Home >Documents >Three-dimensional-interpretation-considering-the-static-and-the-sea-effects-of-magnetotelluric-data-obtained-in-Jeju,-Korea_2013_Journal-of-Applied-Geophysics.pdf...
j ourna l homepage: www.e lsev ie r .com/ locate / j appgeo
Three-dimensional interpretation considering the static and thesea-effects of magnetotelluric data obtained in Jeju, Korea
Jihyang Choi a, Tae Jong Lee a,⁎, Junmo Yang b, Seong Kon Lee a, In Hwa Park a, Yoonho Song a
a Korea Institute of Geoscience and Mineral Resources, 124, Gwahang-no, Yuseong-gu, Daejeon, 305-350, South Koreab Korea Institute of Nuclear Safety, 62, Gwahak-ro, Yuseong-gu, Daejeon, 305-350, South Korea
Article history:Received 13 November 2012Accepted 18 July 2013Available online 27 July 2013
Keywords:Magnetotelluric (MT)Sea-effect3D interpretationStatic shiftsJeju Island
Three-dimensional (3D)magnetotelluric (MT) surveys have been performed in Jeju, the largest volcanic island inKorea to figure out any possible structures or potential anomaly for remnant deep geothermal resources. Variousapproaches have been applied to interpret MT data observed in Jeju. MT dataset shows generally simple stratig-raphy of four layers, though contains the severe static and the sea-effects. In our previous works, the inductionvectors and 3D inversion results have commonly indicated the existence of a conductive anomaly in centralparts of the island, beneath Mt. Halla. The 3D inversion dealt the static shifts as inversion parameters. The JejuMT dataset, however, still contains the effect of conductive sea water surrounding the island.The sea-effect onMT impedance can be represented as a distortion tensor and excluded from the JejuMT datasetby an iterative sea-effect correction. In this study, 3D inversion incorporating static shift parameterization wasconducted usingMT dataset corrected using 1D resistivitymodel obtained from the iterative scheme. Reasonablyreconstructed images are obtained through the 3D inversion and using theMT datasetwith sea-effect correction.The inversion result still shows the conductive anomaly in a similar depth. RMS misfits converged to a lowervalue than that of inversion using MT data before the sea-effect correction. From the fact, it is highly possiblethat the conductive anomaly is not an artifact but a real underground structure. Further investigation aboutthe anomaly including exploration drilling is needed to see if it is from a fracture containing conductive seawater or related to the old volcanic activities.
Magnetotelluric (MT) method is a powerful tool for investigatingvarious kinds of geological structure under the earth's surface. It hasbeen widely used in exploration of not only the conventional energy(e.g. oil and gas) but also the renewable energy such as geothermal re-sources (Goldstein, 1988; Key et al., 2006; Orange, 1989). MT techniqueusually focused on the deeper geologic target than the other EMmethods, though, supplementary high-frequency data from audio-frequencyMT (AMT) can take a part to figure out the shallow structuresand eventually improve the resolution of reconstructed images (Leeet al., 2006). A remote reference processing enhances the quality ofMT data by removing coherent noise (Song et al., 2006a). Combiningthese techniques, several MT (incorporating AMT) surveys have beenconducted in Korea (Lee et al., 2005, 2009b; Song et al., 2006b) for geo-thermal application since 2002.
From 2004 to 2006, research team at Korea Institute of Geoscienceand Mineral Resources (KIGAM) of Korea and National Institute ofAdvanced Industrial Science and Technology (AIST) of Japan performedjoint MT surveys at mid-mountain area of Jeju, the largest volcanicisland in Korea. The purposes of the MT surveys were to see if therestill remains a geothermal regime and if there exist deeply extendedfractures or an aquifer system beneath the mid-mountain region ofMt. Halla. AMT data were also acquired at each MT site, so that wideband, good quality data were acquired from the (A)MT surveys. MTdataset of Jeju, however, suffered from severe static shifts and sea-effect(Lee et al., 2009a).
The effect of static shift is due to the presence of small-scale near-surface inhomogeneities, and manifests itself as a vertical shifting ofthe apparent resistivity curve by a frequency-independent factor with-out any corresponding change in the phase curve (Sasaki, 2004). Con-trol of the static shifts in the inversion can be done by employing thestatic effect as variables. A vector of static-shift parameters in the objec-tive function is assumed to have a Gaussian distribution (Ogawa andUchida, 1996). The sea-effect is due to the sharp electrical contrast be-tween the sea and the land, and cause spurious anomalies in resultantimages of the electrical structure, especially for deeper parts of theearth. Jeju is surrounded by conductive sea, so that the sea-effect yieldsserious effects on the low-frequencyMT data.Many attempts have been
55J. Choi et al. / Journal of Applied Geophysics 98 (2013) 54–61
suggested to effectively remove the sea-effect from observedMT data. Amethod which employed electrical and magnetic distortion tensor tocorrect the sea-effect forMT data (Nolasco et al., 1998) had dimensionallimitation and the other method which used 3D forward algorithm inthe removal of the sea-effect (Santos et al., 2001) needed the preciseprior information before the correction. In this study, we adopted anewly suggested method of an iterative sea-effect correction (Yanget al., 2010), which neither suffers from dimensional limitationsnor requires exact a priori information. Through the comparison ofreconstructed images by MT inversion of observed data and by thoseof corrected data, we could find the feasibility of sea-effect correctiontechnique for the real MT data and derived more reliable subsurfacestructures beneath the Jeju Island.
2. Geological characteristics of Jeju Island
The geology of inland Korea is characterized by old formation cov-ered by thin sedimentary layers. Jurassic or Cretaceous granite can befound at very shallow depths in large part of the country. For more thanthousand years, there were neither any volcanic activities, nor strongtectonic activities in Korea. Thus high temperature geothermal resourcessuch as hydrothermal system can hardly be expected. So far, geothermalutilization in Korea has been confined in direct use, mainly on publicbaths in low-temperature geothermal water down to several hundredmeters. After KIGAM started a research on measuring and analyzing thethermal properties in 2003, utilization of geothermal resources of shallowsubsurface became expanded to district heating and greenhouse uses insome regions (Song, 2004).
Though we can hardly find high temperature geothermal re-sources for power generation in land, Quaternary volcanic rocks areexposed in some islands in the South and East Sea. Jeju is one ofthem and the last volcanic activity of Mt. Halla recorded by historicliterature occurred in 1007 A.D. It is why we focus on Jeju as a candi-date of high temperature geothermal development and MT survey is
3D inversion
Site JJS752
Fig. 1. A bathymetry map around Jeju and MT measurement sites on th
applied in Jeju. The origin of volcanic structure in Jeju is known to behaving no relation either with the ridge of the North Pacific Ocean orthe volcanic front of Japan. Although a recent publication on geother-mal resources in Korea showed that there is no high temperatureanomaly in Jeju Island (Kim and Lee, 2007), it is from the lack ofwell data deep enough and there is no research results yet. The pur-pose of the surveys, thus, is to delineate the deep geological struc-tures including possible deeply extended fractures or volcanic veinbeneath themid-mountain area, whichmay be related with remnantgeothermal regime associated with volcanic eruptions.
Jeju is the largest island located at South Sea of Korea, 31 km longalong the minor axis and 73 km long along the major axis of direc-tion N70°E (Fig. 1). Mt. Halla (a height of 1950 m) at the center ofthe island is formed with a great mass of volcanic rock and morethan 360 parasitic volcanic cones cover the surface of the island.Based on the topography of the bathymetry from ETOP2 data givenby National Geophysical Data Center (Fig. 1, NGDC, http://www.ngdc.noaa.gov), the sea water boundary is getting deeper stepwiseand finally reaches a flat 100 m deep.
Geological structure of Jeju hasmainly formed since the late Pliocene(Yoon, 1997). Most parts of the island are covered with Quaternarybasaltic lava on the Quaternary sedimentary formations which arecomposed of Pleistocene consolidated sedimentary rocks (SeogwipoFormation, SF) and Plio-Pleistocene unconsolidated sediments (U-Formation, UF). SF and UF are electrically indistinguishable from eachother since both are marine-based and electrically conductive(b10 Ω-m). SF has extremely low permeability and lies about 50–60 mbelow the sea level with an average thickness of 100 m (Koh, 1997).Thickness of UF which has formed of well-sorted find quartzose sandand silts before the onset of volcanism varies from 70 to 250 m with anaverage 150 m. Basement rocks, formed of welded tuffs and granite, lieat a depth of about 250–300 m below the sea level (Yoon, 1997). Generalgeological stratigraphy of Jeju has been deduced as a simply four-layeredstructure (Koh, 2007).
Jeju
RR
e island. MT surveys were performed at 108 sites along the 5 lines.
56 J. Choi et al. / Journal of Applied Geophysics 98 (2013) 54–61
3. MT data observed in Jeju
MT and AMT surveys (hereafter, we call MT as combined MT andAMT surveys) were performed along the five lines surrounding themid-mountain area (elevation b 600 m) ofMt. Halla. A total of 108mea-surements were made during the three-year campaign using PhoenixMTU-5A systems as shown in Fig. 1. Frequency range covered over 6 de-cades from3 × 10−4 Hz to 103 Hz. A far remote reference sitewas locat-ed in Japan, approximately 480 km apart from the center of the island.
As mentioned before, good quality data have been acquired but thegeological or geographical setting of the island made the MT dataset ofspecial characteristics. The sounding curves of apparent resistivity andphase generally show typical three-layered responses. Resistive toplayer (basaltic lava), conductive middle layer (SF and UF) and resistivebottom layer (basement) were well-matched with the general stratigra-phy of Jeju Island (Lee et al., 2006). Topography in the mid-mountain isnot very steep and the topographic effect appeared to be nothing butthe reflection of thickness change of the top layer except in the close vi-cinity ofMt. Halla (Namet al., 2006). The sea-effect ofMTdatasetwas val-idated with induction vectors, all of which for frequencies below 0.1 Hzpointed to the surrounding sea or deep ocean. Besides the sea-effect,MT dataset contains severe static shifts (Lee et al., 2009a). In several stud-ies of JejuMT dataset, the existence of a conductive anomaly beneathMt.Halla has been mentioned. Two-dimensional (2D) interpretations alongthe survey lines by different schemes commonly indicated the feature ofconductive anomalies at the middle part of each line (Choi et al., 2007;Lee et al., 2006). The central parts of each line coincide with the locationof Mt. Halla. An isolated conductive anomaly on the central part of the is-land is shown in the reconstructed images from 3D inversion incorporat-ing the static shifts. It is suggested that this anomaly can be related withthe volcanic vent or deeply connected fracture filled with saline water(Lee et al., 2009a). Back to the induction vector, we can identify some in-duction vectors direct to Mt. Halla for frequencies around 1 Hz, which isanother evidence of the conductive body at the center of the island be-neath the SF and UF. Therefore, a correction or removal of sea-effect andapplying 3D inversion to the revised dataset can justify the existence ofthis conductive anomaly at the right location.
4. Iterative sea-effect correction
After an iterative method for correcting land–ocean boundary effectin seafloor voltage data was proposed (Koyama, 2002), Baba and Chave(2005) applied the iterativemethod to topographic correction ofmarineMT data. Yang et al. (2008) proposed an iterative method to correct the
Fig. 2. Sounding curves with error bars for the MT data observed at site
sea-effects in land MT data, on the basis of synthetic data. The iterativemethods have two stages. At the former iteration, to compute and cor-rect either the topographic effect or the sea-effect is the first stage andthe second one is to invert the corrected responses in the model spaceswithout the topography or the sea. The inverted subsurface structuresare used again for the correction of the next iteration. These proceduresare repeated until the constraints are satisfied (Yang et al, 2008).
MT impedance tensor Z can be expressed by themultiplication of theimpedance tensors of the subsurface structure and the sea-effect.
Z ¼ ZsZm ð1Þ
where, Zs and Zm are theMT impedance tensors of the sea-effect and thesubsurface structure, respectively. Both tensors are of 2 × 2 and all theelements of the tensors have complex values. Impedance tensors canbe computed using an initial subsurface structure with (Z′) andwithout(Zm′) the sea-effects, where the superscript denotes numerically com-puted impedance. The sea-effect tensor is given by,
Zs ¼ Z′ Z′m
h i−1: ð2Þ
The impedance tensor of sea-effect-corrected data (Zc) can be ap-plied to the observed data (Z0) as below,
Zc ¼ Zs½ �−1Z0: ð3Þ
Substituting Eq. (2) into Eq. (3) gives
Zc ¼ Z′ Z′m
� �−1� �−1
Z0 ¼ Z′m Z′� �−1
Z0: ð4Þ
In Eq. (4), the sea-effect-corrected data is finally given by the tensorproduct of initial structure model and the observed data. After the firstiteration, Z0 is replaced with Zc in the former iteration and Eq. (4) canbe rewritten as
Z mþ1ð Þc ¼ Z′ mð Þ
m Z′ mð Þ� �−1Z mð Þc ð5Þ
and Zc is getting closer to the impedance tensor of the truemodel as theiteration proceeds. Fig. 2 shows an example of procedures of the itera-tive correction technique. Data observed at JJS752 was distorted bysea-effect especially in the lower frequency range. After sea-effect cor-rection, split between the twomodes at around 1 Hz becamemitigated
JJS752 (Fig. 1) and sea-effect corrected data at each iteration stage.
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
0
-10
-20
-30
0 10 20 30 40 50 60
Fig. 3. Horizontal depth slices of the 3D inversion results for the observed MT data. The seashore boundary of Jeju Island, Mt. Halla and all the measurement sites are superimposed on each slice.
Fig. 4. Horizontal depth slices of the 3D inversion results for the sea-effect corrected MT data. The seashore boundary of Jeju Island, Mt. Halla and all the measurement sites are superimposed on each slice.
58J.Choietal./JournalofA
ppliedGeophysics
98(2013)
54–61
(a) Observed9
rms
lambda
rms
lambda
7
rms
mis
fit
60
40
20
0
lambda
60
40
20
0
lambda
5
3
9
7
rms
mis
fit
5
3
1 3 5 7 9Iteration
11 13 15
1 3 5 7 9Iteration
11 13 15
(b) Corrected
Fig. 5. Curves of RMSmisfits and Lagrangemultipliers (λ) during inversion of the observedMT data (a) and the sea-effect corrected MT data (b).
59J. Choi et al. / Journal of Applied Geophysics 98 (2013) 54–61
in the sounding curves of apparent resistivity. The response of the bot-tom layer (the fourth layer) is clearly recovered after the sea-effectcorrection.
MT data observed at 11 sites in Jeju were corrected using thisscheme and inverted to 1Dmodels. After averaging 1D inversion results,1D resistivity model was suggested in Yang et al. (2012). Putting 1D re-sistivity model as the structure model to Eq. (4), the MT dataset ob-served in Jeju was corrected.
5. MT 3D inversion incorporating static-shifts
The MT inverse problem is non-linear with respect to subsurfaceelectrical properties. The problem incorporating static shifts can be lin-earized as,
Δd ¼ JΔmþ Gs ð6Þ
where Δd is a vector of differences between observed and predicteddata, Δm is a model correction vector, J is a sensitivity matrix, s is a vec-tor of static-shift parameters and G is a matrix which relates s to
observed data (deGroot-Hedlin, 1991; Ogawa and Uchida, 1996). Themodified objective function Φ by Lee et al. (2003) is given by,
Φ ¼ jjWd JΔmþ Gs−Δdð Þjj2
þ λ2 Rmkþ1��� ������ ���2 þ α2 mkþ1−mb
��� ������ ���2 þ β2 sj jj j2� �
ð7Þ
where Wd is a diagonal matrix whose elements are the reciprocal ofmeasurement uncertainties, R is a second-order difference operatorquantifying model roughness, mk + 1 is the (k + 1)th model to mini-mizeΦ andmb can be either a base model or themodel of the previousiteration (mk). λ is the Lagrange multiplier (or the regularization pa-rameter) that is used to control the relative importance betweenmodel smoothness and datafit.α is a trade-off parameter and β is a con-stant controlling the weighting for the static shifts, which is fixedthrough the entire inversion process.
Minimizing Φ in Eq. (7) is equivalent to solving an observationequation
Wd J WdGλR 0λαI 00 λβI
2664
3775 mkþ1
s
� �¼
Wd Jmk þ Δd� �
0λαmb
0
26664
37775 ð8Þ
where I is the identity matrix. The iteration is continued until either aspecific number of iteration ismet or RMSmisfits are reduced to accept-able level. The RMS data misfit is defined as
where M is the number of data, where ρa is the apparent resistivityin ohm-m, ϕ is the phase in degree, superscripts o and p stand for ‘ob-served’ and ‘predicted’, respectively, and w is a constant controllingthe relative importance of the phase to the apparent resistivity(Sasaki, 2004).
6. Reconstructed images
Among the whole dataset of 108 sites, 85 MT data from thecentral part of the island within the rectangle in Fig. 1, were usedfor 3D MT inversion. Fifteen frequencies from sounding data areselected in the range of 0.003 Hz to 300 Hz. A linearized least-squares inversion technique incorporating static-shift parameteriza-tion was performed. The algorithm was modified from the originalone developed by Sasaki (2004). As the first step, the sea and surfacetopography were not considered. The inversion blocks had a dimen-sion of 2 km × 2 km in horizontal direction, while the vertical di-mension varied with depth (20 × 14 × 19 blocks in x, y, and zdirections, respectively). Data weighting was adopted based on theerrors of measured data.
Reconstructed images shown in Fig. 3 arehorizontal depth slices of the3D inversion results for the observed MT dataset including static shiftparameterization. The RMS error was 3.75% at the final (15th) iteration.The conductive characteristics at the shallow slices can be appearance ofconductive sedimentary layers beneath the resistive top layer. Thediscon-tinuous and conductive anomalies along the seashore boundary of Jejuand the peripheries of the survey lines are due to the lack of spatial cover-age. Aspect that the resistive lump in the center of the island is gettingsmaller to the deeper slice is representing the change of the thickness ofthe top layer.
Since 3D inversion in this study assumed the actual surface as asea level, we needed a little imagination to interpret the reconstructedinversion results. The top layer with resistivities higher than a hundredohm-m extends to an average depth of 0.4 km, which is an average
1000Zxy mode
Zyx mode
observed datacalculated by obs datacalculated by cor data
observed datacalculated by obs datacalculated by cor data
Zxy mode
Zyx mode
Station number
100
ρa (
ohm
-m)
10
1000
100
ρa (
ohm
-m)
10
φ (d
egre
e)
0
30
60
90
φ (d
egre
e)
0
30
60
90
0 10 20 30 40 50 60 70 80Station number
0 10 20 30 40 50 60 70 80
(a) apparent resistivity (b) phase
Fig. 6. Comparison of the apparent resistivities and the phases at 6.899 Hz of measured and calculated MT data in the Zxy and Zyx modes at 85 survey sites.
Fig. 7. Static shifts in the Zxy and Zyx modes at 85 survey sites.
60 J. Choi et al. / Journal of Applied Geophysics 98 (2013) 54–61
elevation of themeasurement sites. The conductive sedimentary forma-tion lies beneath the resistive surface layer with several tens of ohm-m.And the resistivity of the inversion blocks gradually decreases as it goesdeeper. Basement rock with welded tuff and granite has resistivityhigher than a thousand ohm-m. Note an isolated conductive anomalynear the location of Mt. Halla. Neither the sea water surrounding the is-land nor the lack of the spatial coverage can produce an isolated con-ductive anomaly in the central part of the island in 3D MT modeling(Lee et al., 2009a). Though it is not an artifact from the surroundingsea, the sea surely affects the MT data especially at the low frequency,and thus deeper part of the inversion result. The sea-effect correctioncan help interpret the depth extension of the conductive anomaly andgeological structure of deep subsurface of the island.
Secondly, Fig. 4 shows the reconstructed images from the inversionusing sea-effect-corrected data. All the parameters for the 3D inversionwere the same as those in Fig. 3 except for the dataset. Comparing Fig. 3and Fig. 4, shallow parts above 2 km deep show similar features witheach other, because the correction doesn't affect high frequency data.To the deeper parts, however, the sea-effect-corrected images (Fig. 4)are getting slightly more conductive than uncorrected images (Fig. 3).Below the depth of 20 km, one can notice that the bottom layer canbe more clearly identified in the sea-effect-corrected image than inFig. 3. The RMS error was 3.34 at the 15th iteration. RMS misfits(Fig. 5) of inversion of the corrected data converged smother thanthose of the observed data. The sea-effect correction also helps the sta-ble inversion procedure when Lagrange multifliers fulfill their roles.
The apparent resistivities and the phases are compared at 85 surveysites in Jeju to verify the aspects of the inversion results. Fig. 6 shows thecurves of the apparent resistivity and the phases at 6.899 Hz of mea-sured MT data and calculated MT data from the final models whichare the results of inversions of MT data without andwith sea-effect cor-rection. Resistivity curves calculated from inversion results of sea-effect-corrected data were getting closer to the observed data thanthose of uncorrected data. Both the inversions extract similar but severestatic shifts at each site as shown in Fig. 7.We can still find the very con-ductive body in the center of the island and it seems to be extendeddown to about 3 km in depth from both of the figures.
From the resistivity data calculated by 3D inversion using the sea-effect-corrected MT data, we can get 2D sections along the surveylines. Vertical slices along the central and southern lines in Fig. 1 areshown in Fig. 8. Pink dots on the surface are the crossing points of thetwo vertical slices. One can recognize the boundaries of the basalt, SF
and UF, and the basement rock with welded tuff and granite as depictedin thefigure. Since the inversion doesn't incorporate the geological topog-raphy, boundaries seem to godown in themiddle parts of the images. Dif-ferences of the elevations atMT sites in the central and southern lines areless than 500 m and 200 m and the effect by the topography can be neg-ligible. But, considering that the average elevation of the middle part ofthe survey line is about 400 m, the boundaries of SF and UF actually goup in themiddle part of the island (Lee et al., 2009a). Note the conductiveanomaly at the central part of each survey line, depicted as ‘A’ in Fig. 8. It islocated next to the top of Mt. Halla, the 2 km to the east from the centralline and 3 km to the north from the southern line. The anomaly seems tobe extended deeper than 5 km, though the image smeared out as it goesdeeper. This surely tells the existence of the real conductive anomaly inthe geological structures under the Mt. Halla. It can be a volcanic vent ora fracture system filled with conductive fluid.
7. Conclusions
MT data observed in Jeju Island has been examined and analyzedwith respect to the sea-effect correction. Since the purpose of the survey
Fig. 8. 2D sections along the survey lines. Reconstructed images are inverted using the sea-effect correctedMT data. Dash lines designate the boundaries of the electrical layers beneath thesubsurface of Jeju; top layer of basalt, marine-oriented layer (Seogwipo formation and U-formation), and bottom layer. Conductive anomaly ‘A’ is distinguishable in the sections. Blacktriangles are locations of the survey stations and pink dots are the crossing point of two sections.
61J. Choi et al. / Journal of Applied Geophysics 98 (2013) 54–61
is tofigure out any possible remnant geothermal regimeunderneath thevolcanic island, deep structure and the depth extension of conductiveanomaly are the major concerns of this study. Among the 108 totalsites that we had performed MT surveys, we selected 85 sites whereMT signal showed better quality for the 3D inversion. Iterative sea-effect corrections have been applied to the data and the inversion re-sults before and after the sea-effects correction have been compared.The method repeats correction of sea-effect by putting 3D forwardmodels including and excluding the surrounding sea-effect for comput-ing the distortion tensor and then by inverting the sea-effect-correctedresponses until the difference between observed data and model re-sponses converges to a criterion. The sea-effect correction generally re-covers the resistivity and phase at low frequencies and the 3D inversionshowed general four layered structure beneath the island.
A linearized least-squares inversion method incorporating static-shifts was used with the 15 frequencies ranging from 0.003 to 300 Hz.Reasonably reconstructed images are obtained through the 3D inver-sion incorporating static shift parameterization and using the MTdataset with sea-effects correction. The reconstructed images derivedfrom 3D inversion of the sea-effect-corrected MT data gave more reli-able results and reaffirmed the conductive body extended down to afew kilometers deep. RMS misfits converged to a lower value thanthat of inversion using MT data before the sea-effect correction. Fromthis result, we can argue that the conductive anomaly is not an artifactbut a real underground structure. Further investigation about the anom-aly including exploration drilling is needed to see if it is from a fracturecontaining conductive sea water or related to the old volcanic activities.
Acknowledgments
This work was supported by national project No. 2010T100200494supported by the Ministry of Knowledge Economy of Korea (MKE)through the Korea Institute of Energy Technology Evaluation and Plan-ning (KETEP) and by the Basic Research Project of Korea Institute ofGeoscience and Mineral Resources funded by the Ministry of Science,ICT and Future Planning of Korea (MSIP).We thankDr. Toshihiro Uchidafrom Advanced Industrial Science and Technology (AIST) of Japan forthe cooperative data acquisitions.
References
Baba, K., Chave, A.D., 2005. Correction of seafloor magnetotelluric data for topographic ef-fects during inversion. Journal of Geophysical Research 110. http://dx.doi.org/10.1029/2004JB003463.
Choi, J., Kim, H.J., Nam, M.J., Lee, T.J., Lee, S.K., Song, Y., Suh, J.H., 2007. Geoelectrical struc-ture of Jeju Island deduced from 2D inversion of AMT and MT data. Proceedings of2007 KSEG-KGS Joint Conference, Jun 7–8, Daejeon, Korea, pp. 287–290.
deGroot-Hedlin, C., 1991. Removal of static shift in two dimensions by regularization in-version. Geophysics 56, 2102–2106.
Goldstein, N.E., 1988. Subregional and detailed exploration for geothermal-hydrothermalresources. Geothermeal Science and Technology 1, 303–431.
Key, K.W., Constable, S.C., Weiss, C.J., 2006. Mapping 3D salt using the 2D marinemagnetotelluric method: case study from Gemini Prospect, Gulf of Mexico. Geophys-ics 71, B17–B27.
Kim, H.C., Lee, Y., 2007. Heat flow in the Republic of Korea. Journal of Geophysical Re-search 112, B05413. http://dx.doi.org/10.1029/2006JB004266.
Koh, G.W., 1997. Characteristics of the Groundwater and Hydrogeologic Implications ofthe Seogwipo Formation in Jeju Island. Ph.D. Thesis Busan National University,Korea (326 pp.).
Koh, G.W., 2007. Hydrogeologic characteristics of groundwater in Jeju Island, Korea, specialsymposium on groundwater, geothermal, and industrial application of geophysics.Proceedings of Korean Society for Geosystem Engineering, Jeju. Korea, pp. 3–17.
Koyama, T., 2002. A Study of the Electrical Conductivity of Mantle by Voltage Measure-ments for Submarine Cables. Ph.D. Thesis University of Tokyo, Japan (129 pp.).
Lee, T.J., Song, Y., Uchida, T., 2005. Two-dimensional interpretation of far-remote ref-erence magnetotelluric data for geothermal application. Exploration Geophysics(Mulli-tamsa) 8, 145–155.
Lee, T.J., Lee, S.K., Song, Y., Uchida, T., 2006. Use of audio-band on the interpretation ofmagnetotelluric data. Exploration Geophysics (Mulli-tamsa) 9, 261–270.
Lee, T.J., Nam,M.J., Lee, S.K., Song, Y., Uchida, T., 2009a. The Jeju dataset: three- dimension-al interpretation of MT data from mid-mountain area of Jeju Island Korea. Journal ofApplied Geophysics 68, 171–181. http://dx.doi.org/10.1016/j.jappgeo.2008.11.006.
Lee, T.J., Song, Y., Han, N., Park, D.W., 2009b. Application ofmagnetotelluric survey for de-velopment of deep geothermal water at Seokmo Island, Korea. Proceedings of 9thSEGJ International Symposium, p. 43.
Nam, M.J., Kim, H.J., Song, Y., Lee, T.J., Suh, J.H., 2006. Three-dimensional FEMmodeling ofmagnetotelluric data observed in Jeju Island, Korea. Proceedings of the 18th EM In-duction Workshop, El Vendrell, Spain, p. S3.
Nolasco, R.P., Smith, J.T., Filloux, J.H., Chave, A.D., 1998. Magnetotelluric imaging of the So-ciety Island hotspot. Journal of Geophysical Research 103, 30,287–30,309.
Ogawa, Y., Uchida, T., 1996. A two-dimensional magnetotelluric inversion assumingGaussian static shift. Geophysical Journal International 126, 69–76.
Orange, A.S., 1989. Magnetotelluric exploration for hydrocarbons. Proceedings of the IEEE77, 287–317.
Santos, F.A.M., Nolasco, M., Almeida, E.P., Pous, J., Mendes-Victor, L.A., 2001. Coast effects onmagnetic and magnetotelluric transfer functions and their correction: application toMTsoundings carried out in SW Iberia. Earth and Planetary Science Letters 186, 283–295.
Sasaki, Y., 2004. Three-dimensional inversion of static-shifted magnetotelluric data. EarthPlanet Space 56, 239–247.
Song, Y., 2004. Recent activities on low-temperature geothermal development in Korea.Proceedings of the 6th Asian Geothermal Symposium, Daejeon, Korea, Oct. 26–29,2004, pp. 13–19.
Song, Y., Lee, T.J., Uchida, T., 2006a. Effect of remote reference on audio-frequencymagnetotelluric data. Journal of Korean Society for Geosystem Engineering 43, 44–54.
Song, Y., Nam, M.J., Lee, T.J., 2006b. Topography and sea effects on MT responses observedin Jeju Island, Korea. Proceedings of the 8th SEGJ International Symposium, Kyoto,Japan, pp. 121–124.
Yang, J., Lee, C., Yoo, H.-S., 2008. Correction of the sea effect in the magnetotelluric (MT)data using an iterative tensor stripping during inversion. Geophysics and GeophysicalExploration 11, 286–301.
Yang, J., Min, D.-J., Yoo, H.-S., 2010. Sea effect correction in magnetotelluric data and itsapplication to MT soundings carried out in Jeju Island Korea. Geophysical Journal In-ternational 182, 727–740.
Yang, J., Lee, C.-K., Min, D.-J., Lee, H., Lee, T.J., 2012. 1-D crustal resistivity structure re-vealed by sea effect corrected magnetotelluric (MT) data obtained at Jeju Island,Korea. Journal of Applied Geophysics 76, 92–101.
Yoon, S., 1997. Miocene–Pleistocene volcanism and tectonics in southern Korea and theirrelationship to the opening of the Japan Sea. Tectonophysics 281, 53–70.
