HOLOCENE OVERWASH OCCURRENCE AGE IN THE ISUMI RIVER LOWLAND, EASTERN BOSO PENINSULA, JAPAN

Abstract

Analysis of the chronological data and observation of a lagoonal sediment core reveal sand washover events between around 2.4 to 2.5 cal. ky BP in the Isumi River lowland. We conducted radiocarbon dating with AMS and constructed an age-depth model using the latest calibration curve and appropriate model routine. In the middle to lower part of the core, dark-gray sand layers are repeatedly deposited. Sand layers may exhibit an erosional surface at the base with fining upward grading. The overwash layers are composed of well-rounded sand with occasional gravel, indicative of transportation. overwash sediment characteristics are consistent with proximal marine deposits, suggesting an ocean origin (though riverine sediment is also similar in character). The age-depth model indicates very high sediment accumulation rates associated with overwash deposits. Based on the amount of accumulated sediment, relatively large-scale redeposition events occurred during this period but more information is needed to constrain the mechanism(s) causing the events. We also present a local reservoir age correction compatible with the Marine20 calibration curve.

INTRODUCTION

The purpose of this study is to create an age-depth model using the sediment core collected from the Isumi River lowland and to confirm overwash layers. Radiocarbon dating was performed on both terrestrial plant samples and bulk samples, allowing discussion of the age offsets. In the Boso Peninsula, overwash may be caused by a variety of factors. This area was originally a drowned valley formed during the post-glacial transgression, and subsequent sedimentation and uplift have turned it into an estuarine lowland (Sakai et al. 2006). The bay was closed off from the open ocean due to the development of a barrier system and sediment transported by the longshore current infilled the bay.

The Boso Peninsula is a tectonically active area on the Pacific coast of central Japan where three plates converge offshore to the southeast. These are the North American Plate, the Eurasian Plate, and the Philippine Sea Plate that form a triple junction. The associated ocean trenches are the Japan Trench, Sagami Trough, Izu-Bonin Trench, and Nankai Trough. On the Pacific Ocean side, river and coastal terraces have developed due to past seismic uplift. The tectonic activity promoted valley-fill sedimentation in the area. Barrier development and valley-fill reduced the influence of seawater, creating an environment where mud and/or silt are slowly deposited. Such an environment is suitable for the study of the overwash deposit, which primarily consists of sand because the deposits may be readily differentiated.

STUDY AREA

The Isumi River lowland is in the eastern part of the Boso Peninsula (Figure 1) and extends approximately 6 km from north to south and 2 km from east to west. The maximum tidal range along the Pacific coast of Boso is 1.0 m. The Isumi River is 67.5 km long and has an about 300 km² drainage area (the largest river of those discharging on the Pacific Ocean side of the Boso Peninsula) and flows east to west through the middle of the lowland to the Pacific Ocean. Terrace developments are recognized in the Isumi River lowland (Nagasawa 1979; Sakai et al 2006; Utsunomiya and Oi 2019). Three major terraces in the area are the Choja, Izumi, and Fukuhara surfaces. Our study area is located in the northeastern part of the Isumi River and is blocked on the seaward side by dunes and is on the Izumi Surface.

Figure 1 (A) Map of Japan using etopo1 elevation data: White and black circles respectively show the approximate location of the epicenters of the 1923 Taisho and 1703 Genroku Earthquakes. Box shows the area in panel B. (B) Shaded terrain map of the Boso Peninsula. Box shows the area in Panel C. Isumi River drainage extent is shown in blue. (C) Isumi region and the on-land location of Core ISL-6 (star) with the Izumi (dunes), Choja (lagoon), and Fukuhara Surfaces shaded in white, red, and blue, respectively (Nagasawa 1979; note that the Izumi lagoon and Choja dune surface exist but are not in range of Panel C.). Shaded terrain data from the Geospatial Information Authority of Japan (https://www.gsi.go.jp/bousaichiri/hillshademap.html). River data is from the National Land Information Division of the Ministry of Land, Infrastructure, Transport and Tourism (https://geoshape.ex.nii.ac.jp/river/resource/120010/). (Please see online version for color figures.)

Isumi River lowland was formed by sediment transport by the longshore current. The depositional process and landforms from about 11.0 cal. ky BP to 3.0 cal. ky BP are reconstructed from valley-fill deposits affected by rapid tectonic uplifts (Sakai et al. 2006). The sediment progradation occurred during a period of rapid sea level rise between 9.0 to 8.5 cal. ky BP. This was due to an increase in sediment supply caused by tectonic uplift associated with a large earthquake. The Paleo-Isumi Bay extended inland around 7.0 cal. ky BP due to postglacial marine transgression. The Choja and Izumi surfaces formed sometime after 6.5 cal. ky BP and around 3.5 cal. ky BP, respectively, and the modern barrier formed about the same time as the Izumi Surface emergence. The barrier enclosed the remnant Paleo-Isumi Bay, forming a small lagoon. The lagoon was filled by the progradation of small delta and by sediments deposited by large waves from the outside of the barrier. The study area’s proximity to the Ocean and large river makes it prone to overwash due to flood and wind waves.

MATERIALS AND METHODS

Sample and Description of the ISL-6

We analyzed a 3.3 m long sediment boring core, ISL-6, recovered from the former lagoon on the northeastern part of the Isumi River lowland, approximately 150 m inland from the modern coastline. The core was collected during a research observation conducted by the National Institute of Advanced Industrial Science and Technology (AIST; Shishikura et al. 2013).

While the core was photographed in the field, the image resolution was relatively low. We therefore re-imaged it using a portable core scanner known as The Namahage, which is a portable core imaging device that produces continuous images, similar to a traditional “line scanner” by using computer vision techniques to create a low distortion composite image from photographs taken at an approximate 5-cm interval. It has been successfully used previously in lacustrine environments (Obrochta et al. 2018). We also obtained CT images of the core at AIST that allow us to examine the internal structure and estimate sediment density. Relatively dense sand beds have a higher CT value, while less dense mud beds have lower CT values. The use of CT images allows more precise identification of overwash deposits. Finally, we conducted CNS elemental analysis using an elemental analyzer (Vario MACRO cube) at AORI (Atmosphere and Ocean Research Institute of Tokyo University).

Age-Depth Model

Radiocarbon dating was conducted on 5 terrestrial plant samples and 11 bulk samples using accelerator mass spectrometry (AMS) at AORI (Atmosphere and Ocean Research Institute of Tokyo University). Three plant fragments were sent to paleo labo for dating shortly after core recovery but have not yet been published and are included here. Bulk samples were selected with high carbon content based on elemental analysis data. Bulk samples were also taken from directly above and below the overwash layers. We used automatic graphitization at AORI that can perform graphitization by simply inserting samples to the elemental analyzer (e.g., Yokoyama et al. 2022). Obtained determinations were calibrated to calendar years by using MatCal (Lougheed and Obrochta 2016) using IntCal20 (Reimer et al. 2020). We used Undatable (Lougheed and Obrochta 2019) for age-depth modeling, which is a very fast, frequentist, age model routine. The reader is referred to Lougheed and Obrochta (2019) and to Nara-Watanabe et al. (2023) for more detailed descriptions and comparison with Bayesian methods. It includes both depth and age uncertainty. A certain percentage of data can be bootstrapped out of each simulation, which is useful when not all determinations are in stratigraphic order. If radiocarbon determinations were performed on different material, those materials can be weighted differently by selectively not bootstrapping higher-quality material, giving for example terrestrial material priority over bulk organic matter.

The following is a summary of age-depth model conditions in this study. Bootstrapping was not performed on leaf samples, which are more reliable than bulk samples.

Conditions of Age-Depth Model

Software: Undatable (Lougheed and Obrochta 2019)

1. 1. Four plant fragments from 270 to 273 cm treated as replicate samples.

2. 2. Bulk determinations: 11 (corrected for offset from plant fragments; see below)

3. 3. Plant fragment determinations: 8

4. 4. 100,000 iteration Monte Carlo simulation

5. 5. 30% bootstrapping of bulk dates (no bootstrapping of leaves)

6. 6. Sedimentation date uncertainty factor 0.1

7. 7. IntCal20 (Reimer et al. 2020) for calibration

Bulk ages are systematically older than the plant fragments. The offset between the plant fragments and bulk dates was assessed by using Undatable to separately model the median radiocarbon age of both types of material (Figure 2a). The above parameters were used for modeling. The median modeled radiocarbon age of the plant fragments was then subtracted from the bulk median. This offset (Figure 2b, solid line) was subtracted from each bulk radiocarbon age prior to age modeling (Figure 2c). The analytical uncertainty of the bulk dates was increased based on the modeled 1 σ range (Figure 2b, dashed lines) using root mean sum of squares.

Figure 2 Bulk organic matter age offset: (A) Median model radiocarbon age and 1σ range for plant fragments (red) and bulk organic matter (blue). (B) Age difference (solid line) and 1σ between plant fragments and bulk organic matter. (C) Plant fragment radiocarbon dates (red) plotted with bulk dates (blue) after subtraction of offset. Horizontal lines indicate 1σ uncertainty for use in age modeling.

Updated Marine Reservoir Age

Though C/N ratios indicate organic matter is primarily terrestrial in origin (See Results), we also re-calculated the local reservoir value (ΔR) to be compatible with the Marine20 calibration curve. This will be of use for future researchers. Uplift during two historical earthquakes, the 1932 CE Taisho and the 1703 CE Genroku subaerially exposed seafloor, creating terraces on the Miura Peninsula. Shishikura et al. (2007) performed radiocarbon dating on shells sampled from the uplifted terraces, obtaining ΔR of 82 ± 33 and 77 ± 32 for the Taisho and Genroku Earthquakes, respectively for the Marine04 calibration curve. We applied the same method as Shishikura et al. (2007) to obtain ΔR relative to Marine20. Uncertainty in ΔR is calculated by root sum of squares.

RESULTS

Dating Results

Radiocarbon dating results are shown in Table 1, corrected age and each offset are shown in Table 2. The range is about 2400 to 4000 ¹⁴C y BP. Calibrated ages range from 2.4 to 3.5 cal. ky BP. Samples generally show stratigraphically consistent ages (Figure 3), but bulk ages tend to be older than plant ages. In the model, a rapid sediment supply occurs around 2.5 cal. ky BP in the 1.8 to 2.7 m as evidenced by reliable plant sample ages. The data density is relatively high in this range, and the probability density is approximately within error. The sediment accumulation rate (SAR) in the deeper portion of the core is less than 5 cm/ky, but the maximum sedimentation rate around 2.5 cal. ky BP is over 1100 cm/ky, indicating rapid sedimentation supply.

Table 1 Radiocarbon dating result. The radiocarbon ages are from 8 plant fragments, 11 bulk samples.

Table 2 Corrected bulk radiocarbon ages and each offset from plant fragment ages (Figure 2B). Errors are instrumental error reported in Table 1 combined with the uncertainty in Figure 2B using root mean sum of squares. Total organic carbon is also reported as it was the basis for bulk sample selection, as is molar C/N and C/S ratios.

Figure 3 Age-depth model. ISL-6 age model based on the IntCal20 (Reimer et al. 2020) calibration curve. The red line is the median age produced by the model routine Undatable (Lougheed and Obrochta 2019). Dashed blue and black lines are the 1σ and 2σ ranges. Shading represents the 1st to 99th percentiles (darkest is the 50th percentile). Calibrated radiocarbon dates are shown as probability density functions (PDFs). Green are plant fragments and brown are bulk ages after correction for offset from plant fragments. Dark shading denotes the calibrated 1σ intervals, while light shading indicates the 2σ region.

Sediment Description

The upper part of the core, from 0 to 1.2 m is reclaimed landfill. A muddy silt is deposited below this from 1.2 to 3.2 m, overlying a peat deposit. The muddy silt interval typically exhibits C/N values of ∼15 or higher, though with some lower values. The typical C/S value for the muddy silt is ∼2–3. Dark-gray sand layers are repeatedly deposited below ∼1.7 m (Figure 4).

Figure 4 Results of event deposit identification. From left to right, the columnar section (Shishikura et al. 2013), core photograph taken on site, CT image, core photograph scanned with core image scanner and CNS element analysis result. The gray zones in the EA result are the locations of the identified nine event layers. C/N > 15 and C/S > 5 indicates terrestrial organic matter and freshwater, respectively. Locations of radiocarbon dates are shown on right vertical axis. Triangles are plant fragment dates, and circles are bulk organic matter dates.

Event Deposits

We identified up to nine sandy event layers. The sand grains are composed of well-rounded sand with occasional rounded gravel that are most likely derived from the beach based on the characteristics of the grains and facies. The layers are composed of a single layer or multiple, thin alternating layers, some of which have erosional surface at the base and exhibit fining upward grading. The multiple layers contain fine alternating sand and silt layers. The maximum event layer thickness is 30 cm, and the minimum is 3 cm. In the lower part of the core, there is an abrupt change in facies from light-colored muddy silt layers to peat layers.

Marine20 Local Reservoir Correction

We obtained ΔR values of –69 ± 76 and –113 ± 81 years for the from the 1923 CE Taisho and 1703 CE Genroku uplifted terraces, respectively (Table 3). The built-in Marine20 reservoir age is generally ∼150 years older than previous calibrations curves. Thus, previously positive local corrections become negative. We recommend use of the average ΔR, which is is –91 ± 111 years.

Table 3 Radiocarbon ages obtained from marine terraces uplifted during the 1923 Taisho and 1703 Genroku Earthquakes that are used to calculate local reservoir correction using the Marine20 calibration curve.

DISCUSSION

We identified the nine sand layers that were repeatedly deposited in an otherwise low-energy, back-barrier muddy lagoon environment. The relatively high C/N ratio suggest organic matter is primarily terrigenous, though with some mixing of aquatic organic matter because C/N ratios of freshly deposited organic matter derived mainly from plankton range between 6 and 9 (Biggs et al. 1983) while terrestrial plants are generally above 15 (Orem et al. 1991).

Variations in C/S ratio likely reflect changes in redox state caused by washover events breaching the barrier. A C/S ratio above five is indicative of freshwater, while values near three are typical of fine seafloor sediment deposited in an oxic environment. Values near one are commonly exhibited by sediment deposited in poorly oxygenated bays and estuaries (Berner and Raiswell 1984; Sampei et al. 1997). The transition from peat to silt deposits is consistent with increasing relative sea level in the lower part of the core. Overall, sulfur exhibits a decreasing upward trend, suggesting a diminishing marine influence over time. This could be caused by both lowered sea level, uplift, or an increasingly well-developed coastal barrier. While regional uplift is documented at ∼3.5 ka (Sakai et al. 2006), the event layers identified here were deposited ∼1000 years later and thus not correlatable to any known uplift events.

The age-depth model indicates that sedimentation occurred rapidly, particularly between 1.8 m and 2.7 m, where the sand layers are concentrated. No significant sand layer is detected above 1.8 m to the start of the reclaimed landfill at 1.2 m. The well-rounded sand is consistent with beach sand and thus likely transported by water flow from the sea. Without additional evidence for concurrent uplift, it is difficult to determine the origin of the overwash with only the information obtained in this study. If these sands were brought in from seaward, then strong currents may have surged over the barrier that had closed the remnants of the Paleo-Isumi Bay.

The age model also indicates that the bulk ages are typically older than the plant fragments by 400–600 years (Figure 2). While this value is similar to the global marine reservoir age in the most recent calibration curve (Marine20, Heaton et al. 2020), C/N ratios are inconsistent with a marine origin. Rather, the offset is more likely due to the inclusion of older organic matter derived from landward sources based on interpretation.

CONCLUSIONS

We conducted the description of core ISL-6 drilled in a former lagoon on the Isumi River lowland and radiocarbon dating for constructing of an age-depth model with the latest calibration curve and appropriate model routine. We identified nine sand layers and revealed that some of which were deposited by overwash between 2.4 and 2.5 cal. ky BP. These events appear associated with changes in redox condition in the lagoon which over time became increasingly isolated from the ocean due to either local seismic uplift, relative sea-level fall, enhanced barrier development, or a combination of each. At this point, the mechanism causing washover events is not yet constrained.