I. Matthew Watson, MTU and University of Bristol, UK
Remote sensing of volcanic emissions using MODIS
Talk structure
• Why quantify volcanic emissions?• MODIS channels of interest• Pre-MODIS algorithms• Recent developments using MODIS
– 7.34 m channel• e.g. Hekla eruption (2000)
– Forward modeling• Water vapor correction
• Future work– Species interaction (spectral)
• Conclusions
Why quantify volcanic emissions:
• Indicators of volcanic activity
• Hazardous to population, wildlife, environment and infrastructure
• Long-lived and climatologically active
• Aircraft hazard mitigation
Why quantify volcanic emissions:
• Indicators of volcanic activity
• Hazardous to population, wildlife, environment and infrastructure
• Long-lived and climatologically active
• Aircraft hazard mitigation
MODIS bands of interest
28 29 30 31 32
MODIS bands of interest
28 29 30 31 32
SO2 band (8.6 m)
MODIS bands of interest
28 29 30 31 32
SO2 band (7.3 m)
MODIS bands of interest
28 29 30 31 32
ash band (11 m)
MODIS channel
applications
Previous (pre-EOS) work
Species Original Sensor [and channels]
Reference Year MODIS channels
ASTER channels
AIRS channel
sets
Ash AVHRR [4,5] Prata 1989 a,b 31, 32 13, 14 12 -14
Ash AVHRR [4,5] GOES [4,5]
Wen and Rose
1994 31, 32 13, 14 12 -14
Ice AVHRR [4,5] GOES [4,5]
Rose et al., 1995 31, 32 13, 14 12 -14
SO2 TIMS [1-6] Realmuto (et al.)
(1994),1997, 2000
29 - 32 10-14 N/A
SO42- HIRS/2 [5-10] Yu and Rose 2000 30 - 34 N/A 10-15
SO2 TOVS1 [8,11,12] Prata et al., 2003 27, 28, 31
N/A 5
Previous (pre-EOS) work
Species Original Sensor [and channels]
Reference Year MODIS channels
ASTER channels
AIRS channel
sets
Ash AVHRR [4,5] Prata 1989 a,b 31, 32 13, 14 12 -14
Ash AVHRR [4,5] GOES [4,5]
Wen and Rose
1994 31, 32 13, 14 12 -14
Ice AVHRR [4,5] GOES [4,5]
Rose et al., 1995 31, 32 13, 14 12 -14
SO2 TIMS [1-6] Realmuto (et al.)
(1994),1997, 2000
29 - 32 10-14 N/A
SO42- HIRS/2 [5-10] Yu and Rose 2000 30 - 34 N/A 10-15
SO2 TOVS1 [8,11,12] Prata et al., 2003 27, 28, 31
N/A 5
Applications of different retrievals for SO2
Retrieval Application Advantages Disadvantages
TOMS
0.3-0.36 m
Large scale eruptions
Spectrally independent/long
archive
Spatial resolution
AIRS
7.34 m
Large scale eruptions
High spectral resolution
Spatial resolution
MODIS/
TOVS
7.34 m
Mid scale eruptions
Climatologically active SO2
High altitude required
MODIS/ ASTER
8.6 m
Mid-scale – passive
degassing
Monitoring possible
Ash and sulfate interference
Applications of different retrievals for SO2
Retrieval Application Advantages Disadvantages
TOMS
0.3-0.36 m
Large scale eruptions
Spectrally independent/long
archive
Spatial resolution
AIRS
7.34 m
Large scale eruptions
High spectral resolution
Spatial resolution
MODIS/
TOVS
7.34 m
Mid scale eruptions
Climatologically active SO2
High altitude required
MODIS/ ASTER
8.6 m
Mid-scale – passive
degassing
Monitoring possible
Ash and sulfate interference
Hekla volcano, Iceland, erupted Feb 26, 2000
Hekla MODIS rgb image (7.3, 11 and 12 m)
7.34 m SO2 retrieval - Prata et al, 2004
MODIS retrieval for Hekla, Feb 26 2000 eruption
Approximate aircraft track
SOLVE (NASA DC-8) in situ SO2 measurements and MODIS retrievals show good agreement
Integrated across plume:
SOLVE=248 ppmV
MODIS=241 ppmV
Water vapor interference
The slopes of these lines are in opposition. The ‘split window’ algorithm BTD (11-12 m) is used operationally by VAACs to detect ash and by climatologists to map water vapor. Hence water vapor masks the ash signal.
Atmospheric effects of water
plume height 1.25 km, plume thickness 1.0 kmground T 300 K, NCEP_00.04.26.
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
6 7 8 9 10 11 12 13 14 15 16
wavelength m
tran
smis
sion Pu'u O'o
Hekla
plume height 11km, plume thickness 2.0 kmground T 300 K, Hekla_28.02.12.
Forward model justification
• User specifies ‘external’ parameters: ground emissivity, ground temperature, atmospheric profile (water vapor, pressure and temperature as a function of height).
• User specifies ‘plume’ parameters: plume (top and base) altitude, effective radius, variance, refractive index and number density of particles
• The forward model uses MODTRAN to calculate ‘external’ effects and Mie scattering code to calculate ‘plume’ effects
• Can be used to– investigate atmospheric effects on IR retrievals– generate a LUT for multi-species spectra – genrate transmission spectra to be used to correct SO2 maps for
ash/sulftae
Forward model rationale
Forward model results – example spectra
200
210
220
230
240
250
260
270
280
290
300
7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0
wavelength (m)
brig
htne
ss te
mpe
ratu
re (
K)
5.0
6.0
7.0
8.0
9.0
10.0
11.0
12.0
13.0
14.0
15.0
bri
ghtn
ess
tem
pera
ture
(K
)
SHV 12 December
Spurr 28 June
BT difference
Forward model results – example spectra
200
210
220
230
240
250
260
270
280
290
300
7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0
wavelength (m)
brig
htne
ss te
mpe
ratu
re (
K)
5.0
6.0
7.0
8.0
9.0
10.0
11.0
12.0
13.0
14.0
15.0
bri
ghtn
ess
tem
pera
ture
(K
)
SHV 12 December
Spurr 28 June
BT difference
Mt. Spurr in tropical atmosphere
0 1
2 3
Brightness temperature
change BTD
Change in cloud area caused by water vapor interference
0
50,000
100,000
150,000
200,000
250,000
300,000
350,000
0 10 20 30 40 50 60 70 80
hours since eruption onset
area
(km
2 )
Spurr
'SHV'
Species Interference
3 m andesite
5 m ice
0.5 m 75 % H2SO4
10 g m-2 SO2
Mid-lat atmosphere
Multi-species algorithm development
• Volcanic emissions group (3 professors, 5 PhD students and 2 MS students at two universities) working on various parts of the problem.
• Empircal and theoretical approaches used– Looking at examples of ash-SO2 separation (e.g. Anatahan
2003)
– Forward modeling of multispecies clouds
– Correction of SO2 for ash and sulfate
– Multi-sensor comparisons (TOMS, AIRS, TOVS, ASTER)
• Several eruptions targeted, work in progress
Conclusions
• The development of the 7.3 m algorithm has been a significant advance in quantifying volcanic SO2 (limited species interaction, works at night).
• Hekla-DC8 interaction provides unprecedented ground-truthing opportunity.
• Forward modeling can be used to quantitatively determine the effects of different water vapor concentrations on the ‘split-window’ signal.
• Water vapor more strongly affects clouds that are optically thinner as relative proportions of signal from the underlying ground (and water vapor) increase.
• A multi-species algorithm is the holy grail of volcanic emission remote sensing. Only through MODIS’s ability to detect and quantify more than one species has the problem been (a) illuminated and (b) potentially solvable.