Imaging of An Uplifted Serpentinite Complex In The Kamuikotan Zone, Northern Japan, From Magnetotelluric Soundings

We conducted magnetotelluric measurements to investigate a large serpentinite complex in the northern Kamuikotan zone that intruded a Cretaceous–Paleocene forearc sedimentary sequence. The resistivity model we derived by three-dimensional inversion clearly shows a low-resistivity zone beneath the outcrop of the serpentinite complex. We interpret the low-resistivity zone to represent aqueous pore uid within a serpentinite mélange derived from the subducting Pacic plate or mantle wedge. Previous geological studies in the area have shown that the serpentinite mélange had uplifted during the early Pleistocene. They indicate that the ultramac rocks and aqueous uids have continued to rise in the area. The uplifting serpentinite body might have formed a zone enriched in pore uid that promoted the occurrence of a previously identied intra-plate slow slip event. These results demonstrate the important role of uid transport during tectonic processes related to uplift in subduction zones. role related to the uplift of metamorphic belts for uid transport in subduction zones, may have to the 2012–2013 intra-plate SSE that occurred between conductive anomalies C-1 and C-2.


Introduction
Serpentinites in subduction zones contain large volumes of aqueous uid in the form of hydrated minerals and pore water (e.g., Hyndman and Peacock 2003), and play essential roles for aqueous uid transport, fault rupturing, and tectonic processes. The serpentinites are formed by the hydration of ultrama c rocks in the mantle wedge, the subducting oceanic plate, or in normal fault zones in the subducting oceanic plate (e.g., Guillot et al. 2015). In outcrop, such serpentinites are commonly associated with uplifted high-P/T metamorphic rocks. However, serpentinites in outcrop are not well understood because the broad structural features of deep subsurface serpentinite bodies have not been directly imaged.
The area of the Shirikomadake serpentinite complex in northern Hokkaido ( Fig. 1) is important in studies on the role of serpentinites in subduction tectonics. The Shirikomadake complex consists of serpentinized ultrama c rocks and gabbro-diorite dykes in a serpentinite mélange (Katoh et al. 1979) and is one of the largest ultrama c bodies in the Japan arc. Faulting of the serpentinite complex against Cretaceous to Quaternary sedimentary rocks implies that the complex experienced large-scale uplift and intrusion (e.g., Igi 1959). In addition, Ohzono et al. (2015) detected a slow slip event (SSE) near the edge of the serpentinite body as the rst discovery of slow earthquake in intra-plate (Fig. 1b). Because slow earthquakes have been attributed to the presence of pore uids (e.g., Shelly et al. 2006; Obara 2020), the intra-plate SSE may also be associated with pore uid derived from the serpentinites. Because of the uniqueness, the Shirikomadake serpentinite complex provides an ideal opportunity to undertake a detailed investigation of this association.
Electrical resistivity measurements can be used to investigate crustal lithologies and the distribution of pore uids within them (e.g., Glover et al. 2000). Resistivity distributions can be determined by the magnetotelluric (MT) method using naturally occurring electromagnetic elds. The MT impedance tensor derived from a measured electromagnetic eld re ects the subsurface resistivity distribution. Exploration depth and width are dependent on the period of MT impedance. Multi-site and broadband MT soundings can be used to image the three-dimensional (3-D) resistivity distribution. In this study, we inverted broadband MT soundings at 48 sites to model the 3-D resistivity distribution in the Shirikomadake area.
We then interpreted the estimated resistivity model and considered its possible association with uplift of the Shirikomadake serpentinite complex and the SSE.

Tectonic And Geologic Setting
The Shirikomadake serpentinite complex is in northern Hokkaido Island, which lies at Kamuikotan Zone in the Northeast Japan arc. The following summary of the geology of the island is based on the work of Ueda (2016). The Sorachi-Yezo belt (Fig. 1a) is characterized by a Jurassic ophiolite overlain in turn by a Late Jurassic to Early Cretaceous ophiolite and siliceous sedimentary sequence (Horokanai Ophiolite and Sorachi Group) and a Late Cretaceous to Paleocene forearc basin sequence (Yezo Group). The Kamuikotan Zone represents the cores of anticlines within the Sorachi-Yezo belt (Fig. 1a) and characterized by high-P/T metamorphic rocks (Kamuikotan metamorphic rocks) and serpentinite mélange. Serpentinite bodies are dispersed through the Kamuikotan Zone. The metamorphic rocks of the Kamuikotan zone originated from the subducting slab and accreted rocks, and were metamorphosed during the early Cretaceous to Paleocene (e.g., Sakakibara and Ota 1994).
The source rocks of the Shirikomadake serpentinite complex were dunite, harzburgite and orthopyroxenite and the complex contains antigorite that formed after lizardite, chrysotile, and brucite (Katoh et al. 1979).
Late Cretaceous (Yezo Group) and Paleogene-Neogene sedimentary rocks distributed around the complex (Fig. 1b) thicken westward, extending to a depth of more than 4,500 m in the western part of the study area (borehole SK-1 in Fig. 1b) (Ogura and Kamon 1992), and are deformed by active folds and thrusts (e.g., Oka 1985). Miocene igneous rocks are distributed in the middle and eastern parts of the study area. Rocks of a Cretaceous accretionary complex in the eastern part of the study area represent the Idonnappu Zone (Suzuki et al. 1997).
Geodetic studies indicate that the study area is presently in a zone of east-west regional compression (e.g., Sagiya et al. 2000), which is consistent with the geology described above. Crustal seismic activity is high in the western and central parts of the study area (e.g., Tamura et al. 2003) (Fig. 1b). Ichiyanagi et al. (2015) identi ed earthquake swarms that occurred in the south of the study area between July 2012 and January 2013. Ohzono et al. (2015) used global national satellite navigation data to identify a SSE (Mw 5.4) that occurred in the eastern part of the zone of earthquake swarms during the same period (Fig. 1b).
The fault mechanisms of both the earthquake swarms and the SSE are consistent with the regional eastwest compressional stress eld. In the eastern part of the study area, seismicity is low and no active faults or folds have been recognized.

Magnetotelluric measurements and impedance tensors
We conducted broadband MT measurements in 2001,2002, and 2018 at 25, 20, and 3 sites, respectively ( Fig. 1b). We measured two horizontal components of the electric eld by using Pb-PbCl 2 electrodes and three orthogonal components of the magnetic eld by using induction coils. These data were recorded by MTU-5 systems (Phoenix Geophysics, Ltd., Toronto, Canada) for 2-6 days. MT impedance tensors were estimated from the obtained electromagnetic elds during 0.00325-to 1,820-s by using SSMT2000 software (Phoenix Geophysics, Ltd.) and remote reference processing (Gamble et al. 1979). For remote reference processing, we used horizontal magnetic eld data obtained at the following sites: Esashi station (operated by the Geospatial Information Authority of Japan) for the 2001-2002 MT data, and Sawauchi station (operated by Nittetsu Mining Consultants) for the 2018 MT data. Both of the remote stations are about 400 km south-southwest of the study area.
Estimated MT impedances are shown in Fig. 2 and Additional File 1. To visualize the spatial trend of impedance, we estimated sounding curves for the sum of the squared elements' invariant impedances (SSQ; e.g., Rung-Arunwan et al. 2016; see Additional File 2), which show trends of low and high resistivity in the western and eastern parts of the study area, respectively.
In the western area (red symbols in Fig. 1b and Additional File 2), phase angles are high and apparent resistivity decreases with increasing period in the short-middle-period band (< 30 s). These results imply the presence of a near-surface conductive layer. In the central area (green and black symbols in Fig. 1b and Additional File 2), apparent resistivity is low and phase angles are neutral in the short-middle-period band (< 30 s). These results imply that the conductive anomaly is distributed in the subsurface. The splitting of off-diagonal components at longer periods implies the presence of deeper 2-D or 3-D structures.
In the eastern area, the sounding curves show considerable variations and large diagonal components (blue symbols in Fig. 1b and Additional File2). These results are indicative of a strong 2-D or 3-D structure.

3-D electrical resistivity inversion
We constructed a regional 1-D electrical resistivity model to use as the initial model of the 3-D inversion to minimize the local minima problem inherent in non-linear inversions. The 1-D modeling was based on the 1-D Occam's inversion procedure developed by Constable et al. (1987). We used the averaged SSQ invariant impedances as input data (Additional File 3a) and set the target RMS mis t of the inversion at 1.00. The modeled 1-D resistivity pro le explained the data reasonably well (RMS mis t: 1.00) and showed an increase of resistivity with depth (Additional File 3).
The 3-D resistivity distribution was then estimated using the inversion code of Tada et al. (2012), which is a modi ed version of the WSINV3DMT 3-D inversion code (Siripunvaraporn et al. 2005) that adopts a data-space variant of the Occam's inversion procedure. The modi ed WSINV3DMT code can incorporate boundaries between land and seawater where high resistivity contrasts can seriously distort MT responses. In this inversion procedure, the following objective function is minimized: where m and m p are model parameters and prior model vectors, respectively, C m is the model covariance matrix that characterizes the expected magnitude and smoothness of resistivity variations relative to m p , and d is the data parameter vector consisting of the observed MT impedances. F[m] is the vector of the forward response to m and λ is a hyperparameter that balances data mis t and model mis t terms, including model roughness. C d is a data covariance matrix representing the observation errors.
MT impedances at 16 periods between 0.0667 and 1,820 s were used as data parameters (d) in the inversion. We added error oors to avoid over tting the calculated responses to observed data for which errors were small. The error oor was de ned as 5% of the SSQ impedance and was applied to all components of the MT impedance. The resistivity model space de ned a volume of 3,268 (x-axis) × 3,268 (y-axis) × 1,049 km (z-axis, without air layers) discretized into 58 × 58 × 37 blocks. The x and y dimensions of the blocks were 2 km within the survey area but were widened outside the study area. The z dimensions of the blocks increased with depth and were the same as the those of the 1-D inversion (Additional File 3b). The inversion procedure began with the 1-D resistivity model (Additional File 3b), except for the area of seawater where we set the resistivity at 0.3 Ω m (RMS mis t: 9.72). The prior model (m p ) was the same as the initial model in this procedure. We iterated the inversion ve times and obtained a minimum RMS mis t model (RMS mis t: 2.13) at the third iteration. We then updated m p as the minimum RMS mis t model and again iterated the inversion ve times. Finally, the minimum RMS mis t model (RMS mis t: 1.64) was obtained at the fth iteration, which we adopted as the "inverted model".
The inverted model shows a shallow conductive layer (C-1, Fig. 3) at 0-5 km depth in the western part of the study area. This layer is also clearly indicated by low apparent resistivities of the off-diagonal components of MT impedance in the western part of the study area (Additional File 2). Also in the western part of the study area, a layer of moderate resistivity (10-100 Ω m) partly extends from the surface to a maximum depth of about 1 km as shown in the section "X = − 5 km" in Fig. 3, as indicated by the moderate apparent resistivity and high phase angle in the short-period band (e.g. site D13 in Fig. 2). The C-1 layer thins eastward and disappears approaching the eastern edge of the study area, consistent with the west-east trend of the sounding curves (Additional File 2).
The inverted model also shows a zone of high resistivity that extends from the surface in the east of the study area. This modeled high resistivity is consistent with the high apparent resistivity in this area.

Sensitivity test for a near-vertical anomalous conductive zone (C-2)
A near-vertical anomalous conductive zone (C-2, Fig. 3) was modeled under the area where the Shirikomadake serpentinite complex crops in the south-central part of the study area. Because the C-2 anomaly was not apparent in our original sounding curves, we used the following sensitivity test to examine the anomaly further.
We replaced modeled resistivity values within the C-2 anomaly that were < 10 Ω m (Fig. 3b) with an approximation of the resistivity of the surrounding (100 Ω m) and recalculated the model response using the modi ed WSINV3DMT code. The sounding curves of this model changed considerably from those of the inverted model and did not explain the observed impedances at sites above the C-2 anomaly (e.g., Sites M18 and S18 in Fig. 2). In particular, the low apparent resistivities at periods > 3 s and moderate phase angles (45° and − 135°) between periods of 1 and 30 s of the off-diagonal components at Site M18 were not explained. These results indicate that the observed data require the presence of conductive anomaly C-2.
To further examine the reliability of the resistivity values of the C-2 anomaly, we calculated the responses of models for a range of replacement resistivities in the same manner as the above test (Fig. 4). F-tests at the 95% con dence level indicate that the models derived using replacement resistivities > 15 Ω m and < 1.0 Ω m were signi cantly inferior to the inverted model (Fig. 4). If it is assumed that the conductive area is uniformly distributed within the anomalous zone, these results indicate that the resistivity range of the C-2 anomaly lies between 1.0 and 15 Ω m. Note that the real conductive zone is possibly narrower and of lower resistivity because of the smoothness constraint of the inversion.

Interpretation of the resistivity model
The C-2 anomaly lies directly below the outcrop of the Shirikomadake serpentinite complex (Fig. 3). A similar conductive anomaly has been identi ed under a serpentinite body in the southern part of the Kamuikotan zone (Ogawa et al. 1994;Ichihara et al. 2019). In a general sense, these relationships can be considered to imply an association between serpentinite bodies and underlying conductive anomalies. However, the resistivity (3-100 Ω m) in the shallow part of our 3-D model of the outcropping serpentinite complex (Fig. 3b) is not particularly low. Near-surface resistivity surveys (0-200 m depth) conducted in the south of the Shirikomadake serpentinite complex (Fig. 1b) by Okazaki et al. (2011) detected a range of resistivities consistent with our results. They also showed that the resistivities of their massive, foliated, and clay-rich serpentinite core samples (massive, foliated, and clay-rich forms) were 250-1,500, 200, and 20-40 Ω m, respectively. Therefore, the resistivities of serpentine group minerals are not the sole control of conductive anomalies in serpentinite, such as our C-2 anomaly.
Candidates for the source of conductive anomaly C-2 include the presence of (1) inter-connected pore uids, (2) interconnected conductive minerals (e.g., magnetite) that are commonly found in serpentinite, and (3) a high-temperature zone or melt that decreases resistivity. The rst of these is the most likely candidate as previous studies of conductive anomalies in subduction zones (e.g., Aizawa  The second candidate (interconnected conductive minerals) has been considered for mid-crustal conductive areas (e.g., Myer et al. 2013) because magnetite contained in serpentinites is highly conductive and can produce a conductive anomaly (Stesky and Brace 1973). However, it is di cult to attribute the C-2 anomaly to magnetite alone because both the conductivity of borehole samples collected by Okazaki et al. (2011) and the surface-measured resistivity of the Shirikomadake serpentinites do not indicate a su cient anomaly, although the presence of magnetite may contribute to the low resistivity of interconnected pore uids. The third candidate (high-temperature zone or melt) is unlikely to be a direct cause of the C-2 anomaly because the study area is a considerable distance from the nearest area of volcanic activity and no high-temperature phenomena, such as hot springs, are known in the study area.
There are two possible explanations for the relatively high resistivity zone between the serpentinite outcrop and the C-2 anomaly (Figs. 3 and 4). It might represent resistive rocks that contain little or no pore uid, such as, massive serpentinite, basaltic lavas (Katoh et al. 1979), or metamorphic rocks of the Kamuikotan zone that do not outcrop in the study area. Alternatively, because the resistivities of aqueous uids increase with decreasing temperature (i.e., with decreasing depth) (Sakuma and Ichiki 2016; Sinmyo and Keppler 2017), the resistivity remains high in the shallow sequence.
Borehole SK-1 near the western margin of the study area (Fig. 1b) initially penetrated Paleogene-Neogene sedimentary rocks at 3,030 m depth and, was within late Cretaceous sedimentary rocks at its total depth (4,505 m) (Ogura and Kamon 1992). Borehole logs indicate that resistivities in the Paleogene-Neogene and late Cretaceous sequences were 1-4 and 4-10 Ω m, respectively (Kanekiyo 1999). The modeled resistivity within anomaly C-1 (Fig. 5) is consistent with that logging data. It is noteworthy that the surface extent of the C-1 anomaly is consistent with the mapped extent of the Cretaceous (Yezo Group) and Paleogene-Neogene sedimentary rocks (Figs. 1b and 3a). Hence, we interpret the C-1 anomaly to represent Cretaceous and Paleogene-Neogene conductive sedimentary rocks. Similar associations between resistivity anomalies and outcrops have been recognized in other parts of middle-western Hokkaido The C-1 anomaly deepens toward the western boundary of the study area (Fig. 5) and the steepest dip of its lower boundary coincides with thrust faults (Fig. 5) that uplifted the eastern part of the study area. The C-1 anomaly is partly overlain by a layer of moderate resistivity (about 100 Ω m) that is similar to a nearsurface layer of moderate resistivity identi ed by Ueda et al. (2014) near sites B13 and B14. On the basis of seismic re ection survey data, they interpreted this layer to be the Quaternary Sarabetsu Formation, which contains relatively fresh groundwater. Thus, we interpret the layer of moderate resistivity that we modeled overlying anomaly C-1 to represent Quaternary sediments.

Uplift of serpentinite and its implications for intra-plate SSEs
Previous geological studies have identi ed faults at the boundaries between the Shirikomadake serpentinite complex and surrounding sedimentary rocks (Igi 1959;Katoh et al. 1979;Oka 1985).

Conclusions
We conducted broadband magnetotelluric soundings at 48 sites around the Shirikomadake serpentinite complex in the Kamuikotan zone in Hokkaido, northeastern Japan arc. The 3-D resistivity inversion of the estimated MT impedance revealed two anomalies: a near-surface conductive layer (C-1) and a vertically elongated conductive zone beneath the surface expression of the serpentinite complex (C-2). We interpreted the C-1 anomaly to represent Cretaceous-Neogene sedimentary rocks that showed evidence of thrust faulting (Fig. 5) and the C-2 anomaly to represent inter-connected pore uids within the serpentinite complex and clayey serpentinite related to the serpentinite outcrop (Fig. 5)

Availability of data and materials
Contact the corresponding author to access the digital data underpinning the 3-D resistivity model.

Competing interests
The authors declare that they have no competing interests regarding this study.

Funding
This research was supported in part by KAKENHI Grant Numbers 11555268, 16H06475, and 18K13638 from the Japanese Ministry of Education, Culture, Sports, Science and Technology.
Authors' contributions HI, TM, TU, HS, YY, MF, SY, and KO contributed to the magnetotelluric observations. HI, TM, TU, HS, and NT analyzed the data. All authors contributed to interpretation of the data and approved the nal manuscript.    line is the boundary between the C-1 and C-2 anomalies. Note: The designations employed and the presentation of the material on this map do not imply the expression of any opinion whatsoever on the part of Research Square concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries. This map has been provided by the authors.

Figure 4
Results of the sensitivity test for the C-2 anomaly. Replacement resistivity values (red diamonds) for anomalous C-2 resistivities versus RMS mis t. The green line is the RMS mis t of the inverted model and