Tsunami history over the past 2000 years on the Sanriku coast, Japan, determined using gravel deposits to estimate tsunami inundation behavior

Tsunami history over the past 2000 years on the Sanriku coast, Japan, determined using gravel deposits to estimate tsunami inundation behavior

Accepted Manuscript Tsunami history over the past 2000 years on the Sanriku coast, Japan, determined using gravel deposits to estimate tsunami inundat...

21MB Sizes 0 Downloads 1 Views

Accepted Manuscript Tsunami history over the past 2000 years on the Sanriku coast, Japan, determined using gravel deposits to estimate tsunami inundation behavior

Tomoko Goto, Kenji Satake, Toshihiko Sugai, Takeo Ishibe, Tomoya Harada, Aditya Riadi Gusman PII: DOI: Reference:

S0037-0738(19)30001-6 https://doi.org/10.1016/j.sedgeo.2019.01.001 SEDGEO 5433

To appear in:

Sedimentary Geology

Received date: Revised date: Accepted date:

26 November 2018 31 December 2018 3 January 2019

Please cite this article as: T. Goto, K. Satake, T. Sugai, et al., Tsunami history over the past 2000 years on the Sanriku coast, Japan, determined using gravel deposits to estimate tsunami inundation behavior, Sedimentary Geology, https://doi.org/10.1016/ j.sedgeo.2019.01.001

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Tsunami history over the past 2000 years on the Sanriku coast, Japan, determined using deposits

to

estimate

tsunami

PT

gravel

inundation behavior

RI

Tomoko Gotoa, Kenji Satakea, Toshihiko Sugaib, Takeo Ishibec, Tomoya Haradaa, Aditya

SC

Riadi Gusmand

NU

a) Earthquake Research Institute, the University of Tokyo, 1-1-1 Yayoi, Bunkyo, Tokyo 113-0032, Japan

MA

b) Graduate School of Frontier Sciences, the University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa City, Chiba 277-8563, Japan Association

for

the

Development

of

Earthquake

Prediction,

1-5-18,

D

c)

PT E

Kanda-Sarugakucho, Chiyoda, Tokyo 101-0064, Japan d) GNS Science, 1 Fairway Drive, Avalon 5010, Lower Hutt, New Zealand

Tomoko Goto

CE

Corresponding Author Information

AC

Affiliation: Earthquake Research Institute, the University of Tokyo Address: 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan Phone: +81-3-5841-5703 E-mail: [email protected]

1

ACCEPTED MANUSCRIPT

Abstract At Numanohama Marsh on the rocky Sanriku coast in northeast Japan, we identified 17 sand layers with sedimentary structures characteristic of event deposits (i.e., a sharp

PT

erosional contact with the underlying unit, a normal and/or inverse grading structure, parallel/cross lamination, mud clasts, and/or a thin mud layer). Abundant marine

RI

nannofossils, coccoliths, and the small catchment area in the narrow valley suggest that

SC

these event layers were likely formed by inundations during tsunamis or storms. Among the 17 sand layers, nine contained gravels derived from the beach, riverbed, or slope

NU

areas. Six event layers contained all three of these gravels, suggesting a strong inflow

MA

current that transported round pebbles from the beach, erosion of wall rock in the valley during the inflow and outflow phases, and a backwash current that caused sub-angular

D

riverbed gravel to accumulate in lowland areas. These six layers were traced up to 490

PT E

m inland from the coastline (present elevation: 7 m above mean sea level) and were several cm thick at almost every survey point. We concluded that these six event layers are tsunami deposits, and dating results revealed that they correlate with the 2011

CE

Tohoku-oki earthquake; the 1933 and 1896 Sanriku-oki earthquakes; any of the 1763

AC

Aomori-oki, 1793 Miyagi-oki, and 1856 Aomori-oki earthquakes; the 1611 Tohoku-oki earthquake; and the 869 Tohoku-oki (Jogan) earthquake. In addition, changes in the sedimentary environment inferred from diatom assemblages were well correlated with the deposition of three of the event layers. Two historical earthquakes — the 1454 and 1611 Tohoku-oki earthquakes — have been previously suggested as being the immediate predecessors of the 2011 Tohoku-oki earthquake, and our dating results favor the 1611 earthquake. Although the source region of the 869 Tohoku-oki (Jogan)

2

ACCEPTED MANUSCRIPT

earthquake has been estimated to have been off the Fukushima and Miyagi prefectures, our results suggest that the source region extended further north along the Japan Trench.

PT

Keywords: Gravel provenance; Late Holocene; Sanriku coast; The 869 Tohoku-oki

AC

CE

PT E

D

MA

NU

SC

RI

(Jogan) earthquake tsunami; Tsunami deposits

3

ACCEPTED MANUSCRIPT

1. Introduction Prior to 2011, paleo-tsunami deposit surveys along the Tohoku coast were primarily conducted on the Sendai Plain, a coastal plain with a sandy beach. Tsunami deposits resulting from the 869 Tohoku-oki (Jogan) earthquake were identified as those

PT

underlying the To-a volcanic ash layer, a tephra ejected from the Towada volcano in 915

RI

(e.g., Minoura and Nakaya, 1991; Sawai et al., 2008). Geological surveys have mapped

SC

five extensive tsunami deposits in sediments spanning 3000 years, with a recurrence interval of 500 to 800 years (Fig. 1) (Sawai et al., 2012; Satake, 2015). Prior to 2011,

NU

few tsunami deposit studies were conducted along the rocky Sanriku coast, located north of the Sendai plain (e.g., Yagishita, 2001; Haraguchi and Ishibe, 2009).

MA

The 2011 off the Pacific coast of Tohoku earthquake (hereafter the 2011 Tohoku-oki earthquake: “oki” means offshore in Japanese) tsunami provided an excellent

D

opportunity to investigate the sedimentary features of modern tsunami deposits. New

PT E

insights for tsunami deposits (e.g., finding of the common presence of tsunami deposits in the form of mud, findings of geochemical signature to document the extent of

CE

inundation even in case of lack of tsunami deposits, or possibility of formation of

AC

tsunami deposits with no or very small component of marine origin sediments) have been obtained through sedimentological, micropaleontological, and geochemical studies (e.g., Abe et al., 2012; Chagué-Goff et al., 2012; Goto et al., 2012, 2014; Jagodziński et al., 2012; Pilarczyk et al., 2012; Szczuciński et al., 2012b; Fujiwara and Tanigawa, 2014; Shinozaki et al., 2015; Goto et al., 2017). In addition, recent comparative studies of modern tsunami and storm deposits have suggested criteria to distinguish tsunami and storm deposits (Nanayama et al., 2000; Goff et al., 2004; Tuttle et al., 2004; Morton

4

ACCEPTED MANUSCRIPT

et al., 2007). However, the criteria remain controversial because tsunami and storm deposits sometimes have similar sedimentological and stratigraphic characteristics (e.g., Piotrowski et al., 2017; Soria et al., 2017; Watanabe et al., 2017). For example,

PT

lamination is considered to be characteristic of storm overwash deposits (e.g., Morton et al., 2007), though laminated tsunami deposits have also been reported (e.g., Srinivasalu

RI

et al., 2007; Szczuciński et al., 2012a, 2012b; Switzer et al., 2012). Recently, Gusman et

SC

al. (2018) developed a numerical method to simulate the grain-size distribution within tsunami deposits and suggested that the wave amplitude and period of the 2011 tsunami

NU

could be estimated from the grain-size distribution.

MA

Following the 2011 tsunami, many geological surveys have been conducted to reconstruct the history and spatial extent of historical and paleo-tsunamis along the

D

Tohoku coast (Fig. 1, Tanigawa et al., 2014; Goto et al., 2015; Ishimura and Miyauchi,

PT E

2015, 2017; Takada et al., 2016; Goto et al., 2017). Along the rocky Sanriku coast, Ishimura and Miyauchi (2015) identified tsunami deposits associated with the 869 Tohoku-oki (Jogan) earthquake at Koyadori, Yamada Town, Iwate Prefecture, the age of

CE

which was re-examined by Ishimura (2017). In addition, four distinct gravelly sand

AC

layers were recently identified in a deposit dated to 1100–2700 cal BP at Noda Village on the northern Sanriku coast (Inoue et al., 2017), with the age of the youngest tsunami deposit correlated with the 869 Tohoku-oki (Jogan) earthquake. In our previous study (Goto et al., 2017), we investigated the sedimentary features and compositions of the 2011 tsunami deposits and characterized autochthonous gravels in beach, slope, and riverbed areas by rock type and clast roundness. We found that the 2011 tsunami layer contains gravel from all three areas and demonstrated that these

5

ACCEPTED MANUSCRIPT

gravels can be used to infer tsunami inundation behavior at the study site. In the present study, we reconstructed the event history of the past 2000 years on the Sanriku coast. To overcome the above-mentioned difficulties in distinguishing tsunami and storm deposits,

PT

we applied our previous approach of classifying event layers by their gravel composition (Goto et al., 2017). We show that careful investigation of the characteristics

RI

of local autochthonous gravels and of the gravel composition within event layers can be

SC

useful for identifying tsunami deposits and reconstructing paleo-tsunami history,

NU

especially along rocky coasts where various gravels are distributed.

2. Historical tsunamis and storms on the Sanriku coast

MA

Historically, the Sanriku coast in the northern Tohoku region (Fig. 1) has been impacted by several devastating tsunamis resulting from earthquakes along the Japan

D

Trench, where the Pacific Plate is subducting beneath northern Honshu at a rate of

PT E

approximately 8 cm yr-1 (e.g., Sella et al., 2002). The most recent tsunami, generated by the 2011 Tohoku-oki earthquake, had a run-up height of up to 40 m along the Sanriku

CE

coast (Tsuji et al., 2011; Mori et al., 2012; Tsuji et al., 2014; Ogami and Sugai, 2018).

AC

Similarly, the 1933 Sanriku-oki earthquake produced a tsunami with a run-up height of 30 m (Earthquake Research Institute, 1934) and the 1896 Sanriku-oki earthquake produced a tsunami with a run-up height of up to 38 m (Iwabuchi et al., 2012; Satake et al., 2017). The severe damage caused by large earthquakes during the non-instrumental period, as well as the heights of the generated tsunamis, are documented in the literature. According to analyses of historical documents (e.g., Usami et al., 2013), the 1677, 1763,

6

ACCEPTED MANUSCRIPT

and 1856 Aomori-oki earthquakes, the 1793 Miyagi-oki earthquake, and the 869, 1454, and 1611 Tohoku-oki earthquakes generated large tsunamis along the Tohoku coast (Table 1). Regarding the 869 Tohoku-oki (Jogan) earthquake, a Japanese history book,

PT

Nihon Sandai Jitsuroku (A chronicle of Japan), records “On the 26th of the 5th month of the Japanese calendar (July 9, 869), a large earthquake occurred in the province of

RI

Mutsu (divided since into the five provinces of Iwaki, Iwashiro, Rikuzen, Rikutyu and

SC

Mutsu). The sky was illuminated like day-time. A little later, people, panic-stricken by the violent trembling, were lying on the ground; some were buried under fallen houses

NU

and others inside wide-opened ground fissures, while horses and cows desperately ran

MA

about and trampled each other. A number of castles, towers, and other tall structures collapsed. Then roarings like thunder were heard towards the sea. The sea soon rushed

D

into the villages and towns, over-whelming a few hundred miles of land along the coast.

PT E

There was scarcely any time for escape, though there were boats and the high ground just before them. In this way about 1000 people were killed. Hundreds of hamlets and villages were left in ruins.” (Imamura, 1934). This earthquake is suggested to be one of

CE

the predecessors of the 2011 Tohoku-oki earthquake (e.g., Sawai et al., 2015). The

AC

magnitude and source region of the 869 earthquake were re-evaluated (e.g., Sugawara et al., 2012; Namegaya and Satake, 2014) using the insights obtained from the 2011 tsunami.

Two historical Tohoku-oki earthquakes (i.e., the 1454 and 1611 Tohoku-oki earthquakes) are suggested as potential immediate predecessors of the 2011 Tohoku-oki event. Namegaya and Yata (2014) searched a historical earthquake and tsunami database for tsunami descriptions they considered reliable and found that the tsunami following

7

ACCEPTED MANUSCRIPT

the earthquake on 12 December 1454 inundated a wide area of the Tohoku coast, resulting in a large number of casualties. In addition, Ebina (2014) examined historical documents describing the 1611 Tohoku-oki earthquake and tsunami and found that

PT

although the earthquake caused ground shaking from Tohoku to Tokyo without damage, the resulting tsunami devastated areas along the Tohoku coast, resulting in numerous

RI

casualties. Supporting these claims, Sawai et al. (2015) identified a sand sheet

SC

containing marine and brackish diatoms on the Sendai Plain, which they dated to AD 1406–1615 (2σ range).

NU

The recurrence of giant earthquakes along the Kuril Trench has been investigated by

MA

surveying tsunami deposits and the most recent devastating tsunami is suggested to have occurred during the 17th century (e.g., Nanayama et al., 2003; Sawai et al., 2009).

D

A recent tsunami simulation has suggested a combined rupture of off-Tokachi and

PT E

off-Nemuro regions, including a slip of 25 m at the shallower part of the plate interface near the trench axis (total seismic moment, 1.7 × 1022 Nm [Mw 8.8]) (Ioki and Tanioka, 2016). However, the tsunami heights at our survey site calculated from the above source

CE

are no higher than 4 m, probably because the survey site is located in the extension of a

AC

long axis of rectangle faults that radiate smaller tsunami energy than that radiated from the extension of a short axis (e.g., Kajiura, 1970; Ben-Menahem and Rosenman, 1972). This implies that tsunami deposits at our survey site are mostly the result of near-field tsunamis originating along the Japan Trench. Trans-Pacific tsunamis generated by giant earthquakes have also caused damage along the Sanriku coast. The 1960 Chilean tsunami struck the Sanriku coast with a height of several meters approximately 23 hours after the earthquake (e.g., Committee

8

ACCEPTED MANUSCRIPT

for Field Investigation of the Chilean Tsunami of 1960, 1961), resulting in 123 casualties in the Tohoku region. Similarly, the tsunami generated by the 2010 Maule (Chile) earthquake also propagated across the Pacific Ocean, striking the Sanriku coast

PT

with a height of approximately 2.0 m (Tsuji et al., 2010). Historical documents also record an orphan tsunami that struck the Tohoku coast without any ground shaking,

RI

which was later found to have been generated by the 1700 Cascadia earthquake (Satake

SC

et al., 1996; Atwater et al., 2005).

Storm disasters along the Sanriku coast are also documented (Arakawa et al., 1961;

NU

Miyako City Education Committee, 1991). Nine storms characterized by heavy rain,

MA

heavy wind, and flooding are recorded to have struck Aomori, Iwate, and Miyagi Prefectures during the Edo era (1603–1867) (Table 2 in Goto et al., 2015). Descriptions

D

of two of the storms (in 1648 and 1850) include the term “high tide”. In modern history,

PT E

two disastrous storms have struck the area: Typhoon Catherine in 1947 and Typhoon Ione in 1948, which produced inundation heights in Miyako City of 4 and 5 m,

3. Study site

CE

respectively.

AC

The survey site was Numanohama Marsh in Taro Otobeno, Miyako City, Iwate Prefecture, on the Sanriku coast (Fig. 2). The Sanriku coast is a typical ria coastline with steep valleys formed by submergence and subsequent flooding of the mountainous terrain, producing numerous deep narrow bays of various sizes and depths (e.g., Koike et al., 2005). The survey site was located in a lowland back marsh surrounded by high, steep cliffs composed of Cretaceous volcanic rock and isolated from the open sea by a beach ridge with a present height of 4.5 m. A felsic volcanic rock zone is distributed

9

ACCEPTED MANUSCRIPT

within 300 m of the coast. A small, narrow river with an estimated catchment area of 2 km × 1 km flows through the central part of the study site. Granitic host rocks are distributed upstream, and sub-angular granite clasts are deposited on the riverbed near

PT

the estuary. The To-a tephra unit (AD 915) is commonly used on the Sendai plain to confirm underlying tsunami deposits as resulting from the 869 Tohoku-oki (Jogan)

RI

earthquake; however, our study site was located outside the fall area of the To-a tephra

SC

(Fig. 1) (Machida and Arai, 2003).

At the study site, the height of the tsunami due to the 2011 Tohoku-oki earthquake

NU

ranged from 16.5 to 30.1 m (Tsuji et al., 2011, 2014). Previously reported tsunami

MA

heights near the study site were 24.2 m (due to the 1896 Sanriku-oki earthquake) (Yamana, 1896; reproduced by Unohana and Ota, 1988), 7.2 m (due to the 1933

D

Sanriku-oki earthquake) (Earthquake Research Institute, 1934), and 2.2 m (due to the

PT E

1960 Chile earthquake) (Committee for Field Investigation of the Chilean Tsunami of 1960, 1961).

CE

4. Materials and methods

4.1. Sampling and sedimentological analysis

AC

In the present study, cores were collected by machine coring at four sampling sites (C1–4) located 260–490 m from the coastline (core C2: diameter 116 mm; cores C1, C3–4: 90 mm; length, 1.3–5.7 m) (Fig. 2). Sites C1 and C2 were only 3 m away from the marsh, site C3 was 15 m from the marsh, and the most inland site, C4, was 220 m from the marsh and 30 m from the river channel. The 15 Handy Geoslicer samples (L1– 15) were obtained in the lowland behind the beach ridge during our previous surveys (Goto et al., 2015).

10

ACCEPTED MANUSCRIPT

A total of 19 sedimentary samples (4 machine cores + 15 Handy Geoslicer cores) were analyzed. After collection and visual inspection of the sedimentary samples, the grain size (clay to gravel), sorting, composition, and structure of each layer were

PT

carefully sketched. Candidate event layers were identified based on grain-size distribution and the presence of a sharp erosional contact with the underlying layer,

RI

normal/inverse grading, parallel/cross laminations, mud clasts, and/or a thin mud layer

SC

(e.g., Moore et al., 2006; Goto et al., 2011; Goff et al., 2012; Fujiwara and Tanigawa, 2014). As core C2 includes the largest number of event layers, event layers were

NU

identified first in core C2, and then correlated between the other cores and sites.

MA

Grain-size analyses of candidate event layers identified in core C2 were performed on samples of ~20 g collected at 2 cm intervals. Briefly, after plant material was removed, particles >4Φ were extracted from the samples by wet sieving. After recording their dry

PT E

D

weight, the remaining sand/pebbles were sieved into 11 size fractions from −6 to 4Φ at 1Φ intervals, and the weight percent of each size fraction was calculated. Sand and gravel particles larger than 4Φ were observed under a stereomicroscope and a

CE

representative photograph of the particle composition of each event layer was taken. compositions

were visually identified under a

AC

Sand particle and mineral

stereomicroscope and used as characteristics of the event layer. Then, the event layers were correlated across the different core samples based on their facies, sand particle and mineral compositions. Magnetic susceptibility was measured for both autochthonous and allochthonous deposits by using a pocket-size magnetic susceptibility meter (SM-30; ZHinstruments), which has a sensitivity of 1 × 10−7 SI units, based on manufactures features.

11

ACCEPTED MANUSCRIPT

4.2. Gravel composition analysis Autochthonous gravels were extracted from the event layers identified in cores C2 and C4. We used all available gravels (i.e., 3–580 clasts per layer) above 2 mm diameter.

PT

The roundness (Krumbein, 1941) and rock type of the gravels were determined by

RI

visual inspection, and the gravel sources were defined as ‘beach’, ‘riverbed’, or ‘slope’

SC

following our previous methods (Goto et al., 2017) (Fig. 3). Beach gravel was defined as well-rounded (0.8–0.9 roundness) and included any rock type, slope gravel was

NU

defined as angular to sub-angular (0.1–0.2 roundness) and included any rock type, and riverbed gravel was defined as sub-angular (0.4–0.5 roundness) granites.

MA

After the source identification, we determined the relative ratios of the three index gravels in each event layer and created triangle diagrams. However, due to the limited

D

amount of gravel in each sample, the relative ratios of the three index gravels might be

PT E

biased. Therefore, we also classified each event layer as one of four types based on the presence or absence of each of the index gravels: Type-A contained all three index

CE

gravels, Type-B only beach and slope gravels, Type-C only riverbed and slope gravels,

AC

and Type-D contained only finer particles without any index gravel.

4.3 Marine nanofossil analysis The presence of coccoliths was used as an index of the historical presence of seawater. Coccoliths are calcareous, photosynthetic, single-celled organisms living at the sea surface and are often used as index fossils (Matsuoka and Okada, 1989). At the survey site, coccoliths are considered to have been brought from the ocean because the host rock around the survey site is Cretaceous volcanic rock. Samples were collected at 5–

12

ACCEPTED MANUSCRIPT

10-cm intervals from core C2, and coccoliths were counted under a microscope (e.g., Okamura and Yamauchi, 1984; Goto et al., 2015). Because coccoliths were heterogeneously distributed in the samples, we counted the total number of coccoliths in

PT

10 microscope fields at 400× magnification.

RI

4.4 Diatom analysis

SC

To examine changes in the sedimentary environment, we counted the number of diatom frustules in samples of autochthonous peaty soil collected at 12 different depths

NU

above or below the event layers in core C2. To remove organic material, 10% hydrogen peroxide was added to approximately 1 cm3 of sample in a test tube. After incubation

MA

for 24 h, diatom frustules were identified from the material remaining in the test tube and counted under a light microscope at 600× to 1500× magnification. Two partially

D

crushed frustules were counted as one diatom frustule. In each sample, more than 200

PT E

diatom frustules were counted (Kosugi, 1986, 1988; Ando, 1990), and the relative abundance and density of each diatom species calculated. Diatom density is reported as

CE

the number of diatom frustules per cubic centimeter. In addition, by referring to Kosugi

AC

(1986, 1988) and Ando (1990), who classify modern diatoms as an environmental indicator, we characterized the identified diatom frustules by their living environment into marine-brackish, brackish, brackish-freshwater, or freshwater taxa. Finally, five diatom zones within core C2 were identified based on the dominant diatom species.

4.5. Dating and estimation of event ages Organic material such as plant/wood fragments and charred material obtained from above and below the identified event layers were used for dating. The samples were

13

ACCEPTED MANUSCRIPT

cleaned in an ultrasonicator and dried at 80 °C for 24 h. The ages of the samples were then estimated by

14

C accelerator mass spectrometry (AMS). The radiocarbon ages of

the samples were calibrated to calendar years with an error range of 2σ using the OxCal

PT

program (ver. 4.2.3; Bronk Ramsey, 2008; Bronk Ramsey and Lee, 2013) and the IntCal13 database (Reimer et al., 2013).

RI

We used Bayesian estimation with the sequence model for age-depth relation, to

program. Based on the

137

Cs and

210

SC

stochastically constrain the event age ranges (Bronk Ramsey, 2009) in the OxCal Pb dating results reported previously (Goto et al.,

NU

2015), we constrained the event ages of three of the upper layers (S1, S4, and S5; see

MA

section 5 for details) to three known event dates: the Tohoku-oki earthquake on 11 March 2011, the Sanriku-oki earthquake on 3 March 1933, and the Sanriku-oki

D

earthquake on 15 June 1896, respectively. We also used the boundaries of the identified

PT E

diatom zones as additional constraints on the event age ranges. Finally, we correlated the dated event layers with historical events.

CE

5. Identification and characterization of event layers

AC

Multiple sandy layers were distinguishable from the autochthonous peaty soils in the back marsh in all of the sedimentary samples collected. Some of these sandy layers include gravels of various shapes and rock types and have sedimentary features characteristic of event deposits, such as a sharp erosional contact with the underlying layer, a normal/inverse grading structure, parallel/cross lamination, mud clasts, and thin mud layer; we therefore concluded that these layers were allochthonous event deposits. As a result of correlation of event layers along the transect, we identified 17 event

14

ACCEPTED MANUSCRIPT

layers (Fig. 4), which we named S1–S17 from top to bottom. Nine of these event layers (S1–S8 and S10) were identified in our previous survey and inferred to have been deposited in the past 500 years (E1 to E9 in Goto et al., 2015). The sand layer S6 (E6 in

PT

Goto et al., 2015) at 140–141 cm depth in Handy Geoslicer sample L4 is composed of very fine to medium sand without gravel. This layer was not observed in cores C1–4.

SC

RI

The remaining eight layers (S9 and S11–S17) were newly identified in the present study.

5.1. Event layers in core C2

NU

In core C2 (length, 5.7 m), we identified 16 sandy layers with sedimentary features characteristic of event deposits (Figs. 4b, 5). The tip of core C2 reached the bedrock

MA

during excavation. Close-up photographs of the event layer show a wide variety of particle compositions (Fig. 6). For example, event layer S1 includes various shapes and

D

sizes of particles, while event layers S9 and S11 are primarily composed of well-sorted

PT E

very fine sand.

The top event layer covering the ground surface (event layer S1: core depth, 0–4 cm) is

CE

normally graded and primarily composed of undecomposed plants and coarse gravelly

AC

sand, including round pebbles of chert, shale, and sandstone, angular rhyolite rocks, and sub-angular granite particles. Event layer S2 (14–16 cm depth) is primarily composed of silty sand rich in tephra, charcoal, and mica fragments. Event layer S3 (30–32 cm depth) is composed of silt to medium sand including tephra, charred material, and mica fragments rich in sub-angular particles, and has a characteristic yellowish-brown color. Event layer S4 (46–51 cm depth) is primarily composed of coarse sand rich in granules and the median grain size shows a normal grading structure.

15

ACCEPTED MANUSCRIPT

Event layer S5 (70–73 cm depth) is composed of very coarse sand rich in round gravel and mud clasts and the median grain size shows an inverse grading structure. Event layer S7 (104–128 cm depth) consists of medium to very coarse sand with mud clasts and a thin

PT

mud layer, and has normal and inverse grading structures and parallel laminations. Event layer S8 (153–170 cm depth) is dominated by very coarse sand containing round and

RI

angular gravels, and has normal and inverse grading structures. A mud layer is present in

SC

the middle of the event layer. Event layer S9 (211–224 cm depth) consists of well-sorted fine sand rich in rounded minerals, mica, and plants, and has parallel lamination. This

NU

layer is not graded and has no sharp erosional contact with the underlying layer. Event

MA

layer S10 (266–267 cm depth) is a very coarse sand layer containing rounded particles of quartz, chert, shale, sandstone, and chlorite, as well as angular granite. Layer S11 (288–

D

328 cm depth) is composed of very fine sand and has a laminated structure rich in mica.

PT E

This layer has no erosional contact with the underlying peaty soil and is not graded. Layer S12 (388–428 cm depth) is primarily composed of medium sand to granules, contains round gravel and pebbles, angular gravels, mud clasts, and a thin mud layer, and

CE

has inverse and normal grading structures. This layer was divided into three units

AC

according to the sedimentary structure (Fig. 7). The lowermost unit (Unit 1) is composed primarily of gravels of various shapes (diameter: −4Φ to −2Φ) and has an erosional contact with the underlying layer. The intermediate unit (Unit 2) contains gravels of similar diameter to those in Unit 1 and is normally graded. The uppermost unit (Unit 3) is composed primarily of coarse to very coarse sand, containing round gravels. Event layers S13 (432–435 cm depth), S14 (456–459 cm depth), and S15 (470–477 cm depth) are composed of very fine sand. Layer S16 (488–500 cm depth) is composed of

16

ACCEPTED MANUSCRIPT

very fine sand abundant in mica and has parallel laminations of sand and mud. Layer S17 (518–540 cm depth) is composed of medium sand rich in plant material. In this layer, cross-laminations of mud and sand are present in the upper unit, whereas

PT

parallel-laminations occur in the lower unit; there is a muddy layer rich in plant

RI

fragments between the units.

SC

5.2. Event layers in other cores

After identifying the event layers in core C2, we correlated each event layer across the

NU

other sedimentary samples (cores C1, C3, C4 and L1–L15) along the survey transect based on the facies, particle composition, and individual characteristics of each event

MA

layer (Fig. 8). We refer to our previous study (Goto et al., 2015) for the correlation of upper event layers: S1 to S9, using 15 Geoslicer samples (L1 to L15). We also confirm

D

that several of the upper sand layers (S1, S4, S5, S7, and S8) are continuously distributed

PT E

from the outcrop observed near the river channel. In core C1 (length: 3.9 m), 11 of the event layers were identified (Fig. 4a). The

CE

correlation of event layers was straightforward due to the proximity of cores C1 and C2.

AC

The top event layer (0–10 cm depth) correlates with event layer S1 and the top sand layer of the other 17 sedimentary samples. The second event layer (16–26 cm depth) is composed of silty sand, has an inverse grading structure, and correlates with event layer S2. The third event layer (30–34 cm depth) is a yellowish-brown layer and correlates with event layer S3. The fourth event layer (48–55 cm depth) consists of rounded sand and correlates with event layer S4. The fifth event layer (85–95 cm depth) is composed of fine to coarse sand and correlates with event layer S5. The sixth event layer (99–102 cm

17

ACCEPTED MANUSCRIPT

depth) is primarily composed of angular clastic particles, has normal and inverse grading structures, and correlates with event layer S7. The seventh event layer (132–150 cm depth) is composed of very coarse sand containing rounded particles and correlates with

PT

event layer S8. The eighth event layer (207–220 cm depth) is composed of well-sorted fine sand and correlates with event layer S9. The ninth event layer (230–231 cm depth) is

RI

a sand layer composed of rounded pebbles and correlates with event layer S10. The tenth

SC

event layer (254–276 cm depth) is composed of very fine sand and correlates with event layer S11. The eleventh event layer (339–390 cm depth) is composed of medium sand to

NU

granules containing rounded gravel and pebbles and very angular gravels and correlates

MA

with event layer S12. All event layers in this core, except for S9 and S11, show erosional contacts with the underlying peaty soil.

D

In core C3 (length, 1.4 m), five of the event layers were identified (Fig. 4c). The top

PT E

event layer (0–14 cm depth) correlates with the top sand layer of all the other sedimentary samples. The second event layer (28–34 cm depth) is composed of round pebbles and correlates with event layer S4. The third event layer (44–62 cm depth) is

CE

composed of very coarse sand abundant in round gravels, has laminated structures, and

AC

correlates with event layer S5. The fourth event layer (82–120 cm depth) contains rounded particles and correlates with event layer S8. The fifth event layer (128–140 cm depth) contains many angular clastic particles and correlates with event layer S12. In this core, event layers S1, S5, and S8 all had sharp basal contacts. In core C4 (length: 1.7 m), eight of the event layers were identified (Fig. 4d). The top event layer (0–20 cm depth) correlates with the top sand layer of all the other sedimentary samples. The second event layer (Sf; 32–36 cm depth) is composed of coarse

18

ACCEPTED MANUSCRIPT

sand and riverbed gravels. The gravel composition of this layer does not match that of the neighboring layers (i.e., layers S2 and S3). The third layer (50–52 cm depth) is composed of medium sand, contains both riverbed and slope gravels, and correlates with event layer

PT

S3. Below 32 cm depth, each event layer is distinguishable from the eroded contact by a clear lower layer incorporating autochthonous sands. The lower contacts of the fourth

RI

(55–70 cm depth), fifth (80–82 cm depth), sixth (92–120 cm depth), seventh (126–132

SC

cm depth), and eighth (140–170 cm depth) layers are similarly assumed to have been founded from sandy sediments, and correlate with event layers S4, S5, S7, S8, and S12,

MA

5.3. Classification of event layers

NU

respectively.

Next, we categorized the events layers based on their gravel composition (Fig. 9). Six

D

event layers (S1, S4, S5, S7, S8, and S12) were classified as Type-A, that is, they contain

PT E

all three index gravels (beach, riverbed, and slope) in variable proportions. Two event layers (S2 and S10) were classified as Type-B, containing beach and slope gravels. One

CE

event layer (S3) was classified as Type-C, containing gravels that originated from

AC

riverbed and slope areas. Eight event layers (S6, S9, S11, and S13–S17) contained no index gravels and were classified as Type-D. Of the six Type-A event layers, the uppermost, S1, covers the ground surface and represents deposits from the tsunami caused by the 2011 Tohoku-oki earthquake (Goto et al., 2017). The 2011 tsunami inundated the valley up to 1 km from the coastline (Haraguchi and Iwamatsu, 2011); therefore, these deposits are likely to be identified further inland of site C4 (490 m from the coast, Fig. 8). The remaining five layers have

19

ACCEPTED MANUSCRIPT

similar sedimentary features to those of S1 (i.e., a sharp erosional contact with the underlying layer, a normal/inverse grading structure, and a similar grain composition), although the proportion of each index gravel varies.

PT

Of the Type-B event layers, event layer S2 was traced up to L10 core (220 m from the shoreline) but was not found in several Geoslicer samples (Fig. 8). Event layer S10 was

RI

traced up to core C2 (260 m from the shoreline). Slope gravel is more abundant than

SC

beach gravel in event layer S2, and beach gravel is more abundant than slope gravel in event layer S10 (Fig. 9). Only one Type-C event layer was identified. Event layer S3 can

NU

be traced up to core C2 (260 m from the shoreline), whereas this layer can possibly be

MA

correlated with the third upper layer at core C4 (490 m from the coast). Of the Type-D event layers identified, the lateral correlations of event layers S6, S9, and S11 were less

D

marked than those of the other types of event layers, and these event layers were found

PT E

only in sedimentary samples collected in the lowland marsh area. Old event layers S13– S17 were only found in core C2; hence the lateral extent could not be traced.

CE

6. Sources of event deposits inferred from nanofossils

AC

The change in the abundance of coccoliths with depth in core C2 is shown in Fig. 4b. The abundance of coccoliths is highest at the bottom of the core, suggesting that the study site was once easily affected by seawater inflow, possibly due to a lower land level (higher sea level) or a less-developed beach ridge. This is consistent with the findings of our diatom analyses discussed in section 7. Coccoliths are more abundant in most of the event layers than in the autochthonous peaty soil, which suggests that these layers were formed by seawater influx (tsunami or storm), and therefore the potential

20

ACCEPTED MANUSCRIPT

that the deposits were formed by river flooding is low. The likelihood that these event layers were formed by tsunamis or storms is also supported by the small size of the catchment area (2 km × 1 km) in the study site.

PT

7. Environmental changes inferred from diatoms

RI

Diatom analyses of core C2 revealed four environmental changes across five diatom

SC

zones (Fig. 10). Diatom zone I is above event layer S17 and below S14. Marsh species (e.g., Epithemia adnate) and fresh-water species (e.g., Fragilaria exigua) are dominant

NU

in this zone, suggesting an aquatic sedimentary environment with flourishing reeds. Diatom zone II extends from just above event layer S14 to just below S12.

MA

Fresh-water species such as Fragilaria sp. are dominant, but marine-brackish (e.g., Thalassiosira bramaputrae), brackish (e.g., Navicula sp.), and river species (e.g.,

D

Achnanthes lanceolata) also occur. This diatom assemblage suggests a brackish

PT E

sedimentary environment where the lowland area was easily inundated by seawater, possibly due to a less-developed beach ridge or lower land level (higher sea level).

CE

Diatom zone III is between event layers S12 and S8. River species (e.g., A.

AC

lanceolata) and terrestrial species (e.g., A. minutissima) are dominant, but marine, brackish, and fresh-water species also occur. This suggests a variable sedimentary environment controlled by changes in the river channels. Diatom zone III can be further divided into sub-zones a, b, and c according to the abundances of brackish and fresh-water species. Diatom zone IV is between event layers S8 and S7. Brackish-freshwater species (e.g., N. schroeterii) are dominant and occur with river and terrestrial species. This suggests a

21

ACCEPTED MANUSCRIPT

brackish sedimentary environment with low salinity, where the marsh and river channel are also distributed. Diatom zone V extends from above layer S7 to the surface. Freshwater species,

PT

including river species (e.g., A. lanceolata) and marsh species (e.g., Cocconeis placentula) are dominant and few other species occur. This suggests a marsh

RI

sedimentary environment fed a small river in which reeds were growing.

SC

Marine and brackish species occur more frequently in diatom zones I and II, i.e., below layer S12, suggesting that the sedimentary environment below layer S12 was

NU

easily affected by seawater, possibly due to a lower land level (higher sea level) or a

MA

less-developed beach system. The event forming layer S12 might have considerably modified the morpho-sedimentary system, which is supported by the results of the

D

gravel analysis, i.e., Type-D events are more frequent in the lower part of core C2,

PT E

whereas Type-A events are more frequent in the upper part. Thus, in the following analyses of inundation behavior (section 8) and dating (section 9), we excluded event layers S13–S17 because the estimated environment was aquatic with the study area

AC

CE

being easily affected by the inflow of seawater.

8. Inundation behavior inferred from event layer gravel composition Three index gravels (beach, riverbed, and slope) in the event layers were used to infer the inundation behavior during events in the valley. In our previous paper (Goto et al., 2017), we found that the tsunami deposits from the 2011 Tohoku-oki earthquake (i.e., event layer S1, Type-A) include all three index gravels and that the layer can be traced more than 490 m inland from the coastline. The presence of slope gravel suggests high

22

ACCEPTED MANUSCRIPT

inundation and flow flux, which eroded bare rock and talus 5 m above sea level and transported them to the valley plain. The mixture of beach and riverbed gravels implies the formation of both an upstream current from the beach, which transported beach

PT

pebbles inland, and a strong backwash current, which transported sub-angular riverbed gravels to the valley plain.

RI

The five remaining Type-A event layers (S4, S5, S7, S8, and S12) can also be traced

SC

up to 490 m inland (present elevation, 7 m) from the coastline and were at least several centimeters thick at almost all survey points (Fig. 11). This suggests that the Type-A

NU

event layers probably represent a record of large events. Diatom analyses indicated that

MA

the sedimentary environment in the area changed after the deposition of three Type-A event layers (S7, S8, and S12). For example, terrestrial species (e.g., A. minutissima)

D

occurred above layer S12, whereas some freshwater species (e.g., Tabellaria

PT E

fenestrata-flocculosa) were not identified after the deposition of event layer S8. After the deposition of layer S7, the abundance of fresh water species such as marsh and river species markedly increased. These changes in sedimentary environments can be caused

CE

by crustal deformation (coastal uplift/subsidence) due to large earthquakes, changes in

AC

the river flow, or changes in sedimentation rate. Thus, these event layers are considered to have been formed by large tsunamis with high inundation and a strong backwash current.

We investigated the ratio of the index gravels versus depth in event layer S12 and found that the three index gravels were initially deposited at almost the same depth, but that the deposition of beach and riverbed gravels ceased prior to that of slope gravels (Fig. 7c). We also found that index gravel proportions peaked at different depths in each

23

ACCEPTED MANUSCRIPT

sedimentary unit comprising event layer S12, indicating a strong correlation between sedimentary structure and inundation behavior during each event. The mixture of beach and slope gravels in the two Type-B event layers (S2 and S10)

PT

suggests events of high inundation and flow flux, which transported beach gravel inland and eroded bare rock and talus, primarily via run-up waves, and transported them to the

RI

valley plain via backwash current. Thus, these layers are also considered to have been

SC

formed by tsunamis.

The single Type-C event layer identified, S3, lacks beach gravel but contains rounded

NU

sand with a grain-size distribution similar to the local beach sand. This suggests an

MA

event with a relatively low flow flux that transported only finer particles from the beach to the inland or a more sandy beach/coastal zone at that time. The percentage

D

composition of riverbed gravel in Type-A and Type-C event layers tended to be higher

PT E

in core C2 than in core C4 because site C2 is closer to the current river channel (Fig. 9). Thus, this Type-C event layer is considered to have been formed by either tsunami or storm. The Type-D event layers (S6, S9, and S11), which did not contain any index

CE

gravels (but did contain rounded sand), may indicate smaller-scale flooding than that

AC

indicated by the other event layer types; however, the Type-D event layers are still considered to have been formed by either tsunamis or storms. Based on the distributions of the index gravels, the inundation distances are inferred to be the longest (at least 490 m from the current coastline) for events resulting in Type-A event layers, intermediate (200–300 m) for events resulting in Type-B and Type-C event layers, and the shortest for events resulting in Type-D event layers.

24

ACCEPTED MANUSCRIPT

9. Dating results and correlations between event layers and historical events 9.1 Dating results and sediment accumulation rate A radiocarbon age of 1950 ± 20 cal yr BP was obtained at a depth of 520 cm in core C2 (Table 2), indicating that all but one of the identified event layers were deposited

PT

during the past 2000 years. The age of each event layer (2σ range) estimated from

RI

radiocarbon ages is shown in Fig. 12 and Table 3. We excluded a possible reworked

SC

sample at a depth of 188 cm, which showed a radiocarbon age of 730 ± 20 cal yr BP. Sediment ages below and above event layer S12 showed a marked gap of

NU

approximately 1100 years (Fig. 12b). The erosive effect of water flow has been reported

MA

for the 2011 Tohoku-oki tsunami (Shinozaki et al., 2015), and the observed hiatus may have been caused by similar erosion of surficial strata at the time of event S12. The sediment accumulation rate excluding event layers was calculated as 3.1 and 4.5 mm

PT E

D

yr-1 below and above event layer S12, respectively. This change in sedimentation rate may have been caused by a change in the crustal uplift rate, or may only be an apparent

CE

rate due to erosion or changes in the degree of compaction.

AC

9.2 Correlations between event layers and historical events According to our dating results, the ages of the upper 12 event layers (S1–S12) correlate with historical events (Fig. 12a), whereas the lower five event layers (S13– S17) are attributed to paleo-tsunamis/storms (327 BC–AD 365). Large tsunamis are likely responsible for the six Type-A event layers. The surficial layer (S1) is the tsunami deposit associated with the 2011 Tohoku-oki earthquake (Goto et al., 2017). We previously attributed two of the Type-A event layers (S4 and S5) to the

25

ACCEPTED MANUSCRIPT

1933 and 1896 Sanriku-oki earthquake tsunamis, respectively based on

210

Pb and

137

Cs

dating (Goto et al., 2015). Type-A event layer S7 (AD 1710–1866) can be attributed to any of the 1763 Aomori-oki, 1793 Miyagi-oki, or 1856 Aomori-oki earthquakes. The

PT

remaining two Type-A event layers (S8, AD 1575–1745; S12, AD 535–1379) can be attributed to large historical tsunamis caused by the 1611 Tohoku-oki and 869

RI

Tohoku-oki (Jogan) earthquakes, respectively.

SC

The 1454 Tohoku-oki earthquake can be correlated with either event layer S10 (Type-B; AD 1456–1610) or S11 (Type-D; AD 1451–1576). This implies that, at the

NU

study site, the 1454 Tohoku-oki tsunami may have been smaller than the 1611 tsunami

MA

or that the source region of the 1454 Tohoku-oki earthquake was located (or limited to) south of the Tohoku region. Further surveys along the Tohoku coast are required to

D

clarify the source areas of these earthquakes.

PT E

Among the Type-A events, the sedimentary age of event layer S12 correlates with the 869 Tohoku-oki (Jogan) earthquake, although the estimated age of the event layer has a wide 2σ range (AD 535–1379). The source region of the 869 Tohoku-oki (Jogan)

CE

earthquake has been estimated to be further south than our study area, in the off-Miyagi

AC

and off-Fukushima regions (Namegaya and Satake, 2014). However, recent studies identified possible Jogan tsunami traces in an alluvial plain on the Sanriku coast (Ishimura and Miyauchi, 2015; Inoue et al., 2017). Our results therefore also suggest that the source region of the 869 Tohoku-oki (Jogan) earthquake extends further north than previously estimated, including the off Sanriku region along the Japan Trench. Event layer S2 (Type-B; after AD 1950) was also likely formed by a tsunami and correlates with the 1960 Chilean or the 1968 Tokachi-oki earthquake tsunami. Layer S3

26

ACCEPTED MANUSCRIPT

(Type-C; AD 1938–1994) correlates with the 1960 Chilean and 1968 Tokachi-oki earthquake tsunamis or typhoons in 1947 and 1948. Event layer S6 (Type-D) can be correlated with either a tsunami or a storm between

PT

1748 and 1890 (within the 2σ range), which means that it correlates with any of the tsunamis generated by the 1763 Aomori-oki, 1793 Miyagi-oki, or 1856 Aomori-oki

RI

earthquakes, or a storm in 1856. Event layer S9 (Type-D) was formed by either a

SC

tsunami or a storm between 1478 and 1649. Event layer S10 (Type-B) may have been

NU

deposited by a tsunami or storm between 1456 and 1610.

10. Conclusions

MA

In the present study, we identified 17 sand layers with sedimentary structures characteristic of event deposits in the peaty soil in a marsh on the rocky Sanriku coast.

D

Among these layers, nine event layers also included gravels. To infer the inundation

PT E

behavior and overcome the difficulty in distinguishing tsunami and storm deposits, we analyzed gravel composition within each event layer following our previous method

CE

(Goto et al., 2017), grouped them into four types (Type-A through D) according to the type of gravel presence and composition of the event deposit, and reconstructed the

AC

event layer history over the past 2000 years using dating results. Six Type-A event layers contained gravels originated from local beach, river channel, and slope areas. These event layers were traced up to the most inland sampling site (490 m inland from the current coastline at present elevation of 7 m). Changes in the sedimentary environment after deposition of three of the Type-A event layers was inferred from fossil diatom analyses. The presence of slope gravel suggests high inundation and flow flux, which eroded bare rock and talus 5 m above sea level and

27

ACCEPTED MANUSCRIPT

transported them to the valley plain. The mixture of beach and riverbed gravels implies the formation of both an upstream current from the beach, which transported beach pebbles inland, and a strong backwash current, which transported sub-angular riverbed

PT

gravels to the valley plain. We concluded that these event layers were formed by tsunamis with high inundation and flow flux, and a strong backwash current. According

RI

to dating results, they are correlated with tsunamis associated with the 2011 Tohoku-oki

SC

earthquake; the 1933 and 1896 Sanriku-oki earthquakes; any of the 1763 Aomori-oki, 1793 Miyagi-oki, or 1856 Aomori-oki earthquakes; the 1611 Tohoku-oki earthquake;

NU

and the 869 Tohoku-oki (Jogan) earthquake. Although the source region of the 869

MA

Tohoku-oki (Jogan) earthquake was previously estimated to be off Fukushima and Miyagi prefectures, the geological evidence presented herein with previous studies

D

(Ishimura and Miyauchi, 2015; Inoue et al, 2017) suggests that the source region may

PT E

extend further north along the Japan Trench. Two historical earthquakes, the 1611 and 1454 Tohoku-oki earthquakes, have been previously proposed as the predecessor of the 2011 Tohoku-oki earthquake. Our dating

CE

result for a Type-A event layer (AD 1575–1745) favors the 1611 Tohoku-oki earthquake,

AC

rather than the 1454 Tohoku-oki earthquake, as the predecessor of the 2011 Tohoku-oki earthquake. Our results also imply that the 1454 Tohoku-oki tsunami may have been smaller than the 1611 tsunami at the study site, or that the source region of the 1454 Tohoku-oki earthquake was located (or limited to) south of the Tohoku region. Sedimentary features of tsunami deposits are sometimes similar to those of storm and/or flood deposits. However, the gravel composition of tsunami deposits is strongly affected by both autochthonous deposits as a supply source and inundation behavior.

28

ACCEPTED MANUSCRIPT

Here, we showed that the use of gravel deposit in the event layers with comparison of autochthonous deposits can infer the inundation behavior and event origin. The methodology is also applicable for other study sites, where characteristic autochthonous

PT

deposits can be studied. Further surveys along the rocky Sanriku coast would provide new insights for more reliable reconstructions of the local tsunami history and the

AC

CE

PT E

D

MA

NU

SC

RI

source areas of these earthquakes.

29

ACCEPTED MANUSCRIPT

Acknowledgments We would also like to thank Kazuomi Hirakawa for his advice and discussions at the survey site. We also thank Jun Muragishi, Satoko Murotani, and Satoshi Kusumoto for their assistance with the field survey. We used Generic Mapping Tools (Wessel and

PT

Smith, 1998), Google Maps, and a Digital Elevation Map provided by the Geospatial

RI

Information Authority of Japan to create the figures. We would like to thank the

SC

editor-in-chief, Catherine Chagué and Brian Jones for editing, and Raphaël Paris and an anonymous reviewer for providing valuable comments and suggestions that were

NU

helpful for improving the manuscript.

MA

Funding: This work was supported by the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan through KAKENHI (Grants 24241060 and

PT E

Mega-earthquake Disasters.

D

26750129) and the Special Project for Reducing Vulnerability for Urban

CE

Declaration of interest

AC

The authors declare that they have no competing interests.

30

ACCEPTED MANUSCRIPT

References Abe, H., Sugeno, Y., Chigama, A., 1990. Estimation of the height of the Sanriku Jogan 11 Earthquake-tsunami (A.D. 869) in the Sendai Plain. Journal of the Seismological Society of Japan (Zisin Ser. 2) 43, 513–525 (in Japanese with

PT

English abstract).

RI

Abe, T., Goto, K., Sugawara, D., 2012. Relationship between the maximum extent of

SC

tsunami sand and the inundation limit of the 2011 Tohoku-oki tsunami on the Sendai Plain, Japan. Sedimentary Geology 282, 142-150.

NU

Ando, K., 1990. Environmental indicators based on freshwater diatom assemblages and its application to reconstruction of paleo-environments. Annuals of the Tohoku

MA

Geographical Association 42(2), 73-88 (in Japanese with English abstract).

272 pp. (in Japanese).

D

Arakawa, Y., Ishida, Y., Ito, T., 1961. Nihon Takashio Shiryo. Yoshikawa-kohbunkan,

PT E

Atwater, B.F., Musumi-Rokkaku, S., Satake, K., Tsuji, Y., Ueda, K., Yamaguchi, D.K., 2005. The orphan tsunami of 1700 — Japanese clues to a parent earthquake in

CE

North America. U.S. Geological Survey 133 pp..

AC

Ben-Menahem, A., Rosenman, M., 1972. Amplitude patterns of tsunami waves from submarine earthquakes. Journal of Geophysical Research 77(17), 3097-3128. Bronk Ramsey, C., 2008. Deposition models for chronological records. Quaternary Science Reviews 27, 42-60. Bronk Ramsey, C., 2009. Bayesian analysis of radiocarbon dates. Radiocarbon 51(1), 337-360.

31

ACCEPTED MANUSCRIPT

Bronk Ramsey, C., Lee, S., 2013. Recent and planned developments of the program OxCal. Radiocarbon 55(2), 720-730. Chagué-Goff, C., Andrew, A., Szczuciński, W., Goff, J., Nishimura, Y., 2012.

PT

Geochemical signatures up to the maximum inundation of the 2011 Tohoku-oki tsunami—implications for the 869 AD Jogan and other palaeotsunamis.

RI

Sedimentary Geology 282, 65–77.

SC

Committee for Field Investigation of the Chilean Tsunami of 1960, 1961. Report on the Chilean tsunami of May 24, 1960, as observed along the coast of Japan. Tokyo,

NU

Maruzen Co., 397 pp..

MA

Earthquake Research Institute, 1934. Field survey report of damage caused by the 1933 Sanriku earthquake, Bulletin of the Earthquake Research Institute, University of

D

Tokyo (supplementary volume), -Papers and Reports on the Tsunami of 1933 on

PT E

the Sanriku Coast, Japan, 1, 9–139 (in Japanese). Ebina, Y., 2014. The 1611 Keicho Oshu earthquake tsunami and reconstruction – A large earthquake and devastating tsunami also occurred 400 years ago – , 69 pp.,

CE

Banzanbou (in Japanese).

AC

Fujiwara, O., Tanigawa, K., 2014. Bedforms record the flow conditions of the 2011 Tohoku-Oki tsunami on the Sendai Plain, northeast Japan. Marine Geology 358, 79-88.

Goff, J., Chagué-Goff, C., Nichol, S., Jaffe, B., Dominey-Howes, D., 2012. Progress in palaeotsunami research. Sedimentary Geology 243-244, 70-88.

32

ACCEPTED MANUSCRIPT

Goff, J., McFadgen, B.G., Chagué-Goff, C., 2004. Sedimentary differences between the 2002 Easter storm and the 15th-century Okoropunga tsunami, southeastern North Island, New Zealand. Marine Geology 204, 235–250.

PT

Goto, K., Sugawara, D., Ikema, S., Miyagi, T., 2012. Sedimentary processes associated with sand and boulder deposits formed by the 2011 Tohoku-oki tsunami at

RI

Sabusawa Island, Japan. Sedimentary Geology 282, 188-198.

SC

Goto, K., Hashimoto, K., Sugawara, D., Yanagisawa, H., Abe, T., 2014. Spatial thickness variability of the 2011 Tohoku-oki tsunami deposits along the coastline

NU

of Sendai Bay. Marine Geology 358, 38-48.

MA

Goto, T., Satake, K., Sugai, T., Ishibe, T., Harada, T., Murotani, S., 2015. Historical tsunami and storm deposits during the last five centuries on the Sanriku coast,

D

Japan. Marine Geology 367, 105-117.

PT E

Goto, T., Satake, K., Sugai, T., Ishibe, T., Harada, T., Gusman, A.R., 2017. Effects of topography on particle composition of 2011 tsunami deposits on the ria-type Sanriku coast, Japan. Quaternary International 456, 17-27.

CE

Gusman, A.R., Goto, T., Satake, K., Takahashi, T., Ishibe, T., 2018. Sediment transport

AC

modeling of multiple grain sizes for the 2011 Tohoku tsunami on a steep coastal valley of Numanohama, northeast Japan. Marine Geology 405, 77-91. Haraguchi, T., Ishibe, T., 2009. Giant earthquake model in the central Sanriku-Oki region estimated from tsunami deposits and event layers accompanying coastal uplift. Monthly Chikyu 31, 223-230 (in Japanese).

33

ACCEPTED MANUSCRIPT

Haraguchi, T., Iwamatsu, A., 2011. Detailed map of the impacts of the 2011 Japan tsunami Vol.1: Aomori, Iwate and Miyagi prefectures. Kokon-shoin Publishers Ltd., Tokyo, 167pp..

PT

Hatori, T., 1987. Distributions of seismic intensity and tsunami of the 1793 Miyagi-Oki earthquake, Northeastern Japan. Bulletin of the Earthquake Research Institute,

RI

University of Tokyo 62, 297-309 (in Japanese with English abstract).

SC

Imaizumi, T., Ishiyama, T., Haraguchi, T., Miyauchi T., Goto, H., Shimazaki, K., 2009. 3.3 Research on earthquake histories based on tsunami deposit survey,

NU

Summarized report of the comprehensive research project for the Miyagi-Oki

MA

earthquake during FY2005-FY2009, The Headquarters for Earthquake Research Promotion, Japan, http://jishin.go.jp/main/chousakenkyuu/miyagi_juten/h17_21/

D

h17-21_3_3.pdf (in Japanese).

PT E

Imamura, A., 1934. Past tunamis of the Sanriku coast. Japanese Journal of Astronomy and Geophysics 11, 79–93.

Inoue, T., Goto, K., Nishimura, Y., Watanabe, M., Iijima, Y., Sugawara, D., 2017.

CE

Paleo-tsunami history along the northern Japan Trench: evidence from Noda

AC

Village, northern Sanriku coast, Japan. Progress in Earth and Planetary Science 4,42.

Ioki, K., Tanioka, Y., 2016. Re-estimated fault model of the 17th century great earthquake off Hokkaido using tsunami deposit data. Earth and Planetary Science Letters 433, 133-138. Ishimura, D., 2017. Re-examination of the age of historical and paleo-tsunami deposits at Koyadori on the Sanriku Coast, Northeast Japan. Geoscience Letters 4,11.

34

ACCEPTED MANUSCRIPT

Ishimura, D., Miyauchi, T., 2015. Historical and paleo-tsunami deposits during the last 4000 years and their correlations with historical tsunami events in Koyadori on the Sanriku Coast, northeastern Japan. Progress in Earth and Planetary Science

PT

2,16 doi:10.1186/s40645-015-0047-4. Ishimura, D., Miyauchi, T., 2017. Holocene environmental changes and paleo-tsunami

RI

history in Onuma on the southern part of the Sanriku Coast, northeast Japan.

SC

Marine Geology 386, 126-139.

Iwabuchi, Y., Sugino, H., Imamura, F., Tsuji, Y., Matsuoka, Y., Imai, K., Shuto, N., 2012.

NU

Development of tsunami trace database with reliability evaluation on Japan coasts.

MA

Journal of Japan Society of Civil Engineers, Ser. B2 (Coastal Engineering) 68, I_1326–I_1330 (in Japanese with English abstract).

D

Jagodziński, R., Sternal, B., Szczuciński, W., Chagué-Goff, C., Sugawara, D., 2012.

PT E

Heavy minerals in the 2011 Tohoku-oki tsunami deposits—insights into sediment sources and hydrodynamics. Sedimentary Geology 282, 57-64. Kajiura, K., 1970. Tsunami source, energy and the directivity of wave radiation,

CE

Bulletin of the Earthquake Research Institute, University of Tokyo 48, 835-869.

AC

Kanamori, H., 1971. Seismological evidence for a lithospheric normal faulting – the Sanriku earthquake of 1933. Physics of the Earth and Planetary Interiors 4, 289-300.

Koike, K., Tamura, T., Chinzei, K., Miyagi, T. (Eds.), 2005. Regional Geomorphology of the Japanese Islands. Geomorphology of Tohoku Region, vol. 3. University of Tokyo Press, Tokyo, 355 pp. (in Japanese).

35

ACCEPTED MANUSCRIPT

Kosugi, M., 1986. Paleoecological analysis based on terrestrial diatoms, and its implications -Introduction to Japan and its prospects-. Japanese Journal of Historical Botany 1, 29-44 (in Japanese).

PT

Kosugi, M., 1988. Classification of living diatom assemblages as the indicator of environments, and its application to reconstruction of paleoenvironments. The

RI

Quaternary Research 27, 1-20 (in Japanese with English abstract).

SC

Krumbein, W.C., 1941. Measurement and geological significance of shape and roundness of sedimentary particles. Journal of Sedimentary Research 11(2),

NU

64-72.

MA

Kusumoto, S., Goto, T., Sugai, T., Omori, T., Satake, K., 2018. Geological evidence of tsunamis in the past 3800 years at a coastal lowland in the central Fukushima

D

Prefecture, Japan. Marine Geology 404, 137–146.

PT E

Machida, H., Arai, F., 2003. Atlas of Tephra in and Around Japan, rev. ed. (in Japanese). University of Tokyo Press, Tokyo, 360pp.. Matsuoka, H., Okada, H., 1989. Quantitative analysis of quaternary nannoplankton in

AC

118.

CE

the subtropical northwestern Pacific ocean. Marine Micropaleontology 14, 97–

Minoura, K., Nakaya, S., 1991. Traces of tsunami preserved in inter-tidal lacustrine and marsh deposits: some examples from northeast Japan. The Journal of Geology 99, 265–287. Minoura, K., Imamura, F., Sugawara, D., Kono, Y., Iwashita, T., 2001. The 869 Jogan tsunami deposit and recurrence interval of large-scale tsunami on the Pacific coast of northeast Japan. Journal of Natural Disaster Science 23, 83–88.

36

ACCEPTED MANUSCRIPT

Miyako City Education Committee, Iwate prefecture, 1991. History of Miyako City (A Chronological Table), 615 pp. (in Japanese). Moore, A., Nishimura, Y., Gelfenbaum, G., Kamataki, T., Triyono, R., 2006.

Aceh, Indonesia. Earth Planets and Space 58, 253-258.

PT

Sedimentary deposits of the 26 December 2004 tsunami on the northwest coast of

RI

Mori, N, Takahashi, T., The 2011 Tohoku Earthquake Tsunami Joint Survey Group,

SC

2012, Nationwide post event survey and analysis of the 2011 Tohoku earthquake tsunami. Coastal Engineering Journal 54(1), 1250001.

NU

Morton, R.A., Gelfenbaum, G., Jaffe, B.E., 2007. Physical criteria for distinguishing

MA

sandy tsunami and storm deposits using modern examples. Sedimentary Geology 200, 184–207.

D

Namegaya, Y., Satake, K., 2014. Reexamination of the A.D. 869 Jogan earthquake size

PT E

from tsunami deposit distribution, simulated flow depth, and velocity. Geophysical Research Letters 41, 2297–2303. Namegaya, Y., Yata, T., 2014. Tsunami which affected the Pacific coast of eastern Japan

CE

in medieval times inferred from historical documents. Journal of the

AC

Seismological Society of Japan (Zisin Ser. 2) 66, 73–81 (in Japanese with English abstract).

Nanayama, F., Shigeno, K., Satake, K., Shimokawa, S., Koitabashi, S., Miyasaka, S., Ishii, M., 2000. Sedimentary differences between the 1993 Hokkaido-nansei-oki tsunami and the 1959 Miyakojima typhoon at Taisei, southwestern Hokkaido, northern Japan. Sedimentary Geology 135, 255–264.

37

ACCEPTED MANUSCRIPT

Nanayama, F., Satake, K., Furukawa, R., Shimokawa, K., Atwater, B.F., Shigeno, K., Yamaki, S., 2003. Unusually large earthquakes inferred from tsunami deposits along the Kuril trench. Nature 424, 660-663.

PT

Ogami, T., Sugai, T., 2018. Effects of longitudinal valley slopes on runup of the 2011 Tohoku tsunami on the Sanriku coast, northeastern Japan. Quaternary

RI

International 471, 253-266.

SC

Okamura, M., Yamauchi, M., 1984. Detailed Survey on Nannofossils at Kansai International Airport in Osaka Bay, Central Japan. Research Institute of Disaster

NU

Science Report (in Japanese).

MA

Pilarczyk, J.E., Horton, B.P., Witter, R.C., Vane, C.H., Chagué-Goff, C., Goff, J. 2012. Sedimentary and foraminiferal evidence of the 2011 Tōhoku-oki tsunami on the Sendai coastal plain, Japan. Sedimentary Geology 282, 78-89.

PT E

D

Piotrowski, A., Szczuciński, W., Sydor, P., Kotrys, B., Rzodkiewicz, M., Krzymińska, J., 2017. Sedimentary evidence of extreme storm surge or tsunami events in the

973-986.

CE

southern Baltic Sea (Rogowo area, NW Poland). Geological Quarterly 61(4),

AC

Reimer, P.J., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Bronk Ramsey, C., Buck, C.E., Cheng, H., Edwards, R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P., Haflidason, H., Hajdas, I., Hatté, C., Heaton, T.J., Hoffmann, D.L., Hogg, A.G., Hughen, K.A., Kaiser, K.F., Kromer, B., Manning, S.W., Niu, M., Reimer, R.W., Richards, D.A., Scott, E.M., Southon, J.R., Staff, R.A., Turney, C.S.M., van der Plicht, J., 2013. IntCal13 and Marine13 radiocarbon age calibration curves 0-50,000 years cal BP. Radiocarbon 55(4), 1869-1887.

38

ACCEPTED MANUSCRIPT

Satake, K., Shimazaki, K., Tsuji, Y., Ueda, K., 1996. Time and size of a giant earthquake in Cascadia inferred from Japanese tsunami records of January 1700. Nature 379(6562), 246-249.

PT

Satake, K, Fujii, Y, Harada, T, Namegaya, Y., 2013. Time and space distribution of coseismic slip of the 2011 Tohoku earthquake as inferred from tsunami waveform

RI

data. Bulletin of the Seismological Society of America 103(2B), 1473–1492.

SC

Satake, K., 2015. Geological and historical evidence of irregular recurrent earthquakes in Japan. Philosophical Transactions of The Royal Society A: Mathematical, and

Engineering

373(2053),

20140375,

MA

doi:10.1098/rsta.2014.0375

Sciences

NU

Physical

Satake, K., Fujii, Y., Yamaki, S., 2017. Different depths of near-trench slips of the 1896

D

Sanriku and 2011 Tohoku earthquakes. Geoscience Letters 4,33.

PT E

Sawai, Y., Fujii, Y., Fujiwara, O., Kamataki, T., Komatsubara, J., Okamura, Y., Satake, K., Shishikura, M., 2008. Marine incursions of the past 1500 years and evidence of tsunamis at Suijin-numa, a coastal lake facing the Japan Trench. The Holocene

CE

18(4), 517-528.

AC

Sawai, Y., Kamataki, T., Shishikura, M., Nasu, H., Okamura, Y., Satake, K., Thomson, K.H., Matsumoto, D., Fujii, Y., Komatsubara, J., Aung, T.T., 2009. Aperiodic recurrence of geologically recorded tsunamis during the past 5500 years in eastern Hokkaido, Japan. Journal of Geophysical Research: Solid Earth 114, B01319, doi:10.1029/2007JB005503.

39

ACCEPTED MANUSCRIPT

Sawai, Y., Namegaya, Y., Okamura, Y., Satake, K., Shishikura, M., 2012. Challenges of anticipating the 2011 Tohoku earthquake and tsunami using coastal geology. Geophysical Research Letters 39, L21309, doi:10.1029/2012GL053692.

PT

Sawai, Y., Namegaya, Y., Tamura, T., Nakashima, R., Tanigawa, K., 2015. Shorter intervals between great earthquakes near Sendai: Scour ponds and a sand layer

RI

attributable to A.D. 1454 overwash. Geophysical Research Letters 42, 4795-4800.

SC

Sella, G.F., Dixon, T.H., Mao A., 2002. REVEL: A model for recent plate velocities

2081, doi:10.1029/2000JB000033.

NU

from space geodesy. Journal of Geophysical Research (Solid Earth) 107(B4),

MA

Shinozaki, T., Fujino, S., Ikehara, M., Sawai, Y., Tamura, T., Goto, K., Sugawara, D., Abe, T., 2015. Marine biomarkers deposited on coastal land by the 2011

D

Tohoku-oki tsunami. Natural Hazards 77(1), 445-460.

PT E

Soria, J.L.A., Switzer, A.D., Pilarczyk, J.E., Siringan, F.P., Khan, N.S., Fritz, H.M., 2017. Typhoon Haiyan overwash sediments from Leyte Gulf coastlines show local spatial variations with hybrid storm and tsunami signatures. Sedimentary Geology

CE

358, 121-138.

AC

Srinivasalu, S., Thangadurai, N., Switzer, A.D., Mohan, V.R., Ayyamperumal, T., 2007. Erosion and sedimentation in Kalpakkam (N Tamil Nadu, India) from the 26th December 2004 tsunami. Marine Geology 240(1-4), 65-75. Sugawara, D., Minoura, K., Imamura, F., 2001. Sedimentation due to the A.D. 869 Jogan tsunami and its numerical modeling (translated to English from Japanese title). Research Report of Tsunami Engineering 18, 1-10.

40

ACCEPTED MANUSCRIPT

Sugawara, D., Goto, K., Imamura, F., Matsumoto, H., Minoura, K., 2012. Assessing the magnitude of the 869 Jogan tsunami using sedimentary deposits: Prediction and consequence of the 2011 Tohoku-oki tsunami. Sedimentary Geology 282, 14-26.

PT

Switzer, A.D., Srinivasalu, S., Thangadurai, N., Mohan, V.R., 2012. Bedding structures in Indian tsunami deposits that provide clues to the dynamics of tsunami

RI

inundation. Geological Society, London, Special Publications 361, 61-77.

SC

Szczuciński, W., Rachlewicz, G., Chaimanee, N., Saisuttichai, D., Tepsuwan, T., Lorenc, S., 2012a. 26 December 2004 tsunami deposits left in areas of various tsunami run

NU

up in coastal zone of Thailand. Earth, Planets and Space, 64,5.

MA

Szczuciński, W., Kokociński, M., Rzeszewski, M., Chagué-Goff, C., Cachão, M., Goto, K., Sugawara, D., 2012b. Sediment sources and sedimentation processes of 2011

D

Tohoku-oki tsunami deposits on the Sendai Plain, Japan—Insights from diatoms,

PT E

nannoliths and grain size distribution. Sedimentary Geology 282, 40-56. Takada, K., Shishikura, M., Imai, K., Ebina, Y., Goto, K., Koshiya, S., Yamamoto, H., Igarashi, A., Ichihara, T., Kinoshita, H., Ikeda, T., River Division Department of

CE

prefectural Land Development, Iwate Prefecture Government, 2016. Distribution

AC

and ages of tsunami deposits along the Pacific Coast of the Iwate Prefecture. Annual Report on Active Fault and Paleoearthquake Researches 16, 1-52 (in Japanese).

Tanigawa, K., Sawai, Y., Shishikura, M., Fujiwara, O., Namegaya, Y., 2014. Event deposits on the Pacific coast of Misawa, Aomori Prefecture, northern Japan. The Quaternary Research 53(1), 55-62 (in Japanese with English abstract).

41

ACCEPTED MANUSCRIPT

Tsuji, Y., Ohtoshi, K., Nakano, S., Nishimura, Y., Fujima, K., Imamura, F., Kakinuma, T., Nakamura, Y., Imai, K., Goto, K., Namegaya, Y., Suzuki, S., Shiroshita, H., Matsuzaki, Y., 2010. Field investigation on the 2010 Chilean Earthquake Tsunami

PT

along the comprehensive coastal region in Japan. Journal of Japan Society of Civil Engineers, Ser. B2 (Coastal Engineering) 66(1), 1346–1350 (in Japanese with

RI

English abstract).

SC

Tsuji, Y., Satake, K., Ishibe, T., Kusumoto, S., Harada, T., Nishiyama, A., Kim, Y.H., Ueno, T., Murotani, S., Oki, S., Sugimoto, M., Tomari, J., Heidarzadeh, M.,

NU

Watada, S., Imai, K., Choi, B.H., Yoon, S.B., Bae, J.S., Kim, K.O., Kim, H.W.,

MA

2011. Field Surveys of tsunami heights from the 2011 off the Pacific Coast of Tohoku, Japan Earthquake. Bulletin of the Earthquake Research Institute

D

University of Tokyo 86, 29-279 (in Japanese with English abstract).

PT E

Tsuji, Y., Satake, K., Ishibe, T., Harada, T., Nishiyama, A., Kusumoto, S., 2014. Tsunami heights along the Pacific Coast of Northern Honshu recoded from the 2011 Tohoku and previous great earthquakes. Pure and Applied Geophysics 171,

CE

3183-3215.

AC

Tuttle, M.P., Ruffman, A., Anderson, T., Jeter, H., 2004. Distinguishing tsunami from storm deposits in eastern North America: the 1929 Grand Banks tsunami versus the 1991 Halloween storm. Seismological Research Letters 75(1), 117–131. Unohana, M., Ota, T., 1988. Disaster records of Meiji Sanriku tsunami by Soshin Yamana. Research Report of Tsunami Engineering 5, 57–379 (in Japanese).

42

ACCEPTED MANUSCRIPT

Usami, T., Ishii, H., Imamura, T., Takemura, M., Matsu’ura, R.S., 2013. Materials for comprehensive list of destructive earthquakes in Japan, 599-2012, University of Tokyo Press, 694 pp. (in Japanese).

PT

Watanabe, M., Bricker, J.D., Goto, K., Imamura, F., 2017. Factors responsible for the limited inland extent of sand deposits on Leyte Island during 2013 Typhoon

RI

Haiyan. Journal of Geophysical Research (Oceans) 122(4), 2795–2812.

SC

Wessel, P., Smith, W.H.F., 1998. New, improved version of generic mapping tools released. EOS Transactions, American Geophysical Union, 79, 579.

NU

Yagishita, K., 2001. Transportation and sedimentation of beach and sea-floor gravels

MA

caused by tsunami: An example at Attari coast, Iwate Prefecture. Journal of Geography (Chigaku Zasshi) 110(5), 689-697 (in Japanese).

D

Yamana, S., 1896. Disaster records of Meiji Sanriku tsunami, reproduced by Unohana

AC

CE

PT E

and Ota (1988). Research Report of Tsunami Engineering 5, 57–379 (in Japanese).

43

ACCEPTED MANUSCRIPT

Figure captions Fig. 1. Space–time diagram of tsunami deposits along the Japan Trench encompassing the past 2800 years. Survey sites of tsunami deposits along the Pacific coast of Japan from Iwate to Fukushima from this and previous studies are shown. The black

PT

triangle and grey dashed line indicate the location of Mt. Towada and the limit of the

RI

fallen area of the To-a tephra (Machida and Arai, 2003), respectively. The estimated

SC

source regions (shown as boxes in the left panel) of the 869 Tohoku-oki (Jogan) earthquake (Namegaya and Satake, 2014); the 1611 Tohoku-oki earthquake (Hatori,

NU

1987); the 1677, 1763, 1856, and 1968 Aomori-oki earthquakes (Hatori, 1987); the 1793 Miyagi-oki earthquake (Hatori, 1987); the 1896 (Satake et al., 2017; shown in

MA

grey) and 1933 Sanriku-oki (Kanamori, 1971) earthquakes; and the 2011 Tohoku-oki earthquake (Satake et al., 2013) are also shown. The contour interval of the 2011

PT E

Tohoku-oki earthquake.

D

Tohoku-oki earthquake slip is 4 m. The grey star indicates the epicenter of the 2011

CE

Fig. 2. (a, b) Location and terrain map of the study area. The survey transect is indicated

AC

by the black line. White circles indicate sampling points along the transect. (c, d) Satellite images of the areas within the squares shown in (b). C1–C4 are the locations where machine coring was conducted in the present study. L1–L15 are the Handy Geoslicer sampling points previously reported (Goto et al., 2015).

Fig. 3. Representative photographs and silhouettes of the three ((a) beach, (b) riverbed, (c) slope) index gravels (Goto et al., 2017).

44

ACCEPTED MANUSCRIPT

Fig. 4. Photographs, facies descriptions, and magnetic susceptibility for the four cores obtained in the present study. (a) C1 (sampled 260 m inland from the coastline), (b) C2 (262 m), (c) C3 (310 m), and (d) C4 (490 m). Black circles indicate the sampling points of charred material, wood fragments, and plant material together with their

RI

PT

conventional 14C age. Coccolith abundance is shown throughout core C2.

SC

Fig. 5. Close-up photographs, sketches, and median grain size of each event layer in core C2 (unless otherwise indicated). (a) S1: depth, 0–4 cm. (b) S2: 14–16 cm. (c) S3:

NU

30–32 cm. (d) S4: 46–51 cm. (e) S5: 70–73 cm. (f) S6: 140–141 cm in Geoslicer

MA

sample L4 (Goto et al., 2015). (g) S7: 104–128 cm. (h) S8: 153–170 cm. (i) S9: 211– 224 cm. (j) S10: 266–267 cm. (k) S11: 288–328 cm. (l) S12: 388–428 cm. (m) S13:

D

432–437 cm. (n) S14: 456–462 cm. (o) S15: 470–475 cm. (p) S16: 488–496 cm. (q)

PT E

S17: 518–540 cm.

Fig. 6. Representative close-up photographs of particles contained within each event

CE

layer. S1: depth, 0–4 cm; S2: 14–16 cm; S3: 30–32 cm; S4: 46–51 cm; S5: 70–73 cm;

AC

S6: 140–141 cm in Geoslicer sample L4 (Goto et al., 2015); S7: 104–128 cm; S8: 153–170 cm; S9: 211–224 cm; S10: 266–267 cm; S11: 288–328 cm; S12: 388–428 cm; S13: 432–437 cm; S14: 456–462 cm; S15: 470–475 cm; S16: 488–496 cm; and S17: 518–540 cm.

Fig. 7. (a) Close-up photograph of event layer S12 in core C2. (b) Grain size variations and (c) index gravel proportions throughout the layer.

45

ACCEPTED MANUSCRIPT

Fig. 8. Correlation of event layers along the transect. The horizontal scale showing the distance from the shoreline between 250 and 300 m is exaggerated. The distribution

PT

of bed rock was estimated from the sedimentary facies of core C2.

RI

Fig. 9. Interpretation of gravel compositions. (a) Representative triangle diagrams of the

SC

four types of event layers (Types A–D) based on relative index gravel proportions. (b) Triangle diagrams showing the relative percentages of the three index gravels in

MA

proposed for each type of event layer.

NU

event layers in cores C2 and C4. (c) Schematic diagram of the inundation behavior

D

Fig. 10. Results of diatom analyses in core C2. Only diatom species with abundances

PT E

greater than 1% are shown. Diatom environments are shown above the diatom taxa.

Fig. 11. Thickness of each event layer along the transect: Type-A (yellow), Type-B

AC

CE

(blue), Type-C (purple), and Type-D (green).

Fig. 12. (a) 14C dating results and probability distributions of the ages of the event layers estimated using a Bayesian approach. The occurrence of historical large earthquakes that generated large tsunamis along the Tohoku coast and can be correlated with Type-A event layers are indicated by vertical lines. (b) Sedimentation rate estimated from sample depth, excluding event layers.

46

ACCEPTED MANUSCRIPT

Table 1. Historical tsunamis that struck the Sanriku coast.

PT

Table 2. 14C dating results.

Table 3. Occurrence age estimated using a Bayesian approach. The gravel composition

RI

type for each event layer and historical tsunamis that can be correlated with the

AC

CE

PT E

D

MA

NU

SC

Type-A event layers are also shown.

47

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5ak

Figure 5lq

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12