Raised shorelines on the Noto Peninsula on the western coast of central Japan suggest a millennial history of recurrent uplift. Emergence of the coastline by up to 50 cm was recently recorded in association with a moderate earthquake (Mw 6.7) on 25 March 2007. The shoreline uplift is indicated by the displacement of sessile organisms such as calcareous tubeworms, which can also be examined as indicators of past shifts in shoreline level. Three levels of pre-2007 paleo-shorelines are thus identified on this stretch of the Japan Sea coast. The lowest of these, dated at AD 1720–1950, tilts northward away from the likely source of a pair of ca. M 6.4 earthquakes in 1892, which occurred in a different area to the 2007 earthquake. The middle paleo-shoreline, dated at AD 1430–1655, records a 30–40 cm emergence that may be aseismic. The high paleo-shoreline, dated at AD 1025–1235, records a coseismic uplift of 50 cm and is limited to the area raised coseismically in 2007. The most recent event preceding the 2007 earthquake thus appears to have occurred approximately 1000 years ago.
 The 2007 Noto Hanto earthquake (Mw 6.7) occurred on the northwestern coast of the Noto Peninsula, on the Japan Sea coast of central Japan. Seismic observations by the Japan Meteorological Agency (JMA) indicate that the hypocenter of the mainshock was located at 37.220°N, 136.985°E at a depth of 11 km (Figure 1), with aftershocks extending along a NE–SW strike over an area 25 km long and 10 km wide. This distribution of events coincides with the hanging-wall region of a previously mapped active submarine thrust fault [Katagawa et al., 2005; Okamura, 2007]. Geodetic data, including leveling, global positioning system (GPS), interferometric synthetic aperture radar (InSAR), and tide gauge observations indicate that the earthquake produced an uplift of up to 50 cm [Ozawa et al., 2008] and a small tsunami with a maximum height of 20 cm [Namegaya and Satake, 2008]. Awata et al.  observed coastal emergence associated with the coseismic vertical displacement of this event, and proposed a fault source model based on the height distribution of uplifted sessile organisms. However, the Holocene deformation and history of paleo-earthquakes in this region remains poorly understood, because it is generally regarded that the traces of moderate earthquakes are rarely preserved due to geological and geomorphological factors. In this study, a detailed paleo-seismological survey of the coast is presented, through which evidence of three relative sea-level fall events was obtained based on the distribution of sessile organisms. The causes of these events and the relationship with the 2007 earthquake are also discussed.
 The emergence of intertidal sessile organisms is an excellent indicator of paleo-sea level change [Pirazzoli, 1996]. The present study focuses on the distribution of Pomatoleios kraussii, a lugworm that constructs rock-attached calcareous tubes and which can only survive within a narrow range of the mid-tidal zone [Miura and Kajihara, 1984]. As the upper limit of the P. kraussii assemblage coincides with the contemporaneous mean sea level, observations of the relative uplift of P. kraussii serve as an excellent indicator of historical mean sea levels.
 The height of P. kraussi assemblages was measured and mapped at 17 sites along the rocky coast between Nagatejima and Wajima in the northwest of the peninsula (Figures 1 and 2) . The elevations of the upper and lower limit of each assemblage were measured with reference to sea level at the time of measurement using either an auto level or a laser distance meter (Impulse 200, Laser Technology Inc.), with accuracies of a few centimeters. The elevation data were then corrected by comparison with the mean sea level recorded over the past 10 years at the tide gauge station in Wajima (Figure 1). As the astronomical tidal range between the high and low water level in this area is 0.4 m or less, the error in the corrected height from the mean sea level is considered to be relatively small. Total measurement error is estimated to be less than approximately 5 cm.
 Eleven P. kraussii shell samples were collected for radiocarbon dating. For each sample, fragments of unweathered shell were detached from the assemblage, cleaned mechanically using a drill bit, and then etched in diluted hydrochloric acid to remove surface staining. The treated samples were then analyzed by accelerator mass spectroscopy (AMS).
 The conventional age results were calibrated with respect to the calendar year using the IntCal04 [Reimer et al., 2004] and Marine04 [Hughen et al., 2004] data sets. As the difference between the global mean and the local marine reservoir effect (ΔR) has not been reported for this area, ΔR was estimated from the 14C ages of marine shell fossils collected from assemblages uplifted during the 14 March 1872 Hamada earthquake (M 7.1), on the Japan Sea coast in southwest Japan (Figure 1), in accordance with the procedure of Shishikura et al. . From the analysis of five samples, ΔR is estimated to be −25 ± 28 years. This value is applied to calibrate the samples from the Noto Peninsula, the results for which are summarized in Tables S1 and S2 of the auxiliary material. The ages referred to hereafter are the calibrated ages with a standard deviation of 2σ probability.
 The 18 sites between Wajima and Nagatejima presented both displaced and stable occurrences of P. kraussii at the time of the 2007 earthquake, whereas the 10 sites between Akakamizaki and Momoura yielded fossilized pre-2007 emerged P. kraussii assemblages (Figures 1 and 2). Further south of Takahama, most of the coastline is consists of sandy beach, and fossilized P. kraussii were not found. Distinguishing between the 2007 and pre-2007 organisms was relatively straightforward at the time of the survey (even 6 months after the earthquake), since the former appeared very fresh like a living organism whereas the latter were obviously stained with dirt or algae.
 The vertical range of the 2007 assemblage at each site reached 20–30 cm, corresponding to the intertidal-constrained habitable range of P. kraussii in this area (Figure 2). The height distribution of its upper limit reveals an asymmetric anticlinal uplift with a steep north-facing frontal limb and a gentle south-facing back limb (Figure 2). This pattern is closely consistent with the geodetic observations reported by Ozawa et al. .
 The pre-2007 emerged P. kraussii assemblages are distributed over a wide range of heights (4–142 cm elevation), yet can be divided into three distinct level ranges (high, middle, low; Figure 2), each comprising an interval of 20–40 cm similar to the habitable range of the 2007 assemblage.
 The high-range assemblage was found only in the vicinity of Sekinohana and Amamisaki, where strong uplift occurred during the 2007 earthquake. This assemblage is dated at AD 1025–1235, and is the oldest of the present assemblages. The uppermost occurrence is located at an elevation of 132–142 cm (82–96 cm above the 2007 emerged assemblage), and is attached to an emerged wave-cut-bench that can be recognized as another paleo-sea level indicator (Figure 3). An emerged wave-cut notch has also been observed at a level of 120–140 cm in this area [Moriya et al., 2007] (Figure 2).
 The middle-range assemblage is distributed at elevations of 45–90 cm between Akakamizaki and Momoura. The uppermost extent of this assemblage becomes gradually lower toward the south, with a gentler gradient than that of the 2007 assemblage. The ages of shell samples from these three sites are AD 1470–1655 at Kurosaki, AD 1430–1625 at Akakamizaki, and AD 1445–1635 at Momoura.
 The low-range assemblage was found only in the southern part of the survey area, between Amamisaki and Momoura. The zonation pattern of this assemblage is parallel and similar to that of the middle-range assemblage, with which it is partially continuous in Hatagoiwa. The ages of the seven samples dated in this area fall in a recent 400-year range (AD 1540–1950). In Momoura and Hatagoiwa, the assemblage is relatively thick and characterized by a vertically layered structure (Figure S1). Samples from the upper (outer) and lower (rock-attached) sides of the assemblage confirm that the upper side is younger than the lower side, consistent with the growth of the assemblage during a period of stable sea level from AD 1540–1810 to AD 1720–1950 in Hatagoiwa.
 The present survey of pre-2007 emerged P. kraussii assemblages suggest that emergence events with three different causes have occurred over the past 1000 years. Figure 4b shows the height differences between the pre-2007 assemblages and the 2007 assemblage, corresponding to three levels of paleo-shoreline prior to the 2007 earthquake. These three emergence events are discussed in detail below.
4.1. Low Paleo-shoreline
 The height distribution of the low paleo-shoreline plunges northward, indicating that the relative sea level change decreases from +40 cm at Momoura to almost zero at Akakamizaki. This local trend suggests a tectonic tilt with maximum uplift of at least 40 cm, but a different pattern to the coseismic vertical displacement of the 2007 earthquake. The emergence process is considered to have been abrupt, probably coseismic, as indicated by the contemporaneous emergence of a wave-cut bench distributed at a similar elevation (30–50 cm) around Momoura [Moriya et al., 2007]. The timing of emergence is constrained to AD 1720–1950 (AD1830–1950 at 1σ probability in calibrated age) based on the upper-side ages of the layered assemblage in Hatagoiwa. In this period, two sequential moderate earthquakes of M 6.4 and 6.3 accompanied by small tidal level changes (tsunami and/or coastal emergence) occurred on 9 and 11 December 1892 south of the source of the 2007 event (Figure 1) [Usami, 2003]. Some records indicate that the area most damaged due to ground shaking was concentrated in and around Takahama [Usami, 2003], at the south end of the present survey area, and the epicenters of the two main shocks have been relocated to approximately 37.1°N, 136.7°E (primary event) and 37.1°N, 136.7°E (secondary event) (Figure 1). These source areas are consistent with the region of uplift deduced from the low paleo-shoreline trace. As no other large earthquakes suitable for explaining the emergence event have been documented in this area, the low paleo-shoreline trace is considered to be associated with the 1892 earthquakes. Although a small number of minor faults near the epicenters of the 1892 events have been mapped onshore by the Research Group for Active Faults of Japan , and offshore by Katagawa et al.  and Okamura  (Figure 1), the source fault model of the 1892 earthquakes has yet to be specified. The present data therefore represent the first quantitative report of constraining the source area for this earthquake.
4.2. Middle Paleo-shoreline
 The trace of the middle paleo-shoreline, with a date of AD 1430–1655, appears to be distributed parallel to that of the low paleo-shoreline, suggesting that a uniform relative sea-level change of approximately 30 cm occurred between Momoura and Akakamizaki in this period. This pattern differs from the coseismic deformation of the 2007 and 1892 events, and would require a large fault mechanism if attributed to a seismic source. However, as no large sources have been discovered on or off the west of the peninsula, a single large fault is unlikely as the cause for this uplift. A multi-segment fault rupture, such as a pairing of the 2007 and 1892 source faults or other inland faults parallel to the coast (e.g., Sakami fault and Fukuura fault), could have produced such a regionally uniform emergence (Figure 1). However, judging from the tectonic setting and the strain rate deduced from recent GPS measurement data [Sagiya etal., 2000]; the frequency of strain release in this area is not sufficiently high to support such a scenario. The vertically continuous zonation of the middle to low assemblages in Hatagoiwa may suggest that the emergence of the middle paleo-shoreline is associated with a relatively gradual motion rather than an abrupt coseismic uplift, indicative of broad aseismic uplift or eustatic sea level fall. Regional aseismic uplift occurs as a result of processes such as stable subducting plate motion [e.g., Sato and Matsu'ura, 1992] and magmatic activity (e.g., Iwojima, southern island in Japan [Kaizuka et al., 1985]), all of which appear to be inapplicable in the present area, which is far from the nearest subduction zone and does not host volcanism. The other conceivable cause is visco-elastic motion of the upper mantle derived from isostasy [e.g., Nakada et al., 1991], which requires further rheological investigation. In the case of eustasy, the 30 cm of change required to lift the paleo-shoreline in the period AD 1430–1655 should be observed not only on the Noto Peninsula but also in other remote areas. However, no other evidence of such phenomena has yet been reported globally.
 Although the data remain insufficient to draw reliable conclusions, it appears that an aseismic cause is a plausible explanation for the emergence of the middle paleo-shoreline trace identified in the present study.
4.3. High Paleo-shoreline
 Although the high paleo-shoreline trace was observed at only two sites (Amamisaki and Sekinohana), it is well associated with an emerged wave-cut-bench and notch identified at the same level. This correlation suggests that the high paleo-shoreline dated at AD 1025–1235 emerged abruptly by tectonic uplift (Figure 3). Based on comparison with the height of the middle paleo-shorelines, the amount of uplift of the high trace is estimated to be approximately 40–50 cm, which corresponds well with that determined for the 2007 earthquake. If this is the penultimate uplift event generated prior to the 2007 event, the recurrence interval is estimated to be 800–1000 years, which is consistent with the mean uplift rate deduced from the height of the Pleistocene marine terrace. The marine terrace in the vicinity of Sekinohana has been assigned to marine isotope stage 5e (120 ka) and reaches 50–60 m in elevation [Ota and Hirakawa, 1979] (Figure 4a). Therefore, if coseismic uplift of 40–50 cm has repeated since 120 ka, the recurrence interval of uplift can be estimated to be approximately 1000 years. Recently, Inoue et al.  analyzed offshore seismic refraction survey data acquired across the source fault of the 2007 earthquake, and revealed that the fault has been displaced cumulatively by ca. 6 m since 10–15 ka. This result suggests an activity rate of 0.4–0.6 m per 1000 years. It can thus be confidently concluded that the event preceding the 2007 earthquake occurred 800–1000 years ago.
4.4. Other Evidence
 Evidence of older coseismic uplifts should also be recorded in the distribution of sessile organism fossils or shoreline emergence above the high paleo-shoreline. At one site near Akakamizaki, a wave-cut notch reaching an elevation of 2.5 m can be identified in a mushroom-like outcrop (Figure S2). Other evidence of raised Holocene shoreline topography has also been identified along the study area, corresponding to several levels up to an elevation of 5 m [Hamada et al., 2007; Moriya et al., 2007]. Although no age data have been obtained for these traces, the existence of the traces implies a net cumulative uplift of a few meters since the middle Holocene.
 Evidence of three emergence events prior to the 2007 Noto Hanto earthquake was found in this study. The three emergence events are attributed to (1) coseismic uplift in AD 1025-1235 associated with the most recent seismic event preceding the 2007 earthquake, (2) probable aseismic motion in AD1430-1655, and (3) coseismic uplift associated with pair of ca. M 6.4 earthquakes in 1892. Based on coastal survey data and recent offshore fault survey data, the recurrence interval of the source fault of the 2007 earthquake is estimated to be approximately 1000 years. The present study has demonstrated that detailed field surveys of emerged sessile assemblages can provide valuable evidence of coastal uplift associated with not only large earthquakes, but also moderate earthquake of M 7 or less.
 Gratitude is extended to Brian Atwater for helpful suggestions on the manuscript, and Akira Ishiwatari for valuable data on emerged shoreline topography. AMS analyses were conducted by Beta Analytic Inc.