nAts

hink

Shallow waterCoastal areaSeawater/freshwater interface

arine a0 totivie these difculties, we selected a MT method that uses natural EM elds withoutent in seawater, which enables the method to be used in areas with active inshore

al areasd socia

ore of Surinam, Southfundamental studies,

ge subsurface structurethe entire hydrological

Journal of Applied Geophysics 100 (2014) 2331

Contents lists available at ScienceDirect

Journal of Applie

j ourna l homepage: www.e lsgeneral lack of habitation areas downstream of the groundwater ow.However, seawater intrusions exist in the ground of coastal areas

system beneath a coastal area that involves both onshore and offshoreregions (Fig. 1). Therefore, in Japan, several national and private researchorganizations have been conducting research and development in orderdepth of at least 300m in a stable geologic formation. Coastal areas areidentied as candidate repository sites for several reasons, such astransportation and logistical advantages for shipping, the lower hydraulicgradient of groundwater ow compared with inland areas, and the

North America (Hathaway et al., 1979) and offshAmerica (Groen et al., 2000). In addition to thesegeoscientic investigations, which seamlessly imafrom land to sea, are important in order to clarifymunicipal, industrial, and agricultural purposes (Mitsuhata et al., 2006).In addition, coastal areas have been considered as candidate

repository sites of high-level radioactive waste (HLW) (e.g., the nuclearwaste created by the reprocessing of spent fuel) (Tanaka et al., 2003).The basic national policy of Japan is to isolate HLW underground at a

(Tanaka et al., 2003). Therefore, evaluating brackish/fresh groundwaterdistributions and hydrological structures in coastal area is essential(e.g., Marui, 2003; Marui and Hayashi, 2001).

The hydrology of coastal areas and related groundwater systems hasbeen investigated extensively, for example, offshore of New Jersey, Corresponding author. Tel.: +81 29 861 7913.E-mail address: takumi.ueda@aist.go.jp (T. Ueda).

0926-9851 2013 The Authors. Published by Elsevier B.Vhttp://dx.doi.org/10.1016/j.jappgeo.2013.10.003l activities in these areas.ater distribution in coastaloundwater utilization for

2003). Furthermore, the subsurface of coastal areas may contain fossilseawater that was trapped during sediment formation. These physicaland chemical changes in groundwater may affect subsurface facilitiesFor example, the importance of the saline/freshwareas has been increasingly recognized in gr1. Introduction

The geological information of coastthe high concentration of economic anwaves, and compact, which enables light-draft small survey boat operation. We conducted offshore dataacquisition using a new MT measurement system and an onshore MT survey at the Horonobe coastal area,Hokkaido, Japan. High-quality data were successfully obtained in both onshore and offshore eld surveys.Two-dimensional (2D) inversion for eld data fromonshore to sea bottom reveals that a quaternary sedimentarylayer of a few hundred meters in thickness, which was determined by well logging to be a freshwater layer,extends horizontally offshore for several kilometers under the sea. The results obtained herein demonstratethat the newly developed marine MT measurement system can be used to clarify the geoelectrical structuresof brackish/fresh groundwater distributions and in coastal areas.

2013 The Authors. Published by Elsevier B.V.

is important because of

(Mitsuhata et al., 2006). The fresh/saltwater boundary in the groundcan move depending on the sea level over long periods of time, e.g.,thousands to tens of thousands of years (Ito et al., 2010; Tanaka et al.,

Open access under CC BY-NC-SA license.Seaoor electromagnetic explorationMarine magnetotellurics sheries. In addition, the developed marine MT system is very short, which reduces motion generated by seaKeywords:waves. In order to overcomtransmitting an electric currA new marine magnetotelluric measuremeenvironment for hydrogeological study

Takumi Ueda a,, Yuji Mitsuhata a, Toshihiro Uchida a,a Geological Survey of Japan, AIST, Central 7, 1-1-1Higashi, Tsukuba, Ibaraki 305-8567, Japanb Geothermal Energy Research & Development Co., Ltd., Shinkawa Nittei Annex Bldg., 1-22-4 S

a b s t r a c ta r t i c l e i n f o

Article history:Received 18 June 2013Accepted 11 October 2013Available online 22 October 2013

We have developed a newmsea wave motions and can bregion with a sea depth of 1areas include (1) shery ac. Open access under CC BY-NC-SA licet system in a shallow-water

unao Marui a, Kenji Ohsawa b

awa, Chuo-ku, Tokyo 104-0033, Japan

e magnetotelluric (MT) measurement system that can reduce noises caused bypplied to measurements under very shallow seawater areas, such as a coastal100m. The difculties of geophysical exploration in shallow water and coastalty, (2) limitations of survey vessel size, and (3) motion noise caused by sea

d Geophysics

ev ie r .com/ locate / j appgeoto describe the geological structures and hydrological systems of coastalareas (Hama et al., 2007; Sugihara, 2009; Tokiwa et al., 2013). Forexample, several geoscientic studies, including geological, geochemical,and hydrological investigations, have focused on groundwater behaviorassociated with internal rocks and geological formations, includingboth saline and freshwater distributions (Ikawa et al., 2012; Kozai

nse.

et al., 2013). We have also been investigating geophysical methodsfor seamlessly evaluating geological structures from land to sea undera nuclear waste repository research project funded by the Ministry ofEconomy, Trade and Industry (METI) of Japan in the Horonobe coastalarea since 2007.

The Horonobe coastal area, which is located along the northwesterncoast of Hokkaido Island (Fig. 2), is part of a sedimentary coastal basin,which is composed of poorly compacted Neozoic sand-, silt-, andmudrocks (e.g., Koshigai et al., 2012). In this area, the gradient of theseaoor is small, e.g., the sea depth is approximately 60m, even 10 kmfrom shoreline. In order to comprehend the subsurface hydrologicaland geological structure, not only seismic and/or borehole loggingmethods but also geoelectrical methods, such as direct current electricand electromagnetic (EM) methods, are important because thesemethods can provide resistivity information related to saline/freshgroundwater (Fig. 1). Furthermore, EM methods are more suitablethan the direct current electric method for geoelectrical investigationof HLW geological disposal because some EM methods can be used toinvestigate deep geoelectrical structure at depths of approximately

EM R

Geophone

Land

Saline/Freshwater

Interface

Fault

Fig. 1. Sketch of geophysical explorations of a coastal area. Geophones and ocean bottom cables

24 T. Ueda et al. / Journal of Applied Geophysics 100 (2014) 233110km using low-frequency EM signals.Both onshore (e.g., Nabighian, 1991) and offshore (e.g., Constable,

1990, 2010), EMmethods have been developed and successfully applied

43

450 100

km

Nor

th la

titud

e

Tenpoku region

HokkaidoIsland

HoronobeStudy area139 141 143 145

41

East longitude

TokyoHonsyuu Island

Fig. 2.Map of Hokkaido Island and location of the study area.in commercial exploration and academic research for the past fewdecades. Although there has been signicant progress, especially indeepwater (water depths greater than approximately 300 m) EMexploration (e.g., Chave et al., 1991; Constable and Srnka, 2007;Constable et al., 1998; Hoversten et al., 1998; Kasaya and Goto, 2009;Kaya et al., 2013; Key, 2011; Key et al., 2013), only a few results andreports of EM surveys in shallow water (water depths of less thanapproximately 300 m) and/or coastal regions (e.g., Andris andMacGregor, 2008; Hoversten et al., 2000; Mittet and Morten, 2013;Weiss, 2007) have been presented. Moreover, very little research anddevelopment has been conducted (e.g., Yoshimura et al., 2006) in seaareas with water depths ranging from 5 to 100m, which is referred toherein as very shallow water.

From a practical point of view, the difculties of EM exploration invery shallow water and in coastal areas lie in (1) shery activity,(2) limitations of the survey vessel size, and (3) motion noise causedby sea waves (Fig. 3) (Fournier and Reeves, 1986; Kinsman, 1965),especially for frequencies from 102 to 100 Hz (period: 1 to 100 s)(e.g., Webb, 1998; Webb and Crawford, 2010). Under these conditions,the MT method is suitable because the MT method simply observesnatural EM elds without transmitting any electric current in water.Therefore, the MT method does not negatively impact shery activityin coastal areas.

However, the MT method requires the measurement of EM signalstypically in the frequency range from 103 to 104 Hz (Vozoff, 1991),including 102 to 100Hz, which are subject to contamination bymotionnoise due to seawaves (Fig. 3). In addition tomotion noise, very shallowwater at coastal areas makes it difcult to operate deep-draft surveyships having adequate payloads for EM survey equipment. For example,in the Horonobe coastal area, the sea depth is approximately 10 m ata distance of 1 km from the shoreline. In order to overcome thesedifculties, we have developed a new marine MT measurement (oceanbottom electromagnetometer, OBEM) system for MT data acquisition invery shallow water, such as coastal regions. The new marine MTmeasurement system is very short in order to reduce motions causedby sea waves and is compact in order to allow light-draft small surveyboat operation.

We began our research by conducting a preliminary hardware testand data acquisition examination using a synthetic signal generatedby a controlled source at test facilities and conducted eld experimentsinvolvingMT data measurement at the coastal area in Horonobe during

eceiver

OBC* Air gun

Offshore

SeaSeafloor

(OBCs) for seismic exploration and receivers for electromagnetic (EM) eldmeasurement.2010 and 2011. We also conducted onshore geophysical investigations,including seismic reection and onshore MT surveys. Using bothonshore and offshore MT data simultaneously, we performed 2Dinversion in order to estimate a 2D resistivity model that is orthogonalto the shoreline. In addition to these geophysical explorations, we alsoconducted a 1000-m-deep drilling and well logging on land near thecoastline of Horonobe eld to obtain detailed geological informationon the area. Resistivity is a function of porosity and pore uid resistivity.Geological and hydrological interpretation based solely on the resistivityobtained by EM methods is difcult. Therefore, all geoscientic infor-mation obtained by various investigations described above have beeninterpreted in order to clarify the geological and hydrological structureunder the Horonobe coastal region.

2. Seaoor MT receiver

2.1. Instrument and its specications

ForMT data acquisition in coastal regions, we have developed a newmarine MT measurement system that is able to operate at sea waterdepths ranging from 5 to 100 m and to survey to depths of a fewkilometers below the seaoor. The new MT receiver system consists ofthree primary components: a data logger, EM sensors, and a baseboardonto which the other components are xed (Fig. 4).

The data logger unit contains an electric eld preamp, an antialiasinglter, a 24-bit analog-digital converter for each of the electric and

For electric eld measurement, we used AgAgCl non-polarizingelectrodes, which are xed at the ends of 5-m exible pipes (Figs. 4and 5) to measure two horizontal electric voltages. ConventionalOBEMs, such as those presented by Constable et al. (1998), Toh et al.(1998), and Kasaya and Goto (2009), have typical 10-m electrodearms. However, the proposed OBEM has 5-m arms in order to preventmotion noise caused by sea waves and to assist loading on the reardeck of small survey boats. Data logger and magnetic sensors in eachpressure-resistant container and two pairs of electrodes are xed to atriangle shape baseboard shown in Figs. 4 and 5. Table 1 lists thespecications for each system.

A small disposable compass is also attached to the baseboard, whichrecords the Earth's magnetic eld strength (in three directions), tilt (in

ag (Fig. 6f) are deployed sequentially using the rope. In order tox the MT receiver on the seaoor, the baseboard and two anchors

10-6 10-5 10-4 10-3 10-2 10-1 100 101 102

Frequency Hz

Ener

gy100 s 10 s 1 s

Fig. 3. Simple sketch of the energy contained in the surface waves of the ocean dependingon period (frequency) based on Kinsman (1965, p. 23).

25T. Ueda et al. / Journal of Applied Geophysics 100 (2014) 2331magnetic components, a CPU, a crystal clock, and a data storagemodulewith a CF card type memory. Two sets of six D-cell (1.5 V) batteries(total: 9V) supply the digital module, and four AA-cell (1.5V) batteries(total: 6V) support the analoglter unit. Four 9-V batteries are connectedin parallel to supply electric power to the analog amplier of themagnetic sensor. The system supports at least 72 h of data acquisition.The data logger and batteries are stored in a pressure-resistant containermade of stainless steel having dimensions of 515mm by 165mm and awater-pressure resistance of 3.5MPa (350m).

We use commercial induction coils (BF-4 sensors manufactured bySchlumberger-EMI) as a magnetic eld sensor having a broadbandfrequency range of 104 to 700 Hz with a length of 1420 mm and adiameter of 60 mm to measure two horizontal magnetic components.The induction coil is stored in the newly developed pressure-resistantcontainer constructed from a non-magnetic GFRP pipe of 1670mm inlength and 65/85mm (ID/OD) in diameter. The container has a doubleO-ring connector and a water-pressure resistance of 3MPa (300m).

860 mm

1670 mm

5000

mm

c

da b

c

b ed1970 mm

Fig. 4. Plan view of the new EM receiver system.(a) Data logger with a pressure-resistantcontainer, (b) magnetic sensor (induction coil) with a pressure-resistant container, (c)AgAgCl electrode, (d) electric dipole pipe, and (e) baseboard.are connected to a 100-m rope (Fig. 6b) that has a visible markerag on the sea-level end (Fig. 6f). This rope is also used to retrievethe anchors and the MT receiver using a winch.

On the seaoor, theMT receiver measures the electric and magneticelds and stores them inash cardmemory. Theorientation andmotionof the receiver are also recorded by a compass during measurement.Typically, after three days (more than 60 h) of data acquisition, thesurvey boat with an empty deck returns to the deployment pointand recovers the equipment. Retrieval of the MT receiver is simplyperformed by reversing the deployment operation. The buoy and agare rst captured, and then the two anchors and the EM receiver areretrieved by pulling up the rope using a winch. The recoverythree directions), acceleration, temperature, and depth. The compass is15mm by 46mm and has non-volatile data storage memory (261,600bytes, each magnetic data: 13 bytes, temp-depth: 3 bytes) and a lifebattery of 1.5 years. A detailed description of the function and thetechnical specication of the compass are provided on the Star-Oddi web site (http://www.star-oddi.com/).

The height of the newly developed MT receiver including thebaseboard, the data logger, the induction coils, and their pressure-resistant containers is less than 250 mm, whereas some conventionalpop-up type OBEMs with glass spheres are 1000 mm in height. Thisshortness makes the proposed MT receiver more stable with respectto motion noise caused by sea waves.

2.2. Field operation

In our eld observation, two sets of MT receivers (each set includestwo anchors, and a buoy tied to a rope, as shown in Fig. 6) are loadedonto the survey boat for each measurement operation. Before loadingthe equipment onto the boat, data acquisition is started by sending aturn-on signal through the control computer. Clock synchronization isalso performed by GPS. The measurement point and the survey boatposition are located using GPS instruments.

At the measurement site, the MT receiver with baseboards isdeployed, and then two anchors (Fig. 6c and d) and a buoy with aFig. 5. Photograph of the new marine MT receiver.

operation without pop-up devices such as glass spheres enables thesystem to be short andmore stable with respect tomotion noise causedby sea waves. Moreover, the method leaves nothing behind, whereasconventional OBEMs are sometimes recovered while leaving concreteanchors on the seaoor. This remains-free data acquisition is importantfor geophysical exploration in areas where coastal shing is active.

disposal through the investigation of the actual deep geological

Table 1MT receiver system specications.

SizeHeight 250mm (max)Weight 143.5 kg (in the air)

44.0 kg (in water)Water resistance Up to 250mDipole length 5m

Data logger/electronicsChannel 4 (Hx, Hy, Ex, Ey)A/D 24bit/1chFrequency band DC 5000HzData storage 8GB Compact ash cardSynchronization GPS (under 1 sec)Power supply D-cell: 12 pieces for digital

AA-cell: 4 pieces for analog

26 T. Ueda et al. / Journal of Applied Geophysics 100 (2014) 2331fg3. Field experiments and data processing

3.1. Study area

We have conducted eld experiments on the proposed marineEM receiver system at the Horonobe coastal area in the northeastpart of Hokkaido Island, Japan. The Horonobe coastal area is locatedin the western part of the Tenpoku region (Fig. 2). This area ispart of a sedimentary coastal basin, which is composed of poorlycompacted Neozoic sand-, silt-, and mudrocks. These formationsare unconformably overlain by terrace deposits, alluvium, andlagoonal deposits (unconsolidated deposits of gravel, sand, and mud)(e.g., Koshigai et al., 2012).

In this region, the Horonobe Underground Research Laboratory(URL) of the Japan Atomic Energy Agency (JAEA) and several researchinstitutes have been conducting scientic and technical research anddevelopment in order to clarify the basic knowledge for geologicala b c d

e

seafloor

Fig. 6. Sketch of the newmarine MTmeasurement system.(a) Electric and magnetic eldreceiver, (b) rope, (c) and (d) anchors, (e) water owmeter, (f) buoy with a ag, and (g)survey boat.environment in sedimentary rocks and various geophysical, geochemical,and hydrological research (Hama et al., 2007; Tokiwa et al., 2013).

Since 2007, we have also been conducting research on hydrological(Ikawa et al., 2012; Ito et al., 2010; Koshigai et al., 2012) and geophysicalmethods for seamlessly evaluating geological structure from land to sea.In addition to the availability of geoscientic information obtainedby these various previous and ongoing studies, the sea depth of theHoronobe coastal region is also important to the present study. In thisarea, the sea depth is shallower than 60m, even at a distance of 10kmfrom the shoreline with a gradient of 1/1000. Therefore, this areais suitable as a test site for offshore, very-shallow-water MT surveysusing the newly developed MT measurement system.

3.2. Field experiment

We conducted eld experiment observations at the Horonobeshallow-water area in order to evaluate the performance of the proposedMT receiver. In 2010 and 2011, we conducted data acquisition at 24measurement sites (MA0MA11, MB0MB11) aligned at approximately1000-m intervals along two parallel survey lines (Lines A and B), eachof which is approximately 12km long (Fig. 7).We set two parallel surveylines that are aligned approximately perpendicular to the shoreline inorder to conrm the two-dimensionality of geological formation, becausethe geologic setting indicates that the coastal plain has an almost 2Dgeologic structure with a northeastsouthwest orientation (e.g., Amoet al., 2007; Tokiwa et al., 2013).

The sea depths at sites MA0 and MB0, nearest the shoreline, areapproximately 4 m, and those of MA7 and MB7 are approximately45 m. The site names and corresponding water depths are shown inTable 2. Because of the very-shallow-water depth of the Horonobecoastal area, the deployment and recovery operations were carried outusing a ve-ton class shing boat having a length of 11.85 m and adisplacement of 4.8 T (Fig. 8). The MT data were acquired for a periodof at least three days (including two nights) atmost of the survey points.However, at MA0 and MB0, because of poor weather conditions, werecovered the EM receivers one day after deployment in order toavoid the risk of being buried by seaoor sand.

3.3. Data processing

At all survey sites, wemeasured the natural EM elds between 104

and 700Hz as the time series data. The observed data, which are storedon a compact ash memory card, are immediately downloaded toa personal computer upon arrival at port. The downloaded data aretransformed into data processing format, and three frequency band datasets are generated at different re-sampling rates. For the high-frequency(HF) band, 500-Hz sampling is selected, which is approximately thesame as the original sampling rate for data acquisition. For the mid-frequency (MF) band and the low-frequency (LF) band, the samplingrates are set to 62.5 and 8.7125 Hz, respectively. Table 3 presents thespecications for the eld test.

For the entire measurement period of the time series data, high-quality data ranges are selected by the operator for use in dataprocessing of the data of each frequency band. The power spectrumof the electric and magnetic elds is computed with the cascadedecimation (Wight and Bostick, 1980) using the selected high-qualitytime series data and then transformed into an impedance tensor. Inorder to obtain better quality data over the entire frequency range, weapplied remote reference processing (e.g., Vozoff, 1991) by setting upa remote site in the northeastern area of Honshu Island, Japan, whichis approximately 500 km from the marine MT measurement sites(shown in Fig. 7 as MT Remote Reference). The remote referenceprocessing successfully improved the MT data quality of both land andmarine data.

The apparent resistivity and phase computed at each frequency

band are combined to obtain the apparent resistivity and the phase

MB7MB0

MA9

MA1Lin

MB4

Land MT

DD1 (Borehole)Onshore seismic

Marine MT (2010)Marine MT (2011)2D MT inversion line

ager

hit

1102

d wasta

27T. Ueda et al. / Journal of Applied Geophysics 100 (2014) 2331curves for the entire frequency range. Finally, we obtained the apparentresistivity, the phase, and the other MT parameters. Fig. 9 shows the

MB11

Im

TesPor

141E

145E42N

0 100km

45NStudy area

MT Remote Reference

Fig. 7.Map of various geophysical surveys including marine MT (open squares for 2010 anline), 2D MT inversion line (solid line), and DD1 drilling location (star) in the Horonobe coapparent resistivity and phase of both onshore sites (1102 and 1119)and offshore sites (MA1,MB4, MB7, andMA9). Although there is almostno human activity noise in the Horonobe offshore region, sea wavemotion noise occurs primarily between 0.1 and 1 Hz, especially atshallow-sea-depth stations. Noise contamination with a period of 0.1to 1 Hz is clear in both the apparent resistivity and phase curves atMA1, where the sea depth is 3.9m (Fig. 9c). Although sea wave motionnoise still exists atMB4 andMB7 (Fig. 9d and e), the effective frequencyrange becomeswide. AtMA9with a sea depth of 51.6m,we can see verylittle sea wave motion noise, even with a period of from 0.1 to 1Hz andobtain high-quality data over the entire frequency range (Fig. 9f).Although remote referencing could improve the data quality, there arestill noise contaminations effected by sea wave motion, especially inthe frequency range from 0.1 to 1 Hz at shallow-water sites such asMA1. This is because remote referencing is not able to eliminate localnoise completely but can reduce noise. For better quality data, including0.1 to 1Hz at shallow-water sites, long-time data acquisition and calm

Table 2Seaoor MT station and sea depth.

Line A site (depth m) Line B site (depth m)

MA0 (3.9) MB0 (3.8)MA1 (7.5) MB1 (8.3)MA2 (13.4) MB2 (13.2)MA3 (19.4) MB3 (19.6)MA4 (26.8) MB4 (26.1)MA5 (32.6) MB5 (35.1)MA6 (38.0) MB6 (40.3)MA7 (44.0) MB7 (43.0)MA8 (48.1) MB8 (47.6)MA9 (51.6) MB9 (51.5)MA10 (54.9) MB10 (54.1)MA11 (57.9) MB11 (57.3)weather conditions are important, as is the case with onshore MTmeasurement.

1230

1129e A

Line B

Teshio River

0 5km

y 2013 TerraMetrics, Map data 2013 Google, ZENRIN

HoronobeTown

TeshioTown

o

JAEAHoronobe URL*

1119

N

hite squares for 2011), land MT (small white dots with gray edges), land seismic (dashedl area. (Google Maps, Imagery (c) 2013 TerraMetrics, Map data (c) 2013 Google, ZENRIN).4. Data inversion and interpretation

In addition to offshore MT measurements, land MT data acquisitionwas conducted at measurement sites (stations 11011129 alongLine A and 12011230 along Line B) aligned at approximately 500-mintervals along two parallel survey lines (Lines A and B in Fig. 7)connecting seaoor MT survey lines.

According to pre-existing results of geological and geophysical studiesof this region (e.g, Amo et al., 2007, Tokiwa et al., 2013), geologicalformation has the two-dimensionality in northeastsouthwest directionthat is orthogonal to the shoreline. Our onshore and offshore MT surveylines are also set to parallel to this two-dimensionality. Then, usingboth onshore and offshore MT data simultaneously, we performed2D inversion of TE, TM, and tipper data for each of the two parallelproles. The total length of MT proles A and B is approximately22km. Each of the sea and land parts is approximately 11km in length.There are 29 and 30 MT sites on land proles, as indicated by Lines Aand B (Fig. 7), respectively. In offshore survey lines A and B, there are12 seaoor MT sites along each line.

In order to invert the land and seaoor MT data, we used asmoothness-constrained 2D inversion (Uchida and Ogawa, 1993).In this inversion scheme, the earth was divided into multiple blocksof constant resistivity, and the logarithm of each block's resistivity wasadopted as the desiredmodel parameter. In the inversion algorithm, theobjective function U to be minimized is given by

U WdobsWF m j jj j2 Dmj jj j2; 1

where dobs is a set of observed data, m is a vector of unknown modelparameters, F is a nonlinear forward modeling function with respectto a specic resistivity distribution, W is a diagonal weighting matrix,

e M

28 T. Ueda et al. / Journal of Applied Geophysics 100 (2014) 2331D is a Laplacian operator matrix that computes the roughness of m,and is the regularization or trade-off parameter. The parameter determines the relative signicance in the objective function of therst and second terms in the right-hand side of Eq. (1). Estimationof m depends on the value of . The most appropriate value of isdetermined using an empirical Bayesian approach, namely, Akaike'sBayesian information criterion (ABIC) minimization (e.g., Mitsuhata,2004; Mitsuhata et al., 2002). The TE-mode and TM-mode a and were adopted as dobs, and the inversion process was performediteratively. For each iteration, the Jacobian matrix is computed for theprevious iteration's model vectorm, andm is updated by minimizationofU, given the optimumvalue of. In order to computeMT responses ofoffshore measurements, a modied version of Uchida and Ogawa(1993), which is able to calculate electric and magnetic elds on theseaoor by Goto (2010, personal communication) is implemented as aforward modeling calculation method.

In inversion, a total of 34 frequencies ranging from 0.0134 to1300 Hz for onshore sites and 29 frequencies ranging from 0.0134 to115Hz for offshore sites are used. At every site, the data contaminatedwith noise, such as the data within the frequency range of 102 to 100

of the sites under shallow water (e.g., MA1 in Fig. 9c), are excluded

Fig. 8. Photograph of the survey boat (left) and the deployment of thfrom inversion. FEM forward modeling includes topography andbathymetry. A rectangular cell having a depth of 2 m and a length of60 m at the surface is used for FEM discretization. The cell sizes invertical direction are extended with respect to depth. The seawaterresistivity is xed at 0.23 -m and was not changed during iterativeoptimization.

The nal 2D resistivity model of Line A calculated by the iterativeinversion of both land and seaoor MT data is shown in Fig. 10 alongwith seismic interpretation. Predicted data computed with the 2Dmodel are presented in Fig. 9 as solid and dashed lines. A 2D resistivitymodel and seismic interpretation clearly indicate that a quaternarysedimentary layer of a few hundred meter thickness, which wasdetermined to be a fresh water layer by the drilling, extends for a

Table 3Field-test specications.

Number of survey sites 16 sites (2010)/8 sites (2011)

Measurement time Maximum 3 of nightsSampling frequency 500HzMeasurement component Hx, Hy, Ex, EyRemote reference Over 500 km far from the test elddistance of several kilometers under the sea. We have also obtainedsimilar results for the 2D inversion of Line B.

In the onshore Horonobe coastal region, a seismic reection surveywas carried out along MT prole line A in this project (Fig. 7). Basedon a P-wave reection survey, Yokota et al. (2012) presented a depth-converted migration section with an estimated geological interface(Fig. 11).

We also performed well logging at a depth of 1000m on land nearthe coastline and obtained detailed geological information for the area(e.g., Ikawa et al., 2012; Koshigai et al., 2012). The location of the1000-m borehole, denoted by DD1, is also shown in Fig. 7. The resultingchlorine concentration in pore water (mg/l) as a function of the porewater sampling depth from DD1 is shown in Fig. 12(a). The resistivitylog data from the surface to a depth of 1000 m at the DD1 drilling isalso shown in Fig. 12(b). It is clear that, below 500 m, the high Clconcentration corresponds to the low-resistivity log results.

In the 2D resistivity model (Fig. 10), beneath the onshore MTsites (11011111), there is a thin (1030m) relatively conductive(110 -m) layer in the resistive background (10100 -m) at adepth of approximately 50 m. We also found this conductive layerthrough our onshore time-domain EM survey. According to Sakai et al.

T measurement system using the rear deck crane and winch (right).(2011, 2012), this layer has been interpreted as a lagoon deposit ofsilt clay. In the present paper, we do not focus on this conductive layerin our hydrogeological interpretation of the entire Horonobe coastalarea. However, this conductive layer is important for shallower andmore local hydrological research (Sakai et al., 2011, 2012).

We then interpreted these geophysical and geological results inorder to evaluate the structures of the Late Neogene and quaternarysedimentary layers and to estimate the distribution of saline andfreshwater aquifers beneath land and beneath the seaoor. Accordingto thewell logging data (Fig. 12) andKoshigai et al. (2012), a quaternarysedimentary layer, the Sarabetsu Formation, has a thickness ofapproximately 500 m at the coastline and is overlain by an alluviumformation of approximately 100 m in thickness. Below the SarabetsuFormation is the Yuchi Formation, which is a late Neogene sedimentarylayer that is approximately 1km thick at the drilling location. The YuchiFormation is underlain by the Neogene Koetoi Formation.

The 2D resistivity model (Fig. 10) and thewell logging data (Fig. 12)also indicate that Neogene sedimentary layers have low resistivitybecause of fossil saline water included during the time of submarinesedimentation (Hama et al., 2007; Koshigai et al., 2012). In contrast,the Quaternary Sarabetsu Formation is resistive on land because theconductive saline water during the sedimentation was later washedout by resistive fresh rainwater.

29T. Ueda et al. / Journal of Applied Geophysics 100 (2014) 233110

100

1000

sistiv

ity

-m

a) 1119YXXYThe geoelectrical model (Fig. 10) shows that upper part of theSarabetsu Formation beneath the seaoor also exhibits relatively highresistivity (approximately 10 to 100-m), although this area is underseawater. This high-resistivity quaternary formation extends morethan 10 km offshore. We estimate that this resistive zone below the

c) MA1YXXY

YXXY

0.01

0.1

1

10

100

1000

Appa

rent

Res

istivi

ty

-m

-90

0

90

180

Frequency Hz

Phas

e de

gree

-180

e) MB7

0.01

0.1

1

10

100

1000

Appa

rent

Res

istivi

ty

-m

-180

-90

0

90

180

Frequency Hz

Phas

e de

gree

YXXY

YXXY

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

0.01

0.1

1

Appa

rent

Re

-180

-90

0

90

180

Frequency Hz

Phas

e de

gree

YXXY

Fig. 9. Apparent resistivity (top) and phase (bottom) curves at MT sites (a) 1119 and (b) 1102,component, and the white squares indicate the yx component. Predicted data computed wicontaminated with noise, such as the data within the frequency range of 102 to 100 of the sitb) 1102

10

100

1000

istivi

ty

-m

YXXYseaoor was created by the regressiontransgression sequence of thecoastline during cycles of glacial and interglacial ages.

Meisler et al. (1984) and Cohen et al. (2010) presented similarmodels for the existence of relatively fresh groundwater beneath theAtlantic continental shelf of North America. The hydrological model of

0.01

0.1

1

10

100

1000

Appa

rent

Res

istivi

ty

-m

d) MB4

-180

-90

0

90

180

Frequency Hz

Phas

e de

gree

YXXY

YXXY

f) MA9

0.01

0.1

1

10

100

1000

Appa

rent

Res

istivi

ty

-m

-180

-90

0

90

180

Frequency Hz

Phas

e de

gree

YXXY

YXXY

0.01

0.1

1

Appa

rent

Res

-90

0

90

180

Frequency Hz

Phas

e de

gree

-180

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

1E-4 1E-3 1E-2 1E-1 1E+0 1E+1 1E+2 1E+3

YXXY

and offshore sites (c) MA1, (d) MB4, (e) MB7, and (f) MA9. The black dots indicate the xyth the inversion result model are shown as solid (XY) and dashed (YX) lines. The dataes under shallow water, are excluded from inversion.

75 75A67-75A81.

-2.0

-1.5

-1.0

-0.5

0.0

Elev

atio

n (km

)

0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0

MA

11

MA

10M

A9

MA

8M

A7

MA

6M

A5

MA

4

MA

3M

A2

MA

1M

A0

1101

1102

1103

1104

1105

1106

1107

1108

1109

1110

1111

1112

1113

1114

1115

1116

1117

1118

1119

1120

1121

1122

1123

1124

1125

1126

1127

1128

1129V.E.=5

Seawater

Sarabetsu

Yuchi

Koetoi

Alluvium

Shoreline Sarobetsu Rv. DD-1 (Borehole) Sarobetsu

Anticline

Offshore Onshore

30 T. Ueda et al. / Journal of Applied Geophysics 100 (2014) 2331the Horonobe coastal area based on our geophysical investigation isconsistent with their models. Further investigations, such as offshoreseismic investigations and drilling investigations, are required inorder to conrm the geological and hydrological structure beneaththe seaoor.

5. Conclusions

Wehave developed a newmarineMTmeasurement system that canbe applied to measurement in a shallow seawater area, such as coastalregions. The difculties of EM explorations in shallowwater and coastalareas include (1) shery activity, (2) limitations of survey vessel size,and (3) motion noise caused by sea waves.

In order to overcome these difculties, we selected a magnetotelluric(MT)method that uses natural EMeldswithout transmitting an electric

Distance (km)

10001001010.1ohm-m

Fig. 10. Two-dimensional electrical resistivity model for the Horonobe coastal areaobtained by inversion of land and seaoor MT data. The stratum boundaries estimatedbased on an existing seismic reection survey (Fig. 11) on land are overlaid (blue lineswith geological layer names)current and/or amagneticeld in seawater,which enables themethod tobe used in areas with active inshore sheries. In addition, the developedmarine MT system is very short in order to reduce motion generated bysea waves and is compact, which enables light-draft small survey boatoperation.

We then conducted an onshore MT survey and offshore dataacquisition using the proposed marine MT system at the Horonobecoastal area, Hokkaido, Japan. High-quality data were successfully

Ikawa, R., Machida, I., Koshigai, M., Nishizaki, S., Uchida, T., Marui, A., 2012. Estimation of

Dep

th (m

)

0

500

1000

1500

2000

Sarabetsu

Yuchi

Koetoi

Distance (m)70006000500040003000200010000 8000

Fig. 11. Depth converted migration section along the P-wave reection survey line shownin Fig. 7 by the dashed line labeled Onshore seismic. The solid line indicates thegeological interface (after Yokota et al., 2012).origin for the water salinization in different aquifers at the coastal area. 22nd SaltWater Intrusion Meeting Proceedings, pp. 7376.

Ito, N., Koshigai, M., Marui, A., 2010. Preliminary model for simulation of groundwaterow and seawater/freshwater interface at coastal area of Horonobe, Hokkaido.J. Groundw. Hydrol. 52, 381394.

Kasaya, T., Goto, T., 2009. A small ocean bottom electromagnetometer and ocean bottomelectrometer system with an arm-folding mechanism. Explor. Geophys. 40, 4148.

Kaya, T., Kasaya, T., Tank, S.B., Ogawa, Y., Tuncer, M.K., Oshiman, N., Matsushima, M., 2013.Electrical characterization of theNorth Anatolian Fault Zone underneath theMarmaraSea, Turkey by ocean bottom magnetotellurics. Geophys. J. Int. 193, 664677.

Key, K., 2012. Marine electromagnetic studies of seaoor resources and tectonics. Surv.Constable, S.C., Srnka, L.J., 2007. An introduction tomarine controlled-source electromagneticmethods for hydrocarbon exploration. Geophysics 72, WA3WA12.

Constable, S.C., Orange, A.S., Hoversten, G.M., Morrison, H.F., 1998. Marinemagnetotelluricsfor petroleum exploration Part 1: a sea-oor equipment system. Geophysics 63,826840.

Fournier, A., Reeves, W.T., 1986. A simple model of ocean waves. Proceedings of the 13thAnnual Conference on Computer Graphics and Interactive Techniques, pp. 7584.

Groen, J., Velstra, J., Meesters, A.G.C.A., 2000. Salinization processes in paleowaters incoastal sediments of Suriname: evidence from d 37 Cl analysis and diffusion modelling.J. Hydrol. 234, 120.

Hama, K., Kunimaru, T.,Metcalfe, R.,Martin, A.J., 2007. The hydrogeochemistry of argillaceousrock formations at the Horonobe URL site, Japan. Phys. Chem. Earth 32, 170180.

Hathaway, J.C., Poag, C.W., Valentine, P.C., Manheim, F.T., Kohout, F.a., Bothner, M.H.,Miller, R.E., Schultz, D.M., Sangrey, D.A., 1979. U.S. geological survey core drilling onthe Atlantic shelf. Science 206, 515527.

Hoversten, G.M., Morrison, H.F., Constable, S.C., 1998. Marine magnetotellurics forpetroleum exploration part 2: numerical analysis of subsalt resolution. Geophysics63, 826840.

Hoversten, G.M., Constable, S.C., Morrision, H.F., 2000. Marine magnetotellurics for base-of-salt mapping: Gulf of Mexico eld test at the Gemini structure. Geophysics 65,14761488.obtained in both onshore and offshore eld surveys. 2D inversionfor eld data from onshore to the sea bottom has revealed that aquaternary sedimentary layer of a few hundred meters in thickness,whichwas determined by drilling to be a freshwater layer, extends to adistance of several kilometers under the sea.

The results of the present study indicate that the proposedmarineMTsystem can be used to clarify brackish/fresh groundwater distributionsand structures in coastal areas for the purpose of HLW site evaluation.

Acknowledgments

The authors would like to thank Captain T. Mori of the survey boatand its survey crew and the Kitarumoi Fishery Cooperative Associationand theGroundwater ResearchGroup of the Institute for Geo-Resourcesand Environment, AIST for their assistance with marine MT dataacquisition. We would also like to thank Horonobe Town and theJapan Atomic Energy Agency (JAEA) for their support in conductingeld surveys, including on/offshore EM exploration, the land seismicsurvey, and the well logging. We would also like to thank the twoanonymous reviewers for their valuable comments and contributionsto this paper. The present study was conducted as a part of a researchproject investigating the interface between freshwater and saltwaterfunded by the Agency of Natural Resources and Energy, Minister ofEconomy, Trade and Industry, Japan.

References

Amo, M., Suzuki, N., Shinoda, T., Ratnayake, N.P., Takahashi, K., 2007. Diagenesis anddistribution of sterenes in Late Miocene to Pliocene marine siliceous rocks fromHoronobe (Hokkaido, Japan). Org. Geochem. 38, 11321145.

Andris, D., MacGregor, L., 2008. Controlled-source electromagnetic sounding in shallowwater: principles and applications. Geophysics 73, F21F32.

Chave, A.D., Constable, S.C., Edwards, R.N., 1991. Electrical exploration methods for theseaoor. In: Nabighian, M.N. (Ed.), Electromagnetic methods in applied geophysics.Applications part B, 2. Society of Exploration Geophysicists, Tulsa, pp. 931966.

Cohen, D., Person, M., Wang, P., Gable, C.W., Hutchinson, D., Marksamer, A., Dugan, B.,Kooi, H., Groen, K., Lizarralde, D., Evans, R.L., Day-Lewis, F.D., Lane, J.W., 2010. Originand extent of fresh paleowaters on the Atlantic continental shelf, USA. GroundWater48, 143158.

Constable, S.C., 1990. Marine electromagnetic induction studies. Surv. Geophys. 11,303327.

Constable, S.C., 2010. Ten years of marine CSEM for hydrocarbon exploration. GeophysicsGeophys. 33, 135167 (ishttp://link.springer.com/article/10.1007/s10712-011-9139-x).

2100

0

allu

vium

nst

one

ba c

al., 2t DD

31T. Ueda et al. / Journal of Applied Geophysics 100 (2014) 2331Key, K., Constable, S., Liu, L., Pommier, A., 2013. Electrical image of passive mantleupwelling beneath the northern East Pacic Rise. Nature 495, 499502.

0 5000 10000 15000Cl mg/l

Dep

th (m

)

1000

900

800

700

600

500

400

300

Sara

betu

con

glom

erat

e, s

ands

tone

, aYu

chi

muds

tone

and

san

dsto

ne

Fig. 12. (a) Geological formation, (b) chlorine concentration in pore water (after Ikawa et(depth: 1 000 m) and resistivity data plot (dashed lines) obtained from 2D MT inversion aKinsman, B., 1965. WindWaves: Their Generation and Propagation on the Ocean Surface.Prentice-Hall, Inc., Englewood Cliffs, New Jersey 23.

Koshigai, M., Marui, A., Igarashi, Y., Akiba, F., Furusawa, A., Oka, T., Hokkaido, N., 2012.Lithofacies, microfossils and tephra observed by deep drilling in the coastal zonesat Horonobe, Northern Hokkaido, Japan (in Japanese). Bull. Geol. Surv. Jpn. 63 (9/10), 233267.

Kozai, N., Ohnuki, T., Iwatsuki, T., 2013. Characterization of saline groundwater atHoronobe, Hokkaido, Japan by SEC-UV-ICP-MS: speciation of uranium and iodine.Water Res. 47, 15701584.

Marui, A., 2003. Groundwater conditions along the seawater/freshwater interface on avolcanic island and a depositional area in Japan. Geol. Q. 47, 381388.

Marui, A., Hayashi, T., 2001. Hydrological research for understanding of 3D shape of theseawater/freshwater interface (in Japanese). J. Min. Mater. 117, 816821.

Meisler, H., Leahy, P.P., Knobel, L.L., 1984. Effect of eustatic sea-level changes on saltwater-freshwater in the Northern Atlantic coastal plain. USGS Water Supply Paper, 2255.U.S. Geological Survey.

Mitsuhata, Y., 2004. Adjustment of regularization in ill-posed linear inverse problems bythe empirical Bayes approach. Geophys. Prospect. 52, 213239.

Mitsuhata, Y., Uchida, T., Amano, H., 2002. 2.5-D inversion of frequency-domain electro-magnetic data generated by a grounded-wire source. Geophysics 67, 17531768.

Mitsuhata, Y., Uchida, T., Matsuo, K., Marui, A., Kusunose, K., 2006. Various-scale electro-magnetic investigations of high-salinity zones in a coastal plain. Geophysics 71,B167B173.

Mittet, R., Morten, J.P., 2013. The marine controlled-source electromagnetic method inshallow water. Geophysics 78, E67E77.

Electromagnetic methods in applied geophysics. In: Nabighian, M.N. (Ed.), Application,Parts A and B, vol. 2. Society of Exploration Geophysicists, Tulsa.

Sakai, T., Ioka, S., Ishijima, Y., Igarashi, T., 2011. Geological analysis of alluvium in theSarobetsu moor (in Japanese). J. Jpn. Soc. Eng. Geol. 52, 213.

Sakai, T., Ioka, S., Igarashi, T., 2012. Hydrogeological structure and groundwater ow ofalluvium in the south Sarobetsu moor (in Japanese). J. Jpn. Soc. Eng. Geol. 53,172182.Sugihara, K., 2009. Geological disposal of high-level radioactive waste and the role of rockengineering. Int. J. Jpn. Comm. Rock Mech. 5, 1924.

0000 0.1 100 10001 10Resistivity (ohm-m)

Resistivity LogMT result

012), and (c) resistivity log results plot (solid line) observed in the borehole location DD11.200 d m

udTanaka, M., Kawabata, J., Toida, M., Hishiya, T., Sugimoto, E., Nishigaki, M., Hasegawa, T.,2003. A groundwater ow analysis at coastal area considering regression andtransgression. In: Kono, I. (Ed.), Groundwater Engineering Recent AdvancesInternational Symposium on Groundwater problems related to Geo-Environment.Taylor & Francis, pp. 501508.

Toh, H., Goto, T., Hamano, Y., 1998. A new seaoor electromagnetic station with anOverhauser magnetometer, a magnetotelluric variograph and an acoustic telemetrymodem. Earth Planets Space 50, 895903.

Tokiwa, T., Tsusaka, K., Matsubara, M., Ishikawa, T., Ogawa, D., 2013. Formationmechanism of extension fractures induced by excavation of a gallery in softsedimentary rock, Horonobe area, Northern Japan. Geosci. Front. 4, 105111.

Uchida, T., Ogawa, Y., 1993. Development of Fortran code for two-dimensionalmagnetotelluric inversion with smoothness constraint. Geological Survey of JapanOpen-File Report.

Vozoff, K., 1991. The magnetotelluric method. In: Nabighian, M.N. (Ed.), ElectromagneticMethods in Applied Geophysics. Applications Part B, vol. 2. Society of ExplorationGeophysicists, Tulsa, pp. 641711.

Webb, S.C., 1998. Broadband seismology and noise under the ocean. Rev. Geophys. 36,105142.

Webb, S.C., Crawford, W.C., 2010. Shallow-water broadband OBS seismology. Bull.Seismol. Soc. Am. 100, 17701778.

Weiss, C.J., 2007. The fallacy of the shallow-water problem in marine CSEM exploration.Geophysics 72, A93A97.

Wight, D., Bostick, F., 1980. Cascade decimation a technique for real time estimation ofpower spectra. ICASSP 80 IEEE International Conference on Acoustics Speech andSignal Processing, 5, pp. 626629.

Yokota, T., Inazaki, F., Mizohata, S., Uchida, T., Ueda, T., 2012. Seismic reection surveys atHoronobe coastal land area (in Japanese). Geophys. Explor. (BUTSURI-TANSA) 65,161172.

Yoshimura, K., Okubo, S., Yamane, K., 2006. A feasibility study for characterizing subsurfacefaults and hydrogeological information with marine electromagnetic prospecting (inJapanese). J. Geogr. (Chigaku Zasshi) 115, 400415.

A new marine magnetotelluric measurement system in a shallow-waterenvironment for hydrogeological study1. Introduction2. Seafloor MT receiver2.1. Instrument and its specifications2.2. Field operation

3. Field experiments and data processing3.1. Study area3.2. Field experiment3.3. Data processing

4. Data inversion and interpretation5. ConclusionsAcknowledgmentsReferences