Reconstruction of Tsunami Occurrence on Okushiri Island, Southwestern Hokkaido, Japan


 The eastern margin of the Japan Sea is located along an active convergent boundary between the North American and Eurasian tectonic plates. Okushiri Island, which is situated off the southwest coast of Hokkaido, is located in an active tectonic zone where many active submarine faults are distributed. Studying the records of past tsunamis on Okushiri Island is important for reconstructing the history and frequency of fault activity in this region, as well as the history of tsunamis in the northern part of the eastern margin of the Japan Sea. Five tsunami deposit horizons have been identified previously on Okushiri Island, including that of the 1741 tsunami, which are interbedded in the coastal lowlands and Holocene terraces. However, these known tsunami deposits date back only ~3,000 years. A much longer record of tsunami occurrence is required to consider the frequency of submarine fault activity. In this study, we cored from 7 to 25 m depth in the Wasabiyachi lowland on the southern part of Okushiri Island, where previous studies have confirmed the presence of multiple tsunami deposits on peat layer surfaces. The results indicate that the Wasabiyachi lowland comprises an area that was obstructed by coastal barriers between the lowland and the coast at ~8.5 ka and consists of muddy sediment and peat layers formed in lagoons and floodplains, respectively. In addition, event deposits and 15 tsunami horizons were observed among the turbidites and peat layers, dating back as far as 3,000 years. Combined with previous findings, Okushiri Island has sustained 20 tsunami events between ~7.5 ka and the present. These findings are critical for investigating the activities of submarine faults off the southwestern coast of Hokkaido, as well as for determining tsunami risks along the coast of the Japan Sea between North Tohoku and Hokkaido.


Depositional facies 222
The OKU-1 and OKU-5 cores were divided into eight depositional facies, 223 excluding the uppermost topsoil and the basement bedrock (Fig. 3). In addition, well-224 sorted sand layers, which indicate a depositional process that is distinctly different from 225 that of the regular depositional environment, were recorded separately as event deposits.      a ~1 mm dark gray extremely fine-grained thinly laminated layer was observed, and is 270 interbedded with a thin layer of clay or clayey silt with a thickness of < 5 cm. 271

Facies LG3 272
This facies was observed in the middle upper part of core OKU-3 (10.8-13.8 273 m depth). This facies consists of a clayey silt to clay layer in which a ~1 mm dark gray 274 extremely fine-grained thinly laminated layer was observed.

Facies BA1 276
This facies was observed in the middle upper part of core OKU-1 (6.8-14.4 m 277 depth). This facies consists of a well-sorted dark gray coarse to fine-grained sand layer 278 with cross-bedding. The facies also has horizons that contain large amounts of light brown 279 to gray silty gravel of granule to fine pebble size. This facies did not contain plant material. were observed (Fig. 4). indicate an overall total sulfur content of ~0.5% between the deepest section and 16.5 m, 313 excluding the data at 17.1 m (Fig. 4). The radiocarbon ages of a total of 50 core samples were determined using plant 319 material, peaty sediment, and shell fragments ( Table 1). All of the radiocarbon ages 320 conformed to the stratigraphic succession of each core. The ages of the horizons in each 321 core were estimated based on the interval sedimentation rates and the horizons for which 322 radiocarbon ages were obtained. Sedimentation rates were calculated for the upper and lower ranges of the calibrated ages (2σ). The thicknesses of the event deposits were 324 excluded from the interval sedimentation rate calculations because their deposition 325 occurred rapidly (Fig. 5).

Event deposits 330
The sand layers observed in all cores were interbedded with the peaty layers 331 and muddy sediment, except for the gravel observed at the bottom of the cores (Figs. 3 332 and 4). A distinct boundary was also observed between the upper and lower facies of the 333 sand layers. This indicates that the sand layers were formed by a process different from 334 the typical depositional process in this environment. In this study, we defined an event 335 deposit as sediment with multiple characteristic facies, including basal erosion, abrupt 336 changes in grain size or degree of sorting, contamination with heterogeneous particles 337 such as plant fragments or mud clasts, and containing sedimentary structures such as 338 parallel or cross-lamination (Table 3).
In core OKU-1, which was located closest to the coast, 16 event deposit 340 horizons were observed to a depth of 6.0 m ( Fig. 3; Table 3). The upper three horizons 341 were fine to very fine-grained or fine to medium-grained sand layers with thicknesses of 342 10-40 mm, while the lower horizons were well-sorted coarse to medium-grained sand 343 layers with thicknesses of 100 mm to more than 300 mm. Core OKU-2 had a shallower 344 bedrock depth compared to the other sites, with an alluvium thickness of ~6.2 m. Thirteen 345 event deposit horizons were observed in this core. While these event deposits included 346 some sand layers with thicknesses of 100 mm at horizons < 2.0 m deep, they consisted of 347 well-sorted fine to medium-grained sand layers with thicknesses of ~5-20 mm. Some thin 348 (7-50 mm) layers were observed at horizons deeper than 2.0 m, but these were composed 349 of well-sorted fine to medium-grained sand layers with thicknesses of 150-340 mm (  Table 3). Eight event deposit horizons were observed in core OKU-4. These event 355 deposits were composed of well-sorted medium-grained sand layers with thicknesses of 356 4-210 mm, the horizons observed at depths of 8.0-8.7 m were interbedded with a coarse 357 to medium-grained sand layer with a thickness of 600 mm ( Fig. 3; Table 3). Ten event 358 deposit horizons were observed in core OKU-5, which was the site located furthest inland. 359 While these event deposits were interbedded with well-sorted medium to coarse-grained 360 sand layers with thicknesses of 20-190 mm, the horizons at depths of 8.6-9.3 m were 361 interbedded with a well-sorted medium to coarse-grained sand layer with a thickness of 362 600 mm ( Fig. 3; Table 3). FL2 represent a fluvial depositional environment (Miall, 1992). freshwater swamp environment (Miall, 1992). In both of these facies, it is possible that 383 the fluvial channels were undeveloped swamps, as these facies did not include 384 interbedded sand layers. 385 Facies FL5 consists of silt layers that contain plant material and sandy silt layers, 386 and is interbedded with poorly sorted coarse-grained sand layers at the landward sites. As 387 no bioturbation was observed and the facies contained thin layers of coarse-grained sand, 388 this depositional facies represents a small fluvial channel and floodplain depositional 389 environment (Miall, 1992). 390 The total sulfur content of the turbidites in facies FL2 and FL3 in core OKU-3 391 were < 1 wt% (Fig. 4), indicating a non-reducing depositional environment that was not 392 affected by seawater. The depositional environment determined from the depositional 393 facies is consistent with implications of the total sulfur content. 394 burrows also indicates a closed and reducing environment (Strum, 1979). LG2, which 398 contained a particularly thinly laminated interval, suggests that reducing conditions 399 persisted at the bottom. LG1 also contained silt, but some bioturbation was also observed. 400 This suggests that the water depth decreased and the number of aquatic organisms 401 increased. The succession and sedimentary features of these facies indicate a coastal 402 lagoon depositional environment.
These facies also provide evidence for the 403 development of a coastal sand barrier that separated the Japan Sea (Aonae Bay) from an 404 inland lagoon. 405 Facies BA1 consists of well-sorted coarse to fine-grained sandy layers that 406 contain cross-bedding structures. Well-sorted sand layers indicate a depositional 407 environment that was strongly affected by a wave environment (Reinson, 1984). Beach 408 ridges that include sand dunes and ridges are located between the current Wasabiyachi Thus, the valley formation of the Wasabiyachi lowland occurred near sites OKU-1 420 through OKU-5 (Fig. 3). Fluvial material containing coarse gravel deposits (FL1 and 421 FL2) were deposited ~9.5 ka. As sea level increased since the last glacial period near the 422 Japanese archipelago, sites OKU-1, -3, and -4 changed rapidly to a lagoon environment, OKU-3 (Figs. 3 and 4). This indicates that the lagoon environment was at its most closed 432 at ~7-7.3 ka. Moreover, the change in lagoon facies (from LG1 to LG3) indicates the 433 early stages of lagoon formation, the progression of aggradation, and the differences in 434 lagoon water depths from seaward to landward. 435 Based on the facies observed in core OKU-1, the coastal barrier of the 436 Wasabiyachi lowland (estuary) was estimated to have formed ~8.5 ka. Furthermore, due 437 to the observed barrier deposits at depths of 6.8-14.4 m in core OKU-1, the estuary 438 barriers can be estimated to have developed at ~6-7.5 ka, then moved seaward thereafter 439 (Fig. 3).  Among the event deposits interbedded in the cores, lateral changes in layer thickness 461 and particle size (Md and sorting) are shown in Figure 7 for deposits OW-4, -5, -8, -9, -462 10, -12, -13, and -14, which are comparable in cores from three or more sites. In general, 463 sand layers with a thickness of 10 cm or more had different individual particle size 464 changes (as observed in the Md), but commonly exhibited upward fining. Sand layers that 465 were thick and could be divided into multiple units often exhibited upward fining in each 466 unit. The sorting of the sand layers and units with thicknesses of 10 cm and greater occurred at many sites, with large variations. However, in some cases, the sorting tended 468 to worsen upward, owing to upward fining (Fig. 7). 469 Among the comparable event deposits, OW-5, OW-8, and OW-13 contained 470 sand layers that were thicker at the OKU-1 site near the coast and became thinner toward 471 the OKU-4 and OKU-5 sites on the landward side. Furthermore, the Md of core OKU-1 472 was 300-400 μm, but became 200-250 μm landward, indicating fining. In particular, 473 the Md of OW-13 in core OKU-5 underwent a sudden change in fining. Meanwhile, the 474 sorting of these event sand layers was ~2-3 in core OKU-1 and ~3 on the landward side, 475 indicating that sorting did not change as much as Md. In addition, the sorting became 476 slightly worse as the layers thinned and grains became finer landward. These changes 477 indicate that sorting may have worsened due to the incorporation of fine terrigenous 478 particles as they flowed upstream (inland). The sand layer in OW-14 has a thickness of 479 ~20 cm in core OKU-1, but became thicker in core OKU-3, and multiple units were 480 observed. The layer thickness decreased by half landward. The Md was 280-360 µm in 481 cores OKU-1 and OKU-3, suggesting upward fining. In contrast, the Md in cores OKU-482 slight fining. However, the upward fining in each unit was unclear, and large variations 484 were observed throughout. Sorting also tended to become poor, as large variations were 485 observed at the landward sites (Fig. 7). 486 Overall, for each event deposit, the layer thicknesses generally became thinner 487 and the grains tended to become finer landward. For reference, the Md of the sand layer 488 at the present beach (foreshore environment) of the Wasabiyachi lowland estuary is 270-489 340 µm, which is similar to the Md of the event layers in cores OKU-1 and OKU-2 near 490 the coast. In addition, the sorting of the present beach sand is ~1.4, which is better sorted 491 than the event sand layers. The lack of uniform event deposit layer thicknesses and grain 492 sizes with distance from the ocean indicates differences due to water mass speed during 493 an event, as well as differences in water depth in the depositional area (microtopography 494 of a lagoon or floodplain). obtaining event deposits at five horizons (Ow-1 to Ow-5). The characteristics of these 501 event deposits included the following: 1) the layers become thinner and the grains become 502 finer landward; 2) they had a grain size composition similar to that of beach sand; 3) the 503 grain fabric of the sand layer indicated a landward paleo-flow direction; and 4) marine 504 dinoflagellate cysts and foraminiferal linings were present in the sand layer. In addition, 505 since the Wasabiyachi lowland has not experienced any flood damage from storm surges 506 or tsunamis in the last 300 years, we can conclude that the origin of these event deposits 507 was multiple tsunamis that occurred over time. Based on radiocarbon dating of the 508 deposits, the age of Ow-1 was 0.7-1.0 ka, Ow-2 and Ow-3 were 1.7-1.9 ka, Ow-4 was 509 ~2.6 ka, and Ow-5 was ~3.0 ka. Kawakami  to IntCal 09 and IntCal 20 is approximately 10 to 80 years, which does not constitute a 520 major change to the chronological outline for Ow-1 to Ow-5 that was established 521

previously. 522
Overall, events deposits OW-6 to OW-20 exhibited thinner layers and finer 523 grains landward. In addition, the grain size composition of the sand layers interbedded at 524 seaward sites OKU-1 and OKU-2 were similar to those of the current beach sand. This 525 strongly suggests that the sand layers between OW-6 and OW-20 were derived from the 526 sea, and not inland regions. In addition, the sand layers of the lower deposits (OW-13 to 527 OW-20) were interbedded with the muddy sediment of an obstructed lagoon environment. 528 This lagoon deposit contains extremely thin laminated facies with no bioturbation, 529 indicating the presence of a closed lagoon in which sand layers that originated in the river 530 have not been transported. In contrast to this depositional environment, the sand layers 531 between OW-13 and OW-20 indicate that they were transported to the lagoon over the 532 beach ridges that developed along the coast. The upper deposits (OW-6 to OW-12) are 533 interbedded with peat layers and muddy lowlands in a similar manner to Ow-4 and Ow-534 5. The sand layers in these deposits (OW-6 to OW-12) have not been examined for marine 535 microfossils, as in Kase et al. (2016). However, these sand layers were interbedded in a 536 depositional environment in a similar manner to Ow-4 and Ow-5, suggesting that the 537 process of sand layer formation was the same. Accordingly, it is likely that event deposits 538 OW-6 to OW-20 observed in this study are tsunami deposits caused by tsunamis beyond 539 the beach ridges that have been surmised to have been present between the Wasabiyachi 540 lowland and the coast. In this study, we clarified the history of tsunami deposits on Okushiri Island 564 and extended the record to ~7.6 ka ( Fig.3 and Table 3 Figure 1 Distribution of the active faults in the offshore between Aomori to Hokkaido, northern part of Japan Sea (Committee for Technical Investigation on Large-Scale Earthquakes in the Sea of Japan, 2014). Okushiri island is located at the offshore of southwest Hokkaido, and active faults is distributed over the surrounding sea. Lage earthquake (1983 Middle Japan Sea earthquake and 1993 Hokkaido Nansei793 Oki earthquake) occurred in this area.

Figure 2
Topography around the Wasabiyachi lowland, Aonae region, southern part of Okushiri island. Index map shows the localities of borehole sites. Geologic and geomorphologic classi cation of landforms is based on Hata et al. (1982) and Koike and Mchida (2001). The Wasabiyachi lowland along the Aonae Bay is barriered by Holocene marine terrace and coastal beach ridge with sand dune. Contour maps are reproduced from online map of the Geospatial Information Authority of Japan (GSJ).   Age-depth curve of the OKU-1 -OKU-5 cores with estimated ages of event deposits. The vertical axis of the graph shows the depth and thickness of event deposits. The horizontal axis shows the calibrated age value to clearly indicate the correlation of the event deposits.  Stratigraphic correlation and changes of lithofacies, Md (median grain size), and sorting (geometric standard deviation) of the event deposits (OW-4, -5, -8, -9, -10, -12, -13, and -14).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.