Геофизики из России (Новосибирск, Томск, Петропавловск-Камчатский), Франции, Саудовской Аравии и Египта описали строение самого опасного супервулкана Земли — Тобы, расположенного на Суматре (Индонезия). Исследование опубликовано в журнале Nature Communications.
Ученые представили модель строения вулкана, основанную на имеющихся сейсмических наблюдениях. По их мнению, на глубине более 150 километров под погружающейся вниз литосферной плитой газы и расплавы формируют потоки магмы, которые поднимаются наверх.
Это приводит к образованию на глубине 75 километров под Тобой магматической камеры объемом 50 тысяч кубических километров. Извержение супервулкана происходит тогда, когда давление в камере достигает критического значения. В этом случае магма поступает наверх, что приводит к расплавлению (разрыву) коры, и изливается на поверхность планеты.
Авторы полагают, что нет никаких оснований полагать, что извержение супервулкана Тоба неминуемо, однако не исключают этого. Геофизики отмечают, что аналогичное строение имеет геологический источник в Йеллоустоуне (США).
Супервулканы считаются одними из самых разрушительных геологических образований, оказывающих влияние на климат планеты. Последний раз Тоба извергался около 74 тысяч лет назад, и, по мнению специалистов, мог существенно сократить биоразнообразие видов на Земле.
The Toba Caldera, located in northern Sumatra, has been the site of several large explosive eruptions over the past million years. The most recent Toba supereruption, ~74,000 years ago, is considered to be the largest terrestrial volcanic eruption of the Pleistocene1, 2. This eruption ejected an enormous volume of material, with the dense rock equivalent estimated between 2,800 and 5,300km3 (refs 1, 3). The scale of the eruption significantly affected the global biosphere and climate4, 5, although some specialists6 suggest that commonly accepted estimates of the catastrophic consequences of the Toba supereruptions are overestimated. In any case, if such an eruption occurred in modern times, it would drastically alter human life. Therefore, it is critical to understand the functioning of the magmatic system that periodically produces such unusually voluminous eruptions.
The Toba Caldera is one of the volcanic complexes of the Sunda Arc, where the Indo-Australian Plate subducts obliquely at a rate of 56mm per year7. In different subduction zones, many traces of caldera-forming eruptions have been recorded8; however, the intensity and repeatability of the eruptions at Toba make it unique. The topography of the Toba area is characterized by a large uplift with an average elevation of ~1,500m and lateral dimensions of 220 × 100km (area indicated by blue dotted line in Fig. 1). This area is transected by the Great Sumatran Fault Zone (GSFZ), which can be clearly identified on the topographic map. Abundant evidence of recent volcanic activity, such as circular caldera structures and cinder cones, is also clearly visible on the topographic map.
In the context of the Toba volcanic activity, several important questions are actively debated: why did several large eruptions occur in approximately the same location in the Toba Caldera area? Why were the supereruptions followed by long periods of quiescence? Is the magma system beneath Toba active and can we expect a new supereruption of Toba in the near future? In this study, we analyse several types of observations, such as topographic/bathymetric maps, geoid transformations, seismicity distributions, and regional and local seismic tomography, and attempt to shed light on some of the above-mentioned questions. Our results show that the anomalous style of magma production beneath the Toba Caldera is primarily caused by perturbations on the slab associated with the subducting fracture zone. In the tomography model, we can identify traces of the magma and volatile pathways in the mantle and two large magma reservoirs on the base of the Moho interface and in the upper crust, which are the major elements responsible for episodic occurrence of supereruptions.
The present-day activity of the Toba supervolcano is fundamentally determined by its geographic location and geological setting. Figure 2 summarizes the locations of the main volcano-related structures and the Great Sumatran Fault, identified from the topographic map and from literature1, 2. The locations of the three most recent calderas, formed within the last million years2, are also shown in Fig. 2.
Consideration of topographic/bathymetric maps may provide some insight into the processes causing the supervolcanism. As shown in Fig. 3a, the Toba Caldera is located above the prolongation of the Investigator Fracture Zone (IFZ), which is a 2,500-km-long transform zone in the Indian Ocean. The IFZ separates the younger northwestern segment of the oceanic plate, which has an age of ~40Ma, from the southeastern segment, which is ~15Ma older9. On the outer rise of the subducting plate, the IFZ splits into several parallel ridges with elevations up to 1,500m above the surrounding seafloor. The bathymetric map shows that the morphology of the accretionary complex at the junction between the IFZ and the Sunda Trench is considerably different compared with the other segments of the trench, which may be explained by stronger shortening of the overriding plate10 in this area.
Figure 3b presents a map of the longitude-directed derivative of the geoid model EIGEN-6C4 (ref. 11). This transformation is effective for revealing latitude-oriented linear structures related to the IFZ. These features represent deep density variations beneath the fracture zone, some of which are continuous across the trench. The possibility of using gravity data to detect the signatures of crustal features in different subduction zones has recently been investigated12. The most prominent continuation of the IFZ to the onshore Sumatran area is observed in the fracture line directed towards Toba. In Fig. 3b, we also plot the slab-related earthquakes identified by local seismic networks. Most of the seismicity defines a linear dipping structure that extends from the lineaments of the IFZ (see also Supplementary Fig. 1). Same earthquakes, but on a larger scale, are presented in Fig. 2, which shows that the narrow seismicity zone associated with the subduction of the IFZ is located directly beneath the southern part of the Toba Caldera.. A similar alignment of seismicity was previously reported by Fauzi et al.13, who performed an analysis of earthquakes recorded by local and regional seismic networks. These authors hypothesized that the IFZ serves as a site of focused volatile release into the overlying mantle wedge.
Regional tomography study
Figure 3c shows P-velocity anomalies in the upper mantle derived as a result of regional tomography studies. An earlier version of this model has previously been presented14; however, we subsequently modified it slightly by tuning the parameters and incorporating new data. This tomography model was constructed using seismic travel time data from the global catalogue of the International Seismological Centre (ISC) and the algorithm developed by Koulakov and Sobolev15. Further details on the data and algorithm are provided in the Methods section. Consistent mantle heterogeneities were reported in another tomography study16 based on similar data and approach. In the model presented in Fig. 3c, we observe a linear high-velocity anomaly representing the subducting lithosphere. We observe that in the area around the IFZ prolongation, the slab has a curved shape, which may indicate strong deformation and tearing.
Local tomography studies
On the local scale, several seismological studies in the Toba region and surrounding areas have previously been performed. The interaction between the IFZ and the overriding plate was investigated based on the offshore seismic network17. The deep structure beneath Toba was studied18, 19 based on data recorded by seismic stations deployed in 1995. Subsequently, a new seismic network was installed in the Toba area in 2008 (ref. 20). Using data from this network, ambient noise tomography studies20, 21 imaged the detailed structure of the upper crust beneath Toba. In addition, the latest study21 revealed strong radial anisotropy at depths >7km, thus providing evidence of a large sill complex beneath the caldera. Receiver function studies22 have identified increased crustal thickness, up to 35–38km, beneath Toba.
Here we present an additional local-scale tomography model, which was calculated based on the combination of seismic travel time data obtained from the two above-mentioned temporary seismic networks in the Toba area. The first data set was recorded by a seismic network deployed around Toba during the period from January to May 1995. The network consisted of 10 broadband and 30 short-period seismic stations (white triangles in Fig. 1). The second data set was recorded by another seismic network deployed in the same approximate area by GeoForschungsZentrum-Potsdam for 5 months from May to October 2008. This network consisted of 42 short-period three-component seismometers (red triangles in Fig. 1). For the second data set, we manually identified the local events and picked the arrival times using SEISAN software23. To select the data for tomography, for both data sets we used only events that had a number of P- and S-picks per event equal to or larger than 8. We rejected the picks with absolute values of the residuals larger than 1 and 1.5s for the P- and S-data, respectively. From the first data set we selected 505 events with 7,058 corresponding arrival times (4,122 P- and 2,936 S-waves) and for the second data set we obtained 4,826 arrival times (2,522 P- and 2,304 S-waves) from 149 local events. Although the number of events in the second case was smaller relative to the older data set, the quality of these data was considerably higher. The average number of picks per event in this case was ~32, whereas for the older data set it was <14. The distributions of seismic stations and events are plotted in Supplementary Fig. 1.
As a result of the tomographic inversion, we obtained the three-dimensional (3D) distributions of the P- and S-wave velocity anomalies with respect to the preferred one-dimensional (1D) velocity models and the updated locations of seismic events. For the main discussion, we selected the distributions of S-wave velocities (Fig. 4), because they are more sensitive to temperature- and fluid-related heterogeneities and, therefore, to the volcano-related features. The P-wave anomalies in the horizontal sections are shown in Supplementary Fig. 3. In Fig. 5, we also present the P- and S-velocity anomalies and Vp/Vs ratio in a vertical section oriented parallel to the displacement direction of the subducting plate. The absolute P- and S-wave velocities are shown for the same section in Supplementary Fig. 4.
At shallow depths corresponding to upper crustal structures (10km depth), the low-velocity S-wave anomalies (Fig. 4) coincide with the locations of the Toba Caldera, other volcanic complexes (yellow crosses) and the GSFZ. The prominent low-velocity anomaly ‘4’ beneath the Toba Caldera appears to be consistent with previous studies that were based on body wave18, 19 and ambient noise20, 21 data. At depths of 30 and 50km, we identify an elongated low S-velocity anomaly ‘3’ with dimensions of ~120 × 20km that probably represents magma storage at the base of the crust. In the vertical section in Fig. 5, we observe that these two anomalies (patterns ‘3’ and ‘4’) are characterized by low S-wave velocity patterns and high Vp/Vs ratio. The deeper anomaly ‘3’ coincides with a seismicity cluster located at the base of the crust. It is noteworthy that a very similar structure was identified beneath the Klyuchevskoy volcano group in Kamchatka, where a Vp/Vs anomaly as high as 2.2 is located at the base of the crust and coincides with an extremely strong seismicity cluster25. Similarly, as in the Toba case, a low Vs and high Vp/Vs anomaly was detected in the middle crust beneath the Klyuchevskoy volcano and interpreted as an intermediate magma reservoir. The synthetic modelling, which will be discussed below, shows that using the higher-quality data in this study provides the possibility to distinguish the patterns ‘3’ and ‘4’, whereas in previous studies they were seen as a single vertically smeared anomaly. Another crustal anomaly, similarly characterized by low S-velocity and high Vp/Vs ratio pattern, is clearly associated with the GSFZ and indicates fracturing of crustal rocks in the fault zone.
At greater depths (80km section in Fig. 4), we observe two clearly separated linear low-velocity anomalies that can also be observed in the vertical profiles of both the P- and S-models (Fig. 5). One of these anomalies (denoted by ‘1’) represents a vertically oriented ‘wall’ that originates in the slab area at a depth of ~150km. A similar feature has been identified in a previous tomography study19. The second low-velocity anomaly, denoted by ‘2’, connects the slab area at ~80km depth with the forearc. This anomaly has not been revealed in previous tomographic studies because of insufficient data coverage.
Several synthetic tests provided in Supplementary Figs 5–8 show that incorporating new data in the model has enabled better coverage compared with that of previous work19, in particular for areas between the Toba Caldera and the coast. The checkerboard test in Supplementary Fig. 5 shows that at shallow depths the anomalies >30km in size are robustly resolved, whereas in deeper sections the minimum size of the resolved anomalies is ~50km. Tests using realistic anomalies, shown in Supplementary Figs 6 and 8, help to assess the leakage of anomalies due to smearing and limited spatial resolution. We defined the model patterns to achieve the best resemblance between the recovered model and the results derived from inversion of the experimental data. Thus, the amplitude values defined in the realistic synthetic model represent the true values expected for anomalies in the Earth’s subsurface. It is important that we can now clearly distinguish relatively complex structures in the crust, as shown in the synthetic test in Supplementary Fig. 8, which illustrates considerable improvement in the model resolution following the addition of new data. However, for the mantle, the synthetic test with vertical checkerboards (Supplementary Fig. 7) demonstrates the limited vertical resolution caused by the trade-off between source and velocity parameters.
It is a general problem of tomography studies that the values of anomalies depend on the data coverage and damping parameters. Synthetic tests using realistic anomalies allow estimation of the leakage of the anomaly values due to smearing and damping, and assessment of their original amplitudes. For example, for anomalies ‘3’ and ‘4’, we estimated the S-wave velocity anomaly values at 16% and 18%, respectively. However, for some parts of the model, we cannot guarantee that the reported values actually do represent the exact seismic parameters in the Earth. Therefore, we should exercise caution when attempting direct conversion of seismic anomalies into petrophysical parameters, such as temperature or melt content. In our interpretation, we generally qualitatively consider the shapes and relative strengths of anomalies without emphasis on their numerical values.
Some of the volatiles may escape from the slab during the first stages of dehydration phase transitions at relatively shallow depths. In particular, the negative anomaly ‘2’ (Fig. 5a,b), which originates from the slab at 80km depth beneath the forearc, may be a signature of serpentinization and/or high fluid release from the slab. High Vp/Vs ratio (Fig. 5c) above this anomaly may indicate fluid saturation in the uppermost mantle and crust beneath the forearc. The temperature of the pathway in this area is not high enough to melt rocks and generate considerable volcanic activity. Such ‘early’ escape of fluids was similarly observed in many other subduction zones, such as in the northern27 and southern Andes28.
In addition to bringing an anomalous amount of fluids to the mantle, subduction of the IFZ could trigger slab tear, which may also intensify the process of magma generation29. Based on the analysis of deep earthquakes beneath Sumatra, Fauzi et al.13 pointed out that the IFZ may correspond to a tear or a step in the slab. Strain along this step is likely to be the cause of the linear zone of seismicity. The same conclusion is drawn from the regional tomography result shown in Fig. 3c. The shape of the slab-related high-velocity anomaly shows curvature of the slab in the area of the IFZ, which may indicate strong deformation and tearing, as highlighted by the blue lines.
We suggest that the difference in oceanic plate age across the IFZ could be responsible for the step or tear. The younger crust to the northwest of the IFZ was formed at a fast-spreading mid-ocean ridge, whereas the older crust to the southeast was formed at a slow-spreading ridge (see the plate model9). This is likely to result in significantly different crustal thickness and thermal structure, and therefore in a difference in slab buoyancy. According to this interpretation, the southeastern segment of the slab is more dense and therefore it subducts more steeply than the northwestern segment. If a slab tear is present, this could make it easier for fluids to enter or exit the deeper part of the slab and could also cause significantly faster heating of the slab. In turn, this may be one of the key factors contributing to the origin of the Toba supervolcanism. Penetrating hotter asthenospheric material may be another factor facilitating the voluminous melting in the mantle wedge. In Fig. 6a, we schematically illustrate the relationship between slab tear and the magma system beneath the Toba Caldera.
The second key question is why the extremely violent eruptions in the Toba Caldera alternate with long periods of relative quiescence. What prevents the continuous occurrence of moderate volcanism? The local seismic tomography model indicates the configuration of the multilevel magma plumbing system beneath Toba and can help to answer this question. The functioning of the Toba magmatic system deduced from the seismic tomographic model is illustrated in Fig. 6b. The subducting lithosphere along the IFZ line is characterized by thicker crust and is more hydrated than ‘normal’ oceanic lithosphere26. Furthermore, it may be additionally heated from a slab window that originated due to slab tear. Both of these factors may lead to the anomalous release of volatiles at a depth of ~150km, which is expressed by high seismicity along the slab. When the fluids penetrate through the mantle wedge, they may react with peridotites and transform them into phlogopite- or amphibole-bearing rocks, which have lower melting temperatures29. These processes lead to the growth of ascending diapirs containing high-temperature volatile-rich basic melts produced by partial melting of the mantle wedge. The ascending partially molten magma pathways are expressed by the low-velocity anomaly ‘1’ in Fig. 6b. At depths of 30–50km, these ascending magma diapirs form a large reservoir (anomaly ‘3’). The strong negative shear velocity anomaly and much weaker P-wave velocity anomaly suggest the presence of significant amounts of partial melt and volatiles inside this reservoir. However, as the seismic S-wave can propagate through the reservoir, the solid component is still dominant. It is difficult to quantify the volume of the liquid phase because of uncertainty in the exact determination of seismic anomaly values and due to ambiguous transitions between seismic and petrophysical properties. We selected a value of −7% of the S-wave anomaly at 50km depth as the most plausible threshold representing the rocks with high fluid content. Following the contour line for this value (violet dotted line in Fig. 4), we estimated the volume of reservoir ‘3’ at 50,000km3. A similar basic magma reservoir has recently been identified at the base of the crust beneath the Yellowstone supervolcano30, which may indicate that this is a common feature of supervolcanic structures.
The basic magma in reservoir ‘3’ is too dense to continue ascending through the lower-density continental crust. Geochemical evidence based on 87Sr/86Sr ratios indicates that the typical Sumatran stratovolcanoes comprise mafic melts (and their derivatives) generated from the mantle wedge, whereas Toba consists almost exclusively of melts derived from continental crust2. From this, we can conclude that there is no considerable contribution of melts from the mantle in forming the upper crust reservoir; however, the reservoir may serve as a powerful heat source and may transfer upward a significant amount of volatiles. The mobile ascending volatiles appear to be a very efficient mechanism of heat transport. The observed seismicity in the lower crust above anomaly ‘3’ possibly indicates the pathways of the ascending volatiles, which provides evidence for the active state of the magma factory at the present day. Reaching the middle and upper crust, the overheated volatiles cause melting of rocks that form the shallow silicic magma reservoir observed as anomaly ‘4’. When a critical amount of molten upper crust material highly saturated with fluids is accumulated, an avalanche-type process may start. Ascending quantities of such material results in decompression and transformation of overheated fluids to gases. In turn, this increases pressure and accelerates fluid and magma ascent. Finally, this avalanche-type process triggers a large explosion causing a supereruption. The accumulation of a sufficient amount of ‘explosives’ in the upper crust takes hundreds of thousands of years, which may give the answer to the second question related to long periods of quiescence between supereruptions. A similar scenario was proposed by Shapiro and Koulakov31 for the case of Yellowstone, based on the tomography results of Huang et al.30 who identified a multilevel plumbing system including a large reservoir at the base of the crust, which appears to be very similar to the results of this study for Toba.
The final question, related to the present-day activity of the magma reservoirs beneath Toba and the possibility of a future supereruption, remains open. It is probable that, in the long term, large eruptions will occur repeatedly until the IFZ, which is the major source of the supervolcanism, subducts beneath Toba. Regarding the present-day state, the intensive seismicity within the reservoir at the base of the crust indicates its current activity. On the other hand, the time that has currently passed since the last supereruption (74Ka) is too short in comparison with the periodicity of large volcanic events at Toba. The critical mass of the molten magmas and volatiles in the upper crust has most likely not yet been achieved and the next supereruption may be expected only in some dozens of thousands or hundreds of thousands of years.
We conclude that the exceptionally voluminous explosive volcanism of Toba results from a combination of several factors. First, anomalously large amounts of volatiles are generated at depth, caused by the subduction of the IFZ. These volatiles cause active melting in the mantle wedge, thereby forming ascending magma diapirs. The direct ascent of the magmas to the surface is not efficient because of the presence of a ~38-km-thick continental crust22, which appears to be an obstacle for basic magmas that are not sufficiently buoyant. Therefore, they are stored in a large magma reservoir at the base of the crust. This reservoir serves as a powerful heat source and a source of upward-migrating overheated volatiles. Reaching the middle and upper crust, these volatiles cause active melting of silicic rocks. These melts form a shallow reservoir of magmas that are viscous and strongly enriched in volatiles and, therefore, are explosive. This reservoir does not produce small frequent eruptions and is spasmodically emptied during recurrent but sporadic catastrophic eruptions that are followed by long periods of quiescence. The results of our tomographic model presented in this study suggest that the Toba magma-generating engine continues to be active at present and, despite its current period of inactivity, this volcano system may generate strong eruptions in the future.