Introduction

The 2011 off the Pacific coast of Tohoku earthquake was the largest earthquake recorded (Mw 9.0) in Japan, and it generated a mega-tsunami that severely damaged the coastal areas (CDMC, 2011). Following this event, the Japanese Government re-assessed the earthquakes and tsunamis expected in the future in the Suruga and Nankai troughs, close to central and western Japan (Fig. 1). Many studies have examined the earthquakes and tsunami events along subduction zones (e.g., Atwater 1987; Minoura et al. 2001; Nanayama et al. 2003; Monecke et al. 2008; Shishikura et al. 2010; Ishimura and Miyauchi 2015; Kitamura 2016; Inoue et al. 2017; Ando et al. 2018; Kitamura et al., 2018a, b; Goto et al. 2018). The Philippine Sea Plate is subducting beneath the Eurasian Plate at these troughs, and megathrust earthquakes of magnitude ~ 8 and ensuing large tsunamis have occurred since an AD 684 Hakuho earthquake, with a recurrence interval of 90–270 years (Ando 1975b; Ishibashi 2004).

Fig. 1
figure 1

Spatiotemporal distribution of great earthquakes along the Nankai and Suruga troughs, from Ando (1975b), Ishibashi and Satake (1998), Ishibashi (2014) and Kitamura et al. (2018a, b). Photographs are used with permission from the Geospatial Information Authority of Japan (https://maps.gsi.go.jp/development/ichiran.html). 1 Akatsuchiyama mound tomb (data source: Sangawa (2013)); 2 Sakajiri (data source: Sangawa (2001)); 3 Tsurumatsu (data source: Sangawa (2001)); 4 Harakawa (data source: Sangawa (2001)); 5 Kawai (data source: Sangawa (2001)); 6 this study

Full size image

Historical documents show that the Suruga and Nankai troughs are divided into six segments based on patterns of seismicity, labeled A–E and Z (Ando 1975b; Ishibashi 1976; Furumura et al. 2011). The 1854 Ansei-Tokai, 1854 Ansei-Nankai, 1944 Tōnankai Earthquake and 1946 Nankai Earthquake were caused by ruptures of segments C–E, A–B, C–D and A–B, respectively (Ishibashi 1976). Segment E (Suruga Trough) has not ruptured since the 1854 Ansei-Tokai earthquake, and this section shows remarkable interseismic crustal movement. In this context, Ishibashi (1976) predicted that a M-8 earthquake (Tokai earthquake) will occur sooner or later upon this segment. A geodetic precursor was reported for the 1944 Tōnankai Earthquake (Fig. 1). The data, taken far from the epicentral region, were interpreted as suggesting uplift of a few centimeters due to pre-slip that occurred one or two days before the main shock (Ando 1975a; Sato 1977; Mogi 1984/1985). Consequently, the Japanese government enacted the Large-Scale Earthquake Countermeasures Act (1978) for preparedness and short-term prediction of the Tokai Earthquake, although the earthquake has yet to occur.

During the past 40 years, many studies have investigated the occurrence of the Tokai Earthquake and the feasibility of short-term predictions of this event. For example, both historical and geological records indicate the 1854 Ansei-Tokai earthquake resulted in 1.5–1.8 m of coseismic uplift to the west of the Suruga Trough (Ishibashi 1984; Kitamura and Kobayashi 2014) (Fig. 2), although uplift in the western part of Suruga Bay was not detected during the 1707 Hoei and 1498 Meio earthquakes (Hatori 1975, 1977). On the other hand, Kitamura et al. (2018a, b) recently found emerged fossils of boring bivalves at Cape Omaezaki, which faces seismic segments D and E, indicating at least 2.5 m of total coseismic uplift resulting from the Tokai earthquake and the AD 1361 Shohei (Koan)–Nankai earthquake. These findings imply that ruptures of segments C, D, and E were simultaneous, and the estimated recurrence interval of ~ 500 years is consistent with the interval of 450–500 years reported by Kuroki et al.’s (2004) based on a numerical simulation. However, since the recurrence intervals and sizes of great earthquakes in many subduction zones worldwide are highly variable (e.g., Satake and Atwater 2007; Shimazaki et al. 2011; Rubin et al. 2017), we must examine the temporal distribution of earthquake prior to the 1361 Shohei Tokai earthquake.

Fig. 2
figure 2

Locality map of the study area. a Estimated coseismic vertical crustal displacement (m) resulting from the AD 1854 Ansei-Tokai earthquake along the coast of Suruga Bay (after Ishibashi (1984) and Kitamura and Kobayashi (2014)). b Locations of Shimizu Plain, Orido Bay, and Miho Peninsula, showing isobaths are based on Tsuchi (1955). c Location of the study area. The figure is modified from a map issued by the Geospatial Information Authority of Japan (http://maps.gsi.go.jp/). d Aerial photograph of the study area, used with permission from the Geospatial Information Authority of Japan (https://maps.gsi.go.jp/development/ichiran.html). e Topographic map of the Shizuoka and Komagoe districts (1:25,000 scale, compiled in 1932), showing the location of the study area

Full size image

Coseismic uplift events have been identified from geological and geomorphological records such as uplifted marine terraces (e.g., Fujiwara et al. 2010; Komori et al. 2017), emerged marine sessile assemblages (e.g., Shishikura et al. 2008; Iryu et al. 2009; Kitamura et al. 2014, 2015, 2017, Kitamura et al. 2018a, b), and changes in sedimentary facies and microfossil fauna in coastal lowland strata (e.g., Atwater 1987; Nelson et al. 1996a, b; Hayward et al. 2004; Sawai et al. 2004; Shimazaki et al. 2011; Kitamura and Kobayashi, 2014; Mannen et al. 2018). However, geomorphological records can be strongly influenced by human disturbances. Examining the sedimentary features of coastal lowland deposits is a promising approach in reconstructing ancient coseismic uplift events.

This study examines geological evidence for late Holocene megathrust earthquakes based on observations of sedimentary features in cored deposits collected from the area around the Kaicho-ji Temple on the Shimizu Plain, Shizuoka Prefecture, which faces seismic segment E (Fig. 2). The temple was built in AD 1011 on a beach ridge (Fig. 2) with an elevation of about 4 m. This timing provides reliable constraints on the age determination of the emergence of the study area. We identified an abrupt change in sedimentary facies that could have been caused by coseismic uplift during the early fourth century.

Study area

The Shimizu Plain is a fluvial-coastal lowland adjacent to Orido Bay, which is protected from the open ocean by Miho Peninsula (Fig. 2), a sand and gravel spit with an elevation of up to 10 m that formed at 5000 years BP (Matsubara, 1989). The Kaicho Temple is located on the beach ridge in the coastal area of the innermost part of Orido Bay. The Miho Peninsula is composed of sediment carried by the Abe River and derived from the Udo Hills to the west of the peninsula (Matsubara, 1989; Ishihara and Mizuno, 2016). The coastline is under a microtidal regime, with a mean tide range of 1.6 m at Shimizu Port. Tidal gauge data during 1955–2016 show that the Shimizu area has subsided at a rate of c. 6.3 mm/year (Coordinating Committee for Earthquake Prediction, 2018). According to Tsuchi (1956), mollusk remains collected at 5 m water depth provide evidence of euxinic and stagnant conditions throughout most of Orido Bay.

The coseismic uplift of the Shimizu Plain resulting from the 1854 Ansei-Tokai earthquake is documented on historical maps compiled immediately after the earthquake and mentioned in interviews conducted 38 years after the earthquake (Hatori 1976, 1977). Based on these accounts, Hatori (1977) and Ishibashi (1984) estimated coseismic uplifts of 1.5–2.0 m and 3 m, respectively. Kitamura and Kobayashi (2014) reported geological and paleontological evidence indicating a maximum coseismic uplift of 1.2 m caused by the 1854 Ansei-Tokai earthquake, on the area of the intertidal flat that emerged during the earthquake, and a total post-earthquake subsidence of ~ 0.6 m. Although historical documents show that the 1707 Hoei and 1498 Meio earthquakes caused damage on the Shimizu Plain, there were no coseismic uplifts (Hatori 1975, 1977). There are no historical records of earthquakes in the region prior to the 1361 Shohei Tokai earthquake. Nevertheless, archeological sites near segments D and E show sand boils and other liquefaction features that indicate intense shaking during certain time periods, including during the mid seventh to early eighth centuries AD, the second and early fourth centuries AD, and in the second century BC (Sangawa 2001, 2013) (Fig. 1). It is possible that these intense shaking events were caused by ruptures of segments D and E.

Historical reports also show that the coastal area of the Shimizu Plain was struck by tsunamis associated with the 1854 Ansei-Tokai and 1707 Hoei earthquakes. The wave heights of the tsunami caused by the 1854 Ansei-Tokai earthquake along the northeast-facing coast of the Miho Peninsula and the coastal area of the northern Shimizu Plain are estimated to have been 6 m and 3 m, respectively (Hatori 1977). No tsunamis have been documented in the inner part of Orido Bay (Hatori 1977). The tsunami related to the 1707 Hoei earthquake along the northeast-facing coast of Miho Peninsula and the coastal area of the northern Shimizu Plain had estimated wave heights of 5 m and 4 m, respectively (Hatori 1977).

Methods

We conducted ground-penetrating radar (GPR) surveys along three lines using an SIR-4000 unit (GSSI Co. Ltd., Nashua, NH, USA) with a 400 MHz antenna, with a vertical resolution of ~ 15 cm (Fig. 3). After acquisition, the GPR data was processed using the GSSI’s software RADAN version 7.4.15. During the processing, the finite impulse response (FIR) filter was used to reduce the noise and the gain correction was used to improve the visibility of the data.

Fig. 3
figure 3

Profiles obtained from a ground-penetrating radar survey. The right-hand vertical axis shows two-way travel time (ns) and the left-hand vertical axis shows estimated depth (m) for a given specific inductive capacity (ν; 14.0, 9.0, and 6.0 for profiles A−B, C−D, and E−F, respectively)

Full size image

Sediment cores (7 m long and 6 cm in diameter) recovered at two sites in the study area (Fig. 2) were split longitudinally and logged according to lithology, sedimentary structures, and the presence of mollusks. X-ray radiographs were taken of slab samples (6 cm wide × 20 cm long × 1 cm thick) from the split cores.

Grain size was measured using the wet-sieving method at 1-phi intervals in the range of − 3 phi (8 mm) and 5 phi (32 μm). A portion of the remaining subsamples was dried and ground following the removal of sandy grains. A 1 N HCI solution was added to 10 mg of sample in a silver container to remove inorganic carbon, and the remaining sample was heated on a hot plate for 2 h. Dried samples were placed in a tin container. The total sulfur content (TS) and the ratio of total organic carbon to TS (the C/S ratio) were determined using a CHNS material analyzer (Flash 2000, Thermo Fisher Scientific). The accuracy of the procedure was controlled using the BBOT standard (2,5-Bis-(5-tert.-butyl-benzoxazol-2-yl)-thiophen). As the C/S ratio of muddy sediments is higher in freshwater than in marine (Berner and Raiswell 1984) and euxinic and brackish coastal waters (Leventhal 1983), C/S ratios have been used to distinguish between ancient marine and freshwater sedimentary deposits (Berner and Raiswell 1984; Kamei et al. 1997; Kitamura et al. 2019) (Fig. 4).

Fig. 4
figure 4

Scatter plots of TS versus TOC for the cored sediments, and marine and freshwater sediments

Full size image

Foraminiferal tests were picked from the > 0.063 mm size fraction of core sediment samples. Taxonomic identification was based on Takata et al. (2006, 2016) and Charrieau et al. (2018). We calculated the number of specimens of the dominant taxa per gram of sediment. Transfer functions are commonly applied to coastal deposits for sea-level reconstruction (Horton et al. 1999); however, the very low abundance of foraminifera precluded the use of this technique in the present study.

Seventeen samples of well-preserved leaves and woody material were subjected to radiocarbon dating at Beta Analytic, using accelerator mass spectrometry (Table 1). The results were translated into the 2σ calendar-year range using the program OxCal4.1 (Bronk Ramsey 2009), based on a comparison with IntCal13 data (Reimer et al. 2013).

Table 1 14C ages of samples from the study area
Full size table

Results

GPR profile

The GPR profile along the line A–B showed seaward dipping reflections between 1 and 2 m in depth (Fig. 3). It suggests that the depositional setting near site 1 corresponds to the seaward slope of the beach ridge. The reflection pattern indicates continuous progradation of the coastline to the northeast. Note that reflections below 2 m were not successfully resolved due to the presence of the water table, and burial of artificial objects is visible in the systematic reflection patterns near the surface. The GPR profile along the lines C–D and E–F, which is shore-normal and shore-parallel directions, respectively, showed horizontal to sub-horizontal reflections below 1 m in depth (Fig. 3). It suggests a different sedimentary environment near site 2 comparing with that near site 1. Note that reflections above 1 m in depth was chaotic, which may correspond to landfills.

Sedimentary facies

We classified the borehole sediments into three facies on the basis of lithology, grain size, color, sedimentary structures, and fossil (Fig. 5; Appendixes 1 and 2).

Fig. 5
figure 5

Columnar sections of two sediment cores collected at the study area, showing two specimens of leaf materials (specimens 3 and 8). Details of radiocarbon dating reported are provided in Table 1

Full size image

Facies A is composed of massive, bluish-gray medium-grained sand and shows a fining-upward trend (Figs. 4, 5, 6 and 7). The lowest part of the borehole sediments contains plant materials and shells of the mollusks Umbonium moniliferum and Batillaria cumingii (Fig. 8). The relative abundance of mollusk shells shows an upward decrease within the facies. A few individuals of benthic foraminifera were found, which were Ammonia spp., Elphidium advenum, and Pseudononion japonicum (Fig. 8, Table 2). The relative abundance of P. japonicum shows an upward decrease and increase in sites 1 and 2, respectively (Fig. 8).

Fig. 6
figure 6

Grain-size distribution of cored sediments at site 1. See Fig. 5 for legend

Full size image
Table 2 List of foraminifera from cored sediment samples
Full size table

Facies B consists of laminated clay and graded sand beds (<10 cm thick) (Fig. 5), and overlies Facies A. The facies contains plant materials, but no mollusks or foraminifera (Fig. 8). The TS content and C/S ratio in the clay are 1–2% and ~ 0.7, respectively (Fig. 8). Clay (< 32 μm) contents are 70–80% (Figs. 6 and 7). The youngest 14C age from the clay is AD 398 to 539, obtained from a sample of a leaf (Fig. 5, Table 1).

Facies C consists of alternating sand and clast-supported gravel beds and occupies the uppermost part of the borehole sediments. The base comprises a layer (10 cm thick) of massive to weakly laminated fine-grained sand, and it overlies Facies B across a sharp boundary. This facies contains mud clasts and woody material, but no mollusks or foraminifera. At site 2, a woody debris bed was identified at a depth of 2.04–2.10 m (Figs. 5 and 7). Leaf material in this bed yielded 14C ages of AD 258 to 428 (Fig. 4, Table 1). GPR reflections show parallel to seaward dipping layers that indicate progradation of the beach ridge and shoreline (Fig. 3).

Fig. 7
figure 7

Grain-size distribution of cored sediments at site 2. See Fig. 5 for legend

Full size image

Discussion

Facies A contains shells of the mollusks U. moniliferum and B. cumingii, which inhabit the intertidal to shallow subtidal zones in tidal flats (Okutani 2000) (Fig. 8). A few individuals of benthic foraminifera Ammonia spp., Elphidium advenum, and Pseudononion japonicum were found in Facies A (Fig. 5). Ammonia spp. are generally found in the innermost bay regions and brackish water (e.g., Matsubara, 1989; Kosugi et al. 1991; Kitamura et al. 2019), and Elphidium advenum and Pseudononion japonicum inhabit the sublittoral zone (Akimoto and Hasegawa, 1989; Oki, 1989). As such, we interpret Facies A to have been deposited in a sandy marine environment, representative of the intertidal to shallow subtidal zones. According to Nomura and Seto (1992), Pseudononion japonicum prefers salinities of > 25‰. The difference in the relative abundance of P. japonicum at sites 1 and 2 might indicate movement of the river mouth over time.

Fig. 8
figure 8

Stratigraphic distributions of the dominant species of mollusk fossils, total sulfur content (TS), ratio of total organic carbon to TS (i.e., the C/S ratio), and foraminifera in cored deposits at sites 1 and 2. See Fig. 5 for legend

Full size image

The preservation of laminations and absence of mollusks and foraminifer shells in Facies B indicate deposition in euxinic bottom water. The C/S value of the clay (~ 0.7) is consistent with that of deposits in euxinic and brackish coastal water, such as the Black Sea (Leventhal 1983), and is much lower than those of intertidal mud-flat deposits (4.9–11.4) (Kitamura et al. 2019). This indicates that Facies B was deposited under euxinic and stagnant conditions below the intertidal zone. As the mean tide range at Shimizu Port is 1.6 m, the water depth of Facies B was estimated to be greater than − 0.8 m. The very low TS content (0.1%) and high C/S ratio (6.3) in the uppermost part of Facies B at site 1 might reflect alteration resulting from the seepage of groundwater from the overlying Facies C.

Facies C is characterized by alternating sand and clast-supported gravel beds and the occurrence of mud clasts. These features indicate a sedimentary environment characterized by high-energy hydrodynamic conditions. According to Kennedy and Woods (2012), wood debris beds on gravel-dominated beaches can be deposited during high wave events and are located at the top of the storm berm. It is therefore likely that Facies C was deposited in a backshore or beach ridge environment at an elevation of > 0.8 m, which corresponds to the sea level at high tide.

The vertical facies succession suggests the sedimentary environment changed from intertidal to shallow subtidal zones (Facies A), through a euxinic subtidal zone (Facies B), and then to a beach area (Facies C), indicating a sea-level rise followed by a drop. Radiocarbon dating shows that the sea-level rise took place between 1257 cal BC and 400 cal AD. Tanabe et al. (2016) estimated about 2 m of relative sea-level rise during 1000–0 cal BC based on the analysis of sedimentary records around Kanto Plain, central Japan. However, eustatic sea level was likely stable during that time (Sloss et al. 2007; Woodroffe et al. 2012). As noted above, the mean subsidence rate is 6.3 mm/year in the Shimizu area. With the duration of the depositional event assumed to be 1660 years (i.e., between 1257 cal BC and 400 cal AD), the mean sedimentation rate is estimated to be about 1.2 mm/year. We therefore interpret that the sea-level rise during the deposition of Facies A–B was caused mainly by local subsidence (Fig. 9).

Fig. 9
figure 9

Reconstruction of water-depth changes and depositional environments in the study area (Shimizu, central Japan) for the periods before and after the AD 400 earthquake

Full size image

Of note, the abrupt change from the subtidal zone (below − 0.8 m) to beach conditions (above 0.8 m) is indicated by lithofacies changes from Facies B to C, with an absence of intertidal Facies A. The fall in relative sea level is estimated to have been up to 1.6 m (Fig. 9). The youngest 14C age within Facies B is AD 398–539, derived from leaf material (specimens 3), while the oldest 14C age within Facies C is AD 258–428, also derived from leaf material (specimen 8) (Fig. 5). As these specimens are thin and delicate leaves, their very well-preserved condition (Fig. 5) indicates that they have not been reworked from older deposits. As such, based on the 14C ages of these specimens, we estimate that sea level fell during the period AD 398–428.

There are no reports of eustatic sea-level decreases at this time; however, Sangawa (2001, 2013) (Fig. 1) reported evidence for intense shaking in west Shizuoka Prefecture during the early fourth century (segment D). The present study area experienced 1.2 m of coseismic uplift during the 1854 Ansei-Tokai earthquake (Kitamura and Kobayashi 2014). We therefore propose that the fall in relative sea level was caused by coseismic uplift associated with the rupture of segment E, possibly during the AD 400 earthquake (Fig. 9). As the base of Facies C is located at mean sea level, the total subsidence was at least 0.8 m greater than the uplift caused by the coseismic uplift events after the AD 400 earthquake.

This study did not identify a tsunami deposit associated with the AD 400 event, for which there are two potential explanations. The first is that Miho Peninsular reduced the tsunami’s energy. Indeed, no tsunami associated with the 1854 Ansei-Tokai earthquake has been documented in the inner part of Orido Bay, even though the wave height of the tsunami along the coastal area of the northern Shimizu Plain is estimated to have been 3 m (Hatori 1977). The second explanation is that an earthquake nucleating in segment E might induce significant uplift but does not cause a large tsunami because the segment is located too close to the shore.

Sangawa (2001, 2013) postulated that a large earthquake occurred around AD 400 within segment B, as inferred from slope failure at Akatsuchiyama mound tomb (Fig. 1). Therefore, it is thought that megathrust ruptures of segments B–E occurred simultaneously or within a few years of each other at ~ AD 400. The interval between the AD 400 and 684 Hakuho earthquakes is about 280 years, slightly longer than the average interval of 220–230 years between events that occurred between the AD 684 and 1361 earthquakes. The data from this study show that segment E ruptured during the 1854 Ansei-Tokai, 1361 Shohei Tokai, and AD 400 earthquakes, and that the recurrence interval varies from about 500–1000 years.

Conclusion

To clarify the history of megathrust earthquakes along the Suruga Trough, central Japan, we studied the sediment, fossil contents, and the C/S ratio of two sediment cores collected at Kaichoji Temple, which was built in AD 1011 on a beach ridge in the coastal area of the Shimizu Plain, which is protected from the open sea by the Miho Peninsula. The results can be summarized as follows:

  1. 1.

    We identified three sedimentary facies: Facies 1 (intertidal to shallow subtidal zones), Facies 2 (stagnant conditions below the intertidal zone), and Facies 3 (backshore and beach ridge). A sharp boundary between Facies 2 and the overlying Facies 3 at 0.20–0.25 m above sea level indicates a rapid fall in relative sea level of 1.6 m at AD 398–428.

  2. 2.

    The age of the boundary between Facies 2 and Facies 3 corresponds to the timing of a large earthquake in west Shizuoka Prefecture around AD 400. The abrupt change in the sedimentary environment was possibly related to coseismic uplift caused by a megathrust rupture along the Suruga Trough.