Talk slides on small-scale deformation in active volcanoes given at 地球計測セミナー at ERI, UTokyo.

1. Small-scale deformation of active
volcanoes measured by
Synthetic Aperture Radar
Yosuke Aoki
Earthquake Research Institute, The University of Tokyo
Email: yaoki@eri.u-tokyo.ac.jp
with
Xiaowen Wang (Southwest Jiatong University, Sichuan, China)
Jie Chen (The Chinese University of Hong Kong)
Wang & Aoki (2019), J. Geophys. Res. Solid Earth, 124, 335-357, doi:10.1029/2018JB016729.
Wang, Aoki, & Chen, Earth Planet. Space, in revision.
2019年10月15日
地球計測セミナー
東京大学地震研究所

2. Small-scale deformation in volcanoes
Volcán de Colima, Mexico
(Salzer et al., EPSL, 2017)
Santiaguito, Guatemala
(Ebmeier et al., EPSL, 2012)
Wolf, Gálapagos
(Xu, Jónsson, Ruch, and Aoki, GRL, 2016)
SAR is capable of measuring small-scale deformation by
taking advantages of high spatial resolution.

3. Volcanoes of Today
Tokyo
Ontake
Usu
Mt. Fuji
Part 1
Usu: Thermoelastic deflation
(Wang & Aoki, 2019)
Part 2
Asama: Flank instability (?)
(Wang, Aoki, & Chen, in
revision)
Asama

4. Inter-eruptive volcano deflation
Nabro, Eritrea
(Hamlyn et al.,
Prog. Earth Planet. Sci., 2018)
Asama, Japan
(Aoki et al., Geol. Soc. Lond. Spec. Publ., 2013)
Kutcharo, Japan
(Fujiwara et al.,
Earth Planet.
Space, 2017;
Yamasaki et al.,
JVGR, 2018)

5. Why studying volcano deflation?
Potential mechanisms of volcano deflation
 Viscoelastic relaxation (Hamlyn et al., 2018; Yamasaki et al.,
2018)
 Contraction of magma reservoir (e.g. Hamlyn et al., 2018)
 Cooling of emplaced lava (Wittmann et al., JGR Solid Earth,
2017)
Temporal evolution of volcano deflation could carry various
information such as rheology of intruded magma and host rock.

6. Toya caldera
 Usu volcano is located at the rim of
Toya caldera which ejected >100 km3
of magma ~114,000 years ago.
 Eruption of Usu volcano in historical
time:
1663: VEI=5
1769: VEI=4
1822: VEI=4
1853: VEI=4
1910: VEI=2
1944: VEI=2
1977: VEI=3
2000: VEI=2

7. Volcanic activity of Usu Volcano
1910
Activity time
Eruptive
Interval (yr)
Location Eruption type
Upheaval
height (m)
July ̶ Nov. 1910 57 North flank Phreatic 170
Dec. 1943 ̶ Sep.
1945
33 East flank Phreatomagmatic 280
Aug. 1977 ̶ Mar.
1982
32 Summit Phreatomagmatic 180
Mar. ̶ Aug. 2000 18 West flank Phreatomagmatic 80

8. A phreatomagmatic eruption in 2000
Showa Shinzan that emerged during
the 1943-1945 eruption
280 m

9. Subsidence of the 1943 vent
 Persistent subsidence observed by leveling survey.
 Subsidence of 54 mm/yr (1965-1975) and 32 mm/yr (1975-1990)
 Current deformation? Spatial variation?
Yokoyama & Seino
(EPS, 2000)

10. SAR data processing
JERS-1ALOS-1ALOS-2
Ascending Descending
 A total of 111 scenes from JERS-1
(1992-1998), ALOS (2006-2011),
and ALOS-2 (2014-2017)．
 Time-series analysis from all
possible interferograms (a total of
239 pairs).

11. LOS changes The 2000 eruption (Nishiyama)
• Two subsidences
• 38 mm/yr of LOS extension (mainly
subsidence) between 2006 and 2011.
• Negligible LOS changes between 2014
and 2018．
 The 1977-1982 eruption (summit)
• LOS extension rate declines from 66
mm/yr (1992-1998) to 45 mm/yr (2006-
2011) and 43 mm/yr (2014-2017).
 The 1943-1945 eruption（Showa
Shinzan）:
• Stationary LOS extension rate of ~20
mm/yr
DescendingAscending
2000
1977
1943

12. Decomposing (quasi-)EW and vertical velocities
• EW contraction and
subsidence.
• The subsidence rate
is higher than the
contraction rate.
1977
19432000
1977
1943
NC KC
ALOS-1 (2006-2011) ALOS-2 (2014-2017)

13. Modeling by thermal contraction
V
d
Sea level
Intruded
magma body
Surface
Temperature
Time elapse
High Low
V: source volume ;
d: depth of the source;
T: magma temperature (1200 K);
a: thermal expansivity ( 2×10-5);
k: thermal diffusivity;
v: poisson ratio (0.25);
u(x, t) = f (x, t, V, d, T, a, k, v)
 Assumed an intrusion of a
spherical body (Furuya,
2004, 2005).
Black: fixed
Blue: model parameters

14. Optimum parameters
 The depth of the intruded magma is shallower than 400 m bsl．
 The apparent thermal expansivity is an order higher than the
lab-derived value except for the 1943-1945 case.
Longitude
(°)
Latitude
(°)
Depth
( m b.s.l)
Volume
(×106 m3)
Thermal
diffusivity
(×10-5 m2/s)
Misfit Data source
2000 site
140.8034 42.5541 213±19 6.67±0.21 8.21±1.01 2.78 ALOS-1 (NC)
140.8118 42.5563 100±13 2.05±0.13 8.06±1.20 2.02 ALOS-1 (KC)
1977 site 140.8353 42.5416 396±29 132.18±5.21 10.05±1.09 5.06 JERS+ALOS-1+ALOS-2
1943 site 140.8662 42.5426 92±12 49.51±2.12 1.65±0.22 1.03 JERS+ALOS-1+ALOS-2

15. Observation vs calculation
2000 vent
1977 vent 1943 vent
ALOS-
1
JERSALOS-1ALOS-
2
JERSALOS-
1
ALOS-
2

16. The vent of the 1943-1945 eruption
KC
NC
Vertical velocity map
P1943
P1977
P2000

17. KC
NC
Vertical velocity map
P1943
P1977
P2000
The vent of the 1977-1982 eruption

18. KC
NC
Vertical velocity map
P1943
P1977
P2000
The vent of the 2000 eruption

19. Why is the apparent thermal diffusivity high?
Hydrothermal convection effectively release heat
from magma right after the intrusion?
 Lake Toya is right next to the volcano, providing groundwater.
 Frequent phreatomagmatic eruptions
Question:
Why is the apparent thermal diffusivity in the 1943 vent normal?
Possible collaboration with IPGP:
Reconstructing hydrothermal circulation beneath Usu volcano by
numerical simulation.

20. Usu Summary
 We measured ground deformation of Usu volcano by SAR images.
 Deformation is concentrated around lava domes that emerged during
previous eruptions.
 The observed deformation is explained by thermal contraction of the
intruded lava dome.
 The inferred apparent thermal diffusivity is larger than the lab-derived
value especially right after the intrusion.
 Hydrothermal circulation effectively cools the intruded magma?

21. Previous eruptions of Mt. Asama
 1108, 1783: Large (VEI=5) eruptions
 1900-60s: Intermittent explosive eruptions (VEI<=3)
 1973, 1982, 1983, 2004: Middle-sized (VEI=2) eruptions
 2008, 2009, 2015, 2019: Small and minor eruptions
15 Sep. 2004
15 Feb. 1973
30 Sep. 2004

22. Monitoring by seismometers and GNSS
Seismometer GNSS

23. Magma pathway of Asama
Eruptions in
Aug. 2008 and Feb. 2009
Enhanced seismicity
Summer 2008-
Shallow inflation
Summer-winter 2008
Nagaoka et al. (EPSL, 2012)
Aoki et al. (Geol. Soc. Lond. Spec. Publ., 2013)

24. Motivation
GPS stations
 Near-summit deformation?
 2 GNSS sites at the summit.
The next closest one is 4 km
from the summit.
 No deformation field observed
by SAR
GPS baseline variations
(Aoki et al., 2013)
KAHGKAWG
AVOG
Eruption EruptionEruption

25. Time span
Num. of
images
Path
number
Off-nadir
angle (°)
Azimuth
angle (°)
Resolution
(Rg.×Azi. m)
Num. of
interferogram
ALOS-2
20141028‒
20180814
20 19 36.2
-169.7
(D)
1.4×2.1 75
Sentinel-
1A
20150411‒
20180302
58 119 33.8
-166.8
(D)
2.3×14.0 187
Sentinel-
1B
20150430‒
20180303
62 39 33.8
-10.5
(A)
2.3×14.0 220
SAR images
 ALOS-2 and Sentinel-1 images between 2014 and 2018.
 We processed Sentinel-1 images to enhance the temporal
resolution.

26. Challenges
Meancoherence
Time (Year)
Summer
Summer
WinterWinter
Winter
 Vegetation
 Snow in winter
Normalized Difference Vegetation Index (NVDI)
derived from Sentinel-2

27. Data processing
t0
tn
t3
t2
t4
…
Distributed ScatterersPersistent Scatterers

28. Data processing
(1) Select SAR interferograms with
small temporal and perpendicular
baselines.
(2) Combine both PS and DS targets
for the phase analysis.
(3) Apply the spatially adaptive phase
filtering to improve the phase
quality.
(4) Exclude temporally noisy pixels.
(5) 3D phase unwrapping (Hooper et
al., 2004).StaMPS

29. Average LOS velocities
 Deformation in NE and
SE flanks.
 NE flank dominated by
subsidence (6 mm/yr)
with negligible (<1
mm/yr) (quasi-)EW
motion.
 SE flank dominated by
(quasi-)eastward
motion (~7 mm/yr)
with smaller subsidence
(~2 mm/yr).
Path 39 Path 119
ALOS-2 Path 19
Sentinel-1ALOS-2
DescendingAscending
NEF
NEF NEF
SEFSEF
SEF

30. Time Series
 Steady state between 2014 and 2018.
 Not affected by the 2015 eruption.

31. SEF
Comparing with geological map
The area of deformation corresponds to
the 1783 lava flow.
Lava with a thickness 20-90 m (Yasui &
Koyaguchi, Bull. Volcanol., 2004) should
give negligible thermoelastic contraction
100 years after the emplacement
(Chaussard et al., J. Volcanol. Geotherm.
Res., 2016; Wang & Aoki, JGR-Solid
Earth, 2019).
Flank instability dominated by north
displacements? (InSAR cannot measure).
1108 eruption

32. Deformation at SE flank
ALOS-2: Path 19
LOS velocity
NEF
SEF
KAHGKAWG
Flank instability?
If this region (~0.5 km3) generates a sector
collapse (a big assumption!), then the volume the
collapse will be up to 107 m3.
（24 ka Kurofu： 4x109 m3,
1783 Lava flow： 1.7x108 m3 ）
Scaling laws between the area and volume of landslides
(Klair et al., JGR, 2011; Blahut et al., Landslides, 2019)
classification of remotely sensed imagery (e.g. Fiorucci et al. 2011) and
their derivatives such as high resolution digital elevation models (e.g.
Břežný and Pánek 2017) and SyntheticApertureRadar (e.g.Strozzi et al.
2018). The rapid preparation of more reliable and representative land-
slide inventories covering increasingly extensive regions is being facili-
tated by these advancing technologies coupled with data mining from
media reports (e.g. Battistini et al. 2013) or social networks (De
Longueville et al. 2010; Yin et al. 2012). Unfortunately, the mapping of
submarine landslides is limited by financial constraints and the time-
consuming nature of seafloor investigations. Therefore, there is a ten-
dency for submarinelandslideinventoriesto comefrom regionsof high
economic importance, including both continental margins (Chaytor
et al. 2009; Katz et al. 2015) and intraplate volcanic islands (e.g. Gee
et al. 2001). In this regard, it is notable that much of the information
about giant landslides on volcanic islands comes from economically
prosperous archipelagos such as the Canary Islands and the Hawaiian
Islands. To some extent, the global distribution of giant landslides on
volcanicislandsoutlined heremay bedistorted by thefact that so much
research focuses on these islands. In the future, it is expected that an
increasing number of giant landslides will be recognised in more re-
mote, lessprosperous islandssuch asthosein theSubantarctic.
Conclusions
In thisreport,aglobal databaseof giant landslideson volcanicislandshas
been presented. Onehundred and eighty-two recordsarelisted: seventy-
fivearehosted in theAtlantic Ocean, sixty-seven arehosted in thePacific
Fig. 4 Volume-area relationshipscalculated using theoutlined database and from
other literaturesourcesV = αAγ
. (1) Giant landslidesonvolcanicislands, α = 0.26,
γ = 1.29 (Blahůt et al., thisstudy); (2) Martian landslides, α = 2.53, γ = 1.23
(Crostaet al. 2018); (3) Martian landslides, α = 0.20, γ = 1.43(Legros2002); (4) all
landslides, α = 0.15, γ = 1.33(Larsen et al. 2010);(5) bedrocklandslides, α = 0.19,
IPL/WCoEactivities

33. Asama Summary
 We measure deformation around the summit of Asama Volcano
between 2014 and 2018 by SAR images.
 Deformation is temporally steady without any perturbations by
the 2015 eruption.
 Subsidence in the NE flank and eastward motion (with smaller
subsidence) in the SE flank.
 Deformation in the NE flank cannot be explained by thermal
contraction.
 Deformation in the SE flank is due to flank instability?

34. Grand Summary
 SAR images are powerful in extracting small-scale deformation.
 Volcano deforms for various reasons including thermoelastic deformation
and flank instability.
 ALOS and ALOS-2 (L-band) images work well even in vegetated regions,
but temporal resolution (>14 days) is not favorable.
 Sentinel-1 (C-band) images do not work everywhere, but the temporal
resolution (down to 6 days) is favorable.
 Future SAR missions (NISAR, ALOS-4, both L-band) will further enhance
temporal resolution.