1 Introduction

Tsunami deposits are reliable evidence for recent, historical, and prehistoric tsunamis and can be exploited to reconstruct tsunami recurrence intervals by dating (Nanayama et al. 2003; Cisternas et al. 2005; Kelsey et al. 2005; Jankaew et al. 2008; Sawai et al. 2009b; Dura et al. 2017; Ishizawa et al. 2017, 2019; Rubin et al. 2017; Fujino et al. 2018; Shimada et al. 2019; Higaki et al. 2021), tsunami inundation areas on the basis of their spatial distributions (MacInnes et al. 2010; Sawai et al. 2012, 2015; Sugawara et al. 2013; Pilarczyk et al. 2021), and the hydraulic conditions of tsunamis by inverse modeling (Jaffe and Gelfenbaum 2007; Jaffe et al. 2012; Goto et al. 2014; Naruse and Abe 2017; Tang and Weiss 2015; Tang et al. 2017; Mitra et al. 2020, 2021). In general, tsunami deposits are initially identified as coarse layers within fine strata by visual observation and soft X-ray imaging and are then diagnosed using multiple approaches (e.g., sedimentological, paleontological, and/or geochemical analyses; Goff et al. 2012). To eliminate other possibilities for the origins of the coarse-grained event deposits, such as river flooding and storm surge events (e.g., Tuttle et al. 2004; Morton et al. 2007), a comparison of paleo- and modern tsunami deposits is essential.

The tsunami deposits associated with the 2011 Tohoku-oki tsunami have been well documented (e.g., Goto et al. 2011, 2012, 2014; Abe et al. 2012, 2020; Chagué-Goff et al. 2012a, b; Jaffe et al. 2012; Naruse et al. 2012; Szczuciński et al. 2012; Pilarczyk et al. 2012; Richmond et al. 2012; Takashimizu et al. 2012; Matsumoto et al. 2016; Tanigawa et al. 2018; Iijima et al. 2021). For example, Goto et al. (2011) and Abe et al. (2012) observed sandy and muddy tsunami deposits on the Sendai Plain and compared their distributions to the area inundated by seawater during the tsunami event. Additionally, Goto et al. (2014) described changes in the thicknesses of the tsunami deposits from approximately 1300 locations, as well as flow-depth and elevation data on the Sendai Plain, to investigate the relations between these parameters. Those reports provided valuable information on the 2011 tsunami deposits as modern analogs for identifying paleo-tsunami deposits; however, detailed descriptions of sedimentary features and grain-size data in each location have not yet been published. Thus, although the 2011 tsunami provided considerable knowledge and abundant data on tsunami deposits have been obtained in previous studies, these data are uneven in their spatial and vertical resolution.

In this paper, we provide archives of visual images including soft X-ray images and thickness and grain-size data for the 2011 Tohoku-oki tsunami deposits, allowing direct comparison between the various locations and areas. The archives of the tsunami deposits were obtained from the Sendai Plain and the Fukushima coast, which were little investigated immediately after the tsunami, to document their sedimentological features. We also provide thickness and grain-size data at 1-cm vertical intervals along leveled transects. This unparalleled dataset will serve as an archive that will improve the identification of paleo-tsunami deposits and enable precise estimates of the hydraulic conditions of inundation flow via inverse modeling.

2 Study area

2.1 Geomorphological setting

Our field investigations were conducted along an approximately 80-km north–south distance along the coasts of the Sendai Plain and the northern part of Fukushima Prefecture (Fig. 1). We considered eight study areas that extend up to 5 km inland (Fig. 1B), which are, from north to south, the Onuma, Arahama, Natori, Watari, Yamamoto, Suijin-numa, Shinchi, and Odaka areas (Fig. 1B). The first five areas are located on the Sendai Plain, which extends about 50 km from north to south. The Sendai Plain is approximately 10 km wide in its northern part around the Onuma and Arahama areas, and narrows toward the south to less than 5 km wide around the Yamamoto area (Fig. 1). The Sendai Plain is a flat lowland, generally has an elevation above sea level of less than 5 m, and is characterized by beach ridges and muddy inter-ridge swales (now used as paddy fields) that lie parallel to the coastline (Tamura and Masuda 2004). The average gradient is gentle within the study areas in the Sendai Plain: about 0.3‰ in the Natori area, 0.4–0.5‰ in the Onuma, Arahama, and Watari areas; and 0.8‰ in the Yamamoto area. The three southernmost study areas are located on lowlands in narrow valleys between the hills (known as the Soso Hills, Fukushima Prefecture), which are up to several tens of meters high. These lowlands are generally at less than 5 m elevation, and each has a narrow (maximum 1 km wide) eastward opening and extends 2–4 km in the east–west direction. The average gradient is relatively high in these three areas: about 3‰ in the Suijin-numa area; 2‰ in the Shinchi area; and 1‰ in the Odaka area.

Fig. 1
figure 1

Location map of study areas along the Pacific coast around the Tohoku district, northeastern Japan. A Index map of northeastern Japan showing the Pacific coast around the Tohoku district. The epicenter of the 2011 Tohoku-oki earthquake (star) and the coseismic slip distribution (dashed lines; Ozawa et al. 2011) are marked. B Detailed index map of the eight study areas along the Pacific coast. Study locations within each area are shown in Fig. 2. The inundation limit of the 2011 tsunami (dashed blue line) follows the “1:100,000 tsunami flood area overview map” on the GSI Web site (Nakajima and Koarai 2011)

Full size image

2.2 The 2011 Tohoku-oki tsunami in the study area

Field surveys of inundation depths, heights, and areas associated with the 2011 tsunami in the study areas have been conducted by many researchers. Inundation depths above the surrounding ground levels in these areas were measured by using watermarks and calibrated to provide inundation heights relative to the mean sea level of Tokyo Bay (Tokyo Peil). These survey results were compiled by the joint survey group (The Tohoku Earthquake Tsunami Joint Survey Group 2011; Mori et al. 2012). The Geospatial Information Authority of Japan provided the inundation map, which was estimated from interpretations of aerial photographs and satellite images (Nakajima and Koarai 2011). The maximum inundation distance and inundation heights within each study area are summarized in Table 1. In addition, Goto et al. (2012) concluded that the first wave was the largest and reached the furthest inundation distance on the Sendai Plain.

Table 1 Summary of maximum inundation distances and heights within the studied areas
Full size table

3 Methods

We conducted field investigations in April and May 2011, January 2012, and November 2015 (Table 2). We selected a total of 123 study locations within the eight study areas: Onuma (ON1–17); Arahama (AR1–21); Natori (NT1–16); Watari (WT1–17); Yamamoto (YM1–27); Suijin-numa (SJ1–9); Shinchi (SN1–8); and Odaka (OD1–8) (Figs. 1, 2; Table 2). We excavated small pits, observed the grain size and sedimentary structures of the tsunami deposits, and measured their thicknesses at every study location (white and pink circles in Fig. 2), except for two locations (AR21 and OD8) at which no tsunami deposits, not even thin mud drapes, were found. We obtained samples of the tsunami deposits for laboratory analysis at 49 locations with a plate-shaped plastic case (pink circles in Fig. 2). Inundation depths were measured using watermarks at seven locations within the study areas (blue squares in Fig. 2; Table 3). We also surveyed the topography along east–west transects within the Shinchi and Odaka areas using VRS-GNSS survey instruments (Viva GS10, Leica Geosystems AG, Heerbrugg, Switzerland) in November 2015 (Figs. 2G, H, 3).

Table 2 Survey locations, survey dates, and deposits in each area
Full size table
Fig. 2
figure 2figure 2

Aerial views of the eight study areas showing the study locations. The position of each area is shown in Fig. 1B. A Onuma area (ON1–17). B Arahama area (AR1–21). C Natori area (NT1–16). D Watari area (WT1–17). E Yamamoto area (YM1–27). F Suijin-numa area (SJ1–9). G Shinchi area (SN1–8). H Odaka area (OD1–8). Aerial photographs are from the Geospatial Information Authority of Japan and were taken between December 2012 and September 2013

Full size image
Table 3 Measured inundation depths and distances from the neighboring coastline
Full size table
Fig. 3
figure 3

Topographic profiles along the transects in the A Shinchi and B Odaka areas; survey transects are indicated in Fig. 2G and H, respectively. Arrows indicate study locations for tsunami deposits. Several tiny peaks indicate footpaths between rice fields

Full size image

In the laboratory, soft X-ray imaging was carried out on the samples from 49 locations (Table 2) with a soft X-ray apparatus (SOFRON SRO-i503-2, SOFTEX Co., Ltd., Tokyo, Japan) and a digital X-ray sensor (NAOMI NX-04SN, RF Co., Ltd., Nagano, Japan). Grain-size analyses of the samples were conducted using an image analyzer (Retsch Camsizer, Verder Scientific, Haan, Germany), which has a wide effective measuring range that can capture grain sizes from silt to pebbles. Prior to the grain-size analysis, all samples were subsampled vertically at 1-cm intervals, and the subsamples were sieved in water to remove the mud component. The mud content of each subsample was calculated from the resulting weight reduction. Large pieces of organic matter such as roots and leaves in the subsamples were removed by hand prior to analysis. The measurement range of grain-size analysis was set from − 5.25 to + 6.25 phi at intervals of 0.25 phi. Finally, grain-size distribution properties such as mean grain size and sorting were calculated from the measurements of each subsample using the logarithmic graphical method of Folk and Ward (1957). The average values of these grain-size properties and the mud contents at each location were re-calculated based on the results of the subsamples.

4 Results

4.1 General characteristics of tsunami deposits in the study area

The tsunami deposits were widely distributed over the lowlands in the study areas, and extended up to 4 km inland from the coastline. The tsunami deposits were composed mainly of sand, with occasional gravels, mud clasts, shells, and plant fragments, and sometimes were overlain by a mud drape. The tsunami deposits were easily recognized because they generally overlie the muddy soil of paddy fields or pavement. In addition, the tsunami deposits generally showed an upward-fining trend and parallel lamination, with occasional upward-coarsening or repeated upward-fining trends (Figs. 4, 5, 6, 7, 8 and 9).

Fig. 4
figure 4

Schematic diagrams of A stratigraphic columns, B soft X-ray images, and C grain-size analyses at study locations within the Arahama and Natori areas. Scale bars represent 5 cm. Tsunami deposits are indicated by double-headed arrows beside the columns. Horizontal dashed lines in C indicate the basal boundary of the tsunami deposits. Mud contents (black circles) are plotted against the bottom logarithmic axis; mean grain sizes (crosses) and sorting values (white squares) are plotted against the top axis

Full size image
Fig. 5
figure 5

Schematic diagrams of A stratigraphic columns, B soft X-ray images, and C grain-size analyses at study locations within the Watari area. Scale bars represent 5 cm. Tsunami deposits are indicated by double-headed arrows beside the columns. Horizontal dashed lines in C indicate the basal boundary of the tsunami deposits. Mud contents (black circles) are plotted against the bottom logarithmic axis; mean grain sizes (crosses) and sorting values (white squares) are plotted against the top axis

Full size image
Fig. 6
figure 6

Schematic diagrams of A stratigraphic columns, B soft X-ray images, and C grain-size analyses at study locations within the Yamamoto area. Scale bars represent 5 cm. Tsunami deposits are indicated by double-headed arrows beside the columns. Horizontal dashed lines in C indicate the basal boundary of the tsunami deposits. Mud contents (black circles) are plotted against the bottom logarithmic axis; mean grain sizes (crosses) and sorting values (white squares) are plotted against the top axis

Full size image
Fig. 7
figure 7

Schematic diagrams of A stratigraphic columns, B soft X-ray images, and C grain-size analyses at study locations within the Suijin-numa area. Scale bars represent 5 cm. Tsunami deposits are indicated by double-headed arrows beside the columns. Horizontal dashed lines in C indicate the basal boundary of the tsunami deposits. Mud contents (black circles) are plotted against the bottom logarithmic axis; mean grain sizes (crosses) and sorting values (white squares) are plotted against the top axis

Full size image
Fig. 8
figure 8

Schematic diagrams of A stratigraphic columns, B soft X-ray images, and C grain-size analyses at study locations within the Shinchi area. Scale bars represent 5 cm. Tsunami deposits are indicated by double-headed arrows beside the columns. Horizontal dashed lines in C indicate the basal boundary of the tsunami deposits. Mud contents (black circles) are plotted against the bottom logarithmic axis; mean grain sizes (crosses) and sorting values (white squares) are plotted against the top axis

Full size image
Fig. 9
figure 9

Schematic diagrams of A stratigraphic columns, B soft X-ray images, and C grain-size analyses at study locations within the Odaka area. Scale bars represent 5 cm. Tsunami deposits are indicated by double-headed arrows beside the columns. Horizontal dashed lines in C indicate the basal boundary of the tsunami deposits. Mud contents (black circles) are plotted against the bottom logarithmic axis; mean grain sizes (crosses) and sorting values (white squares) are plotted against the top axis

Full size image

The average mud contents of the tsunami deposits and the pre-tsunami deposits (underlying soils) over all locations were 10.6% and 32.9%, respectively. The location-averaged mud contents of the tsunami deposits had a range of 0.4–96.6%. In general, the mud content increased landward, although this pattern varied within study areas (Fig. 10A). In contrast, the location-averaged mud contents of the underlying soils had a range of 2.7–70.3%, and showed no discernible relationship with distance from the coastline (Fig. 10B). The average mud content of the muddy post-tsunami deposits that were found only in the Odaka area was 71.1%. The location-averaged values of these deposits were 30.8%–87.0% and appeared to show a positive correlation with distance from the coastline (Fig. 10B).

The maximum observed thickness of the tsunami deposits was 65 cm, and the average thickness was 8.6 cm. The thickness decreased exponentially landward; however, the thickness varied widely within each study area (Fig. 11).

The mean grain size of the tsunami deposits averaged over all locations was 1.6 phi, and the range of location-averaged values was − 0.5 to + 2.7 phi. The tsunami deposits showed a faint landward-fining trend, but this varied widely within the study areas (Fig. 12A). The mean grain size of the underlying soil was 1.7 phi, which was not significantly different from that of the tsunami deposits. The range of location-averaged soil grain sizes was 0.7–2.4 phi, and soil grain size appeared to show no relationship with distance from the coastline (Fig. 12B). The mean grain size of the post-tsunami deposits in the Odaka area was on average 2.5 phi, and the location-averaged values were 1.4–2.9 phi, with a landward-fining trend (Fig. 12B). Additionally, the tsunami deposits usually consisted of moderately well-sorted to moderately sorted sand (Figs. 4, 5, 6, 7, 8 and 9; see Additional file 1: Text S1 and Additional file 2: Data S1 for details), except for muddy layers such as mud drapes.

Normal and inverse grading were common within the tsunami deposits; for example, at location AR1, the tsunami deposits showed inverse and normal grading in the lower and upper parts, respectively, with a gradual upward change in mean grain size from approximately 1.4 phi through 1.1 phi to 2.0 phi (Fig. 4). These grading structures sometimes overlay each other and were associated with mud drapes within the deposits, indicating that the tsunami deposit contained multiple units.

Parallel lamination was also common within the sandy tsunami deposits. The tsunami deposits in this study usually had an erosional contact with the underlying soil, and the lower parts of the tsunami deposits often contained mud clasts and gravel. Below, we describe the sedimentological characteristics of the tsunami deposits in each study area. Detailed descriptions including grain-size data of the tsunami deposits at each location are provided in Table 2, Additional file 1: Text S1 and Additional file 2: Data S1.

4.2 Description of the tsunami deposits in each study area

4.2.1 Onuma Area

The furthest landward inundation limit (5.3 km) and the greatest maximum inundation height (19.7 m) of the study areas occurred in the Onuma and Arahama areas (Table 1). The inundation depth was 155 cm at location ID1 (Fig. 2A, Table 3). Field observations at 17 locations in the Onuma area demonstrated that the tsunami deposits extended about 4 km inland from the coastline and showed an obvious landward-thinning trend: the deposit thickness changed from 14 cm at ON2 (1042 m inland from the coastline) to almost 0 cm at ON17 (4030 m from the coastline) (Fig. 11, Table 2). Visual inspection revealed that the tsunami deposits were generally composed of fine to medium sand, and a mud drape was common at the top of the tsunami deposits. At locations more than 3 km inland from the coastline, the tsunami deposits were predominantly composed of mud.

4.2.2 Arahama area

In the Arahama area, the tsunami reached its furthest inundation limit and highest inundation height, as described above. The inundation depths were 205, 270, 176, and 126 cm at locations ID2, ID3, ID4, and ID5, respectively (Fig. 2B and Table 3). Field observations at the 21 sampling locations revealed that the tsunami deposits extended about 3.6 km inland and showed a landward-thinning trend with a wide variation of thicknesses, from 17 cm at AR1 (262 m), to almost 0.1 cm at AR20 (3617 m), and finally disappearing at AR21 (3744 m) (Fig. 11, Table 2). Visual inspection revealed that the tsunami deposits were generally composed of fine to medium sand, and a mud drape was common at the top of the unit (Fig. 4, Table 2). At locations farther than 3 km inland from the coastline, the tsunami deposits were mostly composed of mud drape. Laboratory analysis of samples from seven locations in this area demonstrated that the tsunami deposits had a mean grain size of 1.6 phi with a mud content of 24.0%, whereas the underlying soil had a mean grain size of 1.5 phi and a mud content of 25.4%. The thick mud drape (maximum thickness 5 cm) at the top of the tsunami deposits greatly contributed to the high mud content relative to the soil. The tsunami deposits showed an obvious trend of landward-increasing mud content and a slight landward-fining trend, whereas the soil showed almost no correlation of those properties with distance from the coastline. An upward-fining trend was also common in the tsunami deposits, but parallel lamination was rare.

4.2.3 Natori area

In the Natori area adjacent to Sendai Airport, field observations at 16 locations demonstrated that the tsunami deposits extended more than 4 km inland, with a wide range of thicknesses from 1 cm at NT6 (1782 m) to 18 cm at NT5 (1730 m) (Fig. 2C, Table 2); thus, in this area, there was almost no trend in thickness with distance from the coastline. Visual inspection revealed that the tsunami deposits were generally composed of medium to coarse sand, occasionally including gravel, and a thin mud drape was commonly present at the top of the tsunami deposits (Table 2). Parallel lamination and upward fining were common in the tsunami deposits, and upward coarsening was rarely observed in this area. At locations more than 3 km from the coastline, the tsunami deposits were predominantly composed of mud. Laboratory analysis of a sample from location NT10 showed that the tsunami deposits had a mean grain size of 1.2 phi and a mud content of 1.3%, whereas the mean grain size of the underlying soil was 1.3 phi and the mud content was 4.3% (Fig. 4, Table 2).

4.2.4 Watari area

In the Watari area, field observations were carried out at 17 locations. The tsunami deposits were found more than 2.6 km inland and showed a landward-thinning trend, from 33 cm at WT1 (164 m from the coastline) to 3 cm at WT14 (1754 m from the coastline) (Fig. 11, Table 2). Visual inspection revealed that the tsunami deposits were generally composed of fine to medium sand, and a mud drape was occasionally observed at the top of the tsunami unit (Fig. 5, Table 2). The tsunami deposits were predominantly composed of mud at locations more than 1.7 km from the coastline. On the basis of samples from nine locations in this area, the tsunami deposits had a mean grain size of 1.6 phi and a mud content of 11.1%, whereas the grain size of the underlying soil was 1.3 phi and the mud content was 26.5%. The tsunami deposits showed an increase in mud content and a slight fining trend landward, but the soil showed a faint increase in mud content and almost no trend in mean grain size landward. An upward-fining trend was common in the tsunami deposits, and parallel lamination was occasionally observed.

4.2.5 Yamamoto area

In the Yamamoto area, our field observations at 27 locations demonstrated that the tsunami deposits extended more than 3.4 km inland and showed a clear exponential thinning trend landward, from 65 cm thickness at YM2 (154 m from the coast) to 0.5 cm at YM27 (3417 m from the coast) (Fig. 11, Table 2). Visual inspection revealed that the tsunami deposits were generally composed of fine to medium sand, and, particularly at landward locations, included a mud drape at the top of the unit (Fig. 6, Table 2). At locations more than 2.7 km from the coastline, the tsunami deposits were predominantly composed of mud. Samples from nine locations in this area were analyzed: the tsunami deposits had a mean grain size of 1.5 phi and a mud content of 5.7%, whereas the mean grain size of the underlying soil was 1.4 phi and the mud content was 17.2%. The tsunami deposits showed a clear increase in mud content and became finer landward; the soil showed a clear increase in mud content but an almost constant mean grain size landward. Upward-fining and parallel lamination was common in the tsunami deposits, and upward coarsening was occasionally visible.

4.2.6 Suijin-numa area

In the Suijin-numa area, we carried out field observations at nine locations and identified that the tsunami deposits extended more than 1.5 km inland and showed a faint thinning trend landward, from 27 cm thickness at SJ1 (211 m from the coast) to 6 cm at SJ7 (999 m from the coast) (Fig. 11, Table 2). Visual inspection revealed that the tsunami deposits were generally composed of fine to coarse sand, with gravelly sediments occurring at SJ1 (Fig. 7, Table 2). A mud cap was commonly present at the top of the tsunami deposits (Fig. 7, Table 2). Based on the laboratory analysis of samples from seven locations in this area, the tsunami deposits had a mean grain size of 1.3 phi and a mud content of 10.1%, whereas the underlying soil had a mean grain size of 1.7 phi and a mud content of 23.3%. The tsunami deposits showed a faint increase in mud content and a faint fining trend landward, albeit with wide variation, but the soil showed a clear increase in mud content and an almost constant mean grain size landward. Upward fining and parallel lamination were common in the tsunami deposits, and upward coarsening was occasionally present.

4.2.7 Shinchi area

In the Shinchi area, a topographic survey demonstrated that the lowland in the narrow valley had a relatively high average gradient of 2‰ within the area of the tsunami deposit (Fig. 3A). Field observations at eight locations demonstrated that the tsunami deposits extended more than 1.5 km inland and gradually thinned landward, from 22 cm at SN1 (304 m from the coast) to 5 cm at SN5 (712 m from the coast) (Fig. 11, Table 2). Visual inspection revealed that the tsunami deposits were mainly composed of medium sand, and a mud cap was common at the top of the unit (Fig. 8). Based on the laboratory analysis of samples from all locations in this area, the mean grain size of the tsunami deposits was 1.7 phi and the mud content was 7.4%, whereas the mean grain size of the underlying soil was 1.4 phi and the mud content was 43.7%. The tsunami deposits showed a faint increase in mud content and almost no trend in mean grain size landward, whereas the mud content and grain size of the soil increased and slightly fined landward, respectively. An upward-fining trend and parallel lamination were very common in the tsunami deposits, and upward coarsening was occasionally observed. These upward coarsening–fining trends were repeated, especially at locality SN3.

4.2.8 Odaka area

In the Odaka area, a topographic survey revealed that the lowland in the narrow valley had a relatively high average gradient of 1‰ within the area of the tsunami deposits (Fig. 3B). Field observations at eight locations demonstrated that the tsunami deposits extended to a distance of 1.7 km inland. Visual inspection revealed that the deposits were mainly composed of fine to medium sand and showed a gradual upward change from sand to muddy sand and/or sandy mud, and finally a thick mud cap at the top, which was characteristic of the tsunami unit in this area (Fig. 9). Distinguishing the tsunami deposits from post-tsunami reworked sediments was difficult in this area, because our survey was conducted 4 years after the tsunami. Here we define the tsunami deposits as layers of sand and muddy sand, and post-tsunami deposits as layers of mud and sandy mud, though some uncertainty about these identifications remains. Abundant rootlets were found penetrating the post-tsunami deposits toward the top of the tsunami deposits. The tsunami deposits showed a faint landward-thinning trend from 12 cm at OD2 (983 m from the coast), to 8 cm at OD7 (1699 m from the coast), and finally disappearing at OD8 (2271 m from the coast) (Fig. 11, Table 2). Samples from all locations in this area were analyzed. The tsunami deposits had a mean grain size of 1.6 phi and a mud content of 11.1%; the corresponding values for the underlying soil were 2.2 phi and 46.9%, respectively. The tsunami deposits contained an almost constant percentage of mud and become slightly finer landward, whereas the soil showed no trend in mud content and almost constant mean grain size. The thick mud cap (mud content 71.1%) at the top had a mean grain size of 2.5 phi and showed a clear increase in mud content and a gradual fining trend landward. Within the tsunami unit, upward fining and parallel lamination were very common, and mud clasts and plant fragments were occasionally present. Additionally, rootlets penetrating down into the tsunami deposits were well developed in this area.

Fig. 10
figure 10

Average mud contents at each location with distance from the coastline within each study area. A Mud contents of the tsunami deposits. B Mud contents of the pre- and post-tsunami deposits

Full size image

5 Discussion

5.1 Spatial distribution of tsunami deposits

The wide extent of tsunami inundation and the spatial distribution of tsunami deposits may be attributed to the low-gradient topography of the study areas, where the average gradient is 0.3–3‰. In contrast, in areas with steeper gradients, the distribution of tsunami deposits is generally limited; for example, Goto et al. (2017) found gravelly tsunami deposits within approximately 1 km of the coastline with an average gradient of ca. 2% in a narrow valley on the ria-type Sanriku coast. Naruse et al. (2012) reported the presence of tsunami deposits within approximately 1.5 km of the coastline in a bay-head delta on the Sanriku coast, where the average gradient is 1%. The tsunami inundation limit and the distribution of tsunami deposits are controlled by the gradient and the surrounding environment.

Our observations revealed that the widely distributed tsunami deposits on the Sendai Plain and the northern coast of Fukushima Prefecture generally showed a landward-thinning trend. This trend is a common feature of the 2011 tsunami deposits (e.g., Goto et al. 2011; Abe et al. 2012, 2020; Richmond et al. 2012; Matsumoto et al. 2016) and other tsunami deposits elsewhere (e.g., Gelfenbaum and Jaffe 2003; Hori et al. 2007; Fujino et al. 2010), suggesting a general landward decrease in the flow capacity of tsunamis (Hiscott 1994). Goto et al. (2011) reported that the 2011 tsunami deposits became thin and mud-dominated landward around Sendai Airport, near the Natori area of this study. Abe et al. (2020) and Iijima et al. (2021) also studied the 2011 tsunami deposits in narrow valleys near Suijin-numa and in the Odaka area, respectively. Both studies concluded that the deposits generally thinned landward; however, there were some fluctuations resulting from the micro-geomorphology and artificial structures. These results are concordant with our findings that there is an overall landward-thinning trend, but some non-negligible fluctuations occur in each area.

5.2 Grain-size composition

The mud content of the tsunami deposits as a whole tended to obviously increase with distance from the coastline (Fig. 10A), concomitant with the landward-fining trend (Fig. 12A), whereas the mud content and mean grain size of the underlying soil did not seem to correlate with distance from the coast (Figs. 10B, 12B). These trends in the tsunami deposits have been reported by previous studies in the Sendai Plain (e.g., Goto et al. 2011; Abe et al. 2012, 2020), and can be explained by the continual entrainment of mud from the surface soil (Nanayama and Shigeno 2006; Matsumoto et al. 2010; Iijima et al. 2021) and the selective sedimentation of coarser grains (Bondevik et al. 1997; Fujino et al. 2010) during landward inundation. However, the mud content in each area did not tend to show a regular increase, especially in the three southern areas. Abe et al. (2020) studied a location very near the Suijin-numa area in this study and showed that the thicknesses of the mud layers of tsunami deposits, such as mud drapes, varied independently of the inundation distance or topographic elevation. They concluded that the thicknesses of the mud layers were affected by sediments from Suijin-numa Pond located in the middle of the valley.

Fig. 11
figure 11

Thicknesses of tsunami deposits with distance from the coastline within each study area

Full size image
Fig. 12
figure 12

Mean grain sizes with distance from the coastline within each study area. The mean grain size is the average value for each location. A Mean grain sizes of the tsunami deposits. B Mean grain sizes of the pre- and post-tsunami deposits

Full size image

Diversity in the grain-size composition of the tsunami deposits was recognized in this study, although the deposits were composed mainly of well-sorted sandy sediments at most locations. For example, the tsunami deposit at SJ1 was characterized by gravelly sediments, and the tsunami deposits in the Odaka area were characterized by thick mud drapes at their tops (Figs. 7, 9). These variations in tsunami deposits are commonly recognized for the deposits formed by modern and paleo events in the Sendai and Fukushima regions and elsewhere (e.g., Hori et al. 2007), and can result from various factors; for example, bedforms formed under inundation flow and influenced by the return flow might enhance vertical and horizontal grain-size deviations. In general, tsunamis hit wide areas that contain a range of environments, so it is difficult to specify the factor(s) causing the diversity; however, differences in sediment sources along coastal zones and variations in the depositional processes owing to topographic differences are possible causes (e.g., Szczuciński et al. 2012; Goto et al. 2017; Abe et al. 2020). Such diversity in grain-size composition and in sedimentary structures, as mentioned below, makes it difficult to identify paleo-tsunami deposits. In most cases, it is necessary to combine multiple features as supporting evidence for paleo-tsunami deposits (Goff et al. 2012; Sawai 2012). Thus, consideration of the diversity of sedimentary features of modern tsunami deposits is important when identifying paleo-tsunami deposits.

5.3 Sedimentary structures

The grain-size properties and sedimentary structures of the tsunami deposits in this study reflect the sedimentary processes under the inundation flow at each location. Normal grading was the most common structure in the tsunami deposits in this study and has also commonly been found in other paleo- and modern tsunami deposits (e.g., Dawson and Shi 2000; Hori et al. 2007; Jankaew et al. 2008). Normal grading results from suspension settling or bedload during the waning of the flow (Naruse et al. 2012). In this study, normal grading was frequently associated with parallel lamination. These combined features were observed within about 1.1 km of the coast on the Sendai Plain and 1.6 km of the coast in the three southern areas, implying that these sediments were deposited from bedload under relatively rapid flow, as mentioned below. In contrast, normal grading without parallel lamination was mainly found at locations further inland and might have resulted from suspension settling under relatively slow flow or in stagnant water. Inverse grading was also observed in this study and has been documented in other modern tsunami deposits (e.g., Hori et al. 2007; Paris et al. 2007; Sawai et al. 2009a; Naruse et al. 2010, 2012). Inverse grading results from flow acceleration during the waxing stage (Naruse et al. 2010; Iijima et al. 2021) or a traction carpet under sheet flow (Sohn 1997; Moore et al. 2011), indicating a considerable flow. This suggestion is consistent with the fact that inverse grading was observed in the relatively thick tsunami deposits within approximately 700 m of the coast. Multiple-unit structures usually consist of repeated sets of normal and/or inverse grading, suggesting multiple stages of inundation and/or return flows (Hori et al. 2007; Paris et al. 2007; Naruse et al. 2010, 2012).

Parallel lamination was commonly detected in this study, including by soft X-ray imaging (Figs. 4, 5, 6, 7, 8 and 9, Table 2). In this study, parallel lamination was typically recognized in tsunami deposits that have a thickness of more than 5 cm. The tsunami deposits showing parallel lamination were detected as far as approximately 1.7 km inland. These structures are formed by low-amplitude bedforms under a plane-bed flow regime (Allen 1984; Best and Bridge 1992), suggesting that the inundation flow had a relatively high velocity during the deposition of these sediments. In contrast, cross-lamination/ripple marks were not found in this study, except at location SJ8 (Table 2), and have rarely been reported in other studies (e.g., Naruse et al. 2012; Takashimizu et al. 2012). The scarcity of cross-lamination/ripple marks might imply that there were almost no cases in which the tsunami inundation flow continuously maintained a lower flow regime.

Mud drapes were recognized in the tsunami deposits at 83 locations over the study areas (Figs. 4, 5, 6, 7, 8 and 9, Table 2). These drapes were formed from suspension during the stagnant stage of the inundation flow. Therefore, the presence of mud drapes suggests that multiple flows were produced by the tsunami (Fujiwara and Kamataki 2007). Mud drapes have previously been found in the 2011 tsunami deposits (e.g., Naruse et al. 2012; Abe et al. 2020) and other modern tsunami deposits (e.g., Nanayama and Shigeno 2006; Choowong et al. 2008; Matsumoto et al. 2008; Naruse et al. 2010).

The lower contact of the tsunami deposits in this study was usually a gently undulating sharp erosional surface, often with mud clasts and basal gravels on the surface, suggesting that the strong flow eroded the existing muddy surface to form the tsunami deposits. Some studies have reported that deformation structures such as load casts occur at the contact between tsunami deposits and underlying soft sediments for both ancient (Minoura and Nakata 1994; Sawai et al. 2015) and modern (Matsumoto et al. 2008) tsunamis; however, such structures were rarely found at the lower contact of the tsunami deposits in this study. This absence is probably because most of the tsunami deposits in this study were deposited on rice fields during the fallow period, a relatively hard preexisting surface that resisted deformation.

5.4 Application to paleo-tsunami research

The differences between the tsunami deposits in the Odaka area and those in the other areas might imply the occurrence of post-depositional alteration, because the field observation in the Odaka area was conducted 3–4 years after that in the other areas. In fact, it is necessary to consider environmental differences: the tsunami deposits in the Odaka area uniquely possessed thick mud drapes at their tops that were possible post-tsunami reworked sediments (Fig. 9). The lower boundary of the tsunami deposits was clearly sharp, as in the other areas; in contrast, the top boundary was vague, possibly because of downward bioturbation by intruding rootlets. Spiske et al. (2020) demonstrated that post-depositional alteration decreased the thickness and mean grain size of the tsunami deposits, and similar processes would have occurred in the Odaka area. Such post-depositional alternation of tsunami deposits could cause problems when identifying paleo-tsunami deposits and using them to reconstruct the magnitude of tsunami events. In this respect, long-term observation of modern tsunami deposits is required for more precise paleo-tsunami research.

Abe et al. (2012) proposed a 1:1 relationship between the maximum inundation distance and the extent of sandy tsunami deposits. They concluded that the relationship was valid up to 2.5 km inland, and the maximum extent of recognizable deposits (≥ 0.5 cm) was limited to within 3 km of the coast. Some differences appear when plotting the data in this study (Fig. 13): the relationship seems to be generally valid up to 3.5 km inland, and the maximum extent of sandy deposits is limited to within 3.4 km of the coast. Moreover, in places where the maximum inundation distance exceeded 5 km and the relationship is totally invalid, a large gap is recognized between the extent of sand and the distribution of mud drapes. Some of these differences might arise from the sparsity of data; however, the fact that a similar tendency was recognized in this study is important for future paleo-tsunami research.

Fig. 13
figure 13

Scatter plot showing the maximum inundation distance and extent of tsunami deposits within the eight study areas. Mud drapes were found further inshore than sand in the Onuma, Arahama, Natori, and Yamamoto areas. The two dashed lines are after Fig. 9 of Abe et al. (2012), which referred to Sugawara et al. (2010) regarding the maximum extent of the 869 CE Jogan tsunami deposits

Full size image

The large amount of data on modern tsunami deposits provided in this study can be used to validate numerical modeling methods, leading to a more precise estimation of the hydraulic conditions of paleo-tsunamis from their deposits. However, some issues still remain, including the precise estimation of maximum inundation distance and post-depositional alteration. Along with the sedimentological data in this study, geochemical (Chagué-Goff 2010) and microfossil (Hemphill-Haley 1996) approaches will help to identify the maximum inundation distance from the deposits, as pointed out by Goto et al. (2011). Finally, it will be possible to evaluate how post-depositional processes have altered deposits by comparing paleo-tsunami deposits and the results of numerical simulations (Spiske et al. 2020). In this regard, the Sendai Plain, where extensive research on the 869 CE Jogan tsunami deposits has been conducted, would be the best place to conduct such a comparative study.

6 Conclusions

We described the 2011 Tohoku-oki tsunami deposits on the Sendai and Fukushima coasts, with quantitative data on their thickness and grain size at 1-cm vertical intervals along transects. The sedimentary characteristics of the 2011 tsunami deposits can be summarized as follows: (1) The tsunami deposits in the study areas extended up to 4 km inland, probably owing to the overall low-gradient topography. (2) The tsunami deposits showed landward-fining and landward-thinning trends with minor fluctuations, and also a trend of landward increase in the mud content. (3) Diversity in the grain-size composition of the tsunami deposits was recognized, resulting from differences in the sediment source and possibly topographic conditions. (4) Normal grading, parallel lamination, and mud drapes were found widely over the study areas, whereas inverse grading and cross-lamination/ripple marks were recognized at only limited locations. (5) The lower contact of the studied tsunami deposits is clearly sharp; however, the top boundary is vague in the Odaka area, possibly owing to the influence of post-depositional alteration. (6) The 1:1 relationship proposed by Abe et al. (2012) is valid only up to 3.5 km inland in this study.

These findings will contribute to a better understanding of modern and paleo-tsunami deposits. Furthermore, the vertically high-resolution dataset representing a range of environments provided in this study will be useful to validate numerical modeling methods, leading to progressive enhancement of the spatial and temporal resolution of estimates of the hydraulic conditions of inundation flows, and improved evaluation of the degree of post-depositional alteration processes.