Application of a modified pattern informatics method to forecasting the locations of future large earthquakes in the central Japan

We propose a modification of the Pattern Informatics (PI) method that has been developed for forecasting the locations of future large earthquakes. This forecast is based on analyzing the space–time patterns of past earthquakes to find possible locations where future large earthquakes are expected to occur. A characteristic of our modification is that the effect of errors in the locations of past earthquakes on the output forecast is reduced. We apply the modified and original methods to seismicity in the central part of Japan and compared the forecast performances. We also invoke the Relative Intensity (RI) of seismic activity and randomized catalogs to constitute null hypotheses. We do statistical tests using the Molchan and Relative Operating Characteristic (ROC) diagrams and the log-likelihoods and show that the forecast for using the modified PI method is generally better than the competing original-PI forecast and the forecasts from the null hypotheses. Using the bootstrap technique with Monte-Carlo simulations, we further confirm that earthquake sequences simulated based on the modified-PI forecast can be statistically the same as the real earthquake sequence so that the forecast is acceptable. The main and innovative science in this paper is the modification of the PI method and the demonstration of its applicability, showing a considerable promise as an intermediate-term earthquake forecasting tool.     

Background seismicity in the Central Apennines of Italy: The Abruzzo region case study

We investigate background seismic activity of the Abruzzo region, a 5000 km² area located within the Central Apennines of Italy, where in the past 600 years at least 5 large earthquakes (I = XI–X) have occurred.Between April 2003 and September 2004, a dense temporary seismic network composed of 30 digital three-component seismic stations recorded 850 earthquakes with 0.9 < M_L < 3.7. We present earthquake locations and focal mechanisms obtained by standard procedures and an optimized velocity model computed with a search technique based on genetic algorithms.The seismicity occurs at a low and constant rate of  2.6 e− 04 events/daykm² and is sparsely distributed within the first 15 km of the crust. Minor increases in the seismicity rate are related to the occurrence of small and localised seismic sequences that occur at the tip of major active normal faults along secondary structures.We observe that during the 16 months of study period, the Fucino fault system responsible for the 1915 Fucino earthquake (M_S = 7.0), and the major normal faults of the area, did not produce significant seismic activity.Fault plane solutions evaluated using P-wave polarity data show the predominance of normal faulting mechanisms ( 55%) with NE-trending direction of extension coherent with the regional stress field active in this sector of the Apennines. Around 27% of the focal solutions have pure strike–slip mechanisms and the rest shows transtensional faulting mechanisms that mainly characterise the kinematics of the secondary structures activated by the small sequences.We hypothesize that the largest known NW-trending normal faults are presently locked and we propose that in the case of activation, the secondary structures located at their tips may act as transfer faults accommodating a minor part of the extensional deformation with strike–slip motion.     

Spatio-temporal stress states estimated from seismicity rate changes in the Tokai region, central Japan

Since unprecedented large-scale silent slip was detected by GPS in 2001 in the Tokai region, evaluating whether such movement is uniquely connected to the expected Tokai earthquake or repeatedly occurs in this area becomes vitally important. Because of short history of GPS observations and the limited areal coverage surrounding the Suruga trough, we take advantage of continuously recorded seismicity that is presumed to be sensitive to the deformation at seismogenic depth. Together with the well-maintained NIED earthquake data, we employ the seismicity-to-stress inversion approach of rate/state friction to infer the spatio-temporal stress changes in and around the presumed hypocentral zone of the future Tokai earthquake. Mapping stress changes inverted from microseismicity year by year, we find that the stress under Lake Hamana, the western expected future Tokai source, has been decreasing since 1999, during which the GPS data showed a normal trend of plate coupling. In contrast, stresses in the surrounding regions are calculated to have increased by transfer from Lake Hamana region. We interpret that this continuous process is associated with the 2000–2004 Tokai slow slip event. The characteristic patterns related to aseismic stress-release are also identified in the early 1980s and during 1987–1989, when slow events are inferred to have occurred on the basis of conventional geodetic measurements. Revisiting the seismotectonics and taking into account the mechanical implications of the inversion results, we argue that the transition zone situated between a deep stable creeping zone and a locked zone undergoes episodic creep and plays an important role in the transfer of stress to the locked zone. Consequently, even though we speculate that the current (2000 to present-day) silent slip event might be one of the repeating events, the inferred enlargement of the stress releasing area is significant and possibly raises the likelihood of the next Tokai earthquake.     

The Bajool earthquake sequence of 1991, and implications for the seismicity of central Queensland

The township of Bajool, situated 30 km south of Rockhampton, Queensland, experienced two minor earthquakes, each of Richter magnitude 2.9 on 10 June 1991. The foci were located 7 km north of Bajool, at a depth of less than 10 km. The felt intensity in Bajool was generally IV‐V, with minor damage occurring to several houses. There were several foreshocks and four aftershocks. The events may represent reactivation of a thrust fault, close to its junction with an inferred transform fault, the Fitzroy River Fault. These events demonstrate that the quiescence of central Queensland is apparent only, and that the recently installed University of Central Queensland Regional Network will provide a useful contribution to the understanding of earthquakes occurring along a passive continental margin.     

Variations of Seismicity in the Avachinsky Gulf (Kamchatka, Russia)

Variations of seismic mode in the region of the Avachinsky Gulf (Kamchatka, Russia) are considered. Observed anomalies (seismic quiescence, the ring seismicity, reduction of the slope of the earthquake recurrence diagram) provide a basis to consider this region as a place of strong earthquake preparation. The Kamchatka regional catalogues of earthquakes between 1962–1995 were used in the analysis. A reduced seismicity rate is observed during 10 years in an area of 150 km × 60 km in size. During the last five years, in the vicinity of the area considered, earthquakes with M > 5 occurred three times more often than the average over thirty years. It is interpreted as ring seismicity. The block of 220 km × 220~km in size, including the quiescence zone, is characterized by a continuous decrease of the recurrence diagram slope, which has reached a minimum value for the last 33 years in this region.     

Asperity distribution of the 1964 Great Alaska earthquake and its relation to subsequent seismicity in the region

The 1964 Alaska earthquake was the second largest seismic events in the 20th century. The aim of this work is the use of surface deformation data to determine asperity and slip distributions on the fault plane of the Alaska earthquake: these distributions are calculated by a Monte Carlo method. To this aim, we decompose the fault plane in a large number of small square asperity units with a side of 25 km; this allows us to obtain plane surfaces with an irregular shape. In the first stage, each asperity unit is allowed to slip a constant amount or not to slip at all, providing the geometry of the dislocation surface that best reproduces the observed displacements. To this purpose, a large number of slip distributions have been tried by the use of the Monte Carlo method. The slip amplitude is the same for all the asperities and is equal to the average fault slip inferred from the seismic moment. In the second stage, we evaluate the slip distribution in the dislocation area determined by the Monte Carlo inversion: in this case, we allow unit cells to undergo different values of slip in order to refine the initial dislocation model. The results confirm the previous finding that the slip distribution of the great Alaska earthquake was essentially made of two dislocation areas with a higher slip, the Prince William Sound and the Kodiak asperities. Analysis of the post-1964 seismicity in the rupture region shows a strong correlation between the larger earthquakes (M_w≥6) and the distribution of locked asperities following the 1964 event, which can be considered as an independent test of the validity of the model. We do not find slip values higher than 25 m for any of the patches, and we determine two separate high-slip zones: one correspondent to the Prince William Sound asperity, and one (18 m slip) to the Kodiak asperity. The slip distribution connected with the 1964 shock appears to be consistent with the following seismicity in the region.     

Paleoseismological evidence for non-characteristic behavior of surface rupture associated with the 2004 Mid-Niigata Prefecture earthquake, central Japan

The 2004 Mid-Niigata Prefecture earthquake sequence (mainshock magnitude, M_JMA 6.8), which occurred in an active fold-and-thrust belt in northern central Japan, generated a small thrust surface rupture (< 20 cm of vertical displacement) along a previously unmapped northern extension of the active Muikamachi–Bonchi–Seien fault zone, on the eastern margin of the epicentral region. To better understand past seismic behavior of the rupture, we conducted a paleoseismic trenching study across the 10-cm-high west-side-up surface rupture at the foot of a pre-existing 1.8-m-high east-facing scarp, which probably resulted from past earthquake(s). A well-defined west-dipping thrust fault zone accompanied by drag folding and displacing the upper Pliocene to lower Pleistocene strata and the unconformably overlying upper Pleistocene (?) to Holocene strata was exposed. The principal fault zone is connected directly to the 2004 surface rupture. From the deformational characteristics of the strata and radiocarbon dating, we inferred that two large paleoseismic events occurred during the past 9000 years prior to the 2004 event. These two pre-2004 events have a nearly identical fault slip (at minimum, 1.5 m), which is ≥ 15 times that of the 2004 event (∼ 10 cm). These paleoseismic data, coupled with the geological and geomorphological features, suggest that the 2004 event represented non-characteristic behavior of the fault, which can potentially generate a more destructive earthquake accompanied by meter-scale surface displacement. This study provides insight into the interpretation of past faulting events and increases our understanding of rupture behavior.     

Application of the Spatially Smoothed Seismicity and Monte Carlo Methods to Estimate the Seismic Hazard of Eritrea and the Surrounding Region

The region of interest is characterized by incomplete data sets and little information about the tectonic features. Therefore, two methodologies for estimating seismic hazard were used in order to elucidate the robustness of the results: the method of spatially smoothed seismicity introduced by Frankel (1995) and later extended by Lapajne et al. (1997) and a Monte Carlo approach presented by Ebel and Kafka (1999). In the first method, fault-rupture oriented elliptical Gaussian smoothing was performed to estimate future activity rates along the causative structures. Peak ground accelerations were computed for a grid size of 15 km × 415 km assuming the centre of the grids as epicentres, from which the seismic hazard map was produced. The attenuation relationship by Ambraseys et al. (1996) was found suitable for the region under study. PGA values for 10% probability of exceedence in 50 years (return period of 475 years) were computed for each model and a combined seismic hazard map was produced by subjectively assigning weights to each of these models. A worst-case map is also obtained by picking the highest value at each grid point from values of the four hazard maps. The Monte Carlo method is used to estimate seismic hazard, for comparison to the results from our previous approach. Results obtained from both methods are comparable except values in the first set of maps estimate greater hazard in areas of low seismicity. Both maps indicate a higher hazard along the main tectonic features of the east African and Red Sea rift systems. Within Eritrea, the highest PGA exceeded a value 25% of g, located north of Red Sea port of Massawa. In areas around the capital, Asmara, PGA values exceed 10% of g.     

The role of Pan-African structures in intraplate seismicity near the termination of the Romanche fracture zone, West Africa

The distribution of epicenters of both historic earthquakes and recent seismic events in southeastern Ghana, compiled from local and teleseismic networks, show strong correlation with the Pan-African structures onshore and indicate an alignment with disruptions on seismic sections offshore. The seismic reflection sections reveal basement structures of the external zone of the Pan-African Dahomeyide orogen and these structures can be traced to offsets of shelf strata and seabottom reflectors, providing direct evidence, for the first time, for neotectonic activity that may be responsible for seismicity in the area. The deep structure of the external zone consists of moderately-dipping reflectors inferred to represent high-strain zones in the variably deformed margin of the West African craton. Taken together, the available data suggest that active tectonics in this intraplate environment may involve inversion of the Pan-African thrust structures but that this activity is apparently not related to reactivation of the nearby Romanche Fracture Zone.     

2023年8月6日山东德州平原M 5. 5地震同震变形及地震活动性变化数值模拟

2023年8月6日,山东省德州市平原县发生M 5. 5地震,造成了严重的经济损失,打破了山东省近40年未发生M≥5. 0地震的平静期。平原地震震中区域人口稠密,经济发展迅速,其引起的区域地震活动性变化与未来强震发生的可能性是社会关注的热点。本文通过有限元数值模拟方法,建立三维模型研究震中孕震环境及其邻区同震形变应力场,利用假定破裂模型模拟地震同震位错,根据库仑应力变化理论探究周围区域地震活动性变化。结果表明,本次地震主要形变集中于发震断层10 km范围内,震中位置库仑应力下降超过500 kPa,未来短期内强震发生概率不大。平原地震对附近地震活动性造成了一定影响,可能促进了北部走滑型中小震活动,抑制了西侧正断型地震的发生。对于华北平原及邻近断层地震活动性变化仍需要更长时间地震数据和更深入的研究工作。     

Spatio-temporal evolution of Coulomb stress in the Pacific slab inverted from the seismicity rate change and its tectonic interpretation in Hokkaido, Northern Japan

We estimated spatio-temporal evolution of Coulomb stress within the subducted Pacific slab in Hokkaido from the analysis of seismicity rate change. For this purpose we used earthquake catalog from the Institute of Seismology and Volcanology (ISV), Hokkaido University for the period 1993/4/1–2006/12/31 after relocating to compensate location errors due to the heterogeneous P- and S-wave structure beneath Hokkaido. We found that spatial pattern of Coulomb stress change inverted from the seismicity rate change is comparable with static change in Coulomb stress estimated from dislocation models. Our results and analyses reveal important insights on spatio-temporal pattern of deformation of the subducted Pacific slab in terms of Coulomb stress change. We found that the 2003 Tokachi Oki earthquake (Mw = 8.0) pervasively perturbed Coulomb stress in a regional scale with a significant impact to trigger the 2004 Kushiro Oki earthquake. The 2004 Kushiro Oki earthquake (Mw = 7.0) is another significant stressing event that changed the pattern of Coulomb stress in the area. We found that stressing events with magnitude smaller than 7.0 has minimal impact on Coulomb stress change in the Pacific slab. Similarly, comparatively deep focused large earthquakes could not change Coulomb stress significantly. Further the pattern of Coulomb stress change after the 2003 Tokachi Oki earthquake correlates the pattern of afterslip distribution in Hokkaido.     

Seismicity anomaly scenario prior to the major recurrent earthquakes off the east coast of Miyagi Prefecture, northern Japan

This paper is concerned with the intermediate-term prediction of the forthcoming M7.4–8.2 earthquake on the plate boundary, off the east coast of Miyagi Prefecture, northern Japan, which has the highest occurrence probability among the long-term forecasted events announced to the public. Seismicity and aftershocks in the regions of stress-shadow preceding each of the previous ruptures in 1936 and 1978 shows significantly lower activity than the predicted rate by the ETAS model (the relative quiescence) during some years preceding the events, whereas the seismicity is normal or even activated in the regions of neutral or increasing Coulomb failure stress (CFS), which leads to the scenario based on the likely precursory slip within or near the source. Assuming such a scenario, a number of sequences of earthquakes or aftershocks during 1979–2004 from various regions in northern Japan are selected to analyze them by fitting the ETAS model. Then the results are examined in relation to the CFS increments in the considered regions using the source models of the 1793, 1936 and 1978 interplate ruptures, and additionally the source model of recently occurred 2003 Miyagi-Ken-Oki intra-slab earthquake of M7.1. It is likely that the results of the normal activity and relative quiescence in the respective activities are due to the preslip of the intra-slab earthquake rather than the preslip of the expected rupture on the plate boundary.     

Seismic velocity discontinuities in the crust and uppermost mantle beneath the Kanto district, central Japan, identified from receiver function imaging and repeating earthquake activity

We constructed vertical cross-sections of depth-converted receiver function images to estimate the seismic velocity structure of the crust and uppermost mantle beneath the Kanto district, central Japan. Repeating earthquake data for the plate boundary were also used to estimate geometries of the subducting Philippine Sea plate and the subducting Pacific plate. As a result, we present images of some major seismic discontinuities. The upper boundary of the Pacific plate dips to the northwest in northern Kanto and to the west–southwest in southern Kanto with some undulations. On the other hand, the upper boundary of the Philippine Sea plate as a whole dips to the northwest. However, it is concave to the northeast in the southern Boso peninsula. We suggest that the low-velocity mantle wedge may be indicated on the top of both subducting plates. Plate thickness gradually decreases to the northeast. The northeastern end of the Philippine Sea plate is interpreted to be at depths of 45–90 km. The Moho discontinuity in the overriding plate is deeper than 25 km in the northern Kanto. It contacts the subducting Philippine Sea plate in the southwestern part near 35.8°N.     

Regional variation in exhumation and strain rate of the high-pressure Sambagawa metamorphic rocks in central Shikoku, south-west Japan

Regional variation in the P–T path of the Sambagawa metamorphic rocks, central Shikoku, Japan has been inferred from compositional zoning of metamorphic amphibole. Rocks constituting the northern part (Saruta River area) exhibit a hairpin type P–T path, where winchite/actinolite grew at the prograde stage, the peak metamorphism was recorded by the growth of barroisite to hornblende and sodic amphibole to winchite/actinolite grew at the retrograde stage. In the southern part (Asemi River area), rocks exhibit a clockwise type P–T path, where barroisite to hornblende core is rimmed by winchite to actinolite. The difference in P–T path could suggest a faster exhumation rate (i.e. more rapid decompression) in the southern than in the northern part. On the other hand, physical conditions of deformation during the exhumation stage have been independently inferred from microstructures in deformed quartz. Recrystallized quartz grains in rocks from the low‐grade (chlorite and garnet) zones are much more stretched in the southern part (aspect ratio ≥ 4.0) than in the northern part (aspect ratio< 4.0), indicating a higher strain rate in the former than in the latter. These facts may indicate that the exhumation and strain rates are correlated (i.e. the exhumation rate increases with increasing the strain rate). The difference in the exhumation rate inferred from amphibole zoning between the northern and southern parts could be explained by an extensional model involving normal faulting, where the lower plate can be exhumed faster than the upper plate due to the displacement along the fault. Furthermore, the model may explain the positive correlation between the exhumation and strain rates, because the lower plate tended to support more stress than the upper plate.     

Robustness of the holistic seismic risk evaluation in urban centers using the USRi

The Urban Seismic Risk index (USRi) published in a previous article (Carreño et al., Nat Hazards 40:137–172, 2007) is a composite indicator that measures risk from an integrated perspective and guides decision-making for identifying the main interdisciplinary factors of vulnerability to be reduced or intervened. The first step of the method is the evaluation of the potential physical damage (hard approach) as a result of the convolution of the seismic hazard with the physical vulnerability of buildings and infrastructure. Subsequently, a set of social context conditions that aggravate the physical effects is also considered (soft approach). According to this procedure, the physical risk index is evaluated for each unit of analysis from existing loss scenarios, whereas the total risk index is obtained by multiplying the former index by an impact factor using an aggravating coefficient, based on variables associated with the socio-economic conditions of each unit of analysis. The USRi has been developed using the underlying holistic and multi-hazard approach of the Urban Risk Index framework proposed for the evaluation of disaster risk in different megacities worldwide. This article presents the sensitivity analysis of the index to different parameters such as input data, weights and transformation functions used for the scaling or normalization of variables. This analysis has been performed using the Monte Carlo simulation to validate the robustness of this composite indicator, understanding as robustness how the cities maintain the ranking as well as predefined risk level ranges, when compared with the deterministic results of risk. Results are shown for different cities of the world.     

Finite element modelling of elastic intraplate stresses due to heterogeneities in crustal density and mechanical properties for the Jabalpur earthquake region,central India

Deep lower crustal intraplate earthquakes are infrequent and the mechanism of their occurrence is not well understood. The Narmada-Son-lineament region in central India has experienced two such events, the 1938 Satpura earthquake and the 1997 Jabalpur earthquake, having a focal depth of more than 35 km. We have estimated elastic stresses due to the crustal density and mechanical properties heterogeneities along the Hirapur-Mandla profile passing through the Jabalpur earthquake region to analyse conditions suitable for the concentration of shear stresses in the hypocentral region of this earthquake. Elastic stresses have been computed by a finite element method for a range of material parameters. The results indicate that the shear stresses generated by the density heterogeneities alone are not able to locally enhance the stress concentration in the hypocentral region. The role of mechanical properties of various crustal layers is important in achieving this localization of stresses. Among a range of material parameters analysed, the model with a mechanically strong lower crust overlying a relatively weak sub-Moho layer is able to enhance the stress concentration in the hypocentral region, implying a weaker mantle in comparison to the lower crust for this region of central India.     

Crustal structure, fault segmentation, and activity of the Median Tectonic Line in Shikoku, Japan

We conducted a seismic tomographic analysis to estimate the crustal structure beneath the Shikoku and Chugoku regions in Japan. The Philippine Sea slab (PHS slab) subducts continuously in a SE–NW direction beneath this region, and the crustal structure is complex. Furthermore, the Median Tectonic Line (MTL), one of the longest and most active arc-parallel fault systems in Japan (hereafter, the MTL active fault system), is located in this area, and the right-lateral strike–slip movement of this fault system is related to the oblique subduction of the PHS slab. The MTL active fault system has ruptured repeatedly during the last 10 000 years, and has high seismic potential. Our tomographic analysis clarified the heterogeneous crustal structure along the MTL active fault system. This fault system in Shikoku can be divided into two segments, an east segment and a west segment, on the basis of the velocity structure. This segmentation model is consistent with other such models that have been determined from geological and geomorphological data such as fault geometry, slip rate, and faulting history. This consistency suggests that the surface characteristics of the MTL active fault system are related to structural properties of the crust. In particular, a prominent low-velocity (low-V) zone is present in the lower crust beneath the east segment. Our tomographic images show that the lower crust structure beneath the east segment is obviously different from that of the other segment. Furthermore, this low-V zone may indicate the presence of fluid, possibly related to dehydration of the PHS slab. As the presence of fluid in the lower crust affects the activity of the fault, stress accumulation and the fault failure mechanism may differ between the two segments of the MTL active fault system.     

Seismic investigation of near-surface geological structure in the Paducah, Kentucky, area: application to earthquake hazard evaluation

Seismics method were used to evaluate shallow geological conditions at 33 sites in the vicinity of Paducah, Kentucky. A combined set of P- and S-wave seismic refraction and reflection soundings were used, in addition to local borehole information, to produce structure maps of (1) a shallow (< 30 m deep) horizon believed to represent an unconformity surface at the top of the Eocene, and (2) the Paleozoic bedrock surface (< 85 to > 160 m deep). Shear-wave velocity contrasts across the shallow unconformity were generally 2-to-1 while the contrast at the top of the Paleozoic bedrock exceeds 5-to-1. These seismic boundaries have been determined to be very important in modelling and interpreting earthquake ground motion amplification in the Paducah area. The quality and accuracy of the data, and the cost effective nature of the methods, suggest that other communities in areas at risk to damage from seismic activity, with foundation conditions comparable to Paducah, might benefit from similar characterization in order (1) to identify seismically hazardous, near-surface, geological conditions, and (2) to develop geological models that could be used in computer simulations of site response.     

Paleoenvironmental analysis of the Mawaki archaeological site,central Japan,in relation to the stratigraphic position of dolphin bones

In order to elucidate the paleoenvironment associated with the early Holocene Mawaki archaeological site on the Noto Peninsula of central Japan, a high‐resolution stratigraphic study was conducted of 17 boreholes drilled at the archaeological site. We selected three boreholes for which lithological and/or chronological data are reported. Initial magnetic susceptibility was utilized for correlation of clastic core samples with the assistance of 26 radiocarbon dates. Four lithological units (A, B, C, and D in ascending order) were identified and interpreted as a sequence in a cycle of marine transgression and regression. Dated coastal horizons were chosen to indicate former sea levels. A Holocene relative sea level curve was generated on the basis of the geological data, and a rapid rise from 8000cal. yr B.P. to 7000cal. yr B.P. and a succeeding minor sea level fall represent the basic eustatic trend around the Sea of Japan because hydroisostatic and tectonic effects are moderate in the study area. Abundant dolphin bones lay just above the top of the marine sequence (boundary between Units C and D), located in the seashore environment. Cultural artifacts are found in a subaerial deposit (Unit D) near the dolphin bone level that is assigned to a period of high, stable sea level after the post‐glacial eustatic highstand. Dolphin bones are associated with stone artifacts (arrowheads, knives, and scrapers) and ritual wood columns, indicating the presence of a longstanding fishery during the early Holocene on the Sea of Japan coast. © 2011 Wiley Periodicals, Inc.     

Pull-down basin in the central part of Japan due to subduction-induced mantle flow

The driving force for the basin subsiding against isostatic balance in and around Lake Biwa in the Kinki district, Japan is discussed. The lake region is characterized by strong negative Bouguer anomalies, especially by a steep horizontal gradient zone of gravity anomaly running along the western margin of the lake. The large negative anomaly (>50 mgal) cannot be explained by low-density sediments beneath it. A down-warping structure extending to the Moho depth should be taken into account. This conjecture has been strongly supported by a short-period receiver function imaging, which shows a clear offset of about 8 km for the Moho discontinuity under the steep gravity gradient zone.A question arises as to what is the driving force to create such a large down-warping structure. We consider that the subduction of the shallow-dipping slab under the region (Philippine Sea Slab) may cause crustal deformation by dragging the viscous mantle downward. In order to verify this model, we simulated the induced mantle flow due to the subduction of the Philippine Sea Slab and the pressure distribution on the crust–mantle boundary. This numerical experiment showed that the induced flow makes a strong negative pressure zone under the lake region if the slab has a vertical offset along the direction of subduction. This offset of the slab is consistent with plate models deduced from hypocentral distributions and Sp phases of the deep-focus earthquakes.