Forecasting activity of Japanese volcanoes through geophysical monitoring

A talk on volcano monitoring of Japanese volcanoes given in the WOVOdat workshop in Yogyakarta, Indonesia.

Transcript

1. Forecasting activity of Japanese volcanoes through geophysical monitoring Yosuke Aoki Earthquake Research Institute, The University of Tokyo Email: yaoki@eri.u-tokyo.ac.jp 27 November 2018 Int’l Workshop on optimizing the use of volcano monitoring database to anticipate unrest Yogyakarta, Indonesia
2. Goal of volcano monitoring and WOVOdat “Just as epidemiological databases help medical researchers to identify factors in the spatial and temporal distribution of diseases, WOVOdat is helping volcanologists to discover new relationships between different variables, “ (Newhall et al., JVGR, 2017) ü Understanding how the volcano is working ü Forecasting what happens next: Unrest or no unrest? Eruption or no eruption? When is the next eruption if any? Magmatic or phreatic eruption? Explosive or effusive? When does unrest end? (not in this talk) ü We could learn from the past, but the record from a single volcano usually is not large enough. We thus need to learn from other volcanoes as well.
3. Volcanoes in Japan Tokyo Recent major eruptions: 2018 Kusatsu-Shirane (phreatic, VEI=1, killed 1) 2015 Hakone (phreatic, VEI=1) 2014 Ontake (phreatic, VEI~2, killed 60+) 2013 Nishinoshima (magmatic, VEI=2-3) 2011 Shinmoe-dake (magmatic, VEI=2) 2011, 2014, 2016 Aso (magmatic, phreatic, VEI=1-3) Quasi-continuous Sakurajima (magmatic, failed eruption in 2015) Recent volcanic activity in Japan is lower than Indonesian volcanoes. No VEI=5 since 18th century and no VEI=4 since 1914. Hakone Kusatsu-Shirane Ontake Nishinoshima Shinmoe-dake Sakurajima Aso
4. Volcano observatories of ERI Asama Izu-Oshima Kirishima Mt. Fuji Asama 2004 (magmatic, VEI=2) 2008-9 (phreatic, VEI=1) 2015 (phreatic, VEI=1) Kirishima 2011 (magmatic, VEI=2) 2018 (magmatic, VEI=2-3) Izu-Oshima 1986-1990 (magmatic, VEI=3) Mt. Fuji 1707 (VEI=5) Not enough number of eruptions to learn from the past.
5. Possible scenarios following unrest Japan, 2000–2001 (Ukawa 2005), is not considered here to represent a failed eruption because magma apparently did not reach the shallow intrusion stage. As magma rises, a variety of processes act to either promote or hinder its reaching the surface (Fig. 3). Magmatic eruption is promoted by: (a) high gas content at depth and minimal degassing enroute to the surface (due, for example, to rapid ascent rate); (b) an increase in driving forces, such as 1996), phreatic explosions (e.g., Barberi et al. 1992), and other potential indicators of magma ascent versus the number of magmatic eruptions, it appears that the majority of intrusions stall at some depth without erupting. Potential “will it erupt/won’t it erupt?” discriminants Bull Volcanol (2011) 73:115–122 117 Fig. 1 Illustration of the four possible outcomes of unrest—deep intrusion; shallow, stalled intrusion; sluggish, viscous magma extrusion; and rapid, often explosive magmatic eruption Moran et al. (Bull. Volcanol., 2011) How can we know which scenario is relevant, given unrest started?
6. Difficulty in forecasting volcanic activity ü Unrest to eruption Seismicity faithfully takes the magma pathway in some cases (e.g., Kilauea, 2000 Usu) but sometimes not (e.g., Asama, Shinmoe-dake). ü Failed eruption Unrest does not always lead to an eruption. Similar deformation pattern can lead to either eruption (2011 Shinmoe-dake) or failed eruption (2012 Shinmoe-dake) ü Blue-sky eruption (?) Some eruptions start with an apparent lack of “short-term” unrest (2015 Kuchinoerabu). The apparent lack of unrest could be real or due to the lack of appropriate monitoring network.
7. The 2000 Usu eruption ctivities were observed during the 1943–45 eruption, hich resulted in the formation of the Showa–Shinzan va dome at the eastern foot (Fig. 3). Minakami et al. horizontally in the case of the foot eruptions. 5.3.2. Eruption locations The craters and upheavals of the past eruptions in the northern foot are distributed from the northwest to the east tracing an arc; however, no traces of past eruptions are found in the southern and southwestern foots (Fig. 3). The surface distribution at the northern foot correlates well to the subsurface structure so as to be enclosed by the subsur- face basement layer, which becomes shallower toward the north. When we compare the distribution with the velocity at 2 km depth, this coincides with a velocity of about 5 km/s (Fig. 15). This correlation indicates that the subsurface structure has also controlled the foot eruption locations. As mentioned above, during the foot eruptions, the magma migratesfrom beneath thesummit to the foot areas. If these migrations are constrained by the structure, as considered in the 2000 eruption, the magma migrating to the northwest, north, and east will be forced to move to shallower depths, because the basement layer becomes shallower so as to enclose the volcano. This may increase the possibility of the occurrence of eruptions. Also, per- haps the basement layer, which becomes shallower, acts as a barrier preventing further migration. This may constrain the eruption locations. On the other hand, even if the magma migrates toward the south, the possibility of g. 14. Earthquake hypocenters during the precursory stage and the quence of the possible magma movements projected on the N–S locity cross section. The dots represent the hypocenters during the ecursory stage. The arrows represent the magma movements. relationship with the subsurface structure have not been considered. Travel time tomography using active seismic sources has successfully revealed shallow velocity structures (e.g., 2. Usu volcano 2.1. Geological setting Fig. 4. Number of precursory earthquakes during the 2000 eruption detected every hour by UVO. The alphabets (a)–(j) indicate the time-slices of the hypocenter distribution shown in Fig. 13. The numbers of earthquakes on March 30 are underestimated bec number of earthquakes. 178 S. Onizawa et al. / Journal of Volcanology and Geothermal Research 160 (2007) 175–194 Onizawa et al. (JVGR, 2007) Time and location of the eruption was well forecasted from seismicity, resulting in a succssful evacuation of local residents.
8. The 2000 Usu eruption S. Onizawa et al. / Journal of Volcanology and Geothermal Research 160 (2007) 175–194 Fig. 13 (continued ). S. Onizawa et al. / Journal of Volcanology and Geothermal Research 160 (2007) 175–194 Onizawa et al. (JVGR, 2007)
9. 2004, 2009 Asama eruptions may indicate migrations of intrusive magma from under the western flank of Mt Asama. While the exact direction of the magma migration is unknown because the change in GPS baseline length between 950221 and 950267 is only sensitive to the infla- tion and deflation under the western flank of Mt Asama, the magma may have either migrated ver- elevated temperature state from the middle of 2002. These observations suggest the essential dif- ferences between the first and later contractions. 2004 eruption The 2004 eruption, the first moderate-sized erup- 2000 2001 2002 2004 2006 2007 2008 2010 2011 2012 0 200 400 # of volcanic EQs (a) 2000 2001 2002 2004 2006 2007 2008 2010 2011 2012 0 5 10 15 # of VT EQs (b) 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 –60 –40 –20 0 20 40 60 Baseline length (mm) Year (c) 0221–0268 KVCO–TASH Fig. 3. Temporal evolution of seismic activity and ground deformation between 2000 and 2011. Eruptive periods are shown by grey shades. (a) Daily counts of volcanic earthquakes with unclear P and S arrivals triggered by the data acquisition system. (b) Daily counts of volcano–tectonic earthquakes triggered by the data acquisition system. (c) Relative changes in baseline length between 0221 and 0268 (dots) and KVCO and TASH (crosses), respectively. Location of the stations are denoted in Figure 2. Note that the distance between 0221 and 0268 increased significantly due to the 2011 Tohoku-oki earthquake. However, the distance between KVCO and TASH did not change much because KVCO–TASH strikes almost north–south while the 2011 Tohoku-oki earthquake generated mostly east–west expansion without any significant north–south expansion or contraction. Outliers in late 2008 in the KVCO–TASH baseline are due to instrumental problems. Y. AOKI ET AL. 70 at University of Tokyo on November 26, 2013 http://sp.lyellcollection.org/ Downloaded from The 2004 eruption started on 1 September 2004 started (Nakada et al. 2005). Combinin Jan. Jan. Jan. Jan. Jan. Jan. Jan. Jan. 2004 2006 2008 2010 138°30' 138°32' 36°24' 36°26' 0 1 2 km –3 –2 –1 0 1 2 3 depth [km] N=1001 –2000 –1000 0 1000 2000 2004/01/01~2011/09/27 (b) (c) (a) Fig. 4. Distribution of relocated hypocentres around Asama Volcano between January 2004 and September 2 hypocentre is coloured by its depth. (a) Distribution of epicentres with the approximate location of dyke in during crises (red line). (b) North–south cross-section of hypocentres. (c) Temporal evolution of hypocent horizontal axis represents the longitude of the hypocentres. MONITORING ASAMA VOLCANO at University of Tokyo on November 26, 2013 http://sp.lyellcollection.org/ Downloaded from Aoki et al. (Geol. Soc. London Spec. Pub., 2013) ü The volcano inflated before an eruptions, but inflations sometimes do not result in an eruption. ü Seismic activity does not seem to correlate with volcanic activity in long term.
10. 2004, 2009, 2015 Asama eruptions 142 0221-950268 および KVCO-TASH 基 の 変化 東北太平洋沖地 したトレンドを していたが 950221-950268 基 の 変化に地 含んでいるため 地 後は異なるトレンドを している これは の余効変動の影 によるものと えられる Monthly number of earthquakes Low-freq. EQs Total EQs Volcano-tectonic Eqs Tornillo
11. The 2011 Shinmoe-dake eruption S. NAKAO et al.: VOLUME CHANGE OF THE MAGMA RESERVOIR OF THE 2011 SHINMOE-DAKE ERUPTION 507 Fig. 3. The velocity field in southern Kyushu measured by GEONET with respect to a site shown as a solid circle (31.416N, 130.136E). The open triangle indicates the location of Shinmoe-dake volcano. It is clear that ground deformation around the volcano is affected by the regional tectonic deformation and dilatation due to Sakurajima volcano (small solid triangle). To estimate the ground deformation originated from Shinmoe-dake volcano, it is important to remove these deformations. The mechanisms of the regional crustal deformation in southern Kyushu are controversial. Nishimura et al. (2004) proposed that block rotation of the Ryukyu arc is the main factor. Takayama and Yoshida (2007) suggested that tem- poral variations in the inter-plate coupling affect the defor- mation rate. Wallace et al. (2009) tried to explain the defor- mation field assuming an east-west striking shear zone in Nakao et al. (Earth Planet. Space, 2013) ➨  ᅇⅆᒣᄇⅆண▱㐃⤡఍                                    Ẽ㇟ᗇ ᪂⇞ᓅ ᚚ㖊 㡑ᅜᓅ ü Inflation followed by eruption ü Increasing seismicity but the distribution is diffuse.
12. Failed eruption: Izu Peninsula ly be extended to near– eading to improved erup- located in central Japan, of more than a dozen 1978. The earthquakes displacements and tilts spanning a 28-day inter- val including the deformation episode (9). The dike was approximated by a uniform opening, rectangular dislocation in a homogeneous, iso- tropic, elastic half-space. The best-fitting dike strikes NW-SE, which is consistent with the 139˚ 00' 139˚ 10' 34˚ 50' 35˚ 00' 3042 3061 0297 5105 2106 2107 3048 4111 3049 2108 3046 3053 3062 KWN0 KWN TNG JIZ J-52 139˚ 00' 139˚ 10' 34˚ 50' 35˚ 00' 0 5 10 km GPS(obs.) GPS(calc.) 20mm Tilt(obs.) Tilt(calc.) 5microrad VOL 286 29 OCTOBER 1999 927 by stress perturbations due to dike growth. Rel- ative tension above the dike tip, normal to the dike plane, would favor left-lateral faulting, as observed in the focal mechanisms. If the ambi- ent stress state was sufficiently close to failure, stress changes due to dilation of a slowly prop- agating, or even static, dike could trigger earth- quakes over a broad area. The rate at which a critical stress state migrates away from the dike depends on the rate of dike opening and the vertical profile of the dike-induced stresses. High-resolution earthquake locations (22) com- pared with geodetic inversions of the type pre- sented here will lead to a better understanding of magma transport through the brittle crust and the causes of volcanic seismicity. References and Notes R E P O R T S 166 Y. Hayashi and Y. Morita Apr. 20 N=1 Apr. 21 N=111 Apr. 22 N=124 Apr. 23 N=50 Apr. 24 N=20 Apr. 25 N=41 Apr. 26 N=44 Apr. 27 N=25 Apr. 28 N=15 Apr. 29 N=25 Apr. 30 N=90 May 01 N=19 May 02 N=27 May 03 N=22 May 04 N=19 May 05 N=53 May 06 N=49 May 07 N=12 May 08 N=13 May 09 N=1 2km Figure 8. Daily distributions of the hypocentres from the front view of the hypocentres plane shown in Fig. 6. The date is written in the left upper part of ea diagram. Solid circles are the hypocentres occurring on the day. Grey dots denote hypocentres occurring prior to the day. Aoki et al. (Science, 1999) Hayashi & Morita (GJI, 2003)
13. 1998 Iwate hquake seemed to decrease the seismic activity around suishi-yama in West Iwate, while the number of earth- kes increased around Onigajo-caldera (Ueki and Miura 2). There was also a dike intrusion around Onigajo- era after the occurrence of the M6.1 earthquake (Sato Hamaguchi 2006), even though seismic activity ually decreased as a whole. In March 1999, a new arole field and dead plants were found close to kura-yama (Doi et al. 2000). The fumarole field rapidly ad for about 2 km in the east-west direction and 0.5 km he north-south direction. The fumaroles released about % H2 O and 3% CO2. in August 1999 (Hirabayashi et al. 6). The activity reached a maximum from January 2000, fumaroles sometimes 300–500 m high. The activity ually declined from the latter half of 2001, and had ost ended in July 2004 (Doi and Saito 2005). mporal changes in the seismic activity and magma ply rate in 1998 shown in the previous section, significant magma ration occurred mainly from February to April, and volcano continued to be inflated after May mainly by spherical pressure source persistent at a shallow depth eath West Iwate. After the M6.1 earthquake in Septem- the volcanic activity gradually decreased. In the present on, we focus on the seismic activity and magma supply before the M6.1 earthquake to understand the mag- magma supply rates obtained for the dike intruded beneath 1 2 3 4 5 6 7 8 9 10 11 12 Month in 1998 0 0 1545 40 3090 80 4636 120 6181 160 7727 200 Cumulative Number Daily Number 1 2 3 4 0 0 218 40 436 80 654 120 872 160 1090 200 Cumulative Number Daily Number Month in 1998 a b 0.53 5.3 15.8 10.8 15.5 Fig. 2 Cumulative and daily numbers of volcanic earthquakes. a From January to December. b From January to April. The numbers in (b) represent the earthquake occurrence rates per day Bull Volcanol (2011) 73:133–142 shallow parts beneath from Onigajo-caldera to Ubakura-yama; (2) from May, a spherical pressure source that expanded at shallow parts beneath Mitsuishi-yama was dominant. Both papers also showed that the volcanic pressure source locations are well correlated spatially and temporally with the hypo- centers, and the authors suggest that the volcanic earthquakes occurred associated with dike intrusions and spherical pressure source activity at shallow parts. The time sequence of the 1998 activity of Iwate volcano is divided into four periods below. We briefly summarize the activity. Before 1998 Volcanic tremor at depths of 8–10 km was first detected beneath East Iwate in September 1995 (Ueki et al. 1996). After that, low-frequency earthquakes and tremor beneath East Iwate were frequently observed. From June 1996, high frequency earthquakes occurred at shallow depths (< 2 km) beneath the summit and flanks of the volcano, and this activity continued until the end of 1997 (Ueki and Miura 2002). From January to April in 1998 In February 1998, minor volcano inflation of up to 0.1 micro radian per month at the foot of the volcano was observed and shallow volcanic earthquakes occurred (Sato and Hamaguchi 2006). Activity of deep and intermediate-depth low-frequency seismic events likely preceded the shallow volcano-tectonic earthquake activity (Nakamichi et al. 2003). Volcano-tectonic earthquakes and low-frequency earthquakes as well as very- long-period seismic events at shallow depths were observed (Nishimura et al. 2000a; Tanaka et al. 2002a). In the middle of March, a shallow seismic swarm was observed, and at the end of April, more than 100 earthquakes occurred in 1 h (Ueki and Miura 2002). There were dike intrusions at a shallow depth beneath Onigajo-caldera from February to April (Sato and Hamaguchi 2006). From May to September 2 in 1998 Seismic activity was high from May to July with intense -3 0 3 6 9 12 15 18 21 Depth (km) Dike opening (cm) 0 10 20 30 140 55' 141 00' 140 55' 141 00' 39 48' 39 51' 39 54' 5 km Yk Kr Ub In Oh Mi Onigajo Caldera 2.4 4.0 5.0 6.06.5 Vp (km/s) VLP Yk Kr Ub In Oh Mi ANS YKB GNB MTK 39 45' Circles: Volcano-tetconic EQs Stars: Low-frequency EQs Red: Very-low-frequency EQs Nishimura & Ueki (Bull. Volcanol., 2011) Fig. 4. The strains at the two stations fluctuate in 1997 with amplitude of 0.1–0.2 micro strain, and then increase to 0.5– 0.7 micro strains from February to April in 1998. The strains also show slightly accelerated changes. To quantify the accelerated changes in the strain data, we extract the strain data at the two stations for phase A (Feb. 14–Mar. 12) and those for phase C (Mar.19–Apr.22), during which dikes intruded into shallow parts from deep regions. We normalize the amplitude and lapse time of strain data by the maximum amplitude and the total length of each phase, respectively, and plot them on a logarithmic scale (Fig. 5). It is found that the strain amplitudes at the two stations for both phases A and C are almost proportional to the square of the time for the normalized time t’ close to 1. The strain at YKB see ANS, and temporal ch discuss the have affect To unde nized in Fi for two ma ascent with This is a c surrounding for ascent growth mo density and that possib become hig cally calcu flow and ga simply exam ascent proc characterize regimes, bu parts to sha is small) t 1 2 3 4 5 6 7 8 9 10 11 12 1997 0.0 2.0 4.0 6.0 8.0 Strain x 10 -7 1 2 3 4 1998 ANS YKB 4.0 6.0 8.0 train x 10 -7 a b phase A phase A phase C 0 Fig. 5 Logar data at YKB periods (phas respectively
14. Coming back to Shinmoe-dake unrest ➨  ᅇⅆᒣᄇⅆண▱㐃⤡఍                                    Ẽ㇟ᗇ ᪂⇞ᓅ ᚚ㖊 㡑ᅜᓅ  ᖺ㸵᭶㸯᪥㹼 ᖺ㸷᭶  ᪥ࡢ㟈※  ᖺ  ᭶㸯᪥㹼 ᖺ  ᭶  ᪥ࡢ㟈※  ᖺ  ᭶㸯᪥㹼 ᖺ  ᭶  ᪥ࡢ㟈※ ୖグࢆ㝖ࡃᮇ㛫ࡢ㟈※ ü The 2010 and 2017 inflations resulted in an eruption, but the 2011 and 2013-2014 inflations resulted in failed eruptions. ü Deformation field between 2010 and 2011 unrest was similar. ü What controls whether an unrest results in an eruption or not? – Not clear or unknown. ü Lab experiments suggest that intruded volume (deformation) is a key (Taisne et al., Bull. Volcanol., 2011), but the observation suggests that other factors may be at work.
15. Blue-sky eruption? 2014 and 2015 Kuchinoerarabu eruptions P Hotta and Iguchi Earth, Planets and Space (2017) 69:173 Fig. 2 Changes in baseline length (top) and monthly number of volcanic earthquakes with indication of geothermal and fumarolic activit tom) between 1992 and 2014 Table 1 Coordinates of GPS benchmarks at Kuchinoerabu-jima, relative displacements with respect to KUCG du uary 2006–April 2014, and uncertainty of change in baseline length Benchmark Longitude (°E) Latitude (°N) Altitude (m asl) Relative displacement with respect to KUCG (m) Uncertainty of change in baseli (m) E–W N–S U–D m were detected at the benchmarks located of Kuchinoerabu-jima. is characteristic the depth of deformation source is quite provided by Kagoshima prefecture are southern foot of Kuchinoerabu-jima. plemented the lacking data using the n of Kuchinoerabu-jima (bottom left) and distribution of GPS benchmarks (solid circles). Numbers 1–15 are cor tively. Circles with dashed lines represent craters of No-ike, Shin-dake, and Furu-dake Hotta & Iguchi (Earth Planet. Space, 2017) ü Both are magmatophreatic eruptions. ü Unrest for ~15 years but no short-term precursors are apparent. ü It is difficult to predict how the volcano evolves when we observed unrest. ü The 2014 eruption (VEI=1) damaged much of the instruments, making incapable of closely monitoring the volcano preceding the 2015 eruption (VEI=3).
16. Blue-sky eruption? The 2014 Ontake eruption −0.4 −0.2 0.0 0.2 0.4 −2.0 −1.5 −1.0 −0.5 0.0 0.5 1.0 0 1000 2000 3000 0 10 20 30 40 50 LP−number −2 −1 0 1 2 −2 −1 0 1 2 Eruption 26 31 06 11 16 21 26 Date 2014 Sep. 2014 Aug. M1 M0 M-1 a b c Magnitude N15W Depth (km BSL) VT−number Distance (km) S15E Long-period earthquake (LP) LP −1.0 −0.5 0.0 Depth (km BSL) d d by matched-filter processing. a Space–time diagram of all detected events before and after the 2014 phreatic eruption, nd Space (2015) 67:111 Page 7 of 11 in the present study (N15W). Whether reactivatio old conduit or coincidental similarity, the similar etries suggest the preferred orientation of pre- cracks along NNW–SSE beneath the summit of Ontake. However, a VLP event was not recorded the elevated VT seismicity before the 2014 eruption During the period of increased LP seismicity, b increased to nearly 1.7 (Fig. 8). Based on spatial m of b-values at many volcanoes (e.g., Wyss et a McNutt 2005; Bridges and Gao 2006), high b have been widely observed in areas with high gradients due to the interaction of hot fluids ( gases, and liquid water) with adjacent magma Thus, we interpret that the high b-values reflect crease in the density of smaller faults/cracks due infiltration of hot fluids (Fig. 9a). Between 11 and 16 September 2014, the comb of an increase in LP events and an increase in b suggests that hot fluids infiltrated into pre-existing cracks above a magma chamber. Fluid pressures wi faults/cracks further increased, resulting in a redu the effective normal stress on faults/cracks (e.g., H and Rubey 1959; Terakawa et al. 2013; Terakawa This reduction in fault strength facilitated the of small faults/cracks into larger ones, producing magnitude earthquakes (Fig. 6d), which can also the gradual decrease in b-values after 16 Septemb (Fig. 8). Based on these observations, we postulate th fluids were pervasive to pre-existing faults/crac pressurized by heat supplied from the underlying chamber from the middle of September 2014 (F The shallow portion beneath the summit perhap stood stress concentration derived from the incr internal pressure of the underlying hydrothermal During the 10 min prior to the phreatic erupti up-dip and lateral migrations of VT earthquakes co with an increase in pre-eruptive tremor amplitu an anomalous tiltmeter signal near the craters. B the tiltmeter signal was consistent with summ heaval (Figs. 1 and 9b), the migrating seismicit −0.4 −0.2 0.0 0.2 0.4 −1.0 −0.5 0.0 −1.0 −0.5 0.0 0.3 m/s 5.0 m/s 0.7 m/s 0.35 m/s 0.70 m/s Distance (km) Depth (km BSL) Eruption −1.0 −0.5 0.0 Depth (km BSL) Velocity (4-12 Hz) Envelope (1-4 Hz) 0.1mm/s 0.02mm/s Tilt: NS Tilt: EW 1nrad 11:30 11:40 11:50 12:00 2014/09/27 Time c d V.ONTN V.ONTN M1 M0 M-1 N, E up b a Kato et al. Earth, Planets and Space (2015) 67:111 Pag ü The eruption took place in the worst possible timing. ü Increasing earthquakes from ~3 weeks before the eruption, it is difficult to judge whether it is considered as unrest. ü Tilt changes started 7 minutes before the eruption. Kato et al. (Earth Planet. Space, 2015)
17. Definition of unrest? The 2006-2009 Mt. Fuji inflation ῎ ῎ ῎ ῎ ῌ ῌ ῎ ῌ ῒῌ ῎ ῌ ῐ ῑ ῒ ῍ ῌ ῌ ῍ ῌ ῍ ῌ ῌ ῌ ῌ ῌ ῌ ῌ ῌ ῍ ῌ ῌ Q Q Q QQ Q Q Q Q   Q   Q Q ῦ  ̳QQ  ῤ Q QQῥ̳ ῭ Q ̱῏̱ ̳ ῱Q Ὶ̯ Q QQ΅ ̮QΊ̯ῢ῔῕Q̮ῡ QQ ̳ῥ ῥ Q ῢῚῠ Q Ῐῢῤ῵ῼ῏̱ῥ ̰Q̰ῤ ῕  ῱ ῿ ῤQ ῡ̮῭ῗῢ῱QῚ̯ Q ΰ ῥ̰ῢQ̮Ύ΅῭QQῤ̯῔ῠ̱ῼ̲̱῱Q ̰ ῟΅῱QQῚ̯ ̮ῧ Q QQῥQ QQ ῤ̯῔ῠῦ Q ῦ ̲QῤQQ ῤῪ̯ῠQ Ῑ΅ Q ῤ̯῔ῠῦ Q Qῥ ̯ QQQῨ Q̳QQ ῥ Q QῥQ  ῡΐ QQ Q ῤ̯῔ῠῦ ῱̱ ῭ ῤ ῟΅̯΅ῥ Q QQ ῤ̯῔ῠ ῝ Ὺ ῷ ̲ ̰ ̮ΌῬ ῝̱̰ ̯̮ῼ ῎ ῏ ῝ ῏ ̮ ῗ ῳῳ ̯ῢ̯ ῌ Ὼ ῴ ῲ̮̲ ῭ ῼ ῌ῜ Ῐ Ῠ Ὶ ̯ Ί ̱ Ί ῭̯ ̲ ῠ ῝ ῌ῜ ῤ ῦ ῤ ̮ ῌ ῌ῜ ῤ ̰ ̰ ῦ ῱ Ῑ ῥ̮ Ύ̰ ῌ῜ ῤ Ῑ ῥ̮ ̯ ῌ Ὶ̮̮῕῎Ῑ̮῏ῒῐῑΐ῍ῖ̮Ῐ̮̮῔ῗ ῌ ῌ ῌ ῌ ῌ ῌ ῌ Same as Fig. except for the dilatational strain around Mt. Fuji. In calculating the strain change, 194 Fig. . Temporal change in the dilatational strain Fig. . around Hakone volcano. we used average coordinates in days and those in the succeeding days at GPS stations in the circle shown in the inlet map. ; ; ( ) GPS GEONET, F Fig. (a) Fig. (b) + + , -* -* ,**, ,**1 ,**3 ,**3 - - +331 +333 ,**, ,**- ,**1 ,**2 - ,**+ +331 +333 ,**1 ,**2 + +332 +333 ,**, ,**- - - Q̲̲̱̯Q̲΅̲̯QQQ῟Q QQ̳ῲQ 195 Fig. . (a) Epicentral distribution of low-frequency earthquakes beneath Mt. Fuji and Hakone volcano. (b) Number of low-frequency earthquakes above respective threshold magnitude. Black and white circles represent number of low-frequency earthquakes beneath Mt. Fuji and Hakone volcano, respectively. (c) Cumulative number and magnitude - time diagram for low-frequency earthquakes - Accelerating seismicity and expansion since ~2006, but the expansion is invisible without extensive post-processing of the GPS time series. Harada et al. (Bull. Volcanol. Soc. Jpn., 2010)
18. Scaling law for “short-term” precursors 128˚ 130˚ 132˚ 28˚ 30˚ 32˚ 34˚ 200m 300m 300m 400m 400m 400m 500m 500m 500m 600m 600m 600m 700m 700m ' Active crater T2 T3 SWA SWB Suwanosejima Kyushu 0m 0m 0m 0m 0m 0m 0m 200m 200m 200m 200m 200m 200m 200m 200m 400m 400m 400m 400m 400m 600m 600m 129˚42' 129˚45 29˚36' 29˚39' 0 1 2 km Ryukyu-Islands Page 3 of 12, 779 emissions, (2) magnitude of explosion earthquake increases with amplitude of uplift, and (3) magnitude of explosion earthquake increases with duration time of pause of continu- ous tremor and duration of uplift. These empirical relation- ships can be explained by the following conduit process just before explosions. Volcanic gases and/or ash are constantly supplied from the actively degassing magma column or cham- ber, because the amplitude and frequency content of continu- ous tremor are almost constant during a sequence of the vulcanian eruption activity. A ‘cap’ is formed at the top of the conduit effectively sealing the conduit and stopping vol- 0 60 t1 t2 te s Explosion Transition Time Continuous Tremor Pause Time Fig. 8 Schematic illustration showing the start and stop times of contin- 779, Page 8 of 12 Bull Volcanol (2013) 75:779 distance of a half observed at a dista T3 (400 m from th SWA (600 m), T2 precursor changes the source depth t should be include termined from the earthquakes at fou beneath the crater of explosion earth deployed close to about 0.5 km (Nak located at the sam which suggests th pressurize the sha stantaneously rele A lava dome ca volcanic gases ben Sakurajima volcan served (Ishihara 19 ma or a plug of cry conduit (Lyons et 2013). Then, remo release of the pre crater to generate a a cap at Suwanose 1 10 100 1000 1 10 100 1000 09/10/01 1 10 100 1000 Pause Time (s) Maximum Amplitude (micron/s) 10/11/11 1 10 100 1000 10/11/20 1 10 100 1000 Pause Time (s) 11/01/14 1 10 100 1000 1 10 100 1000 1 10 100 1000 r=0.66 r=0.42 r=0.75 r=0.63 a b c d Fig. 9 Relation between the maximum amplitude of explosion earth- quakes and the pause time of continuous tremor before explosion. The relations are separately shown for the date of the activity: a Oct.1, 2009, b Nov. 11, 2010, c Nov 20, 2010, and d Jan. 14, 2011. Fine lines are the regression lines. Broken line in each panel represents a linear relation between the maximum amplitude and pause time ü Linear scaling between ”Pause Time” and the maximum amplitude of the explosion earthquake. ü An accumulation of volcanic gases in the shallow conduit might be the main driver of the explosion. Nishimura et al. (Bull. Volcanol., 2013)
19. Sinabung Event Tree Wright et al. (JVGR, in press)
20. Mt. Fuji Event Tree Eruption size Eruptive products Eruption style Vent location Precursor Failed eruption Earthquakes Deformation Gas etc. Summit High flank Low flank Plinian Strombolian Plinian Strombolian Ballistics, Ash Ballistics, Ash, Proclastic flow Ballistics, Ash, Spatter Ballistics, Ash, Proclastic flow, Spatter Ballistics, Ash, Lava flow, Spatter Ballistics, Ash Ballistics, Ash, Spatter Ballistics, Ash, Lava flow, Spatter Large Medium Medium to Large Medium Small Small to Medium Medium to Large Small Large Medium Medium Small Small Medium to Large Phreatic eruption Snow-assisted lahar Lahar Lava flow Spatter Pyroclastic flow Ash Ballistics Possible hazards Aramaki et al. (2007) Red: High probability Yellow: Low probability White: Unlikely
21. Alternative: Introducing Deep Learning Feature 1 2 3 …. Taking time series as input Seismic waveform (Earthquake count?) Deformation Gas Electromagnetics ……. Outcome Explosion (magmatic, (magmato)phreatic, large, small, tiny…) Effusive eruption (large, small, tiny…) Failed eruption …… Start from a volcanoes with a lot of dataset to have the system learn (Sakurajima, Etna, Stromboli, Sinabung, …)? Extension to multi- volacanoes will involve a challenge.
22. Summary ü Assessing precursors to eruptions involve a lot of difficulty for various reasons. ü For example, apparent lack of precursors could be real or due to a lack of optimum monitoring network. ü However, there are some precursors which have repeating features or obey scaling laws. ü Employing data science perspectives will advance the capability of forecasting volcanic activity. What is not discussed here ü Precursors detected by something other than seismic and geodetic instruments (gas, electromagnetics, surface temperature…) ü How can we judge whether the activity ended or not? ü How to include observations in pre-instrument era (eye witness, historical accounts…) ü Data sharing