OSL dating of marine isotope stage 5e marine terrace deposits on southeastern Kii Peninsula, southwestern Japan

Abstract

Optically stimulated luminescence (OSL) dating of feldspar is one of the promising solutions towards reliable chronological constraints on the ages of Pleistocene marine terraces. Marine terraces developed along the southeastern coast of the Kii Peninsula, southwestern Japan, face a seismogenic region along the Nankai Trough. In this study, we determined the emergence age of one of these marine terraces by feldspar OSL dating of the marine terrace sediments. The target marine terrace was previously correlated with marine isotope stage (MIS) 5e based on morphostratigraphy. The samples for post-infrared infrared stimulated luminescence (pIRIR) dating were obtained from foreshore deposits of marine terraces. The first infrared stimulation temperature was fixed at 50°C (IRSL_50/150 and IRSL_50/225), and the pIRIR signals were measured at 150°C (pIRIR₁₅₀) and 225 °C (pIRIR₂₂₅). The resulting IRSL_50/150, IRSL_50/225, pIRIR₁₅₀, and pIRIR₂₂₅ ages with standard errors are 105 ± 4, 120 ± 4, 125 ± 5, and 137 ± 4 ka, respectively. The IRSL_50/225 age was consistent with the pIRIR₁₅₀ age, corresponding to MIS 5e. The IRSL_50/150 and IRIR₂₂₅ ages were younger and older than the IRSL_50/150 and pIRIR₂₂₅ ages, respectively. These results show that the Mm1 surface of the marine terrace corresponds to MIS 5e rather than MIS 5c or 7a–c, which is consistent with the chronological framework based on the morphological features of the terraces in the study area. The underestimation/overestimation of IRSL_50/150/pIRIR₂₂₅ ages can be attributed to the underestimation/overestimation of the fading rate. A more robust and comprehensive chronological framework for the marine terraces of the peninsula could be achieved by conducting OSL dating of the marine terraces above and below the MIS 5e terrace.

Graphical Abstract

1 Introduction

The amounts of uplift of marine terraces since their emergence can be estimated by measuring the height of palaeo-sea-level indicators of marine terraces relative to eustatic sea level (Burbank and Anderson 2011). Pleistocene marine terraces are common in coastal areas worldwide (Miyoshi 1983), including on the Japanese islands (Koike and Machida 2001; Tam and Yokoyama 2021). These terraces have been widely studied in Japan to understand regional uplift rates over timescales of up to a few hundred thousand years (e.g., Yoshikawa et al. 1964; Ota and Omura 1991; Matsu’ura, 2015). Information on such long-term crustal movements is required for various fields of Earth science research, such as the reconstruction of regional landscape evolution (Burbank and Anderson 2011) and the characterization of strain buildup and seismicity in mobile belts (Ikeda 2014).

Marine terraces of the last interglacial period (i.e., marine isotope stage [MIS] 5e, 119–126 ka; Spratt and Lisiecki 2016) are widely distributed along coastal areas, including the Japanese islands (Koike and Machida 2001; Tam and Yokoyama 2021). The identification and classification of marine terraces, as well as the tentative correlation of their formation/emergence ages with particular MISs, has frequently been based on morphostratigraphic features (e.g., Yoshikawa et al. 1964; Yonekura 1968). However, chronological data constraining the timing of formation/emergence are required to establish reliable correlations of marine terraces with MISs (Koike and Machida 2001). Tephrochronology has frequently been used to obtain chronological evidence for MIS 5e terraces in Japan (e.g., Miyauchi 1988; Matsu’ura et al. 2019) by establishing stratigraphic relationships between marine terrace sediments and dated tephra layers. However, in cases where dated tephra layers for the period of MIS 5e, and adjacent periods, are not found in marine terrace sediments, the correlation is less reliable or more tentative. Such cases are found in several coastal regions in Japan (Koike and Machida 2001).

The direct dating of marine terrace sediments is independent of tephrochronology and can be used to establish reliable chronological constraints on the timing of formation/emergence of marine terraces. Optically stimulated luminescence (OSL) dating of feldspar (e.g., Hütt et al. 1988; Thomsen et al. 2008; Buylaert et al. 2009, 2012) is one of the most promising methods for determining the emergence ages of marine terraces on account of the following reasons: (1) feldspar is ubiquitous in Quaternary sediments, including in Japan; (2) the OSL signal is reset during sediment transport and exposure to sunlight, and is therefore capable of constraining burial (depositional) ages; and (3) the applicable dating range extends from several thousand to several hundred thousand years, covering a wide range of the Middle–Late Pleistocene. Previous studies have used feldspar OSL dating to date sediments of MIS 5e marine terraces in Japan (Thiel et al. 2015; Ito et al. 2017; Hayashizaki 2022), as verified by cross-checking with the known tephrochronological framework.

Kii Peninsula is situated close to the Nankai Trough along which the Philippine Sea plate is subducting beneath the Eurasian plate (Fig. 1), generating a seismotectonic cycle with large earthquakes at intervals of 100–150 year (Ando 1975; Yonekura 1975; Matsu’ura 2015). The emergent topography on Kii Peninsula is thought to have been formed mainly by irregular uplift caused by the multi segment earthquake cycles at intervals of 400–600 year (Shishikura et al. 2008) rather than uplift of the seismotectonic cycle at intervals of 100–150 year (Maemoku and Tsubono 1990), as inferred from the heights and ages of Holocene uplifted sessile assemblages. However, short-term deformation trend caused by subduction zone seismic cycle is not necessary corresponding to long-term (e.g., the Pacific coast of northeast, Japan; Ikeda et al. 2012; Niwa and Sugai 2024). Therefore, to robust understanding of uplift histories and their relationship to seismic cycles, it is desirable to compare shorter-term uplift histories with longer-term. However, dated tephra layers are scarce in Pleistocene marine terrace sediments on Kii Peninsula, meaning that a tentative correlation of their emergence ages depends on morphostratigraphic approaches. In this study, we used feldspar OSL dating to constrain the ages of marine terrace deposits on the southeastern coast of Kii Peninsula, to obtain a more reliable chronology for one of the terraces. The target marine terrace has previously been correlated with MIS 5e on the basis of morphostratigraphy (Yonekura 1968, 2001).

Fig. 1

Location of the study area on Kii Peninsula. The distribution of palaeo-sea-level indicators of the marine terrace tentatively correlated with MIS 5e is based on the atlas of marine terraces in Japan (Koike and Machida 2001); data for Kii Peninsula in this atlas were generated by Yonekura (1968, 2001). Major plate boundaries around Japan Islands are delineated based on Loveless and Meade (2010). EUR the Eurasia plate, OHK the Okhotsk plate, PAC the Pacific plate, PHS the Philippine Sea plate. The base map was constructed using ASTER GDEM and digital bathymetric charts (M7000 series; Japan Hydrographic Association)

2 Regional setting

Kii Peninsula, located on the Pacific coast of southwestern Japan, is fringed by marine terraces. Yonekura (1968) clarified their evolution along the southern coast of the peninsula, and identified the Shingu area, located on the southeastern coast, as a type locality of marine terrace classification.

Yonekura (1968) classified marine terraces in the Shingu area into high (H) and low (L) terraces. These two groups of terraces were subdivided into four and three surfaces, respectively: H1, H2, H3, and H4 surfaces of the H terraces at elevations of 125, 113, 95, and 76 m above sea level (asl), and L1, L2, and L3 surfaces of the L terraces at elevations of 61, 49, and 39 m asl, respectively. Of these terraces, L1 surface can be identified almost continuously along the entire southern coast of Kii Peninsula. H1 and L1 surfaces, as well as alluvial plains in the Shingu area, have depositional surfaces that are composed of thick marine and fluvial sediments filling buried fluvial valleys. They are interpreted to have formed by fluvial incision associated with a marine regression and subsequent valley-fill sedimentation associated with a marine transgression. In contrast, H2, H3, H4, L2, and L3 surfaces are erosional in nature.

Regarding the chronology of marine terraces, Yonekura (1968, 2001) proposed that L1 surface was correlated with the last interglacial (MIS 5e), as inferred from sediment interpreted to have been deposited during the transgression prior to that forming the alluvial plains. L2 and L3 surfaces, formed by the caving L1 surface, were correlated with MIS 5c and MIS 5a, respectively. Furthermore, H2 and H3 surfaces and H1 surface were correlated with MIS 7 and MIS 9, respectively, on the basis of the chronological position of L1 surface.

3 Study area, outcrop, and sampling

The study area is located in the Shingu area, specifically on the left bank of the mouth of the Kumano River (Figs. 1 and 2A). We first constructed a terrace classification map of the study area (Fig. 2A) based on field reconnaissance and stereoscopic interpretation of U.S. military aerial photographs (1948; 1:16,000 scale). We identified four marine terrace surfaces in the study area, termed Hm, Mm1, Mm2, and Mm3 in order of descending height.

Fig. 2

A Classification map of marine terraces in the study area. The location of the studied outcrop is 33.707083°N, 135.987001°E. Contour intervals are 20 m. A base map was constructed from a digital elevation model (DEM) with 5 m mesh published by the Geospatial Information Authority of Japan (GSI). B Topographic profile of marine terraces constructed from the 5-m-mesh DEM of the GSI. C Geological columnar section and OSL ages of Mm1 surface. Si silt, fS fine sand, cS coarse sand, Gr granule. The OSL ages are compared with marine isotope stages (MISs) 5c, 5e, and 7a–c (Spratt and Lisiecki 2016; Thompson and Creveling 2021). Integrated OSL ages were calculated from all aliquots of all samples

Hm1 surface, which occurs at elevations of 85–110 m asl, is developed on the slope of the coastal mountain area but not on the coastal alluvial plain of the Kumano River. Mm1, Mm2, and Mm3 surfaces are developed on the coastal alluvial plain of the Kumano River at elevations of 60–65, 36–50, and 20 m asl, respectively. A topographic profile across Mm1 and Mm2 surfaces (Fig. 2B) reveals that (1) Mm1 surface is essentially flat; and (2) Mm2 surface, which has a terrace back scarp with a height of 10 m, tilts towards the sea with a slope of 2%. In relation to the terrace classification map in Fig. 2 of Yonekura (1968), Hm, Mm1, Mm2, and Mm3 surfaces of this study are considered to correspond to H2 and/or H3, L1, L2, and L3 surfaces of Yonekura (1968), respectively.

Sampling for feldspar OSL dating was conducted at an outcrop of sediment of Mm1 surface (Fig. 2C), corresponding to L1 surface, which was tentatively correlated with MIS 5e by Yonekura (1968, 2001). Two sedimentary facies units (the overlying Unit A and underlying Unit B) are distinguished at this site. Unit A is a ~ 5-m-thick, well-sorted, cross-bedded gravel deposit comprising disk-shaped rounded cobbles. This unit is considered to constitute the upper part of terrace Mm1 because similar gravel is exposed along the road extending along the top of the terrace (~ 60 m asl). The sedimentological features of Unit A indicate that it was formed in a high-wave-energy beach where cobble-sized gravels were transported. Unit B, which is stratigraphically located immediately below Unit A, consists of ~ 5-m-thick mud to medium-grained sand with local horizontal laminations. Intercalated gravel lenses comprising rounded pebbles (mainly 2–4 cm in diameter) occur in the upper part of this unit. The trace fossil Skolithos, which inhabits shallow-marine environments (Obata 2005), is common in the muddy layers. In accordance with these features, Unit B is interpreted to have been deposited in a foreshore environment. We obtained three samples for feldspar OSL dating from the foreshore deposits of Unit B (Fig. 2C; Table 1). Samples unexposed to light were extracted using PVC tubes (30 cm long, internal diameter of 4 cm) hammered into the part of the outcrop.

Table 1 Sampling heights, measured grain sizes and dose rate information

4 Sample preparation

The three samples were prepared at the Tono Geoscience Center under subdued-red-light conditions to avoid affecting the luminescence signals. The edge parts (< 1 cm from the ends of the PVC tube) of samples were removed to isolate the light-safe part of the samples. The removed edge parts of samples were used for measurements of water content and radioisotope contents. The remaining parts of samples were sieved to extract grains with diameters of 200–250 μm, which were then treated with hydrochloric acid (10%) and hydrogen peroxide (35%) to remove carbonate and organic matter, respectively. K-feldspar grains with a density of < 2.58 g/cm³ were separated using sodium polytungstate. In this study, we adopted the multi-grain measurement, which has been widely used for feldspar OSL dating of marine terraces in Japan (mounting sample of 2–8 mm diameter area; Thiel et al. 2015; Ito et al. 2017; Hayashizaki 2022; Tamura et al. 2022). The separated K-feldspar grains were mounted using silicon spray on the centre (about 3 mm in diameter) of a stainless-steel disk with a diameter of 9.7 mm. The resulting samples are referred to as “aliquots” below.

5 OSL instrumentation and protocols

Luminescence measurements were performed at the Tono Geoscience Center using a Risø TL/OSL-DA-20 reader equipped with infrared-light-emitting diodes for stimulation and a ⁹⁰Sr/⁹⁰Y beta source (dose rate: ~ 0.1 Gy/s). The beta source was calibrated using the Risø calibration quartz (Batch 98; Hansen et al. 2015, 2018). The dose of Risø calibration quartz was corrected based on Autzen et al. (2022). Infrared stimulated luminescence (IRSL) emissions were measured using a photomultiplier with Schott BG39 and BG3 filters.

Post-IR IRSL (pIRIR; Thomsen et al. 2008) measurements using the single-aliquot regenerative dose (SAR; Murray and Wintle 2000; Buylaert et al. 2009) protocol were adopted for feldspar OSL measurements (Fig. 3; Additional file: Table S1). The first IR stimulation temperature was fixed at 50 °C, and the pIRIR stimulation temperature was set at 150 or 225 °C; hereafter, pIRIR stimulation temperatures are denoted as a subscript (e.g., pIRIR₁₅₀). pIRIR intensities for natural and given doses (L_x) were normalized by those for the test dose (T_x) to correct the change in luminescence sensitivity during the measurement cycle. Aliquots were preheated at 180 or 250 °C for 60 s to remove unstable trapped electrons. At the end of each measurement cycle, an IR bleach was incorporated at 290 °C. The pIRIR intensity was derived from the integral of the first 7 s of signal after subtracting the average intensity of the final 20 s of signal.

Fig. 3

Typical decay curves and dose–response curve (inset) for A IRSL_50/225 and B pIRIR₂₂₅ signals of sample HTK-3

6 Equivalent dose measurements

We measured the D_e values of the three samples (HTK-1, HTK-3, and HTK4) using the pIRIR-SAR protocol (Fig. 3; Additional file: Table S1). The given beta doses were in the ranges of 20–3232 or 21–3424 Gy, and the test dose was 152 or 161 Gy. The recycling and zero-dose (recuperation) ratios were measured to assess the validity of the measurements. Measurement data were fitted using a first-order (exponential) kinetic model (e.g., Guralnik et al. 2015).

Dose recovery tests (Murray and Wintle 2003) and residual measurements were performed using the above protocol. Three aliquots from each sample were measured for each measurement. The aliquots were bleached for 4 h in sunlight on a fine day. For the dose recovery tests for the pIRIR₁₅₀ and pIRIR₂₂₅ signals, the bleached aliquots were irradiated with doses of 455 and 375 Gy, respectively. For HTK-1, -3, and -4, the residual doses of pIRIR₁₅₀ are 4 ± 1, 5 ± 0 and 4 ± 0 Gy, respectively, whereas the residual doses of pIRIR₂₂₅ are 11 ± 0, 10 ± 1 and 10 ± 0 Gy, respectively (Table 2). Dose recovery ratios were calculated after subtracting the residual doses. Although the dose recovery ratios for the pIRIR₂₂₅ signal of HTK-1 extended slightly lower than 0.90 according to the associated uncertainties (standard error), the average dose recovery ratios of all samples were within ± 10% of unity (Table 2). These results allow us to conclude that the adopted pIRIR-SAR protocol is appropriate for equivalent dose measurements for the HTK samples.

Table 2 Results of OSL measurements

7 Fading measurements

Estimation and correction of the anomalous fading effect is needed for accurate feldspar OSL dating. In this study, we applied an approach involving simulation of the unfaded and natural (faded) dose–response curve using a dimensionless recombination centre density ρ′ (Huntley 2006; Kars et al. 2008). This simulation model can provide fading corrected IRSL ages up to the saturation dose of feldspar IRSL (i.e., up to an age of ~ 400 ka; Kars et al. 2008). Therefore, this approach is appropriate for feldspar OSL dating of the studied MIS 5e marine terrace (~ 120 ka).

Fading measurements were performed for all aliquots used for equivalent dose measurements. The measurement protocol followed Auclair et al. (2003). The regenerative dose was 202 or 214 Gy, and the test dose was 101 or 107 Gy. Time intervals between the preheat and IRSL measurements were set between 0 and 48,000 s. The measured data were fitted to determine $\rho {\prime}$ values using a quantum mechanical tunnelling model (Huntley 2006; Kars et al. 2008):

$${I}_{f}={I}_{0}exp\left\{-\rho {\prime}\left[\text{ln}{\left(1.8st\right)}^{3}\right]\right\}$$

(1)

where I_f is the faded luminescence signal intensity, I₀ is the initial luminescence signal intensity, the s is escape frequency of trap (3.0 × 10¹⁵ s⁻¹), and $t$ is the delay time (s), calculated following Auclair et al. (2003).

8 Dose rate calculation

Environmental dose rates of the sampling points were determined on the basis of the natural radioisotope contents and cosmic radiation (Table 1). Contents of uranium, thorium, and rubidium were measured using inductively coupled plasma–mass spectrometry (Agilent 7700x). Potassium contents were analysed using inductively coupled plasma–optical emission spectroscopy (Agilent 5110). Water contents were calculated on the basis of the weights of natural samples. As past changes in radioisotope and water contents are unknown, relative errors of 10% were applied to these measured values, and disequilibrium in the decay chains was not considered. Environmental dose rates were calculated using DRAC software (Additional File: Table S2; Durcan et al. 2015). The alpha and beta dose attenuation factors were from Bell (1980) and Mejdahl (1979), respectively. The a-value used was 0.15 ± 0.05 (Balescu and Lamothe 1994). For the internal dose rate, the potassium content of K-feldspar was estimated at 12.5% (Huntley and Brail, 1997), and the uranium, thorium, and rubidium contents of K-feldspar were assumed to be zero. Cosmic dose rates were calculated using the equations of Prescott and Hutton (1994).

9 Infrared stimulated luminescence ages

IRSL₅₀ signals for pIRIR₁₅₀ and pIRIR₂₂₅ were also measured and analysed. The IRSL₅₀ signals for pIRIR₁₅₀ and pIRIR₂₂₅ are hereafter referred to as “IRSL_50/150” and “IRSL_50/225,” respectively. IRSL₅₀ data are summarized in Additional file: Figures S1–S2 and Tables S3–S4.

The dose recovery ratios for IRSL_50/150 signals were within ± 10% of unity (Table 2). In contrast, the dose recovery ratios for IRSL_50/225 signals for HTK-1 and -3 are lower than 0.90. The difference in dose recovery ratios between the IRSL₅₀ signals can be attributed to differences in preheat temperatures. Although IRSL_50/225 data for HTK-1 and -3 may be less reliable than the other data, these data were also used to determine the ages. The values of g_2days for IRSL_50/150 and IRSL_50/225 are 3.87–5.16%/decade and 3.54–4.51%/decade, respectively (Table 2). Fading correction was carried out with individual $\rho {\prime}$ values of each aliquot. The fading corrected D_e values used to determine IRSL and pIRIR ages were determined using the central age model (Galbraith et al. 1999; Bailey and Arnold 2006) with the R package ‘Luminescence’ (Kreutzer et al. 2012; Fuchs et al. 2015). For HTK-1, -3, and -4, the IRSL_50/150 ages with standard errors are 101 ± 8, 109 ± 6, and 107 ± 6 ka, respectively, and the IRSL_50/225 ages are 111 ± 6, 126 ± 9, and 125 ± 8 ka, respectively (Table 2). The overdispersions of fading corrected D_e for IRSL_50/150 and IRSL_50/225 are 7–12% and 8–15%, respectively.

The samples were collected from the same sedimentary unit (Fig. 2C; Unit B), and the gap in depositional ages among the samples is considered to be a few thousand years and smaller than the errors. Thus, we also calculated the IRSL and pIRIR ages using all aliquots of all samples (i.e., 15 and 30 aliquots for IRSL_50/150 and pIRIR₁₅₀, and IRSL_50/225 and pIRIR₂₂₅, respectively; hereinafter referred to as “integrated IRSL” and “integrated pIRIR”). The fading corrected D_e and ages for integrated IRSL were calculated using the average annual dose rate of samples, which is 3.63 ± 0.08 Gy/ka (standard error). The integrated IRSL_50/150 and IRSL_50/225 ages are 105 ± 4 and 120 ± 4 ka, respectively. The overdispersions of fading corrected D_e for all IRSL_50/150 and IRSL_50/225 are 10% and 12%, respectively.

The IRSL_50/150 ages of each sample were the same within the error ranges, and the overlapped range was 103–109 ka. The IRSL_50/225 ages also overlap at 117 ka. The integrated IRSL_50/150 and IRSL_50/225 ages correspond to the overlapping IRSL_50/150 and IRSL_50/225 ages of the samples, respectively; thus, they are representative values for each sample.

10 Post-infrared infrared stimulated luminescence ages

pIRIR signals were analyzed similarly to the IRSL signals. pIRIR data are summarized in Additional file: Figures S3–S4 and Tables S5–S6. The values of g_2days, which indicate the fading rate, for pIRIR₁₅₀ and pIRIR₂₂₅ are 1.51–2.11%/decade and 1.28–1.41%/decade, respectively (Table 2). These results are consistent with the theory that the athermal stability of IRSL signals increases with increasing stimulation temperature (Li and Li 2011). These values are also within the range commonly reported for K-feldspar (e.g., Kadowaki et al. 2022; Ishii 2024).

For HTK-1, -3, and -4, the pIRIR₁₅₀ ages with standard errors are 119 ± 9, 130 ± 10, and 128 ± 7 ka, respectively, and the pIRIR₂₂₅ ages are 131 ± 7, 133 ± 7, and 146 ± 9 ka, respectively (Table 2). The overdispersions of fading corrected D_e for pIRIR₁₅₀ and pIRIR₂₂₅ are 3–13% and 6–9%, respectively. The integrated pIRIR₁₅₀ and pIRIR₂₂₅ ages are 125 ± 5 and 137 ± 4 ka, respectively. The overdispersions of fading corrected D_e for integrated pIRIR₁₅₀ and pIRIR₂₂₅ are 11% and 8%, respectively.

The pIRIR₁₅₀ ages of each sample were the same within the error ranges, and the overlapped ranges were 121–128 ka. The pIRIR₂₂₅ ages of each sample also overlap at 137–138 ka. The integrated pIRIR₁₅₀ and pIRIR₂₂₅ ages correspond to the overlapping pIRIR₁₅₀ and pIRIR₂₂₅ ages of the samples, respectively; thus, they are representative values of the samples.

11 Discussion and conclusions

The above results show that the integrated IRSL and pIRIR ages are representative values for the IRSL and pIRIR ages of each sample; therefore, we discuss the integrated IRSL and pIRIR ages below. The integrated IRSL_50/150, IRSL_50/225, pIRIR₁₅₀ and pIRIR₂₂₅ ages are 105 ± 4, 120 ± 4, 125 ± 5, and 137 ± 4 ka, respectively (Fig. 2 and Table 2). The integrated IRSL_50/225 age was consistent with integrated pIRIR₁₅₀ age, which correspond to MIS 5e (119–126 ka; Spratt and Lisiecki 2016). The integrated IRSL_50/150 was younger than the integrated IRSL_50/225 and pIRIR₁₅₀ ages and between MIS 5c (90–98 ka; Thompson and Creveling 2021) and 5e. The integrated pIRIR₂₂₅ age is slightly older than the integrated IRSL_50/225 and pIRIR₁₅₀ ages. However, the samples were obtained from marine sediments during the highstand. Thus, the integrated pIRIR₂₂₅ age indicates that the formation period of the marine terrace is MIS 5e.

From these results, we conclude that the Mm1 surface of the marine terrace corresponds to MIS 5e rather than MIS 5c or 7a–c (197–214 ka; Spratt and Lisiecki 2016), which is consistent with the chronological framework of Yonekura (1968, 2001) based on the morphological features of terraces in the study area (Fig. 2). The integrated IRSL_50/150 and pIRIR₂₂₅ ages were slightly younger and older than MIS 5e, respectively. Possible cause of these younger/older ages are overestimation/underestimation of the annual dose rate and underestimation/overestimation of the fading rate. Additionally, partial bleaching can be the cause of overestimated age. Although the IRSL and pIRIR ages were calculated using the same annual dose rate, the IRSL_50/150 and pIRIR₂₂₅ ages are younger and older than the IRSL_50/225 and pIRIR₁₅₀ ages beyond the error range. The overdispersions of fading corrected D_e for each and all samples are ~ 10%, indicating that the samples are well-bleached (Mahan and Dewit 2019). Therefore, the annual dose rate and partial bleaching are unlikely to be the cause of younger and older ages. The fading rates of the pIRIR₂₂₅ are 1.28–1.41%/decade, and such low rates (< 1–1.5%/decade) may have a negligible effect on measured ages (Buylaert et al. 2012). The fading uncorrected pIRIR₂₂₅ age for all aliquots of all samples was 110 ± 3 ka based on measured D_e calculated using the central age model. This age is slightly younger than MIS 5e. From this result, the possible range of integrated pIRIR₂₂₅ age is from 110 ± 3 to 137 ± 4 ka, which supports the conclusion that the Mm1 surface corresponds to MIS 5e rather than MIS 5c and 7. These results demonstrate that the measured fading rate of the pIRIR₂₂₅ may be overestimated. This cause may be that the measured fading rate of pIRIR₂₂₅ is affected by OSL measurement error significantly because the fading rate is low. On the other hand, these results imply that fading correction may be required for more accurate dating even if the fading rate is low. Although the IRSL_50/150 ages were corrected for fading significantly because the fading rate of IRSL_50/150 was high (3.87–5.16%/decade), its validity has not been evaluated. Therefore, more accurate IRSL_50/150 and pIRIR₂₂₅ dating would be needed based on further fading measurements, such as longer delay measurements on week and/or month scales. However, this study aims to correlate marine terraces with marine isotope stages, and the ages generated indicate that the target marine terrace corresponds to MIS 5e. Therefore, further fading measurement was not performed in this study.

In conclusion, we correlated the target marine terrace on the southeastern coast of the Kii Peninsula with MIS 5e using OSL dating of K-feldspar from terrace deposits. A more robust and comprehensive chronological framework for the marine terraces of the peninsula could be achieved in future work involving K-feldspar OSL dating of marine terraces above and below the MIS 5e terrace, as the applicable dating range of this method extends from several thousand to several hundred thousand years. However, pIRIR ages commonly have relative errors of ~ 10%. Thus, to distinguish the corresponding marine isotope substages using pIRIR dating, it is necessary to evaluate the reliability of the pIRIR age, e.g., by cross-checking the pIRIR age at several stimulation temperatures. It is also necessary to evaluate the possible effects of the annual dose rate, fading rate, and partial bleaching on pIRIR ages.