Last interglacial sea-level proxies in the Korean Peninsula

. Like most of the world’s coastlines, the Korean Peninsula experienced higher-than-present sea levels during the Last Interglacial (LIG) otherwise known as Marine Isotope Stage (MIS) 5e. However, the expression of that highstand in the geological record differs across the eastern and western Korean Peninsula. The tectonically active east coast of the Korean Peninsula is characterized by broad uplifted marine terraces, while the stable west coast is characterized by tidal flats and rias. 15 In this study, we used a standardized database template to review and extract the existing constraints on LIG sea levels along both the east and west coasts of the Korean Peninsula. A total of 62 LIG constraining data points were compiled including 34 sea-level indicators, 22 marine limiting records, and 6 terrestrial limiting records. The ages from these data points are based on 61 optically stimulated luminescence (OSL) measurements and 1 paleomagnetic-based age. Along the uplifted east coast, LIG sea-level indicators based on marine terraces are at elevations ranging from +9 to +32 m. The uplifted marine terraces are 20 cut or otherwise deformed by faults developed under a compressional regime due to backarc closing of the East Sea since the early Pliocene. As a result, tectonic uplift likely has affected the elevations of the east coast LIG shorelines. In contrast, LIG sea-level records on the west coast of the Korean Peninsula are found at heights of between +3 and +6 m and include marine and terrestrial elevation-limiting records as well as true sea-level indicators. The LIG sea-level constraints along the west coast of the Korean Peninsula are likely unaffected by vertical movement or experienced minor subsidence during the Quaternary.


Bathymetric contour inveral is in meters.
The East Sea is a semi-enclosed marginal sea with an average water depth of about 1350 m and a maximum water depth of about 3700 m west of the Japanese Island of Hokkaido (Chough et al., 2000). The eastern continental shelf is narrow and rapidly transitions into a deep basin (Fig. 1). Along the east coast of Korea, the tides are microtidal ranging from 10 to 30 cm, based on tide-gauges around the coast and satellite altimeter-derived (TOPEX/Poseidon) sea-surface heights (Nam et al., 2004;80 2015).
The Yellow Sea is a semi-enclosed shallow epicontinental sea with an average water depth of about 55 m and a maximum depth nearing 100 m at its southeastern margin ( Fig. 1; Chough et al., 2000). The seafloor of the Yellow Sea is flat and broad ( Fig. 1). Based on tide-gauge observations, the tides are semidiurnal, and the tidal range varies from mesotidal (2-4 m) along the open coast to macrotidal (>4 m) within embayments (Oh and Lee, 1998;Cummings et al., 2016). Presently, high-tide 85 beaches, multiple swash bars, cheniers, and sandy tidal-flat deposits are found along the open coast, while mud-rich tidal deposits dominate the embayed coastlines (Figs. 7,8).
Tectonically, the Yellow Sea basins formed due to both India-Eurasia and Pacific-Eurasia plate interactions, which led to repeated extension and rifting since the Late Mesozoic (Ren et al., 2002). Extension-driven regional subsidence formed marginal basins in the Yellow Sea (Watson et al., 1987). Since the Late Miocene, the Yellow Sea is thought to have undergone 90 little to no tectonic subsidence, and, at the least, has not experienced uplift (Chough et al., 2000;Li et al., 2016). In the eastern Yellow Sea, the Korean Peninsula is characterized by rias and over 3000 islands along its western and southern coasts (Chough, 2013). The seafloor deepens progressively to the southeast along the NNW-SSE axis of the former Late Pleistocene lowstand shorelines ( Fig. 1; Chough et al., 2000;2004). Eustatic sea-level fluctuations during the Quaternary had a great effect on sedimentation in the Yellow Sea (Chough et al., 2000;Jin et al., 2002;Shinn et al., 2007;Yoo et al., 2016). Korean and Chinese 95 onshore and offshore drill cores have revealed alternating terrestrial and shallow marine deposits formed during repeated Pleistocene transgressions and regressions (Li et al., 1991;Marsset et al., 1996;Jin et al., 2002;Chang et al., 2014;Li et al, 2016;S.H. Hong et al., 2019;H.H. Yoon et al., 2021). The present Yellow Sea formed during a large-scale Holocene transgression of the pre-Holocene terrestrial lowlands between Korea and China (Chough et al., 2000). Many vibracores and drill cores along the west coast of Korea sample those Holocene transgressive intertidal deposits unconformably overlying the 100 pre-Holocene semi-consolidated, oxidized floodplain deposits, forming a retrograding, coarsening-upward succession (Y.A. Park et al. 1998;Y.H. Kim et al., 1999;Lim et al. 2004;K. Choi and Kim 2006;Chang et al., 2014;H.H. Yoon et al., 2021).

Overview of previous studies on the Korean Peninsula
We divided the east coast of the southern Korean Peninsula into two regions: the northern region extending from Gangneung through Uljin and to Yeongdeok (38° to 36.3°N) and the southern region extending from near Pohang to Ulju (36.3° to 35°N) 105 ( Fig. 2). Within the southern region, S.W. Kim (1973) was the first to publish 14 C ages from the marine terraces of Korea. He divided the marine terraces into 6 elevation groups ranging from 3 to 130 m above mean sea level (MSL) and suggested the highest two groups may have formed during a Pleistocene interglacial period. A separate study across much of the same region    Table 3. Bathymetric contour inveral is in meters. Seomseok river, located in Gangneung (Fig. 2), using amino acid racemization dating of peats (S.G. Choi, 1993). In the southern region, another LIG age (125 ka B.P.) was obtained from a separate fluvial terrace, located near the Pohang shoreline ( Fig. 2), using the same method (S.G. Choi, 1996). However, due to inadequate descriptions of the sampled materials and age dating techniques within the original papers, the two ages were not included in our dataset shown in Table 3 or in the openaccess spreadsheet (Ryang and Simms, 2021). Since this initial work, most research has focused not on numerical dating, but 150 documenting the elevations, sedimentary characteristics, and stratigraphy of deposits overlying the marine terraces along the eastern shorelines (e.g., S.G. Choi, 1995aChoi, , 1995bChoi, , 2016aChoi, , 2016bChoi, , 2018Choi, , 2019S.G. Choi and Chang, 2019;Yoon, 1996, 2020;Hwang et al., 2012;S.O. Yoon et al., 1999S.O. Yoon et al., , 2003S.O. Yoon et al., , 2014. In the 2000s, OSL dating started to be applied to the sandy deposits overlying the marine terraces of the eastern Korean Peninsula. The first numerical age derived by OSL dating of paleo-beach sediments overlying a marine terrace was obtained 155 near the Ulju shoreline within the southern region (dated to 112±7 ka) (J.H. Choi et al., 2003). Other studies soon added more OSL ages on the LIG marine terraces along the eastern shoreline of the Korean Peninsula (J.H. Choi, 2004;J.W. Kim et al., 2005aJ.W. Kim et al., , 2007aJ.W. Kim et al., , 2007bS.J. Choi et al., 2008;S.C. Hong, 2014;S.Y. Lee et al., 2015;S.J. Choi, 2016;C.S. Park et al., 2017; G.Y. Lee and Park, 2019b). Shim (2006) applied paleomagnetic analysis to outcrop sections of paleo-beach sand overlying a marine terrace along the 160 Pohang shoreline within the southern region (Fig. 2), which was interpreted as MIS 5e deposits (114 to 120 ka).
Tephrochronology and tephrostratigraphic correlation methods have also been used to suggest an MIS 5e age on deposits overlying a marine terrace at an elevation 15-20 m above MSL in the Gyeongju area of the southern region ( Fig. 2; Inoue et al., 2002), but the age interpretation was quickly challenged because the dated deposits may have been reworked from sediments of a higher terrace (S.J. Choi, 2003Choi, , 2009S.J. Choi et al., 2008). 165 Other efforts have been underway to identify LIG marine terraces or shoreline deposits along the southern and western shorelines of the Korean Peninsula (Lee G.Y. and Park, 2006;S.G. Choi, 2006;J.H. Yang, 2008J.H. Yang, , 2011J.H. Yang et al., 2013;S.O. Yoon et al., 2015;W.J. Shin et al., 2016;J.S. Oh, 2018;Lee G.Y. and Park, 2018). However, most of these studies relied on elevation or stratigraphic correlation in the absence of numerical ages or analyzed the alluvial deposits overlying the shoreline deposits. Recently, LIG shoreline deposits were discovered on the basis of OSL ages in the Sacheon, Wando, and 170 Haenam areas (Figs. 2, 7;D.Y. Yang et al., 2016;J.Y. Shin, 2018;G.Y. Lee andPark, 2018, 2019a;W.J. Shin et al., 2019).
Arguably the best sea-level constraints from these studies are those of W.J. Shin et al. (2019). W.J. Shin et al. (2019) identified paleo-intertidal/beach deposits composed of alternating gravel and sand beds with shell fragments along the southwest Korean shoreline near the Wando shoreline (Fig. 7). Their OSL ages were obtained from intertidal/beach deposits and dated to between 115.9 ± 9.7 and 127.5 ± 8.5 ka at an elevation of 3.5 to 2.8 m above MSL (W.J. Shin et al., 2019). 175

Datums
The Korean geodetic horizontal point is based on the World Geodetic Reference System (ITRF2000 & GRS80). This horizontal and vertical datum has been well managed for the national territory by the National Geographic Information Institute (2021). 180 The Korean official vertical datum is based on mean sea level within Incheon Bay from 1913to 1916. Using this datum, the institute installed a series of nationally unified control points and benchmarks across the entire Korean peninsula and territorial islands. All elevation data in this paper were measured using this national geodetic system.

Elevation measurements
All surveys include GPS or DGPS surveying methods. Most surveys also used an electro-optical distance measuring system 185 (total station) to determine the elevation at specific points from a local benchmark of the Korean official vertical datum. Some locations were determined by comparing the GPS coordinates of the sampling point with a 1:5000 topographic map. Marine surveys used a virtual reference station linked to a GPS (VRS-GPS) or DGPS and a high-resolution echosounder aboard the drilling ship.

Coring 190
Cores were acquired using a hydraulic-powered drill on a barge or a vibrocorer deployed from a ship. Sediment sampling was performed using a standard thin-walled 55-mm diameter tube sampler. In the laboratory, cores were split lengthwise, described, photographed, and sampled. Core descriptions were based on sediment characteristics, including color, lithology, texture, grain size, and structure.

Optically stimulated luminescence
Sixty-one of the 62 numerical ages in this dataset were acquired using OSL. Fifty-one OSL age measurements were conducted on coarse-grained sand or fine-grained silt quartz separates. Seven additional ages were obtained from K-feldspar minerals (IRSL, Infrared Stimulated Luminescence) isolated from coarse-grained sand and another 3 ages were obtained from Kfeldspar minerals of gravel surfaces (rock surface OSL dating) (e.g., J.H. Choi et al., 2004;S.C. Hong et al., 2013;S.C. Hong, 200 2014S.C. Hong, 200 , 2016. All OSL ages were conducted using the single-aliquot regenerative-dose (SAR) procedure (Murray and Wintle 2000) with error ranges listed as either ± 1σ SE or ± 2σ SE in accordance with the original studies.

Paleomagnetism
For the paleomagnetic study, sediment samples were collected almost continuously from an outcrop using non-magnetic 22 × 22 × 22 mm 3 plastic cubes. The remanent magnetic moment of each sample was measured by a high-temperature 205 superconducting magnetometer (F.I.T Messtechnik GmbH, HSM2) at Hanyang University (Shim, 2006). Alternating field demagnetization (Molspin Co.) and anhysteric/isothermal remanent magnetization experiments were performed to isolate the characteristic remanent magnetization of each sample (Shim, 2006). Magnetic susceptibility was measured by a susceptibility meter (Bartington Co., MS2F) for determining the types and amounts of ferrimagnetic minerals in the sediments.

Paleo sea level from marine terraces
The approach used to reconstruct paleo RSLs is in part dependent on the landform used (Tables 1, 2). The most commonly applied approach to reconstructing past RSLs from marine terraces is to use the elevation of the shoreline angle (SA in Fig. 6; "inner edge" of Bradley and Griggs, 1976 or junction between the marine platform and the paleo sea cliff of Muhs et al., 1990) as a proxy for paleo sea levels. Unfortunately for most of the data collected to date across the marine terraces of the east coast 215 of the Korean Peninsula, the shoreline angle (SA) elevation is unknown because the paleo-sea cliffs are covered by overlying deposits (Fig. 6). On the only 2 among the 13 subsites along the east coast ( Fig. 2), the SA elevations of the marine terraces were measured in the original studies (Table 2; Fig. 5). In the case of an unknown shoreline angle elevation but where a paleo wave-cut platform elevation (PWPE) and a paleo seacliff onlap elevation (PSOE) were measured (Table 2; Fig. 6), we expressed the estimated shoreline angle elevation (SAe) as the mid-point elevation between the PWPE and the PSOE with an 220 error range (δSAe) of ½ that elevation difference according to the following equations: This approach allowed us to estimate RSLs from three additional subsites where the PWPE and the PSOE were measured in the field (Table 2; Fig. 6). 225 RSL is calculated from all sea-level indicators using the following equations: Where E represents the elevation of the sea-level indicator, RWL represents the reference water level, and δRSL accounts for the uncertainties in the paleo RSL (Rovere et al., 2016). Furthermore, where Ul represents the upper limit of the modern analogue landform's elevation, Ll represents the lower limit of the modern analogue landform's elevation, SWSH represents storm wave swash height, and db represents the breaking depth. Each of these values was obtained from local sources (Table 1). Breaking depth was approximated using the following equation: where Hs represents an average significant wave height during one year with the constant 0.78 commonly used for wave 235 breaking criteria on a smooth, flat slope (Table 1; U.S. Army Corps of Engineers, 1984;Rovere et al., 2016). The uncertainty in the paleo RSL (δRSL) was determined via: where Ee and IR represent an error in the elevation measurement (standard deviation) and the indicative range, respectively (Rovere et al., 2016). For the case of marine terraces whose shoreline angle was measured or estimated, their RSL equivalent 240 (Paleo SA RSL) would be: In cases along the east coast of the Korean Peninsula where both the elevation of the shoreline angle or paleo wave-cut platform are unknown, we base our estimates of RSL by treating the dated deposits overlying the marine terraces as beach sands using Eq. 8 such that 245 Paleo RSL at marine-terrace data points (Landform Type 1 in Table 1 Similar to the earlier estimates, we also used the local storm wave swash height (SWSH), average significant wave heights (Hs), and breaking depths (db) to calculate 30 paleo RSLs from the elevations of the beach sands overlying the marine terraces along the east coast (Table 3). 250

Paleo sea level calculated from beach and tidal deposits
Beach deposits and beach rock are also a reliable RSL marker because the formative zone is in close proximity to sea levels (Mauz et al., 2015). Five paleo RSLs of beach deposits (Landform Type 2 in Table 1) and an additional five paleo RSLs of beach rock (Landform Type 3 in Table 1) were used to determine paleo RSLs from LIG-aged deposits along the western and southern Korean coast. RSLs from beach rock were calculated using the following equations: 265 Paleo RSL at beach-deposit data points (Landform Type 2 in Table 1) where (ob) represents the ordinary berm height, which was estimated using the average significant wave height over one year (Hs) and MHHW represents the mean higher high water heights (Table 1; Mayer and Kriebel, 1994;Rovere et al., 2016). For beach rocks, we used a similar expression: 270 Paleo RSL at beach-rock data points (Landform Type 3 in Table 1) where (sz) represents the elevation of the top of the spray zone, which was calculated using the average significant wave height over one year (Hs) ( Table 1; Rovere et al., 2016).
Twenty-two OSL-dated clastic tidal-flat deposits from nearshore cores were also used to estimate paleo RSLs based on Eq. 275 (11).
Paleo RSL at tidal-flat clastic data points (Landform Type 4 in Table 1) where MLLW represents mean lower low water heights (Table 1).

Relative Sea-level Indicators 280
In the following sections, we discuss each characteristic of the RSL indicators from the Korean Peninsula, identified by thier 'WALIS RSL ID' in the text, which have been entered into the WALIS database. The ID number corresponds with the WALIS database identification numbers. Similarly, we use 'WALIS LUM ID' followed by a number to reference an optically stimulated luminescence (LUM) age within the database.
The east coast of the southern Korean Peninsula was divided into two regions, a northern and a southern region, based on 285 latitude. The northern region encompasses the area from Gangneung through Uljin and the Yeongdeok area (38°N to 36.3°N) while the southern region encompasses the region between Pohang and Ulju (36.3°N to 35°N) (Fig. 2). Along the west and southwest coast of the southern Korean Peninsula, the sea-level data and indicators were divided into two groups, those found onshore and those found within the nearshore.

Gangneung area
Two areas have been studied within the Gangneung area: Gangneung-Saemokee and Gangneung-Anin (Fig. 2). The marine deposits overlying the paleo wave-cut platform of the marine terrace at Gangneung-Saemokee are found 27~31 m above MSL and contain rounded cobbles and some sand deposits of paleo beach origin (S.C. Hong, 2014). At Subsite Gangneung-Saemokee, two quartz OSL ages were interpreted as minimum ages of >85 ka and >92 ka. These deposits appear to exceed 295 the upper age limit of the methodology because the signal is saturated (e.g., Rhodes, 2011). Fortunately, two IRSL ages of 128.3±24.5 ka (WALIS LUM ID #432) and 124.1±25.3 ka (WALIS LUM ID #433) and one cobble surface OSL age of 133.7±13.9 ka (WALIS LUM ID #434) were obtained from these terrace deposits (Table 3; S.C. Hong, 2014 (Table 3; WALIS RSL ID #4009, 4010). 300 Gangneung-Anin is located 23 m above MSL and consists of beach sand and gravel deposits overlying a paleo wave-cut platform. The two ages from the beach sand were 117±6 ka (WALIS LUM ID #435) and 129±8 ka (WALIS LUM ID #436) (Table 3; S.Y. . At this site, although the paleo shoreline angle elevation is unknown, we estimated the SA elevation to be approximately 27.5±2.5 m above MSL using a value of 25 m for PWPE and 30 m for PSOE in Eqs. (1) and (2) (  Fig. 6). 305

Donghae area
Donghae-Eodal-dong is located 26 m above MSL (Fig. 2) and consists of beach sand and pebble deposits overlying a paleo wave-cut platform. The three samples from the beach sand were analyzed using paired OSL and IRSL methods. The three OSL/IRSL age sets from the beach sand were 126.   Fig. 6).

Youngduk area
Subsite Youngdeok-Geumgok-ri is located 20-25 m above MSL and contains rounded cobbles, pebbles, and sand deposits overlying a paleo wave-cut platform (Figs. 2, 3; S.C. Hong, 2014). Two OSL ages from quartz sand in the overlying marine deposits were interpreted as minimum ages of >44 ka and >41 ka. They appear to exceed the upper age limit of the 325 methodology in these sediments as the traps are saturated (e.g., Rhodes, 2011). Two additional IRSL ages of 124.5±25.3 ka (WALIS LUM ID #444) and 122.1±24.9 ka (WALIS LUM ID #445) were obtained from the sand deposits (Table 3; S.C. Hong, 2014). Without a PWPE we estimated RSL using Eq. (8) based on its use as a beach RSL indicator (Landform Type 1 in Table 1). For that calculation, we arrives at RSL estimates of 19.4±1.7 m and 22.1±1.7 m, respectively (Table 3; WALIS   RSL ID #4018, 4019). 330
Subsite Pohang-Yonghan-1 (silica mine) is located on a marine terrace 32 m above MSL and contains alternating wellrounded pebble and sand beds (~1.2 m thick) (J.W. Kim et al., 2005a). A single OSL age of 123±9 ka (WALIS LUM ID #448) was obtained from the sand deposits (J.W. Kim et al., 2005a). The paleo RSLs of this sample yielded 30.2±1.7 m above MSL using Eq. (8) ( Table 3;  Subsite Pohang-Masan-ri is located 23 m above MSL and contains paleo-beach sand beds (~0.24 m thick) and overlying aeolian sand beds (~5 m thick) above a paleo wave-cut platform (J.W. Kim et al., 2005b). The significance of the terraces in this region is also discussed by Thompson and Creveling (2021) who focus more on the MIS 5c and MIS 5a ages from this 355 site and adjacent areas. The four OSL ages from the paleo-beach sand sediments were 119±8 ka (WALIS LUM ID #451), 111±5 ka (WALIS LUM ID #452), 116±7 ka (WALIS LUM ID #453), and 107±8 ka (WALIS LUM ID #454) (Table 3; J.W. Kim et al., 2005b). Using Eq. (8), the paleo RSLs of these samples without a PWPE yielded 21.2±1.7 m above MSL (Table 3; WALIS RSL ID #4025). A total of 158 sediment samples were also collected almost continuously from approximately 3.8 m of the Masanri (MS) outcrop section for a paleomagnetic study (Shim, 2006). Remanent magnetic moment, alternating field 360 demagnetization, anhysteric/isothermal remanent magnetization, and magnetic susceptibility of each sample were measured to isolate characteristic remanent magnetization (Shim, 2006). The global Blake Excursion Event was discovered in the MS section on the Masan-ri marine terrace in the northern Pohang area (Fig. 2). The elevation of this outcrop section is 22 m above MSL. Considering the Blake Event (111.8 to 117.1 ka), the paleomagnetic age of the paleo-beach sediments overlying the marine terrace suggests a numerical age of 117.6±2.7 ka with an error range of between 114.9 and 120.2 ka (Table 3; Shim, 365 2006). At this site, the shoreline angle elevation is estimated to be 23.5±1.5 m above MSL using 22 m for PWPE and 25 m for PSOE (Table 2; Fig. 6).
Subsite Pohang-Hajeong-ri is located 35 m above MSL, and the section crops out in the footwall of the Hajeong fault. The section contains a wedge-shaped mix (98 to 20 cm thick) of rounded pebbles, sand, and angular alluvial pebbles overlying a marine terrace (S.J. Choi, 2016). The single OSL age from the sand was 128±12 ka (WALIS LUM ID #455) ( Table 3; S.J. 370 Choi, 2016), which is interpreted as a terrestrial limiting data point.

Gyeongju area
Subsite Gyeongju-Jinri is located on a marine terrace 36 m above MSL (Fig. 2) and contains thin (~ 15 cm thick) paleo-beach gravels underlying fine-grained sand beds (~0.5 m thick) (J.W. Kim et al., 2007a). The two OSL ages from the sand deposits were 116±6 ka (WALIS LUM ID #456) and 126±10 ka (WALIS LUM ID #457) (Table 3; J.W. Kim et al., 2007a). Using Eq. 375 (8), the paleo RSLs of these samples without a PWPE yielded 32.3±2.6 m above MSL (Table 3; WALIS RSL ID #4027). In this area, a land seismic survey was conducted to identify the elevation of the buried paleo wave-cut platform and shoreline angle. Based on seismic velocities of 600 m/s for the overlying deposits and 2100 m/s for the wave-cut platform, the elevation of the paleo wave-cut platform ranges from 27 to 29 m above MSL with a slope of 1.5° ( Fig. 5; J.W. Kim et al., 2007a). The shoreline angle elevation is estimated to be 28±1 m above MSL (Table 2). 380

Sacheon area
The Sacheon area is located in the south-facing Korean Peninsula straddling the eastern and western coasts (Fig. 2). Subsite Sacheon-Daepo-dong lies at ~6 m above MSL and unlike the LIG shorelines of the eastern Korean Peninsula is not marked by a distinctive marine terrace geomorphology (J.Y. Shin and Hong, 2018). The paleo-shoreline deposits (2-3.5 m thick) are characterized by clast-supported well-rounded cobble and pebble deposits with little to no matrix. Their sedimentary 400 characteristics are similar to the sandy gravel bars of the modern upper tidal flats in the region (J.Y. Shin and Hong, 2018).

Onshore paleo-beach and terrestrial deposits of the west coast
At the 6 m locality, 4 m of sand and well-rounded cobbles and boulders are thought to represent a paleo-beach deposit 415 (G.R. Lee and Park, 2018). The single age from the sand deposits was 108±18 ka (WALIS LUM ID #464) (Table 3; G.R. Lee and Park, 2018and Park, , 2019a. Based on Eq. (9), we arrive at a RSL estimate of 5.8±0.9 m (Table 3; WALIS RSL ID #4032).

Buan area 430
The Buan area on the west coast of the Korean Peninsula (Fig. 7) has two localities, Buan-Daehang-ri and Buan-Mapo-ri, located 11 m and 9 m above MSL, respectively (G.R. Lee and Park, 2018). At the two localities, the outcrop exposes a mixture of silt, sand, and well-rounded pebbles and cobbles, approximately 2 m thick (G.R. Lee and Park, 2018). These deposits were originally interpreted as alluvial or marine sediments (G.R. Lee and Park, 2018) with little distinctive sedimentological or geomorphologic evidence for a marine origin. The two OSL ages from the sandy sediments were 112±24 ka (WALIS LUM 435 ID #468) and 130±20 ka (WALIS LUM ID #469) (Table 3; G.R. Lee andPark, 2018, 2019a), which are interpreted as terrestrial limiting records.

Seocheon area
Core DSR-C03 was collected onshore of a modern high-tide beach near Seocheon on the west coast of the Korean Peninsula 440 (Fig. 7;Chang et al., 2017). MIS 5e deposits occur lower than about 1.5 m below MSL (Fig. 9). These occurrences of LIG deposits at approximately -1.5 m in elevation are the shallowest LIG deposits found in the core sediments acquired on the west coast of the Korean Peninsula. Four depositional units labeled as Units A, B1, B2, and C, in ascending stratigraphic order, were identified based on sediment texture and structures (Figs. 9, 10).

Unit A 445
The lowermost Unit A is mud-rich with sandy beds showing an overall coarsening-upward trend and attains a thickness of up to 12.5 m in core DSR-C03 (Fig. 9). This unit is comprised of lower muddy beds and upper more sandy beds (Fig. 9), both of which contain tidal couplets of mud and sand (Fig. 10D, E). The lower muddy beds are characterized by intense bioturbation, no shell or rare shell fragments, and lenticular to wavy tidal bedding, while the upper more sandy beds show moderate bioturbation, few shell fragments, and flaser to wavy and lenticular tidal bedding. Considering the presence of mud-sand tidal 450 bedding and muddy nature of the sediments with no shell or rare shell fragments, the lower muddy beds are interpreted as being deposited in upper mud-flat or salt-marsh environments (e.g., Klein, 1985;Dalrymple et al., 1992;Dalrymple, 2010).
The upper moderately bioturbated more sandy beds are suggestive of being deposited in an intertidal mixed-flat environment.
The uppermost part of Unit A resembles the pre-Holocene semi-consolidated oxidized beds commonly occurring on top of tidal-flat sequences throughout the west coast of Korea (e.g., Y.H. Kim et al., 1999;Lim et al. 2004;H.H. Yoon et al., 2021). 455 The two OSL ages of 116±10 ka and 108±8 ka suggest that the tidal deposit formed during the MIS 5 period (Table 3; Fig. 9; e.g., Chang et al. 2014;Baek et al. 2017;H.H. Yoon et al., 2021). Two OSL ages were obtained from Unit A in core DSR-C03. The one MIS 5e OSL age obtained from a coarse quartz sand bed was 116±10 ka (WALIS LUM ID #470) at -3.01 m below MSL. Another MIS 5e/5c OSL age of 108±8 ka (WALIS LUM ID #471) was obtained at a core depth of -4.95 m below MSL (Table 3; Chang et al., 2017). The shallowest occurrence of LIG deposits in Unit A represents the highest among marine limiting records of LIG RSL. 475 OSL-dated clastic tidal-flat deposits were used as paleo RSL markers, identified based on sediment texture and sedimentary structures, and interpreted as tidal environments (e.g., Mauz and Bungenstock, 2007). Although some interpretations are of salt-marsh deposits or intertidal mixed-flat deposits, no microfaunal work has been conducted on the deposits to confirm their marsh and mixed-flat interpretations and thus tightly constrained relationship with paleo-sea levels as is normally done in other salt-marsh or intertidal based RSL studies (e.g., Shennan et al., 2015). Thus the tidal-flat data points are treated as marine 480 limiting records because their relationship to past tidal-datums is not constrained by biological indicators.

Unit B
Unit B is approximately 2 m thick and is separated from the underlying Unit A by a distinct erosional boundary (Figs. 9, 10C).
This unit can be further subdivided into two subunits: the lower unit B1 consists of completely oxidized gravel and sand devoid of muddy sediments and the upper unit B2 contains muddy sediments with less oxidized gravel and sand sediments compared 485 to unit B1. The sediments are composed of poorly sorted gravel and coarse sand (Fig. 10B, C). These units contain neither shells, foraminifers nor any other marine indicators. Muddy deposits of Unit B2 are commonly associated with coarser sand layers and contain wood fragments, fine peat, and rootlets. The erosional unconformities at the base of Units B1 and B2 and the presence of oxidized sediments point to an extended period of subaerial exposure. The contact is interpreted as a sequence boundary (Figs. 9, 10B). The coarse-grained sediments without a muddy matrix probably originated from gravel bars or 490 channel lag deposits in a fluvial environment (e.g., K. Choi and Kim 2006;Chang et al. 2014;Baek et al. 2017).

Unit C
Unit C is characterized by an upper massive sand bed and a lower crudely stratified sand bed showing a fining-upward trend ( Fig. 9). The sand beds are mud-free and contain shell-rich beds (Fig. 10A). The upper sand beds are a few decimeters thick, structureless, and comprised of very poorly sorted medium to coarse yellowish sand. The lower crudely stratified sand beds 495 contain granules and pebbles (Fig. 10A). These characteristics suggest a swash deposit in the beach face. Unit C rests on a sharp erosional boundary separating it from the underlying sand/mud deposits of Unit B2 (Figs. 9, 10A). The erosional boundary is interpreted as a transgressive ravinement surface, formed by landward shoreface retreat and wave action in a shoreline during transgression and thus these deposits likely post-date the last glacial maximum (Fig. 9).

Younggwang area 500
Two cores were obtained on a modern tidal flat at Baeksu and Baeksu-Duuri near Younggwang ( Fig. 7; Chang et al., 2014;Baek et al., 2017). Core 11YG-C4 at Baeksu has approximately 18 m of LIG deposits between -38 and -20 m below MSL (Fig.   11). The LIG deposit is characterized by tidal rhythmites, sand-mud couplets, and a lower stiff tidal mud with fine peats, rootlets, and wood fragments . The characteristics of the presumed LIG deposit are indicative of deposition within a tidal mud-flat and salt-marsh environment . The twelve OSL LIG ages from 4-11 μm quartz grains 505  (Table 3; Chang et al. 2014). These occurrences suggest marine limiting records of LIG RSL.

Gochang area
Core 15DH-C03 in the Gochang area was obtained from the modern nearshore environment and contains LIG deposits between 520 -32 and -39 m below MSL (Figs. 7, 8;H.H. Yoon et al., 2021). The LIG deposit attains a thickness of up to ~7 m in the core, which is characterized by muddy beds containing no shell or rare shell fragments (Fig. 11;H.H. Yoon et al., 2021). Sediments in the lower part of the core mainly consist of laminated mud beds with some mud-sand couplets or lenticular bedding. The sediments contain a few bioturbated beds. Sediments in the upper part of the core are largely mottled, intensely bioturbated, and composed dominantly of silty mud. The sediments contain some dark-gray organic-rich beds with wood fragments. Based 525 on the presence of mud-sand couplets and lenticular bedding, the lower laminated mud beds are interpreted as being deposited in an upper mudflat environment (H.H. Yoon et al., 2021). The upper bioturbated mud with some organic material suggests deposition within a tidal salt-marsh environment (H.H. Yoon et al., 2021). The presence of tidal mud-sand couplets and ages between MIS 5e and 5d suggests that the tidal deposit formed during or shortly after the LIG period (e.g., Chang et al. 2014;Baek et al. 2017). The three OSL MIS 5 ages from 4-11 μm quartz grains were 107.5±7.6 ka (WALIS LUM ID #487) at -35.2 530 m below MSL, 107.6±7.3 ka (WALIS LUM ID #488) at -37.4 m below MSL, and 113.3±7.2 ka (WALIS LUM ID #489) at -38.2 m below MSL (Table 3; H.H. Yoon et al., 2021), suggesting marine limiting records of LIG RSL.

Jindo area
Core H-C11 was obtained from the region offshore and south of Jindo (Fig. 7). A LIG age was obtained from sediments at an elevation of -48.5 m (S.H. Hong et al., 2019). This massive sandy shell bed was interpreted as a shelf deposit, based on the 535 presence of swaley cross strata and abundant oyster fragments (Fig. 11;S.H. Hong et al., 2019). The single OSL LIG age from 4-11 μm quartz grains was 124.4±10.0 ka (WALIS LUM ID #490) at an elevation of -48.5 m (Table 3; S.H. Hong et al., 2019).
This deposit provides a marine limiting record of LIG RSL.  Table 3). For location, see Fig. 7.

Heuksan Mud Belt area 545
Core HMB-102 of the Heuksan Mud Belt was obtained from the modern offshore environment (Fig. 7). One LIG age in the core was obtained at -84 m from a tidal deposit of laminated silt and clay and mottled mud (Fig. 11;Chang and Ha, 2015). The single OSL LIG age from 4-11 μm quartz grains was 125.1±9.9 ka (WALIS LUM ID #491) at an elevation of -84 m (Table 3; J. C. Kim et al., 2019). This deposit represents a marine limiting record of LIG RSL.

Uplift
The seismic stratigraphy and geology of the Gangneung and Donghae areas of the east coast of South Korea were summarized by Kwon et al. (2009) and Ryang et al. (2014). Intense compressional deformation during the early Pliocene, accompanied by the formation of reverse faults, strike-slip faults, and anticlinal folds, occurred mainly along the western margin of the submerged South Korea Plateau of the East Sea ( Fig. 12; Ryang et al., 2007;Kwon et al., 2009). This deformation resulted in 555 partial uplift and erosion of Late Pliocene and Early Quaternary deposits (Fig. 13). The lower boundaries of the post-Miocene stratigraphic units represent a progressive onlap termination against the apices of anticlinal folding (Fig. 13;Kwon, 2005;Kwon et al., 2009).
The southern region of the east coast may have also experienced considerable vertical displacement and deformation reflecting tectonic uplift during the Late Pleistocene along regional faults such as the Hupo, Yangsan, and Ulsan faults (J.H. 560 Choi et al., 2003Choi et al., , 2009S.J. Choi et al., 2008). All of the differential uplift is interpreted to be the result of backarc closing under a compressional regime since the early Pliocene (5 Ma) (S.H. Yoon and Chough, 1995;Chough et al., 2000). Although not all of the cross-coast faults along the east coast have been documented, local vertical movement and deformation probably caused marine terraces to experience uplift at different rates (Figs. 14,15). This suggests their elevations largely reflect differential tectonic uplift (e.g., J.H. Choi et al., 2003Choi et al., , 2009S.J. Choi et al., 2008S.J. Choi et al., , 2016. 565 We estimated rates of tectonic uplift of the marine terraces along the east coast of the Korean Peninsula using Eq. (12) and the five shoreline angle elevations listed in Table 2.
where SA represents the elevation of the shoreline angles and MHHW represents the mean higher high water at the subsites.
The ages of the deposits associated with each of the SA can be found in Table 4. From Eq. 12, we arrive at uplift rates of 0.189 570 ± 0.031 m/ka, 0.195 ± 0.037 m/ka, 0.251 ± 0.024 m/ka, 0.172 ± 0.024 m/ka, and 0.198 ± 0.024 m/ka for Gangneung-Anin, Donghae-Eodal-dong, Pohang-Yonghan-2, Pohang-Masan-ri, and Gyeongju-Jinri, respectively, if assuming a LIG RSL of +3 m for the Korean Peninsula, and 0.166 ± 0.030 m/ka, 0.171 ± 0.035 m/ka, 0.228 ± 0.022 m/ka, 0.146 ± 0.023 m/ka, 0.174 ± 0.022 m/ka, respectively, in the case of a +6 m LIG RSL for the Korean Peninsula (Table 4). The error ranges were calculated based on both the range of SA elevations and the age uncertainty (Table 4). The uplift rates for the northern and southern 575 regions of the east coast are sufficiently different as to suggest differential uplift across the region (Table 4; Fig. 2). Table 2

Subsidence
The currently-dated LIG sites along the west coast of Korea are all subject to potential subsidence rather than tectonic uplift.
LIG deposits are found at lower elevations with greater water depth but these likely reflect deposition in progressively deeper waters during either the preceding rise in sea levels (MIS 6) or the fall in sea levels following the LIG highstand (Figs 1, 11).
LIG deposits appear to be better preserved basinward (Fig. 15). This preservation may be a result of erosion of the higher LIG 585 deposits during subsequent sea-level falls (e.g., MIS 5d -MIS 2). Preservation of the deeper LIG deposits may have also been aided by their tectonic setting with increasing subsidence basinward. However, constraining the magnitude of Quaternary subsidence independent of the LIG elevation has yet to be attempted. Regional tectonic studies independent of the LIG shoreline elevation are needed to determine subsidence rates and correct LIG sea levels from its influence.    Table 3 for more details.

LIG sea-level fluctuations 610
Most ages of the LIG features in the Korean Peninsula were obtained via OSL. The error range of the OSL method is too large for the confident identification of fine-scale oscillations in LIG sea levels. The relatively few occurrences of LIG deposits also limit tests for LIG sea-level oscillations along the Korean Peninsula.

Earlier highstands
Four IRSL ages ranging from 185 to 221 ka were obtained from paleo-beach deposits on marine terraces of earlier highstands 615 at 59 to 63 m above MSL in the Gangneung area of the east Korean coast (Fig. 2; S.C. Hong, 2014). In this area, three ages between 211 and 253 ka were also dated by cosmogenic 10 Be dating of paleo-beach deposits on marine terraces at 6l to 66 m above MSL (S.Y. . More work is needed to better document earlier sea-level highstands on the Korean Peninsula. Representative Holocene sea-level curves are well constrained along the west coast of the Korean Peninsula based on intertidal features and other sea-level indicators (Bloom and Park, 1985;Y.H. Kim et al., 1999;Chang and Choi, 2001;Chough et al., 2004;E. Lee and Chang, 2015). However, the curves are limited by the absence of preserved tidal deposits as well as sparse ages on tidal-flat sea-level indicators from 6500 to 3000 yrs ago, dissimilar to those of the Yangtze delta plain in China (Chough et al., 2004;S.J. Choi, 2018;H.H. Yoon et al., 2021). K.H. Choi (2009) suggested the possibility of a sea-level highstand 625 around 6000 yrs ago based on OSL-dated onland sand dune deposition, but he also noted the absence of age data covering the period between 5000 and 3500 yrs ago in his study. Another suggestion is the possibility of a higher-than-present mid-Holocene highstand based on palynological data on land (Hwang and Yoon, 2011;Song et al., 2018), but these deposits lack a proper sedimentological description and stratigraphic framework. Further work remains to be done in reconstructing mid-Holocene sea levels along the Korean Peninsula. 630

Uncertainty and data quality
As all the summarized age data are based on OSL, with the exception of a single paleomagnetic age, the ages of the LIG shoreline features in the Korean Peninsula are only accurate and precise enough to establish a MIS 5 age, not necessarily which interstadial the features were deposited during. The age uncertainty is partially controlled by the mineral dated, either quartz or K-feldspar, as well as assumptions related to complete bleaching of the grains and their water content through time. The 635 average error of the 42 OSL measurements using quartz minerals is 9.7 ka, and that of the 10 OSL measurements on K-feldspar minerals is 20.7 ka. The large age uncertainty in dating K-feldspar minerals is more than double that of dating quartz. This range of 9.7 to 20.7 ka is thought to be far too large to determine exactly within MIS 5 the deposits formed. However, based on their stratigraphic occurrences as the highest deposits from the region, we interpret most of the OSL-dated deposits as MIS 5e in age. Although the basement massif in the west coast of Korea has been assumed to be stable or undergoing minor 640 subsidence during the Quaternary (Chough et al., 2010), the uncertainty in LIG sea levels is probably not free of additional error from possible subsidence or uplift around the Korean Peninsula.

Concluding remarks
The LIG shoreline is well developed as a marine terrace along the east coast of the Korean Peninsula and MIS 5 deposits are prevalent within the onshore and nearshore west coast. The east coast contains prevalent beach deposits on many marine 645 terraces along the peninsula. The east coast LIG deposits suggest RSLs between +9 and +32 m (Table 3; Fig. 15). However, the uplifted terraces upon which these ages were obtained are likely inflenced by faulting under a compressional regime as a result of backarc closing of the East Sea (Table 2; Fig. 14). As a result, tectonic uplift likely biases the elevation of the east coast LIG shorelines. On the contrary, LIG sea levels appear to be well constrained to between +3 and +6 m by marine limiting records, sea-level indicators, and terrestrial limiting records along the west coast of the Korean Peninsula (Table 3; Fig. 15); 650 the most stable side of the Peninsula, although minor subsidence of the western coast cannot be ruled out. Further work is needed to establish credible sea-level indicators within the onshore portions of the west coast of the Korean Peninsula.

Data availability
The Korean Peninsula Last Interglacial sea-level database is available open access, and updated as necessary, at the following link: https://doi.org/10.5281/zenodo.4974826. The files at this link were exported from the WALIS database interface on June 655 17, 2021. A description of each field in the database can be found at this link: https://doi.org/10.5281/zenodo.3961543 (Rovere et al., 2020) and is accessible and searchable here: https://walis-help.readthedocs.io/en/latest/. More information on the World Atlas of Last Interglacial Shorelines can be found here: https://warmcoasts.eu/world-atlas.html.

Author contributions
WHR read the papers for the data, compiled the data, and wrote the initial manuscript. ARS initiated this work and revised the 660 manuscript and dataset. HHY wrote the section of core DSR-C03 and drew most of the modified and original figures in this paper. Further input and discussion on the data and manuscript were provided by SSC and GSK. All authors revised the final text and agree with its contents.