrater
rim. Itrim. wasItcon wa
approximately 100–130 meters below the Halemaʻumaʻu Crater withrim. fresh
with
It was fresh
lava upwelling
lava upwelling
continuously incirculating
thein north
the north
that migrated
that migrated to theto south,
the south,
cooled cooled
and for an
5.1. ASTER Data
with fresh lava upwelling in the north that migrated to the south, solidifiedsolidified
cooled lava.
and formed The
lava.lava The then
plateslavaof sank
then in
cooler,sankthein south,
the south,
distinguished
distinguished by theby occurrence
the occur
solidified lava. The lavaTwothenregions
sank inof the south, (ROI)
interest distinguished
of cooling and
by the
lavadegassing
and degassing
occurrence
surfaces [34].ofThe
(the [34].
strong
lava lava
The flows
lava
splattering
lake flows
at thefrom the
from
Halema Puʻutheuma‘Ō’ō
Puʻu uvent
‘Ō’ō
Craterwere
vent active
were active for ~30 forye~
and degassing [34].andThethe
lava flows
lava flowsfrom the Pu
from Puʻuu ‘Ō’ō
‘Ō’ōvent
on thewere locations locations
activeplain)
coastal over
for ~30are over
numerous
years
chosen numerous
erupting eruptive eruptiveepisodes, episodes,
in manythe retrieval of surface
to evaluate producingproducing mostly mostly
pāhoehoe pāhoe l
locations over numerous
radiance, eruptive episodes, producing
kinetic temperature, occasional
mostly
and emissivity occasional
pāhoehoe
from the‘a’āASTER
lava
‘a’ā
lavaflow lava
flows
data. [33,35].
flow with
Within[33,35].
This
the
these long
ThisROIs,eruption
long mixtures
eruption
finallyfinally
ofendedended in 2018 in with
2018
occasional ‘a’ā lava flow [33,35]. This long eruption finally lower
ended lower
in East
2018 Rift
East
with
both the cool crust and molten lava surfaces are represented in most pixels (Figure 3). Zone
Rift
the Zone
(LERZ)
cessation(LERZ)
eruption
of eruption
the in the inLeilani
the Leilani
Estates Estates
[14]. The
[14]. lava
The flows
lava f
lower East Rift Zone (LERZ) eruption in the Leilani Estates [14]. 2017–2018
The2017–2018
lava studyflowsstudy were were
observed partduringof
part theofour
61g
theepisode
61g episode that erupted
that erupted from fromthe east theflankeast
2017–2018 study were part of the 61g episode that erupted from flows theflows
propagated
eastpropagated
flank down of Puʻu down
the‘Ō’ō. Pulama
theThese
Pulama
pali and palientered
and entered the oceanthe ocean
at Kamokuna
at Kamoku as
flows propagated down the Pulama pali and entered the oceansurface surface
at Kamokuna lava flow lavaas aand
flow
series lava
and of tubes
lava tubes
pāhoehoe (Figure (Figure
1) [35]. 1) The
[35].majority
The majority of theofflows the flow obs
surface lava flow and lava tubes (Figure 1) [35]. The majoritypāhoehoe, of the pāhoehoe,
flows both both
sheet-like
observed sheet-like
were and ropey
and ropey
tube-fed in texture.in texture.Previous Previous
remote remote
sensing sensinstu
pāhoehoe, both sheet-like and ropey in texture. Previous remote Kīlauea Kīlauea
eruption
sensing eruption
haveofhave
studies investigated
the investigated
pre-2018 lava discharge
lava discharge rates rates
[36], lava
[36], flow
lava emplac
flow em
Kīlauea eruption have investigated lava discharge rates [36], and lavalava andpathway
flow lava pathway
emplacement mechanisms mechanisms
tracking[38]. [37], The
[38].areasThe areasfor this forstudy
this study
were werechosen chosen
for th
observing observing
and lava pathway mechanisms [38]. The areas for this study were chosen for the high probability of molten molten
lava surfaces
lava surfaces
combined combined with the
with availability
the availability
of a variety
of a varie
of
observing molten lava surfaces combined with the availability datasets.
of adatasets.
variety of remote sensing TIR
datasets.
 he maximum temperature or representative emissivity spectrum. There [23])are thatrelatively
[23]) have
that have few
investigated studies
investigated the (i.e.,
influence
the influence of spatialof spatial
and spectral
and spectral resolution
resolutio on
23]) that have investigated the influence of spatial and spectral resolution temperatures
temperatures
on these where where
a molten
calculations a molten lava surface
over lava surface coolscools (<1450 (<1450
K). Therefore,
K). Therefore,simply sim d
emperatures where a molten lava surface cools (<1450 K). Therefore,fraction simplyfraction deconvolving
that that
has the has the themolten
maximum maximum temperature
temperature within within
each each pixel pixel
will willimp
raction that has the maximum temperature within each pixel uncertainty will uncertainty
improve of the ofemissivity,
the the emissivity,
accuracy kinetic
and kinetictemperature,
temperature, and radiant
and radiant flux. flux.
uncertainty of the emissivity,
Remote Sens.kinetic
2020, 12,temperature,
193 and radiant flux. Measuring Measuring accurate accurate
thermal thermalproperties
properties of a of molten
8 aof molten
21 lava lava surface surface
is al
Measuring accurate
Remote thermal
Sens. 2019,properties
11, x REVISIONof a molten lava surface propagation
propagation
is also models critical models [15,24].
to lava[15,24].
With
flowWith the increasing
the increasing number 8number
of of21 spectral
of spectral
bandsband in m
propagation models [15,24]. With the increasing number of spectral (e.g., bands (e.g.,
HyTES
in more HyTES [20]
recent and
[20]TIR the
and Mineral
the Mineral
imagers and Gas and Identifier
Gas Identifier (MAGI) (MAGI)[25]),[25]),
the rad th
e.g., HyTES [20] and the Mineral and Gas Identifier (MAGI) [25]), theemissivity
emissivity radiative
of anof temperature
object’s
an object’s surface and
surface
can be can extracted
be extracted with with increasing
increasing
accuracyaccuracy[26–
emissivity of an object’s surface can be extracted with increasing accuracy emissivity
emissivity
[26–28]. can Athencan then
be used
well-constrained be used with with approachesapproaches such such as linearas linear
spectralspectral
deco
emissivity can then be used with approaches such as linear spectral deconvolution
quantitatively
quantitatively determine modeling
determine possible to spectral
possible spectral
end-memberend-member that defines
that definesthe min the
quantitatively determine possible spectral end-member that defines thermal
thethermal fractions
mineralogical, fractions [29–32]. [29–32].
textural, Additionally,
and Additionally, kinetic kinetic
temperature
temperature (and (andto a to lesse a
hermal fractions [29–32]. Additionally, kinetic temperature (and required to a required
lesser to determine
to determine
degree, the runout
emissivity) the runout
is distance distanceand hazardand hazard potentialpotential
usingusing radian r
equired to determine the runout distance and hazard potential using radiant
propagation
propagation heat
models flux
models in lava
[15,24]. flow
[15,24].
However, However, the accuracy
the accuracy of these of these
derived derived
param p
propagation models [15,24]. However, the accuracy of these derived temperature
temperature
parameters range range
over of typical
theof cooling
typical
lavaslavas is lessiswell less constrained
well constrained at theatvarious
the various spati
emperature range of typical lavas is less well constrained at the various spatial
TIR instruments.
TIR resolutions
instruments. Therefore, of current
Therefore, improving improving the accuracy
the accuracy of theofkinetic
the kinetictemperatu
temp
TIR instruments. Therefore, improving the accuracy of the kinetic temperature previously-determined
previously-determined
and emissivity molten molten
of fraction
the fraction
should should
then reduce
then reduce the uncertainty
the uncertaint in f
previously-determined molten fraction should then reduce the uncertainty directly inrely
directly flow on model
rely these
on theseanalyses
thermal thermal that properties.
properties.
directly rely on these thermal properties.
3. Data 3. Data
3. Data
3.1. Study
3.1. Study Area Area
3.1. Study Area
This This studystudy was conducted
was conducted during during
two field two field campaignscampaigns at Kīlauea
at KīlaueaVolca
This study was conducted during two field campaigns at Kīlauea January/February
January/February
Volcano 2017 2017
in Hawai’i, and USA, 2018,
andin2018, a perioda period whenwhen both both the summit
the summit lava lake
lava ala
anuary/February 2017 and 2018, a period when both the summit lava lavalake lava
flows and flowscoastal
were were plain
active. Itsurface
active. focused
It focused primarily primarily on the on lava the lava
lake lakein the in Hale
the
ava flows were active. It focused primarily on the lava lake inpropagating thepropagating
Halemaʻumaʻu lava flowslava Crater
flows
fromfrom the
andPuʻu the Puʻu‘Ō’ō ‘Ō’ō vent vent(Figure (Figure
1). Kīlauea
1). Kīlauea
is a basais a
propagating lava flows from the Puʻu ‘Ō’ō vent (Figure 1). Kīlauea has is been
ahasbasaltic
been
erupting shield
erupting volcano
nearly nearly that
continuously
continuously for the forpastthe past
500 years500 years[33]. [33].
The lava The
has been erupting nearly continuously for the past 500 years [33].during Theduring longsurfaces
lava long
sustained sustained
areeruptions eruptions
produced where where
pāhoehoe pāhoehoe (tube-(tube-and surface-fed)
and surface-fed) and
during long sustained eruptions where pāhoehoe (tube- and surface-fed) [33]. The[33].
andlava The
‘a’ālake
lava
flows in
lake thein
are Halemaʻumaʻu
the Halemaʻumaʻu
emplaced Crater Crater
observedobserved during during
this study
this stud wa
33]. The lava lake in the Halemaʻumaʻu Crater observed during this2018 study 2018
and was active
varied
and infrom
varied size 2008
in with
size withuntil
maximum maximum dimensionsdimensions of 160ofmeters160 meters widewide and 2
2018 and varied in size with maximum dimensions of 160 meters wide 1) [34]. 1)
and [34].
During
225During this period,
meters this
longperiod, therethere
(Figure werewere fluctuations
fluctuations in lava in lake
lava activity
lake activitywith
1) [34]. During this period, there were fluctuations in lava lake activity with
and irregular continuous
and irregular smallsmall gasexplosions,
explosions,plumes finally finally
ending ending
with withthe summitthe summit collapsecollapse
in M
and irregular small explosions, finally ending with the summit collapse time time of
in May theof field
the field
2018 campaigns,
[13,34]. campaigns,
At thethe lava the lava lake lake level level was was relatively
relatively high
ime of the field campaigns, the lava lake level was relatively high but 100–130
approximately
approximately not 100–130
overflowing,
meters meters
below below
the Halemaʻumaʻu
the Halemaʻumaʻu Crater Crater
rim. It rim.was It cw
approximately 100–130 meters below the Halemaʻumaʻu Crater rim. with with
freshcontinuously
It was fresh
lava upwelling
lava upwelling in theinnorth
circulating the north
that migrated
that migrated to thetosouth,
the south,
cooled cooled
and
with fresh lava upwelling in the north that migrated to the south, cooled solidified
solidified
and lava.formed lava.
The lava
The of
plates then
lava then
sank sank
cooler, in theinsouth, the south,distinguished
distinguished by the byoccurren
the occ
olidified lava. The lava then
Figure sank
3. in the
Examples south,
of thedistinguished
thermal by
infrared the and
(TIR) occurrence
degassing
and
surface of
degassing
radiance strong
[34].
Figure 3. Examples of the thermal infrared (TIR) surface radiance data acquired of the Halemaʻumaʻu The
[34].
data splattering
lava
The
acquiredflows
lava flows
of from
the from
the
Halema Puʻu
the umaPuʻu
‘Ō’ō u vent
‘Ō’ō were
vent were
active active
for ~30 fo
and degassing [34]. The lava
Craterflows
lava from
lake the
and Puʻu
Pu u ‘Ō’ō
‘ Ō’ ō vent
lava were
flows from locations
active the forlocations
~30
different over
years over
numerous
erupting
datasets
Crater lava lake and Puʻu ‘Ō’ō lava flows from the different datasets at ~11.3 μm. Data from numerous
at in
~11.3 eruptive
many
µm. eruptive
Data episodes,
from episodes, producing
01/20/2017, producing mostly mostlypāhoeho pāh
ocations over numerous eruptive 01/30/2018,
01/30/2018,
01/20/2017, and
episodes, producing
02/06/2018
andwere
mostly
acquired
02/06/2018 were
pāhoehoe
occasional
during occasional
acquired the day lava
‘a’ā
during
lava
‘a’ā
(~10:00flowslava
flow
the day HST)with
flow
[33,35].
and
(~10:00
the
[33,35].
This
data
HST)from
This
long long
eruption
and 01/20/2017
eruption
data from
finally finally
ended ended
in 2018 in 20 w
occasional ‘a’ā lava flowand [33,35].
01/20/2017
This
02/08/2018 long
and were
eruption
acquired
02/08/2018 were
finally
during ended
the night
acquired
lower
in lower
2018
during(~21:45
East with
the night
Rift
East
HST). theZone
Rift Zone
(LERZ)
cessation
The Hyperspectral
(~21:45
(LERZ)
of eruption
the
HST). The Hyperspectral
eruption
in the
Thermal EmissioninLeilani
the
Thermal
Leilani
Estates Estates
[14]. The
[14]. lava
The flow
lav
ower East Rift Zone (LERZ) eruption(HyTES)
Spectrometer in the Leilani
instrumentEstateswas [14].
not The
flown lava
2017–2018
in flows
2017–2018
2017. observed
study study
werewere during
part of part our
theof61gthe episode
61g episode that erupted
that erupted fromfrom the eastthe flaea
Emission Spectrometer (HyTES) instrument was not flown in 2017.
2017–2018 study were part of the 61g episode that erupted from the flows flows
east propagated
flank propagated
of Puʻu down down
theThese
‘Ō’ō. Pulama
the Pulama pali and palientered
and entered the oceanthe ocean
at Kamokuna
at Kamo
lows propagated down The the ASTERpali
ThePulama
surface
ASTER surface and radiance
entered
radiance
of ocean
the the active lavasurface
surface
at Kamokuna
of the active lava is
is lava
loweras lava
lower flowthan
a series flow
thanand expected
lava
and
ofexpected lava
tubes
pāhoehoe fortubes
for
molten
(Figure
molten
basalt
(Figure
1) basalt 1)(Figure
[35]. [35].
The
(Figure
4) majority
majority
The of theofflowsthe flo o
with an tubes
average of 19.74 W·mThe −2 ·sr−1 ·µm−1 and a variability of 5.85 W·m−2 ·sr−1 ·µm−1 . The emissivity
urface lava flow and lava (Figure 1) [35]. majority of the flows
pāhoehoe,pāhoehoe, observed
both both were
sheet-like tube-fed
sheet-likeand ropey
and ropey
in texture.
in texture.
Previous Previousremote remote
sensing sens
4) with an average of 19.74 W·m-2·sr-1·µ m-1 and a variability of 5.85 W·m-2·sr-1·µ m-1. The emissivity
spectra and
pāhoehoe, both sheet-like haveropeyan absorption
in texture. feature
Previousatremote 8.63 µm Kīlaueaand
sensingKīlauea higher
eruption
studies spectral
eruption have
of the have contrast
investigated
investigated
pre-2018 inlavathelava lavadischarge
discharge lakerates data. rates
[36], [36],
lava lava
flow flowemp
spectra have an absorption feature at 8.63 µ m and higher spectral contrast in the lava lake data. The
Kīlauea eruption haveTheinvestigated lava dischargeemissivity
average pixel-integrated rates [36], lava flowlava
and
is 0.898 andemplacement
with pathway
lava pathway
a variability tracking
mechanisms
mechanisms
of 0.077 [37],[38]. [38].
The areas
(Figure The areas
for this
4), whereas forstudy
this study
the werewere chosen chosefor
average pixel-integrated emissivity is 0.898 with a variability of 0.077 (Figure 4), whereas the average
and lava pathway mechanisms [38]. The areaskinetic
average pixel-integrated for thistemperature
study were chosenobserving
is 354.1 observing
for molten
the high
K with moltenlava lava
surfaces
probability
a variability surfaces
ofof combined
23.8 combined
K; bothwith with
the availability
of which the
areavailability of a variety
of a va
pixel-integrated kinetic temperature is 354.1 K with a variability of 23.8 K; both of which are
observing molten lava surfaces combined
significantly lower thanwith the availability
expected for moltenofbasaltic
a variety
datasets.datasets.
lavas of remote
[44,45].sensing TIR
significantly lower than expected for molten basaltic lavas [44,45].
datasets.
RemoteRemote Remote
Sens. 2019,
Sens.11, xSens.
2019, 11, x2020,
REVISION 12, 193
REVISION Remote Sens. 2019, 11, x REVISION 3 of 213 of 21 9 of 21 3 of 21
Remote Sens. 2019, 11, x REVISION 9 of 21
currently
currently
available
available
low spatial
low spatial
resolution
resolution
currently
(>500available
(>500
m) TIR m)datasets
low
TIR spatial
datasets
thatresolution
are
thatnot areable
not
(>500
to
able
accurately
m)toTIR
accurately
datasets
resolve resolve
that are not able to accurately resolve
the maximum
the maximum temperature
temperature or representative
or representative
the maximum emissivity
emissivity
temperature
spectrum. spectrum.
or There
representative
There
are relatively
are relatively
emissivity
few studies
few
spectrum.
studies
(i.e., There
(i.e., are relatively few studies (i.e.,
[23]) [23])
that have
that have
investigated
investigatedthe influence
the[23])
influence
that
of spatial
have
of spatial
investigated
and spectral
and spectral the
resolution
influence
resolutiononof these
on
spatial
these
calculations
and
calculations
spectral
overresolution
over on these calculations over
temperatures
temperatureswherewherea moltena molten
lava surface
lava
temperatures
surface
coolscools
(<1450
where
(<1450
K).a Therefore,
molten
K). Therefore,
lavasimply
surface
simply
deconvolving
cools
deconvolving
(<1450 theK). molten
Therefore,
the molten simply deconvolving the molten
fraction
fraction
that that
has thehas maximum
the maximum temperature
fraction
temperature
thatwithin
haswithin
theeach maximum
each
pixel pixel
willtemperature
will
improve improvethe
within
accuracy
the each
accuracy and
pixeland will improve the accuracy and
uncertainty
uncertainty
of theofemissivity,
the emissivity,kinetickinetic
uncertainty
temperature,
temperature,
ofand
the emissivity,
radiant
and radiant flux.kinetic
flux. temperature, and radiant flux.
Measuring
Measuringaccurateaccurate
thermal thermal
properties
properties
Measuring
of a ofmolten
accurate
a molten lavathermal
lava
surface surface
properties
is also is critical
also
of acritical
molten
to lava to lava
lava
flowsurface
flow is also critical to lava flow
propagation
propagation
models models
[15,24].[15,24].
With With
thepropagation
increasing
the increasingnumber
modelsnumberof
[15,24].
spectral
of spectral
With bandsthebands
increasing
in more in more
recent
number
recent
TIR of imagers
TIR
spectral
imagers bands in more recent TIR imagers
(e.g., (e.g.,
HyTES HyTES
[20] and
[20] the
andMineral
the Mineraland
(e.g.,Gas
and
HyTES Identifier
Gas [20]
Identifier
and
(MAGI) the(MAGI)
Mineral
[25]),[25]),
theandradiative
the
Gasradiative
Identifier
temperature
temperature
(MAGI) and[25]),
and the radiative temperature and
emissivity
emissivity
of an ofobject’s
an object’s
surface surface
can be
emissivity
canextracted
be extracted
ofwith
an object’s
with
increasing
increasing
surfaceaccuracy
canaccuracy
be[26–28].
extracted
[26–28].
A well-constrained
with
A well-constrained
increasing accuracy [26–28]. A well-constrained
emissivity
emissivity
can thencan then
be used be used
with with
emissivity
approaches
approachescan
suchthen
such
as be linear
asused
linear
spectral
with spectral
approaches
deconvolution
deconvolution
suchmodeling
as linear
modelingtospectral
to deconvolution modeling to
quantitatively
quantitatively
determine
determinepossible
possible
spectral
quantitatively
spectral
end-member
end-member
determine that defines
possible
that defines the
spectral
mineralogical,
the mineralogical,
end-member textural,
that
textural,
and
definesandthe mineralogical, textural, and
thermalthermal
fractions
fractions
[29–32].[29–32].
Additionally,
Additionally,
thermal kinetic
fractions
kinetic
temperature
temperature
[29–32].(and Additionally,
(and
to a lesser
to a kinetic
lesser
degree,degree,
temperature
emissivity)
emissivity)
(and
is to is a lesser degree, emissivity) is
required
required
to determine
to determinethe runout
the runout
distance
required
distance
andto hazard
determine
and hazardpotential
thepotential
runout
usingusing
distance
radiant radiant
heat
and hazard
flux
heat influx
lava
potential
in flow
lava using
flow radiant heat flux in lava flow
propagation
propagationmodels models
[15,24].[15,24].
However,
However,
propagation
the accuracy
the models
accuracy
of these
[15,24].
of these
derived
However,
derivedparameters
the
parameters
accuracy
over over
the
of these
cooling
the cooling
derived parameters over the cooling
temperature
temperature
rangerangeof typical
of typical
lavaslavas
istemperature
lessiswell
less constrained
wellrange
constrained
of typical
at theatvarious
lavas
the various
isspatial
less well
spatial
resolutions
constrained
resolutions
of current
at of
thecurrent
various spatial resolutions of current
TIR instruments.
TIR instruments. Therefore,
Therefore,
improving
improving
TIRthe instruments.
accuracy
the accuracy
of
Therefore,
theofkinetic
the improving
kinetic
temperature
temperature
the accuracy
and emissivity
and ofemissivity
the ofkinetic
theof temperature
the and emissivity of the
previously-determined
previously-determined molten molten
fraction
fraction
previously-determined
should should
then reduce
then reduce the
molten
uncertainty
the fraction
uncertainty inshould
flowin model
flow
thenmodel
reduce
analysesanalyses
thethat
uncertainty
that in flow model analyses that
directly
directly
rely onrelythese
on these
thermal thermal
properties.
properties.
directly rely on these thermal properties.

3. Data
3. Data 3. Data

3.1. Study
3.1. Study
Area Area 3.1. Study Area
This study
This studywas conducted
was conducted during duringtwo
Thisfield
two
study field
campaigns
was campaigns
conductedat Kīlaueaat during
Kīlauea
VolcanotwoVolcano
field
in Hawai’i,
campaigns
in Hawai’i, USA,atUSA,in
Kīlaueain Volcano in Hawai’i, USA, in
January/February
January/February 2017 2017
and 2018,and 2018,a period
January/February
a period
whenwhen both both the
2017summit
the
andsummit
2018,lavaa lakelava
period and
lake when
coastal
and both
coastal
plaintheplain
surface
summit surface
lava lake and coastal plain surface
lava lava
flowsflowswere were active.active.
It focusedIt focused
lava
primarilyflows
primarily on
were theonactive.
lava
the lava lake
It focused
lake
in theinprimarily
Halemaʻumaʻu
the Halemaʻumaʻu on theCrater lavaCrater
lake
and in andthe Halemaʻumaʻu Crater and
propagating
propagating lava flows
lava flows
from from the Puʻuthe
propagating
Puʻu
‘Ō’ō vent‘Ō’ō lava vent
(Figure flows
(Figure
1). from
Kīlauea
1). Kīlauea
the isPuʻu a basaltic
is‘Ō’ōa basaltic
vent
shield(Figure
shield
volcano volcano
1). Kīlauea
that thatis a basaltic shield volcano that
has been
has been
erupting erupting
nearly nearly
continuously
continuously
has been for the for
erupting
past
the 500 past
nearly
years
500 years
continuously
[33]. [33].The lava Theforsurfaces
lavathesurfaces
pastare 500 produced
are
yearsproduced
[33]. The lava surfaces are produced
duringduring
long sustained
long sustained eruptions
eruptionswhere during
where
pāhoehoe
long
pāhoehoe
sustained
(tube-(tube-and eruptions
surface-fed)
and surface-fed)
where and pāhoehoe
‘a’ā
andflows
‘a’ā(tube-
flows
are emplaced
and
are emplaced
surface-fed) and ‘a’ā flows are emplaced
[33]. The
[33].lava
The lake
lavain laketheinHalemaʻumaʻu
the Halemaʻumaʻu [33].Crater
The lava Crater
observed
lake observed
in the
during Halemaʻumaʻu
during
this studythis studywas Crater
active
wasobserved
active
from from 2008
during
until
2008this untilstudy was active from 2008 until
2018 2018
and varied
and varied in size inwith
size with
maximum maximum
2018dimensions
and dimensions
varied of in 160
size ofmeters
with
160 metersmaximum
widewide anddimensions
225
andmeters
225 meters long
of 160 long
(Figure
meters(Figure
wide and 225 meters long (Figure
1) [34].
1) During
[34]. During this period,
this period,
theretherewere 1) were
[34].
fluctuations
During
fluctuations this
in lavaperiod,
in lake
lavathereactivity
lake activity
were with fluctuations
with
continuous
continuous
ingas lava plumes
gas
lakeplumes
activity with continuous gas plumes
and irregular
and irregular smallsmall
explosions,
explosions, finallyand
finally
ending
irregular
endingwith smallwith
the summit
explosions,
the summit collapse
finally
collapse
inending
May in May
2018
with 2018
[13,34].
the [13,34].
summit
At the Atcollapse
the in May 2018 [13,34]. At the
time time
of the of field
the field
campaigns,
campaigns, thetime lava
the oflavalakethelakelevel
fieldlevelwas
campaigns,
was
relativelyrelatively
thehigh lavahigh butlakenot
butleveloverflowing,
not was overflowing,
relatively high but not overflowing,
approximately
approximately 100–130 100–130
meters meters
below approximately
below
the Halemaʻumaʻu
the Halemaʻumaʻu100–130 Crater
metersCrater
rim. belowIt
rim.wasthe
It continuously
was
Halemaʻumaʻu
continuously circulating
Crater
circulating
rim. It was continuously circulating
with with
freshfresh
lava upwelling
lava upwelling in theinnorththewithnorth
that fresh
migrated
thatlava migrated
upwelling
to thetosouth,the insouth,
the
coolednorthcooled
andthat formed
and
migrated
formed
plates toplates
the
of cooler,
south,
of cooler,
cooled and formed plates of cooler,
solidified
solidified
lava. lava.
The lava
The4.
Figure
Figure then
lava sank
then sank
4.Average
Average in solidified
theinsouth,
mixed
mixed the
pixel
pixel lava.
south,
distinguished
The
surface
surface distinguished
lavacolumn)
(left
(left thenby the
column) sankby occurrence
the
in the
radiance
radiance occurrence
south,
andof
and strong
distinguished
(right
(right ofcolumn)
strong
splattering
column) splattering
by the occurrence
emissivity
emissivity spectra of strong splattering
spectra
and degassing
and degassing [34]. The
[34].lava
acquired
acquired The offlows
of lava flows
from
the(blue)
the andHalema
(blue) from
the
degassing
Puʻu
theuma
Halemaʻumaʻu Puʻu
‘Ō’ō[34].
u vent
‘Ō’ō
Thewere
Crater
Crater vent
lava
lava
lava were
active
flows
lakeactive
lake for
from
and~30
and for
the
(red)
(red) years
~30
Puʻu
Puʻu
Pu years
uerupting
‘Ō’ō
‘Ō’ō
‘Ō’ ōerupting
vent inflows
lava
lava were
many
flowsin active
many
derived
derived forfrom
~30 years erupting in many
from
locations
locations
over over numerous
ASTER
ASTER numerous
TIR,
TIR, eruptive eruptive
locations
MODIS/ASTER
MODIS/ASTER episodes, episodes,
overproducing
(MASTER)
(MASTER) numerous
producing
TIR,mostly
TIR, eruptive
HyTES,
HyTES, mostlypāhoehoe pāhoehoe
episodes,
andMiniature
and Miniature lavaMultispectral
producing
lava
flowsflows
Multispectral withmostly
with
the the
Thermal
Thermal pāhoehoe
Camera lava flows with the
Camera
occasional
occasional
‘a’ā lava
‘a’ā flow
lava flow
(MMT-Cam)
(MMT-Cam) [33,35]. [33,35].
data.
data. Thisoccasional
The
The long
This
errorlong
error eruption
bars
bars ‘a’ā
eruption
lavafinally
represent
represent flowfinally
the
the ended
[33,35].ended
standard
standard inThis
2018inlong2018
witheruption
deviation
deviation with
the cessation
the finally
variation
variation cessation
ininthe
theofregions
ended
theof the
regions in
of 2018
ofinterestwith the cessation of the
interest
lowerlower
East Rift
EastZoneRift Zone
(LERZ)
(ROI).
(ROI). Note
Note (LERZ)
eruption
thereeruption
there lower
are
are in
nonothe East
inLeilani
HyTES
HyTES the
Rift Leilani
data
data Zone
Estates (LERZ)
Estates
between
between [14].9.92
9.92 The
[14].
eruption
and
and lava
The flows
10.75
10.75 lava
inµm µthe
mflows
observed
onLeilani
on observed
Estates
02/08/2018
02/08/2018 duringso during
so[14].
ourThe
these
these our
data
data lava
areflows
are not observed during our
not
2017–2018
2017–2018
studystudy were
includedwere
includedpart inof
in part
the the
the of61g
lava
lava the
2017–2018
lake episode
lake 61g
plots.episode
plots. study
that erupted
that
wereeruptedpart
from offromthe
the east61g
the episode
flank
east flankof that
Puʻu of erupted
Puʻu
‘Ō’ō. ‘Ō’ō.
These
from These
the east flank of Puʻu ‘Ō’ō. These
flowsflows
propagated
propagated downdown the Pulamathe Pulamaflows
pali and pali
propagated
entered
and entered the
down ocean
the theocean
atPulama
Kamokuna
at Kamokuna
pali and as aentered
series
as a series
ofthe pāhoehoe
ocean
of pāhoehoeat Kamokuna as a series of pāhoehoe
surface lava5.2.
surface lavaMASTER
flow
5.2. flow
and lava
MASTER andand lavaHyTES
tubes
and tubes
(Figure
HyTES Data
surface(Figure
Data1)lava
[35].1)flow
[35].
The and majority
Thelava majorityof
tubestheof(Figure
flows
the flows observed
1) [35].
observed
The
weremajority
were
tube-fed tube-fed
of the flows observed were tube-fed
pāhoehoe,
pāhoehoe,both bothsheet-like
The sheet-like
and ropey
and pāhoehoe,
ropey
in texture.
in texture.
both Previous
sheet-like
Previous remoteand remoteropey
sensing sensing
in studies
texture.
studies
of Previous
the of pre-2018
the remote
pre-2018 sensing studies of the pre-2018
The same
same two two ROIsROIs were wereselected
selected in inthe theMASTER
MASTER and and HyTES HyTES data; data; however,
however, the the number
number of of
Kīlauea
Kīlauea
eruptioneruption
pixels have
that have
investigated
investigated Kīlauea
lavaofdischarge
lava eruption
discharge rates
have rates
[36],
investigated
[36],
lava flow
lavaof lava
flow
emplacement
discharge
emplacement tracking
rates tracking
[36],[37],
lava[37],
flowtoemplacement tracking [37],
and lava
and pathway
lava
pixels
pathway
thatconstitute
constitute
mechanisms
mechanisms
each
[38].
each
and
The
[38].
of
lava
these
these
areas
The pathway
are
areas
for
arethis
higher
higher
for mechanisms
study
this
by
by
study
were
aafactor
factor
were
chosen
[38].
of approximately
chosen
The
approximately
for areas
theforhigh
the
for this
high
three
three
probability
study
compared
compared
probability
were
of chosen
of
ASTER
tofor
ASTER
the high probability of
(Figure
(Figure 3).
3).surfaces
observing
observing
molten molten
lava lava surfaces combined
observing
combined with molten
with
the availability
the
lavaavailability
surfaces of acombined
variety
of a variety ofwith
remote
of the
remotesensing
availability
sensing
TIR ofTIR a variety of remote sensing TIR
The
The MASTER
MASTER surface surface radiance radianceisisalso also lowerlower than than expected
expected for temperatures
for the the temperatures of moltenof molten
basalt
datasets.
datasets. datasets. −2 −1 −1 −2 −1 −1
basalt (Figure
(Figure 4) with4) an
with an average
average of 25.77
of 25.77 W·m-2W·m
·sr-1·µ m·sr-1 and
·µmvariability
and variability of m
of 8.94 W· 8.94 -1·µ m-1·sr
W·m
-2·sr . The·µm
pixel-.
The pixel-integrated emissivity spectra have a narrow absorption feature at 8.63 µm
integrated emissivity spectra have a narrow absorption feature at 8.63 µ m and a broader absorption and a broader
absorption feature
feature around around
10.63 10.63 µm
µ m (Figure 4).(Figure 4). The
The average average emissivity
emissivity is 0.723
is 0.723 with with a variability
a variability of
of 0.130. The
0.130. Thepixel-integrated
average average pixel-integrated kinetic temperature
kinetic temperature is 425.2 K is 425.2
with K with a variability
a variability of 64.3 K. of 64.3 K.
The HyTES surface radiance data are lower than those extracted from MASTER over the lava
flows but greater over the lava lake (Figure 4). The average surface radiance is 34.83 W·m-2·sr-1·µ m-1
with a variability of 13.81 W·m-2·sr-1·µ m-1. The emissivity spectra have an absorption feature at ~9.48
Remote Sens. 2020, 12, 193 10 of 21

The HyTES surface radiance data are lower than those extracted from MASTER over the lava flows
but greater over the lava lake (Figure 4). The average surface radiance is 34.83 W·m−2 ·sr−1 ·µm−1 with
a variability of 13.81 W·m−2 ·sr−1 ·µm−1 . The emissivity spectra have an absorption feature at ~9.48 µm
(Figure 4). The average pixel-integrated emissivity is 0.805 with a variability of 0.114, and lower values
are derived from the lava lake; whereas the average pixel-integrated kinetic temperature is 407.8 K
with a variability of 32.5 K. These values are also lower than the MASTER-derived temperatures, but at
a lower variability.

5.3. MMT-Cam Data

The MMT-Cam data presented here are extracted from ROIs covering similar activity to the
airborne and orbital data ROIs, especially at the lava lake (Figure 3). The lava flow ROIs, however, are
significantly smaller but more spatial detail is observed in these data due to the high spatial resolution
(<0.1 m). The radiance and kinetic temperatures derived from the MMT-Cam data are higher than the
other datasets analyzed in the study, as expected.
The radiance derived from the MMT-Cam ROIs is closer to the values for molten basalt
at the wavelengths and bandwidths of the MMT-Cam instrument (Figure 4). The average is
162.91 W·m−2 ·sr−1 ·µm−1 with a variability of 76.31 W·m−2 ·sr−1 ·µm−1 . The emissivity spectra show a
strong absorption between 8.55 and 9.55 µm, with both a single broad feature and two narrow features
(Figure 4). The average pixel-integrated emissivity is 0.739 with a variability of 0.087, and the average
pixel-integrated kinetic temperature is 736.2 K with a variability of 163.0 K.

5.4. Mixed Pixel Derivation

All the TIR datasets were next subjected to a thermally-mixed pixel separation analysis to extract
the values associated with the maximum thermal fraction within each pixel (i.e., molten lava). This step
evaluates the ability to measure an accurate molten fraction (if one is present) regardless of spatial
scale within a given pixel and the radiance values associated with it. The background radiance values
applied to this analysis are calculated from the average values of pixels at the background temperature
for each channel of each dataset and observation. The largest difference between the original data and
the unmixed counterparts was observed in the ASTER data and the smallest was seen in the MMT-Cam
data, as might be expected based on their pixel sizes.
Following the mixed pixel derivations using an average background radiance of
8.18 W·m−2 ·sr−1 ·µm−1 , the average ASTER surface radiance for the molten fraction increases to
493.22 W·m−2 ·sr−1 ·µm−1 as does the variability to 191.57 W·m−2 ·sr−1 ·µm−1 (Figure 5). The emissivity
spectra have an absorption feature at 8.63 µm (except in one lava flow dataset) with a strong decrease
in emissivity at shorter wavelengths (Figure 5). The absorption feature observed at 8.5–9.0 µm is
associated with a molten Si-O absorption and is likely absent in the ASTER lava flow data due to the
very low molten fractions (<0.05) observed with pixels at this target. The average molten fraction
emissivity is 0.752 with a variability of 0.099; whereas the average molten fraction kinetic temperature
is 1242.3 K with a variability of 337.0 K.
 Remote
RemoteRemote Sens.11,
Sens. 2019,
Sens. 2020,
2019, 12,x 193
x REVISION
11, REVISION Remote Sens. 2019, 11, x REVISION 3 of 2111 of 21
3 of 21
Remote Sens. 2019, 11, x REVISION 11 of 21
currently
currently
available
available
low spatial
low spatial
resolution
currently
resolution
(>500 available
(>500
m) TIR m)low
datasets
TIRspatial
datasets
that
resolution
are
thatnot areable
(>500
not toable
m)
accurately
to
TIRaccurately
datasets
resolve that
resolveare not able to accurately re
the maximum
the maximum temperature
temperature or representative
or representative
the maximum emissivity
emissivity
temperature
spectrum.
spectrum.
or representative
ThereThereare relatively
areemissivity
relatively
few studies
few
spectrum.
studies
(i.e.,There
(i.e., are relatively few studie
[23]) [23])
that have
that have
investigated
investigated the influence
the[23])
influence
that
of spatial
have
of spatial
investigated
and spectral
and spectral
the
resolution
influence
resolution
onofthese
spatial
on these
calculations
and calculations
spectralover resolution
over on these calculations
temperatures
temperatureswhere where
a moltena molten
lava surface
lava
temperatures
surface
coolscools
(<1450
where (<1450
K).
a molten
Therefore,
K). Therefore,
lavasimply
surfacesimply
deconvolving
coolsdeconvolving
(<1450 K). theTherefore,
molten
the molten simply deconvolving the m
fraction
fraction
that that
has thehas maximum
the maximum temperature
fraction
temperature
that within
has within
theeachmaximum
each
pixel pixel
willtemperature
will
improve improvethe
withinaccuracy
the eachaccuracy
pixel
and and will improve the accuracy
uncertainty
uncertainty
of theofemissivity,
the emissivity, kinetickinetic
uncertainty
temperature,
temperature,
of the
andemissivity,
radiant
and radiantflux.
kinetic
flux. temperature, and radiant flux.
Measuring
Measuringaccurateaccurate
thermal thermal
properties
properties
Measuringof a of molten
accurate
a moltenlava
thermal
lava
surface surface
properties
is also is critical
also
of a critical
molten
to lava tolava
lava
flow surface
flow is also critical to lava
propagation
propagation
models models
[15,24].[15,24].
With With
the
propagation
increasing
the increasingmodels
number number
[15,24].
of spectral
ofWith
spectral
bands
the bands
increasing
in more in more
recent
number recent
TIRofimagers
TIR
spectral
imagersbands in more recent TIR im
(e.g., (e.g.,
HyTES HyTES
[20] and
[20] the
andMineral
the Mineral
(e.g.,
and Gas HyTES
and Identifier
Gas[20]Identifier
and(MAGI)
the(MAGI)
Mineral
[25]),[25]),
the
andradiative
the
Gasradiative
Identifier
temperature
temperature
(MAGI) and [25]),
andthe radiative temperatur
emissivity
emissivity
of anof object’s
an object’s
surface surface
canemissivity
becanextracted
be extracted
of an
with object’s
with
increasing
increasing
surface accuracy
can accuracy
be [26–28].
extracted
[26–28].
Awith
well-constrained
Aincreasing
well-constrainedaccuracy [26–28]. A well-constr
emissivity
emissivity
can thencan then
be used be usedwith emissivity
with
approaches
approaches
can suchthensuch
as belinear
as
usedlinear
spectral
withspectral
approaches
deconvolution
deconvolution
suchmodeling
as linear
modeling to
spectral
to deconvolution modeli
quantitatively
quantitatively
determine
determinepossiblepossible
spectral
quantitatively
spectral
end-member
end-member
determine that possible
defines
that definesthe
spectral
mineralogical,
the end-member
mineralogical, textural,
that
textural,
defines
and and the mineralogical, textura
thermalthermal
fractions
fractions
[29–32].[29–32].
Additionally,
Additionally,
thermal kinetic
fractions
kinetic
temperature
[29–32].
temperature (and
Additionally,
(and
to a to lesser
akinetic
lesser
degree,temperature
degree,
emissivity)
emissivity)
(and
is toisa lesser degree, emissiv
required
required
to determine
to determinethe runout
the runout
distance
required
distance
and
to determine
hazard
and hazard potential
thepotential
runout
usingdistance
using
radiant radiant
and
heathazard
flux
heat in flux
potential
lavain flow
lavausing
flow radiant heat flux in lava
propagation
propagationmodels models
[15,24]. [15,24].
However,
However,
propagation
the accuracy
themodels
accuracy
of [15,24].
these
of these
derived
However,
derived
parameters
the
parameters
accuracy
over overof
thethese
cooling
the derived
cooling parameters over the co
temperature
temperature
rangerangeof typical
of typical
lavaslavas
is
temperature
lessiswell
less constrained
well
rangeconstrained
of typical
at theatlavas
various
the various
is less
spatial
well
spatial
resolutions
constrained
resolutions of at
current
the
of current
various spatial resolutions of cu
TIR instruments.
TIR instruments. Therefore,
Therefore,improving
improving
TIR instruments.
the accuracy
the accuracy Therefore,
of theofkinetic
theimproving
kinetic
temperature
temperature
the accuracy
and emissivity
andofemissivity
the kinetic
of theoftemperature
the and emissivity
previously-determined
previously-determined molten molten
fraction
previously-determined
fraction
should should
then reduce
then reducethe
moltenuncertainty
thefraction
uncertaintyin
should
flowin model
flow
then model
reduce
analysesanalyses
the that
uncertainty
that in flow model analyse
directly
directly
rely onrelythese
on these
thermal thermal
properties.
properties.
directly rely on these thermal properties.

3. Data
3. Data 3. Data

3.1. Study
3.1. Study
Area Area 3.1. Study Area
This study
This study was conducted
was conducted during during
two
Thisfieldtwo
study field
campaigns
wascampaigns
conducted at Kīlaueaatduring
KīlaueaVolcano
twoVolcano
field
in Hawai’i,
campaigns
in Hawai’i, USA, atUSA,Kīlauea
in in Volcano in Hawai’i, US
January/February
January/February 2017 2017
and 2018,and 2018,
a January/February
period a period whenwhen both 2017both
the summit
and
the summit
2018, lava a period
lake
lava and lake
when coastal
andboth coastal
plain
the plain
summit
surface surface
lava lake and coastal plain su
lava lava
flowsflowswerewere active.active.
It focused
It focused
lava
primarily
flows
primarily were
on the on
active.
lava
the lavaIt
lake focused
lake
in the inprimarily
Halemaʻumaʻu
the Halemaʻumaʻu on the Crater lavaCrater
lake
and inand the Halemaʻumaʻu Crate
propagating
propagating lava flows
lava flows
from from the Puʻuthe
propagating
Puʻu
‘Ō’ō vent ‘Ō’ōlava vent
(Figureflows
(Figure
1).from
Kīlauea
1).theKīlauea
Puʻu
is a basaltic
is
‘Ō’ō
a basaltic
ventshield
(Figure
shield
volcano 1).
volcano
Kīlauea
that that is a basaltic shield volcan
has been
has been
erupting eruptingnearly nearly
continuously
continuously
has been for the erupting
forpastthe 500pastnearlyyears
500 continuously
years
[33]. [33].
The lava The forsurfaces
lava
the past
surfacesare
500produced
are
years produced
[33]. The lava surfaces are prod
duringduring
long sustained
long sustained eruptions
eruptionswhere
during
where
pāhoehoe
longpāhoehoe
sustained
(tube-(tube- anderuptions
surface-fed)
and surface-fed)
whereand pāhoehoe
‘a’ā
andflows‘a’ā(tube-
flows
are emplaced
and
are surface-fed)
emplaced and ‘a’ā flows are emp
[33]. The
[33].lava
The lake
lava in laketheinHalemaʻumaʻu
the Halemaʻumaʻu[33]. The Crater lava
Crater
observed
lakeobserved
in the during
Halemaʻumaʻu
during
this study
this study was
Crater active
was observed
active
from from 2008
during 2008
until this
untilstudy was active from 2008
2018 2018
and varied
and varied in size in with
size with
maximum maximum
2018 dimensions
and varied
dimensions in
of size
160ofwith
meters
160 meters
maximum
widewide anddimensions
225
andmeters
225 meters oflong160long
(Figure
meters (Figurewide and 225 meters long (F
1) [34].
1) During
[34]. During this period,
this period,
theretherewere
1) [34].
were
fluctuations
During
fluctuations this
in lava period,
in lake
lavathere activity
lake wereactivity
withfluctuations
with
continuous
continuous
in gaslavaplumesgas
lakeplumes
activity with continuous gas pl
and irregular
and irregular smallsmall explosions,
explosions, finally
and finally
ending
irregularendingwith small
with
theexplosions,
summit
the summit collapse
finally
collapse
inending
May in May2018
with2018[13,34].
the summit
[13,34].
At the At
collapse
the in May 2018 [13,34]. A
time time
of the of field
the field campaigns,
campaigns, the time lava
theoflava lake
the lake level
field level campaigns,
was was relativelyrelatively
the highlavahigh but
lake not
butlevel overflowing,
notwas overflowing,
relatively high but not overflo
approximately
approximately 100–130 100–130
meters meters
below approximately
below
the Halemaʻumaʻu
the Halemaʻumaʻu
100–130Crater meters Crater
rim.
belowIt
rim.was
theIt continuously
was
Halemaʻumaʻu
continuously circulating
Crater
circulating
rim. It was continuously circu
with with
freshfresh
lava upwelling
lava upwelling in theinnorth
thewith north
that
fresh migrated
that lavamigrated
upwelling
to thetosouth, thein south,
thecooled
northcooled
and
thatformed
and
migratedformed
platesto plates
the
of south,
cooler,
of cooler,
cooled and formed plates of c
solidified
solidifiedFigure
lava.
Figure 5.5.lava
lava.
The Average
The
Averagethen molten
lavamolten
sank
then sank
in pixel
solidified
pixel fraction
theinfraction
south,
thelava.
south, surface
distinguished
The
surface (left
distinguished
lava
(left thencolumn)
bysank
column) the radiance
byoccurrence
in
thethe
radiance occurrenceand
south,
and (right
of(right
distinguished
strong column)
of strong splattering
column) emissivity
splattering
by the occurrence of strong splat
emissivity
and degassing
and degassing [34].
spectra
spectra The
[34].lava
acquired
acquired The flows
ofoflava flows
from
andHalema
the(blue)
the (blue) degassing
from
the Puʻu theuma
Halemaʻumaʻu Puʻu
‘Ō’ō
[34].
u vent
‘Ō’ō
The
Crater
Crater were
vent
lava lava
lava were
flows
active
lakeactive
lake from
for ~30
and(red)
and the
foryears
(red) Puʻu
~30
PuPuʻuuyears
erupting
‘Ō’ō
‘Ō’ō
‘Ō’ ō erupting
vent
lava
lava in were
many
flows
flows inderived
active
manyfor ~30 years erupting in
derived
locations
locations
over
fromover
from numerousnumerous
thethermal
the thermal eruptive
mixed
mixed eruptive
locations
pixel
pixel episodes,
molten episodes,
molten over producing
numerous
fraction
fraction producing
of of the
the mostly
ASTER eruptive
ASTERmostly
TIR,pāhoehoe
TIR, episodes,
pāhoehoe
MASTER MASTER lava
TIR,producing
lava
flows
TIR,
HyTES, flows
with
andmostly
HyTES, with
the MMT-
and
MMT-Cam the
pāhoehoe lava flows wit
occasional
occasional
‘a’ā
Cam lava
‘a’ā
data. flow
lava
The flow
[33,35].
error [33,35].
This
bars occasional
This
long
represent long
eruption
‘a’ā
the eruption
lava finally
standard flow finally
ended
[33,35].
deviation ended
data. The error bars represent the standard deviation variation in the ROI data. Note there are no HyTES inThis
2018in
variation long2018
with
in eruption
with
the
the cessation
ROI the finally
cessation
data. of
ended
Note the of
there the
in 2018 with the cessation
are
lowerlower
Eastdata
Rift
East Zone
no HyTESRift Zone
between (LERZ)
data
9.92 (LERZ)
eruption
between
and eruption
10.75lower
9.92 inand
µm the
East
on in10.75
Leilani
the
RiftLeilani
02/08/2018 µZone
mEstates
onso(LERZ)
Estates
[14].
02/08/2018
these eruption
data The
[14].solava
are The
notin
these flows
lavatheflows
data
included Leilani
observed
arein observed
not Estates
the during
included
lava during
[14].
lake our The
inplots.
the ourlava flows observed durin
lava
2017–2018
2017–2018study
lake studywerewere
plots. part of part theof61g
2017–2018
theepisode
61g episode study
that erupted
that
wereerupted part fromof from
the
the 61geast
theepisode
flank
east flankofthat
Puʻu oferupted
Puʻu
‘Ō’ō. ‘Ō’ō.These
fromThese the east flank of Puʻu ‘Ō’ō.
For the down
unmixed MASTER data using an down
average background −2 ·sr−1 ·µm−1 ,
flowsflows
propagated
propagated down
the Pulama
the Pulama
flows
pali and
propagated
pali entered
and entered the ocean
the
theoceanPulama
at Kamokuna pali radiance
at Kamokuna and as of
as aentered
series 8.38
a series
of W·m
thepāhoehoe
ocean
of pāhoehoeat Kamokuna as a series of pāho
−2 ·sr
surfacethelava
average
surface For
flow
lava surface
theflow
and andradiance
unmixed
lava tubes
lava MASTER
tubes derived
(Figure
surface data
(Figure
1)lava for
using
[35]. the
Thean
1)flow
[35]. maximum
and average
majority
The lava
majorityofthermal
background
tubes theof thefraction
(Figure
flows 1)radiance
flows
observed
[35].is The
300.59
observed were W·m
ofmajority
8.38
were W·
tube-fed tube-fed
of ·−1
m -2the sr·µm
flows
-1 −1-1
·µ m ,
observed were tub
with
pāhoehoe,
pāhoehoe,a variability
both both
sheet-like of 111.48
sheet-like
and ropey
and W·m
pāhoehoe,
ropey
in−2texture.
·srin−1both
·µm
texture.−1
Previous(Figure
sheet-like
Previous remote5).
and The
remote
ropey emissivity
sensing sensing
in studies
texture. spectra
studies
of
Previous
the of from
pre-2018
the the
remote
pre-2018 lava
sensinglake studies of the pre
the average surface radiance derived for the maximum thermal fraction is 300.59 W·m ·sr ·µ m with -2 -1 -1

Kīlauea
Kīlauea
showeruption
aneruption
a variability have
absorption have
investigated
of 111.48 investigated
W·mKīlauea
feature lava
at
-2 ·sr8.63
-1 ·µdischarge
lava
m eruption
µm -1 discharge
and rates
(Figure have
the rates
[36],
5).lavainvestigated
The [36],
lava
flow flow
lavalava
ROIs
emissivity flow
emplacement
show discharge
emplacement
spectra fromrates
a broad tracking
thetracking
[36],
lava[37],
absorption lava
lake [37],
flow
showemplacement
feature at
an tracking
and lava
and
10.63 pathway
lava
µmpathway
absorption with mechanisms
mechanisms
a decrease
feature at 8.63 [38]. and
mThe
[38].lava
at µwavelengths
and areas
The pathway
the areas
for this
for
shorter
lava mechanisms
flow study
this
than study
ROIs were
9.09 showwere
chosen
[38].(Figure
µm achosen
The forareas
broad thefor
5). high
for
thethis
The
absorption high
probability
average study
probability
feature were
molten ofat chosen
of µfor
fraction
10.63 m the high probabi
observing
observing
withmolten
emissivity molten
is lava
a decrease 0.584lava
surfaces
with surfaces
combined
observing
a variability
at wavelengths combined with
of molten
0.141;
shorter with
thethanavailability
lava
the 9.09
whereas availability
surfaces
the
µm of combined
a(Figure
average variety
of a molten
variety
ofwith
5). remote
oftheremote
fraction
The availability
sensing
average sensing
kinetic TIR
molten ofTIRa fraction
temperature variety of remote sensin
datasets.
datasets.
isemissivity
1128.2 K with is 0.584a variability datasets.
of 408.0 of
with a variability K.0.141; whereas the average molten fraction kinetic temperature
UsingKan
is 1128.2 withaverage calculated
a variability background
of 408.0 K. radiance of 8.01 W·m−2 ·sr−1 ·µm−1 , the average HyTES
surface radiance derived for the molten fraction is 402.64 W·mW· −2 ·sr−1 -1·µ−1
·µm
Using an average calculated background radiance of 8.01 m-2·sr m-1with
, the aaverage
variability of
HyTES
146.68 W·m −2 −1 −1
·sr derived
·µm . The emissivity spectra show an absorption
surface radiance for the molten fraction is 402.64 W·m -2·sr-1·µfeature
m-1 withataaround 9.5 µm
variability and
of 146.68
aW·
broader
m-2·sr-1feature centered
·µ m-1. The at 9.75
emissivity µm, with
spectra showa decrease in overall
an absorption emissivity
feature at around at 9.5
shorter wavelengths
µ m and a broader
feature centered at 9.75 µ m, with a decrease in overall emissivity at shorter wavelengths (Figure 5).
Remote Sens. 2020, 12, 193 12 of 21

(Figure 5). The average molten fraction emissivity and kinetic temperature are 0.604 and 1266.1 K with
variabilities of 0.260 and 404.0 K, respectively.
Lastly, using an average background radiance of 15.26 W·m−2 ·sr−1 ·µm−1 , the average molten
surface radiance derived from the MMT-Cam data is 454.27 W·m−2 ·sr−1 ·µm−1 with a variability of
158.12 W·m−2 ·sr−1 ·µm−1 . The derived lava lake emissivity spectra show a single absorption feature
between 8.55 and 8.99 µm or a double feature at 8.55 and 9.55 µm (Figure 5). The lava flow emissivity
spectra decrease in emissivity below 10.05 µm and is centered at 8.55 µm (Figure 5). The average
molten fraction emissivity and kinetic temperature values are 0.711 and 1225.9 K with variabilities of
0.078 and 329.6 K, respectively.

5.5. Comparisons and Trends

Qualitatively, significant variations are observed between the four datasets as a consequence of
differences in spatial resolutions (Figure 3). The low spatial resolution ASTER data provide the least lava
surface detail, whereas the high-resolution MMT-Cam data provide the greatest details. For example,
crustal plates and molten spreading margins at the lava lake are observed in the MMT-Cam data
but not in the ASTER data (Figure 3). However, the entire lava flow field is underrepresented in the
MMT-Cam data due to the proximity of the instrument to the target. Additionally, greater spatial
details of lava surface features are discerned in the HyTES data compared to the MASTER data, likely
the result of a higher number of spectral channels (Figure 3). In general, the spatial resolution of the
instrument strongly correlates to the scale of spatial detail qualitatively observed of the lava surfaces.
In all datasets, the surface radiance increases after the TMP separation analysis as would be
expected. The ASTER surface radiance increases the most in the lava lake (2100%) and lava flow
(2700%) data, compared to only a 70% and 330% increase in the MMT-Cam data, respectively (Figure 6
and Table 3). The MASTER and HyTES surface radiances increase by 500% and 200% in the lava lake
data and 2800% and 3200% in the lava flow data, respectively. The highest increases are observed at
shorter wavelengths (Figure 6), which is consistent with Wein’s Law where the peak radiance shifts
to shorter wavelengths with increasing temperature [41]. The TMP separation analysis appears to
provide a consistent and reasonable method for extracting the higher temperature fractions and allows
more accurate values of surface radiance, kinetic temperature, and emissivity to be extracted from the
lower spatial resolution datasets.

Table 3. The combined spatial and spectral average emissivity and kinetic temperature values pre-
and post-thermal mixed pixel (TMP) separation analysis for each dataset, including the percentage
improvement. The values in parenthesizes represent the standard deviation variation in the data.

ASTER TIR MASTER TIR HyTES MMT-Cam

Pre-TMP 19.74 (5.85) 25.77 (8.98) 34.83 (13.81) 162.91 (76.31)
Radiance
Post-TMP 493.22 (191.57) 300.59 (111.48) 402.64 (146.68) 454.27 (158.12)
(Wm−2 sr−1 µm−1 )
Change 2400% 1070% 1060% 180%
Pre-TMP 0.898 (0.077) 0.723 (0.130) 0.805 (0.114) 0.739 (0.087)
Emissivity Post-TMP 0.752 (0.099) 0.584 (0.141) 0.604 (0.260) 0.711 (0.078)
Change 19% 23% 33% 4%
Pre-TMP 354.1 (23.8) 425.2 (64.3) 407.8 (32.5) 736.2 (163.0)
Temperature
Post-TMP 1242.3 (337.0) 1128.2 (408.0) 1266.1 (404.0) 1225.9 (329.6)
(K)
Change 250% 170% 210% 70%
Remote Sens. 2020, 12, 193 RemoteRemote
Sens. 2019,
Sens.11,
2019,
x REVISION
11, x REVISION 13 of 21
Remote
RemoteSens.
Sens.2019,
2019,11,
11,xxREVISION
REVISION 3 of1421of 21
currently currentlyavailable availablelow spatial
low spatial resolutionresolution (>500(>500 m) TIR m)datasets
TIR datasets that are thatnot areab n
currently available low spatial resolution (>500 them)maximum
the
TIRmaximum
datasets temperature
temperature
that are not or representative
ableor representative
to accurately emissivityemissivity
resolve spectrum. spectrum.ThereThere are rea
the maximum temperature or representative emissivity that spectrum.
[23]) [23]) have
that have There
investigated
investigated arethe relatively
influence
the influence fewof studies
spatial (i.e., spectral
of spatial
and and spectral resolution
resolutio on
[23]) that have investigated the influence of spatial temperatures
temperatures
and spectral wherewhere a molten
resolution a molten lavathese
on surface
lavacalculations
surface
coolscools (<1450over(<1450
K). Therefore,
K). Therefore, simply sim d
temperatures where a molten lava surface cools (<1450
fraction fractionK). Therefore,
that that
has the simply
has maximum
the maximum deconvolving
temperaturetemperaturethe within
molten within
each each pixel pixelwill willimpr
fraction that has the maximum temperature uncertainty
uncertainty
within of thepixel
each ofemissivity,
thewillemissivity, kinetic
improve kinetic
temperature,
the temperature,
accuracyand and radiant
and radiant flux. flux.
uncertainty of the emissivity, kinetic temperature, and radiant
Measuring Measuring flux.
accurate accurate thermal thermal properties properties of a of moltena molten lava lava
surface surface
is al
Measuring accurate thermal properties propagation of a propagation
moltenmodels lavamodels [15,24].
surface [15,24].
isWith alsoWith
the increasing
criticalthe increasing
to lava number
flow numberof spectral
of spectral
bandsband in m
propagation models [15,24]. With the increasing (e.g., (e.g.,
numberHyTES HyTES
of [20]
spectraland
[20]bandsthe
andMineral the
in more Mineralandrecent Gas
and TIR Identifier
Gas Identifier
imagers (MAGI) (MAGI)[25]),[25]),
the rad th
(e.g., HyTES [20] and the Mineral and Gas Identifier emissivityemissivity(MAGI)
of an of [25]),
object’s
an the radiative
object’s
surface surface
can be cantemperature
extracted
be extracted with andwith
increasingincreasingaccuracy accuracy[26–
emissivity of an object’s surface can be extracted emissivity
with emissivity can then
increasing canaccuracy
then
be used be[26–28].
used
with with approachesapproaches
A well-constrained such such as linear as linear
spectral spectral
deco
emissivity can then be used with approaches such
quantitatively as linear
quantitatively spectral
determine determine deconvolution
possible possible
spectral modeling
spectral
end-member end-memberto that defines that defines the min the
quantitatively determine possible spectral end-member thermal thermalfractions
that fractions
[29–32].
defines [29–32].
the Additionally, Additionally,
mineralogical, kinetic
textural,kinetic
temperature
and temperature (and (andto a lesseto a
thermal fractions [29–32]. Additionally, kinetic required required
temperature to determine
to (and
determine thea runout
to the runout
lesser distance
degree, distanceand hazard
emissivity) and hazard ispotential potential
usingusing radian r
required to determine the runout distance and hazard
propagation potential
propagation models using
models [15,24]. radiant
[15,24].
However, heat
However, flux
theinaccuracy
lavaaccuracy
the flow of these of these
derived derivedparam p
propagation models [15,24]. However, the accuracy temperature
temperature
of theserangederived
range
of typical of parameters
typical
lavaslavas is less iswell
over less the constrained
well constrained
cooling at theatvarious
the various spati
temperature range of typical lavas is less wellTIR constrained
instruments. at theTherefore,
TIR instruments. various spatial
Therefore, improving resolutions
improving the accuracy of current
the accuracyof theofkineticthe kinetictemperatutempe
TIR instruments. Therefore, improving the accuracy previously-determined
previously-determined
of the kinetic temperature molten molten
fraction
andfraction
should
emissivity should
thenofreduce
then
the reduce the uncertainty
the uncertaint in f
previously-determined molten fraction shoulddirectly thendirectly
reduce
rely on the
rely uncertainty
these
on thesethermal inproperties.
thermal flowproperties.
model analyses that
directly rely on these thermal properties.
3. Data 3. Data
3. Data
3.1. Study3.1. StudyArea Area
3.1. Study Area
This studyThis study was conducted
was conducted during during
two field two field
campaigns campaigns at Kīlauea
at KīlaueaVolca
This study was conducted during two field January/February
January/February
campaigns 2017 2017
at Kīlauea andVolcano 2018,
and 2018, a period a period
in Hawai’i, whenUSA, when
bothin both
the summit the summit lava lakelava ala
January/February 2017 and 2018, a period when lavaboth lava
flows the summit
flows
were were lava
active. lake
active.
It and
focused
It coastal
focused primarily plainon
primarily surface
theon lava the lavalake lakein thein Hale the
lava flows were active. It focused primarily propagating
on propagating
the lava lavalake flows
lava in flowsfrom
the from
the Puʻu
Halemaʻumaʻu the Puʻu ‘Ō’ō Crater
vent
‘Ō’ō vent(Figure
and(Figure 1). Kīlauea
1). Kīlaueais a basais a
propagating lava flows from the Puʻu ‘Ō’ō vent has (Figure
been
has been 1). erupting
erupting Kīlauea nearly isnearly
acontinuously
basaltic shield
continuously forvolcano
theforpastthethat500
pastyears 500 years[33]. [33].
The lava The
has been erupting nearly continuously for the during
past during
long
500 yearssustained
long sustained
[33].eruptions
The eruptions
lava where surfaces where
pāhoehoe
arepāhoehoe (tube-(tube-
produced and surface-fed)
and surface-fed) and
during long sustained eruptions where pāhoehoe [33].(tube-
The
[33].lava The
andlake
lava in
laketheinHalemaʻumaʻu
surface-fed) the and Halemaʻumaʻu
‘a’ā flows Crater
are Crater observed
emplaced observed during during
this studythis stud wa
[33]. The lava lake in the Halemaʻumaʻu Crater observed
2018 2018
and variedandduring in this
varied size instudy
with
size with was active
maximum maximum fromdimensions
dimensions 2008 until of 160ofmeters 160 meters widewide and 2
2018 and varied in size with maximum dimensions 1) [34]. 1) of
During
[34].
160During this period,
meters this
wide period,there
and 225there
were meterswere
fluctuations
fluctuations
long in lava
(Figure in lake
lava activity
lake activity with
1) [34]. During this period, there were fluctuations in lava
and irregular
and lake
irregularsmallactivity
small
explosions, with continuous
explosions, finally finally
ending gas
ending plumes
with with
the summitthe summit collapse collapse
in M
and irregular small explosions, finally endingtime with time
ofthethe of field
summit the field campaigns,
collapse campaigns,
in May the2018 lava
the[13,34].
lava
lake lake level
At the level
was was relatively
relatively high
time of the field campaigns, the lava lake level
approximately was 100–130
approximately relatively
100–130 metershigh meters but below
below not overflowing,
the Halemaʻumaʻu
the Halemaʻumaʻu Crater Crater
rim. It rim.
wasIt cw
approximately 100–130 meters below the Halemaʻumaʻu with with freshfresh lavaCraterupwelling
lavarim. upwelling
It inwas thecontinuously
innorth
the north that migrated
that migrated
circulating to thetosouth, the south,
cooled cooled
and
with fresh lava upwelling
Figure6.6.Comparison
Figure Comparisonof in the
ofsurfacenorth
surface(left that
(leftcolumn) solidified
migrated
column) radiance solidified
to
radiance and lava.
the lava.
The
south,
and (right lava
The
cooled
(right column) then
lava and sank
column) emissivitythenformedsank
in
emissivity pre-the insouth,
the
plates
pre- and south,
of distinguished
cooler,
and post-
post- distinguished by the byoccurren
the occ
solidified
TMP lava. Theof
TMPseparation
separation lava
of thethen
the sank
molten
molten in the derived
fraction
fraction south,and
derived distinguished
degassing
fromandall
from degassing
all the[34].
the by The
datasets the
[34].occurrence
datasets lava
Theflows
acquired
acquired lavaof of of
flows
from
the
thestrong
from
the Puʻu
Halema splattering
theuma
Halemaʻumaʻu Puʻu
‘Ō’ōu vent‘Ō’ō were vent wereactiveactivefor ~30 fo
andCrater
degassing
Crater lava [34].
lavalake
lakeon The lava flows
onFebruary
February from
6,6,2018,
2018,andthe
andPuʻu
Pu locations
u ‘Ō’ō
Puʻu locations
‘Ō’
‘Ō’ōō vent
lavaover
lava wereover
flows numerous
flows active
on onnumerousforeruptive
January
January ~30 30,
30,eruptive
years
2019.
2019. episodes,
erupting
The
The episodes,
error inproducing
error barsproducing
many
bars mostly mostlypāhoehoe pāh
locations
representover
represent the numerous
thestandard
standard eruptive
deviation
deviation episodes,
variation
variation in the
in producing
occasional
ROI
the occasional
ROIdata. ‘a’ā
Note
data. mostly
lava
‘a’ā
there
Note flow
lava pāhoehoe
areflow
there [33,35].
no
areHyTESno lava
[33,35].
This
HyTES data flows
long
This long
between
data with
eruption
between9.92 the
eruption finally finally
ended ended
in 2018 in 20w
occasional
and 10.75 ‘a’ā
µm lava
on flow [33,35].
02/08/2018 so This
these long
data lower
eruption
are not lower
East
included Rift
finallyEast
in
9.92 and 10.75 µ m on 02/08/2018 so these data are not included in the lava lake plots. Zone
Rift
ended
the Zone
lava (LERZ)
in (LERZ)
2018
lake eruption
with
plots. eruption
the in the in
cessation Leilani
the Leilani
of Estates
the Estates
[14]. The
[14]. lava
The flow
lav
lower East Rift Zone (LERZ) eruption in the Leilani 2017–2018 Estates
2017–2018 study [14].
study
wereThewere lavaof
part flows
part theofobserved
61g
theepisode during
61g episode that ourerupted
that erupted from fromthe east the flaea
In general,
6.2017–2018
Discussion the pixel-integrated
study were part of the 61g episode temperaturesflows
thatflows derived
propagated
erupted propagatedfrom
fromdown data
thedownthe prior
eastPulama the
flank to
Pulama the
ofpaliPuʻu TMP
and
pali separation
entered
andThese
‘Ō’ō. entered
the ocean the ocean
at Kamokuna
at Kamo
analysis show an inverse
flows propagated down therelationship
Pulama pali with spatial
and surface
entered surface resolution,
lava
the flow
lava flow
ocean and with
lava
and tubes
at Kamokuna the larger
lava tubes
as(Figure
a series pixels
(Figure
1)of[35]. potentially
1) [35].
pāhoehoeThe majority
The majority of theofflows the flo o
The ability
containing
surface lavathe flow to and
smallestremotely
fractionmeasure
lava tubes of molten accurate
(Figure lava.
1) [35].
surface
The
pāhoehoe,The ASTER radiance,
bothdata
majority
pāhoehoe, bothofhave
kinetic
the
sheet-like the
flows
sheet-like and
temperature,
lowestand kinetic
observed
ropey in were
ropey
and
texture.
in
emissivity
temperatures
tube-fed
texture.
Previous Previous remote remotesensing senss
ofpāhoehoe,
and athe
molten
MMT-Camlava
both surface
sheet-like
have the within a thermally-mixed
and ropey
highest. Kīlauea
in texture.
Compared with pixel
Kīlauea
eruption
Previous
the is
35-m important.
eruptionhave have
remote
spatial sensing Developing
investigatedinvestigated
resolution studies lavaof
HyTES adischarge
methodology
lava
thedata, discharge
pre-2018 rates rates
however, to
[36], [36],
lava flowlava flow emp
extract
Kīlauea only the molten
eruption have fraction within
investigated lava every
discharge pixel
and lava of
rates these
[36], datasets
lava flow and constrain
emplacement the
tracking uncertainty
[37],
the 50-m spatial resolution MASTER data do have aand
higherpathway
lavaderived
pathway mechanisms mechanisms
pixel [38]. The
integrated [38]. areas
The areas
temperature for this for
butstudy
this
at study were werechosen chose
for
aimproves
and lava
higher the subsequently-derived
pathway
variability. mechanisms
All the kinetic [38]. thermal
The areas
temperatures data
observing
forderived
thisrequired
observing molten
study from for
were themonitoring,
molten
lava surfaces
lava surfaces
chosen
mixed-pixel for the scientific
combined combined
high
data withanalysis,
probability
are with
significantly of andavailability
the availability
the of a variety
of a va
later modeling
observing studies.
molten lava The results from the TMP separation analysis show that greatly improved
below that expected for a surfaces combined
molten basaltic lavawith the
datasets.
surface availability
datasets.
(~1450 K). Inofcontrast,
a variety theofaverage
remote sensing
derived TIR
molten
radiance,
datasets. kinetic temperature, and emissivity values can be extracted
fraction temperature in every dataset is above 1100 K, significantly closer to what is expected for molten at different spatial and spectral
scales. This methodology can be implemented with a variety of data quickly and uncertainties
basaltic lava. Unmixing the HyTES data produced the highest average molten fraction temperature
quantified. The TMP separation analysis is not new but has had limited application in volcanic hazard
(1266.1 K) but with a high variability (404 K), implying there is still significant uncertainty at 35-m
prediction models and assessments. Prior studies have shown that a typical lava surface has multiple
resolution. The MMT-Cam derived molten fraction temperatures span the liquidus temperatures
thermal fractions (up to 8) at high spatial scales [7,23,47]. However, the processing required to analyze
these fractions can be daunting at the scale of an entire flow field or slow in the case of an ongoing
eruption. Furthermore, there is a somewhat limited applicability for this level of multiple fraction
Remote Sens. 2020, 12, 193 14 of 21

of basaltic lava at the lowest variability, which provides the highest constraint on the derived data.
The integration of the TMP calculation into the derivation of temperature improves the accuracy of the
measured kinetic temperature associated with the molten lava fraction in all the datasets.
Emissivity spectra derived from these data before the TMP separation analysis have similar
spectral morphologies to laboratory-derived results of molten basalts [44–46]. However, the spectral
contrast is less than expected from laboratory results by up to 40% for ASTER (Figure 6). The MMT-Cam
emissivity values are less than 15% shallower than laboratory data, with the MASTER and HyTES
emissivity values being within 20%–30% (Figure 6). Following the TMP separation analysis, the spectral
depths and contrasts increased to values consistent with those derived from laboratory experiments
(Figure 6) [44–46]. The spectral morphology remained similar with a decrease in emissivity at shorter
wavelengths and an increase at longer wavelengths. These results are also more exaggerated in the lava
flow data (Figure 6), implying that there is high thermal mixing within a pixel representing the lava flow
than the lava lake. For example, the improvement in the ASTER data highlights an absorption feature
centered at ~8.5 µm with an increase in emissivity at longer wavelengths compared to absorption
features centered at ~8.0 µm and ~10.5 µm in the improved MMT-Cam data. The improvements are
attributed to variations in spatial and spectral resolution between the two instruments, as well as the
difference in channel locations that resolve slightly different Si-O bonding. Similar trends are also
detected in the HyTES and MMT-Cam data.

6. Discussion
The ability to remotely measure accurate surface radiance, kinetic temperature, and emissivity
of a molten lava surface within a thermally-mixed pixel is important. Developing a methodology to
extract only the molten fraction within every pixel of these datasets and constrain the uncertainty
improves the subsequently-derived thermal data required for monitoring, scientific analysis, and later
modeling studies. The results from the TMP separation analysis show that greatly improved radiance,
kinetic temperature, and emissivity values can be extracted at different spatial and spectral scales.
This methodology can be implemented with a variety of data quickly and uncertainties quantified.
The TMP separation analysis is not new but has had limited application in volcanic hazard prediction
models and assessments. Prior studies have shown that a typical lava surface has multiple thermal
fractions (up to 8) at high spatial scales [7,23,47]. However, the processing required to analyze these
fractions can be daunting at the scale of an entire flow field or slow in the case of an ongoing eruption.
Furthermore, there is a somewhat limited applicability for this level of multiple fraction analysis in
current lava flow propagation models where only the maximum molten fraction has the greatest
influence on model results [15,24]. Therefore, improving the analysis of remote sensing data to provide
rapid kinetic temperature and emissivity values of the highest temperature fractions within TIR image
pixels will greatly improve and further constrain lava flow propagation models.

6.1. Emissivity
The spatial resolution of the TIR dataset has only a limited effect on the morphology of the
emissivity spectra but does strongly influence the spectral depth. This result shows that the efficiency
of radiative heat loss from a molten surface is overestimated more in TIR data with lower spatial
resolutions. The emissivity spectra of the molten fraction show improved accuracy in the spectral
contrast to those values expected for molten basaltic lava surfaces within the uncertainty calculated
(Figure 6). The average minimum emissivity decreases by 20%, with the largest decrease observed in
the HyTES data (33%) and the smallest decrease observed in the MMT-Cam data (4%). Additionally,
larger decreases in emissivity are observed at the shorter wavelengths in all the datasets (Figure 6),
which highlights the non-uniform influence of kinetic temperature on emissivity and the non-uniform
mixing of thermal anomalies. Our results show that for future thermal studies of molten basaltic
surface, a more appropriate value for the minimum emissivity would be 0.66 rather than the common
values of 0.95 to 1.0 used in prior studies (i.e., [18,44,48–50]) and thermorheological models of lava flow
[33]. The lava lake in the Halemaʻumaʻu Crater observed during this study was active from 2008 until
2018 and varied in size with maximum dimensions of 160 meters wide and 225 meters long (Figure
1) [34]. During this period, there were fluctuations in lava lake activity with continuous gas plumes
and irregular small explosions, finally ending with the summit collapse in May 2018 [13,34]. At the
time of the field
Remote Sens.campaigns,
2020, 12, 193 the lava lake level was relatively high but not overflowing, 15 of 21
approximately 100–130 meters below the Halemaʻumaʻu Crater rim. It was continuously circulating
with fresh lava upwelling in the north that migrated to the south, cooled and formed plates of cooler,
solidified lava. The lava(i.e.,
propagation then[15,24]).
sank in the
Thissouth, distinguished
minimum by the
emissivity valueoccurrence of strong
is consistent withsplattering
a previous study of 0.55
and degassing
for a[34]. The lava
cooling flows
basaltic from
lava the Pu
from Puʻu
u ‘Ō’ō
‘Ō’ōvent were
at 1323 Kactive for ~30
[44] and years erupting
a laboratory studyinofmany
melts (albeit silicic
locations ones)
over numerous eruptive episodes, producing mostly pāhoehoe lava
that measured emissivity of 0.68 at 1573 K [46]. However, the use of these higher flows with the maximum
occasionalemissivity
‘a’ā lava flow [33,35].
values usedThis long past
in these eruption finally
studies ended basaltic
of cooled in 2018 with
lava the cessation
surfaces of the
is appropriate given the
lower Eastresults
Rift Zone (LERZ) in
calculated eruption in the
this study. Leilani
The TMPEstates [14]. analysis
separation The lavaidentifying
flows observed during fraction
the molten our provides a
2017–2018useful
study approach
were part forof the 61g episode
deriving that erupted
and evaluating the from
actualthe east flankofofmolten
emissivity Puʻu ‘Ō’ō. These
material seen in a variety
flows propagated down the Pulama pali and entered the ocean at
of TIR instruments independent of spatial and spectral resolution. Kamokuna as a series of pāhoehoe
surface lava flow and lava tubes (Figure 1) [35]. The majority of the flows observed were tube-fed
pāhoehoe,6.2.both sheet-like
Kinetic and ropey in texture. Previous remote sensing studies of the pre-2018
Temperature
Kīlauea eruption have investigated lava discharge rates [36], lava flow emplacement tracking [37],
and lava pathway Themechanisms
maximum [38]. temperatures
The areas derived fromwere
for this study all the datasets
chosen following
for the the TMP
high probability of separation are
observing molten lava surfaces combined with the availability of a variety of remote sensing TIR data (Table 3).
closer to the liquidus temperatures of Hawaiian basalt [51], within the variability of the
datasets. The analysis improves the derived kinetic temperatures by up to 250% (ASTER TIR) and 70% in the
MMT-Cam data. Although the eruption temperature of Hawaiian basalt is well-known from past
direct measurements and petrologic analysis, the variability in the temperatures from this approach
is critical for evaluating the uncertainty in subsequent studies that rely upon these measurements.
The quantification of the temperature uncertainty improves the confidence and understanding of those
model predictions, allowing more informed conclusions to be drawn from their forecasting estimates,
for example.

6.3. Accuracy Assessment

The ability to measure accurate thermal properties of a surface from calibrated TIR data is
influenced by numerous instrument factors including the spatial, spectral, and temporal resolution of
those data as well as external factors such as the spatiotemporal variability of the surface itself. Molten
lava surfaces vary, over seconds to minutes and centimeters to meters, either by lava flow propagation
(breakouts) or lava lake overturning. To analyze the influence of the spatial and temporal scales,
a variety of TIR datasets with different resolutions were needed. Most importantly, the new MMT-Cam
data are used to validate the lower resolution datasets and determine whether both the anomalies
and processes (e.g., lava lake overturning and flow crustal formation/morphology) are captured [12].
Typically, lower spatial resolution increases aggregation that leads to an increase in variability and a
decrease in the ability to quantify the small-scale details.
Our analysis shows that the airborne and orbital data provide reliable and accurate results for the
larger-scale anomalies and processes. Typically, anomalies 1.5 to 2 times the size of the spatial resolution
of the dataset are required for more accurate estimates of kinetic temperature and emissivity, whereas
even larger volcanic processes are required to be discerned individually. For example, the ~200 m
diameter lava lake and >100 m long lava flows provide good targets for anomaly detection; however,
the overturning and crustal formation observed within these anomalies are not identified in ASTER
data and rarely in the airborne TIR data.
The temporal resolution also significantly influences the reliability of the TIR data, especially if
the repeat time is greater than hours (which is the case for the ASTER and airborne instruments). As a
result, these data underestimate the percentage of molten lava on the surface and its accurate radiance,
kinetic temperature, and emissivity. Compared to the MMT-Cam TMP separation analysis, the same
analysis of the airborne and orbital data underestimates the surface radiance by as much as 430% and
800%, respectively (Figures 4, 7 and 8). This translates to a lava emissivity error of 5% and 20% with a
corresponding kinetic temperature error of 80% and 120%, respectively (Figures 4, 7 and 8). Separately
evaluating the lava lake and lava flow data reveals the airborne and orbital data of the lava lake are
30% better compared to the lava flow data due mainly to the larger spatial scale and more uniform
surface state of the lake (Figures 4, 7 and 8).
 the same analysis of the airborne and orbital data underestimates the surface radiance by as much as
currently
currently
available
available
low spatial
low spatial
resolution
resolution
(>500(>500
m) TIR m)datasets
TIR datasetsthat are
thatnot areable
not toable
accurately
to accurately
resolve
resolve
430%
the maximum
and 800%,
the maximum
respectively
temperature
temperature
(Figures 4, emissivity
or representative
or representative
7, and 8). This
emissivity
translates
spectrum.
spectrum.
to
ThereThere
a lava
are relatively
emissivity
are relatively
error(i.e.,
few studies
of 5%
few studies
and
(i.e.,
20% with
[23]) [23])
that have a corresponding
that have
investigated
investigated kinetic
the influence temperature
the influenceof spatial error
of spatial of
and spectral 80%
and spectral and 120%,
resolution
resolution respectively
on theseon these (Figures
calculations
calculations 4, 7,
over over and
8). Separately
temperatures
temperatureswhereevaluating
where
a molten the
a molten lava
lava surfacelake
lava surfaceand lava
coolscools
(<1450flow
(<1450 data reveals
K). Therefore,
K). Therefore, the
simply airborne
simply
deconvolving and
deconvolvingorbital the moltenthe
data
the molten of
lava
fraction
Remote lake
thatare
fraction
Sens. has30%
that
2020, 12,the
has better
193 maximum compared
the maximum to the lava
temperature
temperature flow
within within data
each duepixel
each
pixel mainly
will will toimprove
improve the the
larger thespatial
accuracy scale
accuracy
and and
and
16 of 21
more
uncertaintyuniform
uncertainty surface
of theofemissivity, state
the emissivity, of the
kinetic lake
kinetic (Figures
temperature,
temperature, 4, 7,
and radiantand
and radiant8).
flux. flux.
Measuring
Measuring accurate accurate
thermal thermal
properties
propertiesof a of
molten
a moltenlava lava
surfacesurface
is also is also
critical
critical
to lavato lava
flow flow
propagation
propagation
models models[15,24].[15,24].
WithWiththe increasing
the increasingnumber number
of spectral
of spectral
bandsbandsin more in more
recentrecent
TIR imagers
TIR imagers
(e.g., (e.g.,
HyTES HyTES
[20] and
[20] the and Mineral
the Mineral and Gasand Identifier
Gas Identifier(MAGI) (MAGI) [25]),[25]),
the radiative
the radiativetemperature
temperatureand and
emissivity
emissivity
of anof object’s
an object’ssurface surface
can be can
extracted
be extracted
with with
increasing
increasing
accuracy
accuracy[26–28].[26–28].
A well-constrained
A well-constrained
emissivity
emissivity
can thencan then be used be usedwith with
approaches
approaches such suchas linear
as linear
spectral
spectral
deconvolution
deconvolution modeling
modeling
to to
quantitatively
quantitativelydeterminedeterminepossiblepossible
spectral
spectral
end-member
end-member that defines
that defines
the mineralogical,
the mineralogical, textural,
textural,
and and
thermalthermal
fractions
fractions
[29–32]. [29–32].
Additionally,
Additionally, kinetickinetic
temperature
temperature (and (and
to a to lesser
a lesser
degree,degree,
emissivity)
emissivity)
is is
required
required
to determine
to determine the runout
the runoutdistance
distance
and hazard
and hazardpotential
potential
usingusing
radiant radiant
heat heat
flux influxlava
in flow
lava flow
propagation
propagationmodels models[15,24]. [15,24].
However,
However,the accuracy
the accuracyof theseof these
derivedderived
parameters
parameters over over
the cooling
the cooling
temperature
temperature
rangerange of typical
of typical
lavaslavas
is lessiswell
less constrained
well constrained at theatvarious
the various
spatial spatial
resolutions
resolutions
of current
of current
TIR instruments.
TIR instruments. Therefore,
Therefore,improving
improvingthe accuracy
the accuracy
of theofkinetic
the kinetic
temperature
temperature and emissivity
and emissivity of theof the
previously-determined
previously-determined molten molten
fraction
fraction
should should
then reduce
then reducethe uncertainty
the uncertainty in flow in flow
model model
analyses
analyses
that that
directly
directly
rely onrelythese
on these
thermal thermal
properties.
properties.

3. Data
3. Data

3.1. Study
3.1. Study
Area Area
This This
studystudy was conducted
was conducted during during
two fieldtwo field
campaigns
campaigns at Kīlauea
at Kīlauea Volcano Volcano
in Hawai’i,
in Hawai’i, USA,USA, in in
January/February
January/February 2017 2017
and 2018,
and 2018, a perioda periodwhenwhen both both
the summit
the summit lava lakelava and
lake coastal
and coastalplainplainsurface surface
lava lava
flowsflowswerewere active. active.
It focused
It focusedprimarily primarilyon the on lavathe lava
lake lake in the in Halemaʻumaʻu
the Halemaʻumaʻu Crater Crater
and and
propagating
propagating lava lavaflowsflowsfromfrom the Puʻuthe Puʻu‘Ō’ō ‘Ō’ō vent vent
(Figure (Figure
1). Kīlauea
1). Kīlauea is a basaltic
is a basaltic
shield shield
volcano volcano
that that
has been
has beeneruptingerupting nearly nearly
continuously
continuously for the forpast
the past
500 years500 years[33]. [33].
The lavaThe lava
surfacessurfaces
are produced
are produced
duringduring
long long
sustainedsustainederuptions
eruptionswhere where
pāhoehoe pāhoehoe(tube-(tube-
and surface-fed)
and surface-fed) and ‘a’ā
andflows
‘a’ā flows
are emplaced
are emplaced
[33]. The
[33].lava
The lake
lava in laketheinHalemaʻumaʻu
the Halemaʻumaʻu Crater Crater
observed
observed during during
this studythis study
was active
was active
fromfrom 2008 2008
until until
2018 2018
and varied
and varied in size in with
size with
maximum maximum dimensions
dimensionsof 160 ofmeters
160 meters widewide and 225andmeters
225 meterslong long (Figure (Figure
1) [34].
1) [34].
During Duringthis period,
this period,
theretherewerewere fluctuations
fluctuations
in lava in lake
lava activity
lake activity with with
continuous
continuous gas plumes
gas plumes
and irregular
and irregular smallsmall explosions,
explosions, finally finally
ending ending
with with
the summit
the summit collapse collapse
in May in May
2018 2018
[13,34]. [13,34].
At the At the
time time
of the of field
the field campaigns,
campaigns, the lavathe lava lake lakelevel level
was was relatively
relatively high high
but not but overflowing,
not overflowing,
approximately
approximately 100–130 100–130
meters meters
below below
the Halemaʻumaʻu
the Halemaʻumaʻu Crater Crater
rim. It rim.was It continuously
was continuously circulating
circulating
with with
freshfresh
lava upwelling
lava upwelling in theinnorththe north
that migrated
that migrated to thetosouth,
the south,cooled cooled
and formed
and formed platesplates
of cooler,
of cooler,
solidified
solidified
lava.
Figure
Figure 7.lava.
7. The
The lavaThe then
ratio lava
betweenthen
sankthe sank
inpre-
theinsouth,
pre- thepost-TMP
and
and south,
distinguished
post-TMP distinguished
separation
separation by theofby
of occurrence
the
the the occurrence
molten
molten of strong
fraction
fraction of
forstrong
for splattering
the
the (a)splattering
(A) surface
surface
and degassing
and degassing
radiance [34]. The
[34]. lava
The flows
lava flows
from from
the Puʻu
the Puʻu
‘Ō’ō vent
‘Ō’ō were
vent
radiance and (b) emissivity for the Halemaʻumaʻu Crater lava lake acquired on February 6,Values
and (B) emissivity for the Halema uma u Crater lava were
active
lake active
for
acquired ~30
for years
on ~30 years
erupting
February erupting
6, in
2018. manyin many
2018.
locations
locations
Values closer to 1.0 require less TMP separation processing. The error bars represent the standardthe
over
closer to over
numerous
1.0 numerous
require eruptive
less TMP eruptiveepisodes,
separation episodes,producing
processing.producing
Themostly
error mostlypāhoehoe
bars pāhoehoe
represent lava
the lava
flows
standard flows
with with
the
deviation
occasional
occasional‘a’ā lava
variation
deviation ‘a’ā
in thelava
flow
variationROIflow
[33,35].
data.
in the[33,35].
NoteThisthere
ROI This
longare
data. long
eruption
Note eruption
nothere
HyTES finally
aredata finally
no ended
between
HyTES ended
in 2018
9.92
data in 2018
andwith
between10.75with
the µm
9.92cessation
the
on cessation
and 10.75ofµthe
02/08/2018 mof sothe
on
lowerlower
East
these Rift
East Zone
data are
02/08/2018 Rift Zone
(LERZ) (LERZ)
not included
so these eruption
data areinnot eruption
in
theincludedthein Leilani
the
lava lakeinplots.Leilani
Estates Estates
[14].
the lava lake plots. [14].
The lava
The flows
lava flows
observed
observed during during
our our
2017–2018
2017–2018studystudy werewere part of part theof61g
the episode
61g episode that erupted
that erupted fromfrom the east the flank
east flank
of Puʻu of Puʻu
‘Ō’ō. ‘Ō’ō.
TheseThese
Larger
flowsflows
propagated pixel
propagated downsizes
down
thealso
Pulama
the have
Pulamaa
pali higher
and
pali probability
entered
and entered
the of
ocean
Larger pixel sizes also have a higher probability of integrating more than one surface the integrating
ocean
at Kamokuna
at more
Kamokuna as athan
series
as a one
series
of surface
pāhoehoe
of thermal
pāhoehoe
thermal
property,
surface
surface
lava which
lava
flow leads
flow
and to
lava
and errors
lava
tubes in
tubesthe
(Figure data
(Figure
1) analysis
[35].
1) [35].
The and ultimately
majority
The
property, which leads to errors in the data analysis and ultimately subsequent results that aremajority
of the subsequent
of flows
the flows
observedresults
observed that
were are
were inaccurate.
tube-fedtube-fed
The dual-band
pāhoehoe,
pāhoehoe,
inaccurate. both The mixed
both
sheet-like
dual-band pixel
sheet-like
and approach
ropey
mixedand pixel inistexture.
ropey one possible
in texture.
approach Previous solution
is Previous
one remote
possible and
remote results
sensing show
sensing
solution studies that
andstudies of it
results the does
of theimprove
pre-2018
show pre-2018
that the
it does
Kīlauea
Kīlauea
deviation eruptioneruption
of thermal have have
investigated
investigated
properties lava
of molten lava
discharge
discharge
lava of rates
surfaces rates
[36], [36],
lava lava
flow flow
emplacement
emplacement tracking tracking
[37], [37],
improve the deviation of thermal properties molten(Figureslava surfaces7 and 8). The molten
(Figures 7 andsurface8). Theradiance
molten
and lava
and are
values pathway
lavaall pathway mechanisms
within mechanisms
25% of the [38].MMT-Cam
[38].
The areas The areas
for this
forstudy
values, this study
werewere
an improvement chosen chosen
for
of upthefor
tohigh
the
300% high
probability
probability
(Figures of
7 andof8).
observing
Theobserving
molten molten moltenlava lava
emissivity surfaces
andsurfaces combined
kinetic combined with with
temperature thevalues
availability
the availability
derived of from
a variety
of athevariety
of remote
airborneof remotesensing
and sensing
orbital TIR TIRare
data
datasets.
datasets.
all within 25% and 10% of the MMT-Cam values (Figures 7 and 8), respectively, and more consistent
with laboratory and field measurements [44–46,51].
 Remote
currently Sens. 2019,
available 11,spatial
low x REVISION
resolution (>500 m) TIR datasets that are not able to accurately resolve 17 of 21
the maximum temperature or representative emissivity spectrum. There are relatively few studies (i.e.,
surface
[23]) that haveradiance values
investigated theare all within
influence 25% of
of spatial the
and MMT-Cam
spectral values,
resolution an improvement
on these of up to 300%
calculations over
(Figureswhere
temperatures 7 anda 8). The lava
molten molten emissivity
surface and kinetic
cools (<1450 temperature
K). Therefore, values derived
simply deconvolving thefrom the airborne
molten
and
fraction orbital
that
Remote data
the12,are
has2020,
Sens. 193 all within
maximum 25% and 10%
temperature of the
within MMT-Cam
each pixel willvalues
improve(Figures 7 and 8),and
the accuracy respectively,
17 of 21
and more
uncertainty of theconsistent
emissivity, with laboratory
kinetic and field
temperature, measurements
and radiant flux. [44–46,51].
Measuring accurate thermal properties of a molten lava surface is also critical to lava flow
propagation models [15,24]. With the increasing number of spectral bands in more recent TIR imagers
(e.g., HyTES [20] and the Mineral and Gas Identifier (MAGI) [25]), the radiative temperature and
emissivity of an object’s surface can be extracted with increasing accuracy [26–28]. A well-constrained
emissivity can then be used with approaches such as linear spectral deconvolution modeling to
quantitatively determine possible spectral end-member that defines the mineralogical, textural, and
thermal fractions [29–32]. Additionally, kinetic temperature (and to a lesser degree, emissivity) is
required to determine the runout distance and hazard potential using radiant heat flux in lava flow
propagation models [15,24]. However, the accuracy of these derived parameters over the cooling
temperature range of typical lavas is less well constrained at the various spatial resolutions of current
TIR instruments. Therefore, improving the accuracy of the kinetic temperature and emissivity of the
previously-determined molten fraction should then reduce the uncertainty in flow model analyses that
directly rely on these thermal properties.

3. Data

3.1. Study Area

This study was conducted during two field campaigns at Kīlauea Volcano in Hawai’i, USA, in
January/February 2017 and 2018, a period when both the summit lava lake and coastal plain surface
lava flows were active. It focused primarily on the lava lake in the Halemaʻumaʻu Crater and
propagating lava flows from the Puʻu ‘Ō’ō vent (Figure 1). Kīlauea is a basaltic shield volcano that
has been erupting nearly continuously for the past 500 years [33]. The lava surfaces are produced
during long sustained eruptions where pāhoehoe (tube- and surface-fed) and ‘a’ā flows are emplaced
[33]. The lava lake in the Halemaʻumaʻu Crater observed during this study was active from 2008 until
2018 and varied in size with maximum dimensions of 160 meters wide and 225 meters long (Figure
1) [34]. During this period, there were fluctuations in lava lake activity with continuous gas plumes
and irregular small explosions, finally ending with the summit collapse in May 2018 [13,34]. At the
time of the field campaigns, the lava lake level was relatively high but not overflowing,
approximately 100–130 meters below the Halemaʻumaʻu Crater rim. It was continuously circulating
with fresh lava upwelling in the north that migrated to the south, cooled and formed plates of cooler,
solidified lava.
FigureThe lava
8. The thenbetween
The ratio
ratio sank in the
between thepre-
the south,
pre- anddistinguished by the occurrence
post-TMP separation
separation of the molten
molten offraction
strong for
fraction splattering
for the
the (A)
(a) surface
surface
and degassing [34]. and
radiance The(B) lava
(b) flows from
emissivity
emissivity forthe
for thePuʻu
the u ‘‘Ō’ō
Puʻu
Pu ‘Ō’ō
Ō’ vent were
lavaflows
ō lava active
flowsacquired
acquiredforon~30 years 30,
onJanuary
January erupting
30, 2018.
2018. inValues
many
Values closer
closer to
locations overto
1.0 1.0 numerous
require
require lesseruptive
less TMP TMP
separation episodes,
separation producing
processing.
processing. The error mostly
The error
bars pāhoehoe
bars represent
represent lava the
the standard flows withvariation
standard
deviation the
deviation in
occasional variation
‘a’āROI
the lavadata.
flow
in the [33,35].
ROI data. This long eruption finally ended in 2018 with the cessation of the
lower East Rift Zone (LERZ) eruption in the Leilani Estates [14]. The lava flows observed during our
6.4. Implications
6.4.
2017–2018 Implications
study wereand and Reasons
partReasons
of the 61g for Uncertainty
for Uncertainty
episode that erupted from the east flank of Puʻu ‘Ō’ō. These
flows propagated down the Pulama
The results pali and entereddata the ocean atand
Kamokuna as a series of pāhoehoe
The results extracted
extractedfrom fromthe thelava
lavaflow
flow data before
before after
and thethe
after TMP TMPseparation
separationanalysis are less
analysis are
surface lava
uniform flow and
across lava
all tubes
wavelengths (Figure 1)
where [35]. The
compared majority
to the of the
lava flows
lake observed
data (Figureswere7 tube-fed
and 8). The ratio of
less uniform across all wavelengths where compared to the lava lake data (Figures 7 and 8). The ratio
pāhoehoe,
pre- both sheet-like
versus post-TMP and ropeyisinhigher
analysis texture.
at Previous
shorter remote sensing
wavelengths in thestudies
surface ofradiance
the pre-2018
data and lower
of pre-
Kīlauea versus
eruption havepost-TMP
investigated analysis is higher at
lava discharge shorter
rates [36], wavelengths in the surface
lava flow emplacement radiance
tracking [37], data and
at shorter
lower wavelengths
at shorter wavelengths in the emissivity data as expected from the relationship modeled in the Planck
and lava pathway mechanisms [38]. in Thethe emissivity
areas for this data
studyaswere
expected
chosen from thehigh
for the relationship
probabilitymodeled
of in the
equation.
Planck This
equation. ratio
This is closer
ratio isto one
closer at
tolonger
one atwavelengths
longer for
wavelengths all the thermal
for all the properties,
thermal implying
properties,
observing molten lava surfaces combined with the availability of a variety of remote sensing TIR
the thermal
implying themixing
thermalismixingless of is a factor
less of at these at
a factor wavelengths, again, as
these wavelengths, one might
again, as onepredict based on
might predict the
based
datasets.
mixing
on of multi-temperature
the mixing emissions.
of multi-temperature The non-uniform
emissions. ratio in the
The non-uniform lava
ratio in flow data flow
the lava (mostdata
notable at
(most
larger pixel sizes) implies a more complex mixing and distribution of thermal fractions
notable at larger pixel sizes) implies a more complex mixing and distribution of thermal fractions and and a greater
auncertainty in these properties
greater uncertainty at shorter wavelengths.
in these properties Hence, the Hence,
at shorter wavelengths. greatesttheerrors are observed
greatest errors areat
shorter wavelengths
observed in the mixed
at shorter wavelengths in pixels and pixels
the mixed are more
andextreme
are moreinextreme
lower spatial
in lower resolution datasets
spatial resolution
(Figures 7 and 8).
datasets (Figures 7 and 8).
Although the
Although the greatest
greatest uncertainty
uncertainty and
and errors
errors are
are observed
observed inin the
the mixed
mixed pixels
pixels acquired
acquired atat lower
lower
spatial resolutions; these uncertainties fall within the expected values following the TMP
spatial resolutions; these uncertainties fall within the expected values following the TMP separation separation for
all the datasets. The variability within each dataset is used to quantify the uncertainty
for all the datasets. The variability within each dataset is used to quantify the uncertainty of the of the derived
thermal properties of molten basaltic lava surfaces. Although the dual-band mixed pixel separation
approach does provide data more similar to laboratory results, it ultimately increases the variability in
Remote Sens. 2020, 12, 193 18 of 21

the derived properties that results in a decrease in precision and therefore, an increase in uncertainty.
This uncertainty is related to the spatial resolution of the dataset with lower resolution data having
higher uncertainty. This is a function of smaller proportions of the molten lava fraction being present
within a given larger pixel, which can be less than 5% in an ASTER pixel. Figures 7 and 8 show the
ratio in radiance and emissivity between pre- and post-TMP separation analysis, with values close to
one requiring the least separation processing. In the majority of instances, the MMT-Cam data requires
the least processing and the ASTER data requires the most. However, the HyTES lava lake surface
radiance data requires the least separation processing, a function of the hyperspectral resolution, which
offsets the lower spatial resolution (Figure 7a). The HyTES variability in the unmixed surface radiance
data is significantly lower than that from MASTER at a very similar spatial resolution. Therefore,
where spatial resolution is similar, data uncertainty drops by using higher spectral resolution data
(decrease variability by ~40%), which is highly relevant for future orbital instrument design.

7. Conclusions
The accuracy and uncertainty in the thermal properties derived from remotely acquired TIR
data of active lava surfaces were investigated using a variety of instruments acquiring data of two
active basaltic lava surfaces (lava lake and lava flow). The effect of spatial and spectral resolution
on the measured surface radiance and derived emissivity and kinetic temperature were quantified
by comparing them to values expected for basaltic lava at liquidus temperatures based on prior
laboratory and field results. Because a majority of currently available TIR instruments do not have
the radiometric range or spatial resolution to derive the thermal properties of a molten lava surface
accurately, the application of a dual-band mixed pixel separation analysis approach is one solution to
improve results. This thermal unmixing can deconvolve the signature of the molten fraction within a
pixel. By determining the accuracy and uncertainty in these thermal properties across four different TIR
datasets acquired at the same time and with different spatial and spectral resolutions, the temperature,
emissivity, and radiance results are compared and constrained. For example, prior to extracting the
molten fraction within each pixel, it was impossible to compare thermal properties (Figure 4) as they
are strongly dependent on the instrument position with respect to the surface and their individual
specifications (Table 2). However, post-unmixing, the molten fraction (Figure 5) is more directly
comparable between these datasets (Table 3).
Mixed pixel surface radiance values derived from the ASTER data are ~2400% underestimated,
with the MASTER and HyTES data underestimated by ~1000%. Similar underestimates are seen in
the extracted emissivity and kinetic temperature by approximately 20% and 250% in the ASTER data,
and 25% and ~200% in the MASTER and HyTES data, respectively. However, this impact of spatial
resolution is mitigated to a degree by improved spectral resolution. Following the TMP separation
analysis, all surface radiance values are within 15% of the expected values, whereas the emissivity and
kinetic temperature are within 8% and 12% of the expected values, respectively [44–46]. These results
quantify the inherent TIR data uncertainty in the measured and derived thermal properties, demonstrate
a significant improvement on previous estimates, and further constrains the errors associated with
these values. The more accurate constraint of lava temperature, emissivity, and the emitted radiance
from active surfaces derived from TIR measurements will ultimately improve the accuracy and reduce
the unknown uncertainty in future flow models that rely upon these properties (e.g., [11]). Furthermore,
these datasets can be directly compared with other measurements (i.e., terrain elevation, in situ
thermocouple temperatures, and deformation change) to improve analysis of the synoptic eruption
process and quantify the uncertainty in the results and conclusions.

Author Contributions: The following contributions were provided by each author: Conceptualization, J.O.T.
and M.S.R.; data curation, J.O.T.; formal analysis, J.O.T.; funding acquisition, J.O.T. and M.S.R.; investigation,
J.O.T.; methodology, J.O.T.; project administration, M.S.R.; resources, J.O.T. and M.S.R.; supervision, M.S.R.;
validation, M.S.R.; Visualization, J.O.T.; writing—original draft, J.O.T.; writing—review and editing, J.O.T. and
M.S.R. All authors have read and agreed to the published version of the manuscript.
Remote Sens. 2020, 12, 193 19 of 21

Funding: This research is funded by NASA grants 80NSSC18K1001 (M.S.R.) and 80NSSC17K0445 P00001 (J.O.T.).
Acknowledgments: The authors would like to thank the USGS HVO for their assistance in conducting the
field campaigns, especially Matthew Patrick, Greg Vaughan, and Tina Neal. Additional thanks to the NASA
HyspIRI Preparatory Campaign Group, the NASA ground and flight teams, the ASTER science team, and Hawai’i
Volcanoes National Park for facilitating the field campaigns in 2017 and 2018. We would also like to thank the
Editors and anonymous reviewers for their comments and suggestions that greatly improved this manuscript.
Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the
study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to
publish the results.

References
1. Spampinato, L.; Calvari, S.; Oppenheimer, C.; Boschi, E. Volcano Surveillance Using Infrared Cameras.
Earth-Sci. Rev. 2011, 106, 63–91. [CrossRef]
2. Cashman, K.V.; Stephen, R.; Sparks, R.S.J. How Volcanoes Work: A 25 Year Perspective. Bull. Geol. Soc. Am.
2013, 125, 664–690. [CrossRef]
3. Ramsey, M.S.; Harris, A.J.L. Volcanology 2020: How Will Thermal Remote Sensing of Volcanic Surface
Activity Evolve over the next Decade? J. Volcanol. Geotherm. Res. 2013, 249, 217–233. [CrossRef]
4. Calvari, S.; Lodato, L.; Steffke, A.; Cristaldi, A.; Harris, A.J.L.; Spampinato, L.; Boschi, E. The 2007 Stromboli
Eruption: Event Chronology and Effusion Rates Using Thermal Infrared Data. J. Geophys. Res. Solid Earth
2010, 115, 1–20. [CrossRef]
5. Murphy, S.W.; Wright, R.; Oppenheimer, C.; Filho, C.R.S. MODIS and ASTER Synergy for Characterizing
Thermal Volcanic Activity. Remote Sens. Environ. 2013, 131, 195–205. [CrossRef]
6. Yamaguchi, Y.; Kahle, A.B.; Tsu, H.; Kawakami, T.; Pniel, M. Overview of Advanced Spaceborne Thermal
Emission and Reflection Radiometer (ASTER). IEEE Trans. Geosci. Remote Sens. 1998. [CrossRef]
7. Flynn, L.P.; Harris, A.J.L.; Wright, R. Improved Identification of Volcanic Features Using Landsat 7 ETM+.
Remote Sens. Environ. 2001, 78, 180–193. [CrossRef]
8. Handcock, R.N.; Gillespie, A.R.; Cherkauer, K.A.; Kay, J.E.; Burges, S.J.; Kampf, S.K. Accuracy and Uncertainty
of Thermal-Infrared Remote Sensing of Stream Temperatures at Multiple Spatial Scales. Remote Sens. Environ.
2006, 100, 427–440. [CrossRef]
9. Roberts, D.A.; Quattrochi, D.A.; Hulley, G.C.; Hook, S.J.; Green, R.O. Synergies between VSWIR and TIR
Data for the Urban Environment: An Evaluation of the Potential for the Hyperspectral Infrared Imager
(HyspIRI) Decadal Survey Mission. Remote Sens. Environ. 2012, 117, 83–101. [CrossRef]
10. Western, L.M.; Watson, M.I.; Francis, P.N. Uncertainty in Two-Channel Infrared Remote Sensing Retrievals
of a Well-Characterised Volcanic Ash Cloud. Bull. Volcanol. 2015, 77. [CrossRef]
11. Guanter, L.; Brell, M.; Chan, J.C.W.; Giardino, C.; Gomez-Dans, J.; Mielke, C.; Morsdorf, F.; Segl, K.; Yokoya, N.
Synergies of Spaceborne Imaging Spectroscopy with Other Remote Sensing Approaches. Surv Geophys 2019,
40, 657–687. [CrossRef]
12. Walker, G.P.L. Basaltic-Volcano Systems. Geol. Soc. Lond. Spec. Publ. 1993, 76, 3–38. [CrossRef]
13. Global Volcanism Program. Report on Kilauea. In Bulletin of the Global Volcanism Network; Crafford, A.E.,
Venzke, E., Eds.; Smithsonian Institution: Washington, DC, USA, 2018. [CrossRef]
14. USGS Hawaiian Volcano Observatory. Overview of Kı̄lauea Volcano’s 2018 Lower East Rift Zone Eruption
and Summit Collapse. U.S. Geol. Surv. Gen. Inf. Prod. 2019, 1, 1–2.
15. Harris, A.J.L.; Rowland, S.K. FLOWGO: A Kinematic Thermo-Rheological Model for Lava Flowing in a
Channel. Bull. Volcanol. 2001, 63, 20–44. [CrossRef]
16. Crisci, G.M.; Rongo, R.; Di Gregorio, S.; Spataro, W. The Simulation Model SCIARA: The 1991 and 2001 Lava
Flows at Mount Etna. J. Volcanol. Geotherm. Res. 2004, 132, 253–267. [CrossRef]
17. Negro, C.; Fortuna, L.; Herault, A.; Vicari, A. Simulations of the 2004 Lava Flow at Etna Volcano Using the
Magflow Cellular Automata Model. Bull. Volcanol. 2008, 70, 805–812. [CrossRef]
18. Ramsey, M.S.; Chevrel, M.O.; Coppola, D.; Harris, A.J.L. The Influence of Emissivity on the Thermo-
Rheological Modeling of the Channelized Lava Flows at Tolbachik Volcano. Ann. Geophys. 2019, 62.
[CrossRef]
Remote Sens. 2020, 12, 193 20 of 21

19. Thompson, J.O.; Ramsey, M.S.; Hall, J.L. MMT-Cam: A New Miniature Multispectral Thermal Infrared
Camera System for Capturing Dynamic Earth Processes. IEEE Trans. Geosci. Remote Sens. 2019, 57, 7438–7446.
[CrossRef]
Remote Sens. 2019, 11, x REVISION 3 of 21
20. Johnson, W.R.; Hook, S.J.; Mouroulis, P.; Wilson, D.W.; Gunapala, S.D.; Realmuto, V.; Lamborn, A.; Paine, C.;
currently
Mumolo, J.M.;available
Eng, B.T.low spatial
HyTES: resolution
Thermal Imaging (>500 m) TIR datasets
Spectrometer that are not
Development. able to accurately
In Proceedings resolve
of the IEEE
the maximum
Aerospace temperature
Conference, Big Sky,or MT, representative
USA, 5–12 March emissivity spectrum. There are relatively few studies (i.e.,
2011. [CrossRef]
21. [23])S.J.;
Hook, thatMyers,
have investigated
J.J.; Thome, K.J.;the influence
Fitzgerald, ofM.;
spatial andA.B.
Kahle, spectral resolution on these
The MODIS/ASTER calculations
Airborne Simulator over
temperatures where a molten lava surface cools (<1450 K). Therefore,
(MASTER)—A New Instrument for Earth Science Studies. Remote Sens. Environ. 2001, 76, 93–102. [CrossRef] simply deconvolving the molten
22. fraction
Ramsey, M.S.;that hasA.J.L.;
Harris, the maximum
Crown, D.A.temperature
What Can Thermal withinInfrared
each Remote
pixel will improve
Sensing the accuracy
of Terrestrial Volcanoes and
Telluncertainty of the emissivity,
Us about Processes kineticon
Past and Present temperature, and radiant
Mars? J. Volcanol. flux.Res. 2016, 311, 198–216. [CrossRef]
Geotherm.
23. Wright, Measuring
R.; Flynn, L.P. accurate
On thethermal
Retrievalproperties
of Lava-Flow of aSurface
moltenTemperatures
lava surfacefrom is also critical
Infrared to lava
Satellite flow
Data.
propagation
Geology 2003, 31,models
893–896.[15,24]. With the increasing number of spectral bands in more recent TIR imagers
[CrossRef]
24. (e.g.,M.;
Favalli, HyTES
Pareschi,[20]M.T.;
and Neri,
the Mineral
A.; Isola,and Gas Identifier
I. Forecasting Lava Flow(MAGI) Paths[25]),
by a the radiative
Stochastic temperature
Approach. Geophys.and
Res.emissivity
Lett. 2005.of[CrossRef]
an object’s surface can be extracted with increasing accuracy [26–28]. A well-constrained
25. emissivity
Hall, can then
J.L.; Boucher, R.H.;beBuckland,
used withK.N.; approaches
Gutierrez, suchD.J.;asHackwell,
linear spectral deconvolution
J.A.; Johnson, modeling
B.R.; Keim, E.R.; to
quantitatively
Moreno, N.M.; Ramsey, determine possible
M.S.; Sivjee, M.G.;spectral end-member
et al. MAGI: that defines the mineralogical,
A New High-Performance textural, and
Airborne Thermal-Infrared
thermal
Imaging fractions [29–32].
Spectrometer for EarthAdditionally,
Science Applications. kinetic temperature
IEEE Trans. Geosci. (and Remote
to a lesserSens. degree,
2015, 53,emissivity)
5447–5457. is
required
[CrossRef] to determine the runout distance and hazard potential using radiant heat flux in lava flow
26. propagation
Planck, M. On the models
Law of[15,24]. However,
Distribution of Energythe accuracy
in the Normalof these derived
Spectrum Onparameters over the cooling
the Law of Distribution of
temperature
Energy in the Normalrange Spectrum.
of typical Ann.
lavas Phys.
is less1901,
well4,constrained at the various spatial resolutions of current
553–563. [CrossRef]
27. TIR instruments.
Realmuto, V.J. SeparatingTherefore, improving
the Effects the accuracy
of Temperature of the kinetic
and Emissivity: temperature
Emissivity and Normalization.
Spectrum emissivity of the
In Proceedings of the 2nd TIMS Workshop; Abbott, E.A., Ed.; Jet Propulsion Laboratory: Pasadena,analyses
previously-determined molten fraction should then reduce the uncertainty in flow model CA, USA, that
directly rely
1990; pp. 31–37. on these thermal properties.
28. Gillespie, A.; Rokugawa, S.; Matsunaga, T.; Steven Cothern, J.; Hook, S.J.; Kahle, A.B. A Temperature and
3. Data Separation Algorithm for Advanced Spaceborne Thermal Emission and Reflection Radiometer
Emissivity
(ASTER) Images. IEEE Trans. Geosci. Remote Sens. 1998, 36, 1113–1126. [CrossRef]
3.1. Study Area
29. Ramsey, M.S.; Christensen, P.R. Mineral Abundance Determination Quantitative Deconvolution of Thermal
Emission This studyJ. was
Spectra. conducted
Geophys. Res. 1998, during two field[CrossRef]
103, 577–596. campaigns at Kīlauea Volcano in Hawai’i, USA, in
30. January/February
Ramsey, M.S.; Fink, J.H.2017 and 2018,
Estimating a period
Silicic when bothwith
Lava Vesicularity theThermal
summitRemotelava lake and coastal
Sensing: A Newplain surface
Technique
forlava flows
Volcanic were active.
Mapping It focusedBull.
and Monitoring. primarily
Volcanol.on the61,
1999, lava lake[CrossRef]
32–39. in the Halemaʻumaʻu Crater and
31. propagating
Carter, A.J.; Ramsey,lavaM.S.;
flows from the
Durant, A.J.;Puʻu ‘Ō’ōI.P.;
Skilling, vent (Figure
Wolfe, 1). Kīlauea Roughness
A. Micron-Scale is a basaltic ofshield
Volcanicvolcano
Surfaces that
fromhasThermal
been erupting
Infrarednearly continuously
Spectroscopy and Scanningfor theElectron
past 500Microscopy.
years [33]. J.The lava surfaces
Geophys. Res. Solidare produced
Earth 2009,
114,during long sustained eruptions where pāhoehoe (tube- and surface-fed) and ‘a’ā flows are emplaced
1–13. [CrossRef]
32. [33].S.R.;
Rose, TheWatson,
lava lakeI.M.; in the Halemaʻumaʻu
Ramsey, M.S.; Hughes, CraterC.G.observed
Thermal during this study was
Deconvolution: active from
Accurate 2008 until
Retrieval of
2018 and varied
Multispectral Infrared in size with maximum
Emissivity dimensions ofVolcanic
from Thermally-Mixed 160 meters wide Remote
Surfaces. and 225Sens.meters long (Figure
Environ. 2014,
140,1)690–703.
[34]. During this period, there were fluctuations in lava lake activity with continuous gas plumes
[CrossRef]
33. and irregular
Holcomb, small History
R.T. Eruptive explosions, finally ending
and Long-Term withofthe
Behavior summit
Kilauea collapse
Volcano. US in May
Geol. 2018
Surv. [13,34].
Prof. At the
Pap. 1987,
time
1350, of the field campaigns, the lava lake level was relatively high but not overflowing,
261–350.
34. approximately
Patrick, M.R.; Orr, T.R.;100–130 meters
Sutton, A.J.; below
Elias, T.;the Halemaʻumaʻu
Swanson, D.A. The Crater rim.
First Five It was
Years continuously
of Kı̄lauea’s Summit circulating
Eruption
with fresh lava
in Halema’uma’u upwelling
Crater in theFact
2008–2013; northSheetthat2013–3116;
migrated to USthe south, cooled
Geological Survey: andReston,
formedVA, plates
USA, of 2013.
cooler,
solidified lava. The lava then sank in the south, distinguished by the occurrence of strong splattering
[CrossRef]
35. Orr,and degassing
T.R.; Heliker, [34]. The lava
C.; Patrick, flows
M.R. Thefrom
Ongoing the Puʻu
Pu u ‘‘Ō’ō vent were
Ō’ō eruption active for
of Kı̄lauea ~30 years
Volcano, eruptingYears
Hawai’i—30 in many
of
locations over numerous eruptive episodes, producing
Eruptive Activity; Fact Sheet 2012–3127; US Geological Survey: Reston, VA, USA, 2013; p. 6. mostly pāhoehoe lava flows with the
36. occasional ‘a’ā lava flow [33,35]. This long eruption finally ended
Poland, M.P. Time-Averaged Discharge Rate of Subaerial Lava at Kı̄lauea Volcano, Hawai’i, Measured from in 2018 with the cessation of the
lower East
TanDEM-X Rift Zone (LERZ)
Interferometry: eruption
Implications forinMagma
the Leilani
Supply Estates [14]. The
and Storage lava 2011-2013.
during flows observed during
J. Geophys. Res.our
2017–2018
Solid Earth 2014. study were part of the 61g episode that erupted from the east flank of Puʻu ‘Ō’ō. These
[CrossRef]
37. flows propagated
Dietterich, H.R.; Poland, down M.P.;the PulamaD.A.;
Schmidt, pali and entered
Cashman, theSherrod,
K.V.; ocean atD.R.;Kamokuna
Espinosa, as A.T.
a series of pāhoehoe
Tracking Lava
surface lava flow and lava tubes (Figure 1) [35]. The majority
Flow Emplacement on the East Rift Zone of Klauea, Hawaii, with Synthetic Aperture Radar Coherence. of the flows observed were tube-fed
pāhoehoe,
Geochem. Geophys. bothGeosyst.
sheet-like
2012.and ropey in texture. Previous remote sensing studies of the pre-2018
[CrossRef]
38. KīlaueaW.C.;
Koeppen, eruption have
Patrick, M.;investigated
Orr, T.; Sutton, lavaA.J.;discharge
Dow, D.; rates [36], R.
Wright, lava flow emplacement
Constraints tracking of
on the Partitioning [37],
and lava
Kı̄lauea’s Lavaspathway
between mechanisms
Surface and [38].
Tube The areasEstimated
Flows, for this studyfrom were
Infraredchosen for the
Satellite Data,high probability
Sulfur Dioxide of
observing
Emission Rates, molten
and FieldlavaObservations.
surfaces combined with the
Bull. Volcanol. availability
2013. [CrossRef]of a variety of remote sensing TIR
datasets.
Remote Sens. 2020, 12, 193 21 of 21

39. Tonooka, H. An Atmospheric Correction Algorithm for Thermal Infrared Multispectral Data over Land-a
Water-Vapor Scaling Method. IEEE Trans. Geosci. Remote Sens. 2001, 39, 682–692. [CrossRef]
40. Young, S.J.; Johnson, B.R.; Hackwell, J.A. An In-Scene Method for Atmospheric Compensation of Thermal
Hyperspectral Data. J. Geophys. Res. Atmos. 2002, 107, 1–20. [CrossRef]
41. Harris, A.J.L. Thermal Remote Sensing of Active Volcanoes: A User’s Manual, 1st ed.; Cambridge University
Press: New York, NY, USA, 2013.
42. Dozier, J. A Method for Satellite Identification of Surface Temperature Fields of Subpixel Resolution. Remote
Sens. Environ. 1981, 11, 221–229. [CrossRef]
43. Matson, M.; Dozier, J. Identification of Subresolution High Temperature Sources Using a Thermal IR Sensor.
Photogramm. Eng. Remote Sens. Eng. Remote Sens. 1981, 47, 1311–1318.
44. Abtahi, A.A.; Kahle, A.B.; Abbott, E.A.; Gillespie, A.R.; Sabol, D.; Yamada, G.; Pieri, D.C. Emissivity Changes
in Basalt Cooling after Eruption from Puu Oo, Kilauea, Hawaii. Trans. Am. Geophys. Union 2002, 83, F1442.
45. Lee, C.T.A.; Luffi, P.; Plank, T.; Dalton, H.; Leeman, W.P. Constraints on the Depths and Temperatures of
Basaltic Magma Generation on Earth and Other Terrestrial Planets Using New Thermobarometers for Mafic
Magmas. Earth Planet. Sci. Lett. 2009, 279, 20–33. [CrossRef]
46. Lee, R.J.; Ramsey, M.S.; King, P.L. Development of a New Laboratory Technique for High-Temperature
Thermal Emission Spectroscopy of Silicate Melts. J. Geophys. Res. Solid Earth 2013, 118, 1968–1983. [CrossRef]
47. Davies, A.G.; Calkins, J.; Scharenbroich, L.; Vaughan, R.G.; Wright, R.; Kyle, P.; Castańo, R.; Chien, S.; Tran, D.
Multi-Instrument Remote and in Situ Observations of the Erebus Volcano (Antarctica) Lava Lake in 2005:
A Comparison with the Pele Lava Lake on the Jovian Moon Io. J. Volcanol. Geotherm. Res. 2008, 177, 705–724.
[CrossRef]
48. Harris, A.J.L.; Dehn, J.; Patrick, M.; Calvari, S.; Ripepe, M.; Lodato, L. Lava Effusion Rates from Hand-Held
Thermal Infrared Imagery: An Example from the June 2003 Effusive Activity at Stromboli. Bull. Volcanol.
2005, 68, 107–117. [CrossRef]
49. Wright, R.; Garbeil, H.; Harris, A.J.L. Using Infrared Satellite Data to Drive a Thermo-Rheological/Stochastic
Lava Flow Emplacement Model: A Method for near-Real-Time Volcanic Hazard Assessment. Geophys. Res.
Lett. 2008, 35, 1–5. [CrossRef]
50. Harris, A.J.L.; Favalli, M.; Steffke, A.; Fornaciai, A.; Boschi, E. A Relation between Lava Discharge Rate,
Thermal Insulation, and Flow Area Set Using Lidar Data. Geophys. Res. Lett. 2010, 37, 1–6. [CrossRef]
51. Putirka, K. Magma Transport at Hawaii: Inferences Based on Igneous Thermobarometry. Geology 1997,
25, 69–72. [CrossRef]

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).