Studying Volcanoes With InSAR: Where Have We Been and Where Are We Going?
Howard Zebker, Cody WorthamStanford University
10 Years Ago: Where Were We?• Anticipated several new space radar systems for monitoring
and forecasting of volcanic events• Developing inverse methods to reveal the details of faulting
or pressure changes at depth, and better describe precise magma chamber geometries
• Beginning reliable interferometric imaging, m-scale resolution of mm-scale deformation over wide areas
• Proposed high-res stereo radar for 3-D topographic maps fast events, e.g. rapid dome growth
• Foresaw data collection over all of Earth’s 600 potentially active volcanoes weekly or even daily
• Designing satellite constellations in along-track interferometric formation to map flow velocities
Reality: What actually happened• Anticipated several new space radar systems for monitoring
and forecasting of volcanic events• Developing inverse methods to reveal the details of faulting
or pressure changes at depth, and better describe precise magma chamber geometries
• Beginning reliable interferometric imaging, m-scale resolution of mm-scale deformation over wide areas
• Proposed high-res stereo radar for 3-D topographic maps fast events, e.g. rapid dome growth
• Foresaw data collection over all of Earth’s 600 potentially active volcanoes weekly or even daily
• Designing satellite constellations in along-track interferometric formation to map flow velocities
Reliable imaging• Technology has advanced to where we know
how to do this• Want long wavelength, high resolution, rapid
repeat times• Systems beginning to reflect community
knowledge• Still waiting for the “perfect” system
Longer wavelengths yield higher correlation
ALOS Kilauea interferogram: 460 day separation, 1490 m baseline
Wortham et al., 2010
Time series volcano observationsEyjafjallajokull
PS deformation
SBAS deformation
Time series 1993-2000
From Hooper, 2008
PS image of San Andreas Fault - ERS satellite
PS spacing is ~1km
PS performance
RMS error ~1 mm/yr
Precision of PS method
Higher resolution promotes time series analyses
A coarse resolution cell
Brightest scatterer Brightest
scatterer
A fine resolution cell
• Same brightest scatterer, much less background
• Pixel does not scintillate, is “persistent”
•Bright scatterer perhaps less than rest of background
•Pixel scintillates over time from background signal
New spaceborne systems
• Radar systems launched this decade have expanded the data modalities for InSAR
• New frequency bands and orbit repeat geometries
• ALOS PALSAR (Japan) launched 2006• TerraSAR-X (Germany) launched 2007,
Tandem-X satellite launched 2010• Some commercial systems as well
ALOS/PALSAR
• ALOS satellite, PALSAR radar instrument• L-band, wavelength 24 cm• Repeat period 46 days• Multipolarization• 10-20 m resolution• Very high correlation• Just ended successful
mission
TerraSAR-X
• X-band, wavelength 3 cm• Repeat period 11 days• Multipolarization• 1-20 m resolution• Ideal for time-series observations
Current research …
• Most exciting area is time series analysis• Modeling continues to advance• Tandem satellites supersede previous desire
for radar stereo• New measurements and methods will yield
new descriptors
Time series mimics GPS imaging
From Lundgren et al.
Modeling of Yellowstone caldera
Interferograms: C-band
Interpretation: Inflating sill
From Wicks et al., 2006
Modeling Uzon caldera, Kamchatka
Radarsat measurements2000-2005
(a) Distributed opening model, (b) distributed crack model,
(c) depth slice of model bLundgren and Lu, 2006
Probabilistic modeling of Etna activity• Predict eruptive activity from
observed deformation and thermal flux
• Highest activity from coincident increases
• PDF derived from spaceborne data only
From Patrick et al., 2006
Tandem observations for DEMs
• Higher resolution and accuracy than traditional stereo• Could produce radar stereo, but this method is superior
Mt. Merapi Digital Elevation Model from Tandem TerraSAR-X observations
From DLR
New local measurements
From C. Werner, Gamma Res.
… but some things still lacking
• Enabling technology is coverage, temporal and geographic
• Optimizing designs for InSAR– Orbits: poor north retrieval– System parameters– High resolution, long wavelength helps
• Future mission prospect good/bad/?
Vector deformationSBAS vector solution for average deformation rate at Kilauea, HI
•ALOS yields fairly high correlation over 2 year time span•Near polar orbit results in poor northing component retrieval
Wortham et al., 2010
Poor retrieval – northing componentKilauea: GPS – black line, InSAR – Red symbols
Up
N
orth
E
ast
Wortham et al., 2010
Multiple Aperture InSAR (MAI) Method
N. Bechor, PhD Thesis (2006)
• SLCs formed from forward and backward squinted beams
• Beam filtered in Doppler• Interferograms formed from each
beam
• MAI phase gives along-track displacement from
differencing forward/backward interferograms €
φforward = −4πxλ
sin(θ sq + β )
€
φMAI = −2πλx 1 m ≈ 0.6 rad
Average north displacement using MAI
North component of displacement averaged over all InSAR acquisitions
Coverage•ALOS L-band yields high
correlation over challenging volcanoes
•South America data show comprehensive coverage possible
•46 day repeat is too long- misses many signals
•PALSAR data rate/volume too low to monitor 600 volcanoes
From: Fournier et al., 2010
NASA DESDynI-R Mission• Launch in 20XX• L-band, potential 2 m
resolution• Free and open data policy• Specifically designed for InSAR• Volcano hazards one of the
major science objectives
Artist’s concept from JPL
Additional exciting missions• ALOS-2 (Japan): L-band follow-on to ALOS,
launch 2013?, 1-10 m resolution, 14 day repeat• But likely will be commercially oriented with data
hard to get• Tandem-L (Germany): Similar in philosophy to
Tandem-X, but companion to DESDynI, no money yet
• Sentinel-1 (ESA): C-band heir to Envisat, 12 day repeat, 5 m resolution, 201w? Launch
• There are others…
Summary and looking ahead
• InSAR continues to evolve better accuracy and temporal/spatial coverage
• Volcano hazard applications benefit• Future satellites converging on ~12 day
repeats and m-scale resolution• Limiting factor is probably data policy-
agencies still don’t get the science message and pursue commercialization
• If data are acquired, volcanologists will come