Geophysical Images of the North Anatolian Fault Zone in the Erzincan Basin, Eastern Turkey, and ...

23 Pages·2013·2.16 MB·English

PureAppl.Geophys.170(2013),409–431
(cid:2)2012SpringerBaselAG
Pure and Applied Geophysics
DOI10.1007/s00024-012-0521-5
Geophysical Images of the North Anatolian Fault Zone in the Erzincan Basin,
Eastern Turkey, and their Tectonic Implications
U¨MI˙T AVS¸AR,1 ERS¸AN TU¨RKOG˘LU,2 MARTYN UNSWORTH,3 I˙LYAS C¸AG˘LAR,1 and BU¨LENT KAYPAK4
Abstract—The collision between the Arabian and Eurasian to the Aegean Sea in the west (Fig. 1a). The NAF
plates in eastern Turkey causes the Anatolian block to move
forms the northern boundary of the Anatolian block,
westward.TheNorthAnatolianFault(NAF)isamajorstrike-slip
which is moving westward as a result of the colli-
faultthatformsthenorthernboundaryoftheAnatolianblock,and
theErzincanBasinisthelargestsedimentarybasinontheNAF.In sion between the Eurasian and Arabian plates
the last century, two large earthquakes have ruptured the NAF (MCKENZIE, 1972; S¸ENGO¨R, 1979; DEWEY and S¸ENG-
withintheErzincanBasinandcausedmajordamage(M =8.0in
1939andM =6.8in1992).TheseismichazardinErzinscanfrom O¨R, 1979). Recent geodetic measurements suggest
s
future earthquakes on the NAF is significant because the uncon- that about 70 % of the Arabian-Eurasian plate con-
solidatedsedimentarybasincanamplifythegroundmotionduring vergence is accommodated by the westward
an earthquake. The amount of amplification depends on the
thicknessandgeometryofthebasin.Geophysicalconstraintscan extrusion of the Anatolian block (REILINGER, 2006)
be used to image basin depth and predict the amount of seismic with the slip rate on the NAF estimated from GPS
amplification. In this study, the basin geometry and fault zone data at approximately 24 mm/year (MCCLUSKY
structurewereinvestigatedusingbroadbandmagnetotelluric(MT)
et al., 2000).
datacollectedontwoprofilescrossingtheErzincanBasin.Atotal
A series of westward-propagating earthquakes
of24broadbandMTstationswereacquiredwith1–2kmspacing
in2005.InversionoftheMTdatawith1D,2Dand3Dalgorithms have ruptured the NAF over the last century (STEIN
showed that the maximum thickness of the unconsolidated sedi- and BARKA, 1997), and a significant number of
mentsis*3kmintheErzincanBasin.TheMTresistivitymodels
earthquakes have occurred in and around the Erzin-
showthatthenorthernflanksofthebasinhaveasteeperdipthan
thesouthernflanks,andthebasindeepenstowardstheeastwhereit can Basin over the last millennium (BARKA et al.,
hasadepthof3.5km.TheMTmodelsalsoshowthatthestructure 1987). Severe damage occurred in the city of Erzin-
oftheNAFmayvaryfromeasttowestalongtheErzincanBasin.
can (Fig. 1b) in both the 1939 (M = 8) and 1992
s
Keywords: Magnetotellurics,electricalresistivity,faultzone (Ms = 6.8) earthquakes (BARKA and KANDISKY-CADE,
conductor,ErzincanBasin,EasternTurkey. 1988; FUENZALIDA et al., 1997). The 1992 event
claimed 541 lives (BARKA and EYIDOG˘AN, 1993) and
had a peak ground acceleration of 0.5 g, and the
1. Introduction
Mercalli intensity was estimated as IX (ERDIK et al.,
The North Anatolian Fault (NAF) is a major 1992; GU¨NDOG˘DU et al., 1992).
The analysis of BAYRAK et al. (2005) suggested
strike-slip fault, extending from Karliova in the east
that earthquakes up to magnitude 7.5 could be
expected in the Erzincan Basin, and HARTLEB et al.
Electronic supplementary material The online version of this (2006) estimated the earthquake recurrence interval
article (doi:10.1007/s00024-012-0521-5) contains supplementary
as 210–700 years from paleoseismic studies.
material,whichisavailabletoauthorizedusers.
Although the city of Erzincan has been rebuilt, the
1 DepartmentofGeophysicalEngineering,FacultyofMines,
seismic hazard remains high because of the high slip
Istanbul Technical University, Maslak, 34469 Istanbul, Turkey.
rate on the NAF and the fact that the sedimentary
E-mail:[email protected]
2 QuantecGeoscience,NorthYork,ON,Canada. basin can significantly amplify ground motion. The
3 Department of Physics, University of Alberta, Edmonton, strength of ground motion depends primarily on the
ABT6G2G7,Canada.
thickness of sediments in the basin, with the largest
4 DepartmentofGeophysicalEngineering,FacultyofEngi-
neering,AnkaraUniversity,Ankara,Turkey. amplification occurring in the deepest parts of the
410 U¨.Avs¸aretal. PureAppl.Geophys.
Figure1
a Topographic map showing the major fault systems of Anatolia; arrows show the directions of block motion relative to Eurasia. The
rectangleindicatesthestudyarea.bFaultsystemsinthestudyareaaretakenfromBARKAandGU¨LEN(1989);FUENZALIDAetal.(1997)and
AKTARetal.(2004).Thedotsindicatetheearthquakeepicentresfrom1960to2009(KandilliObservatoryandEarthquakeResearchInstitute
andKAYPAKandEYIDOG˘AN,2005).ThefocalmechanismsolutionisfromBERNARDetal.(1992)
basin and at the edges (OLSEN et al., 1995; OLSEN, the upper sedimentary layer and basement was at a
2000). Therefore, the thickness and shape of the depth of 6–12 km. (AKTAR et al., 2004; GO¨KALP,
sedimentarybasinareimportantparameterstopredict 2007; KAYPAK, 2008).
the amplification ofthe ground motion. Furthermore, The magnetotelluric (MT) method involves the
the depth of the basin is an important parameter for measurementofthetimevariationsoftheorthogonal
improving 3D earthquake relocation (AKTAR et al., components of natural electric and magnetic fields,
2004).Severalseismictomographyexperimentshave which contain information about the electrical resis-
attemptedtomapthethicknessoftheErzincanBasin. tivity structure from crustal to upper mantle depths.
Some of these studies used the 1992 aftershocks and Magnetotelluricsiswellsuitedtoimagethepresence
reportedthethicknessofunconsolidatedsedimentsas of fluids within the fault zones as well as the thick-
2–4 km. Theydeterminedthattheboundarybetween nessandtheshapeofthesedimentarybasin(BOERNER
Vol.170,(2013) GeophysicalImagesoftheNorthAnatolianFaultZoneintheErzincanBasin 411
et al., 1995; POMPOSIELLO et al., 2002; PADILHA and ruptured both segments S-2 and S-3 (AMBRASEYS,
VITORELLO, 2002). 1970; BARKA and GU¨LEN, 1989).
Seismic and magnetotelluric methods provide
Twoleft-lateralstrike-slipfaults,theOvacıkFault
imagesofacousticvelocity(V andV)andelectrical
p s (OF) and the North East Anatolian Fault (NEAF),
resistivity (q) respectively on similar spatial scales
intersect the NAF at the southern and northern edge
(BEDROSIANetal.,2007).Thus,jointinterpretationsof
of the Erzincan Basin, respectively, (Fig. 1b). The
resistivityandseismicvelocitycanbeusedinstudies
NE-SW trending Ovacik Fault (ARPAT and SAROG˘LU,
where both parameters are controlled by the same
1975)orMalatya-OvacıkFaultZone(WESTAWAYand
lithological parameters (MARQUIS and HYNDMAN,
ARGER, 2001) has been interpreted as being inactive
1992;JONES,1987,1998;UNSWORTHetal.,2005).The
at present. It has been suggested that the OF formed
main factor controlling the elastic and electrical
the Africa-Turkey plate boundary between 5 and
propertiesofsolid–liquidrockmixturesistheamount
3 Ma,andbecameinactivewhentheleftlateralstrike
of liquid present, i.e., the porosity. The clay content
slipEastAnatolianFault(EAF)developedeastofthe
and pore geometry can also influence the resistivity
OF 3 Ma ago. Since the EAF became the location of
(KOZLOVSKAYAandHJELT,2000).Becauseofthisfact,
motion between the Anatolia and Arabian plates
unconsolidated sedimentary basins and fault zones
(Fig. 1a) (OVER et al., 2004c), the OF has notmoved
can have correlated low seismic velocity and high
significantly (WESTAWAY and ARGER, 2001). How-
electricconductivity anomalies.
ever, GROSSER et al. (1998) indicated that the fault
The purpose of this study is to use MT data to
cutstheQuaternarybasinfillintheOvacıkBasinand
image the geometry of the Erzincan Basin and asso-
classified it as an active fault. The NEAF (TATAR,
ciated faults. The resulting models are used to
1978) has a NE-SW trend and defines the northern
evaluatetheregionaltectonicsandseismichazardsin
boundary of the part of Eastern Anatolia that is
the Erzincan area.
escaping eastward towards the Caucasus (BARKA and
GU¨LEN, 1989).
2. Geological and Tectonic Setting The tectonic evolution of the Erzincan Basin is
not fully understood. The basin was initially
2.1. Regional Tectonics of the Erzincan Area described as a pull-apart basin (ALLEN, 1969; AYDıN
and NUR, 1982; HEMPTON and DUNNE, 1984). How-
The Erzincan Basin is 15 km wide and 50 km
ever, BARKA and GU¨LEN (1989) proposed a revised
long; it is bounded on the northern side by the right
model that suggested that the NAF contributes to the
lateral North Anatolian Fault (NAF). In this region,
growth of the basin in length, while the OF has
theNAFhasbeendividedintoanumberofsegments.
increased the width and depth of the Basin (S¸ENGO¨R
From east to west these are labelled as S-1, S-2 and
et al., 2005).
S-3 in Fig. 1b (BARKA and GU¨LEN, 1989). TheErzincanBasincontinuestogrowinanENE-
• Segment S-1 has a strike direction of N110(cid:3)E and WSW direction because of motion on the NAF and
is located east of the Erzincan Basin. OF. The OF splays into several small segments at its
• Segment S-2 has almost the same strike direction northernmost end, and active normal faulting has
as S-1, forms the northern boundary of the beenmappedalongthesouthernandwesternmargins
Erzincan Basin and consists of a series of sub- of the Erzincan Basin (FUENZALIDA et al., 1997).
parallel faults with an average strike direction of Several dacitic-rhyolitic volcanic cones with ages
N133(cid:3)E (BARKA and GU¨LEN, 1989; AKTAR et al., between 0.1 and 1.06 Ma are found on both the
2004). The 13 March 1992 earthquake ruptured northernandsouthernmarginsofthebasin(HEMPTON
thissegment(GROSSERetal.,1998;BERNARDetal., and DUNNE, 1984; KARSLı et al., 2008). Although the
1992; FUENZALIDA et al., 1997). Erzincan Basin lies within the regionally thickened
• SegmentS-3strikesN105(cid:3)Eandextendswestward crustofeasternAnatolia,highratesofextensionhave
about 110 km. The 1939 Erzincan earthquake caused thinning of the crust in the Erzincan area
412 U¨.Avs¸aretal. PureAppl.Geophys.
Figure2
Geological map of the study area modified from RICE et al. (2009). The fault structures are the same as in Fig.1. The circles show the
magnetotelluricmeasurementsites
(AYDıN and NUR, 1982; FUENZALIDA et al., 1997; limestone and volcaniclastic sedimentary rocks. The
GROSSER et al., 1998; KOC¸YIG˘IT, 2003). Su¨tpınar Formation consists of a 1,500-m-thick,
upward-coarsening succession of mixed carbonate
siliciclastic sedimentary rocks and subordinate vol-
2.2. Regional Geology
canogenic rocks.
Thegeologicalstructureofthe Erzincan region is Thebasementrocksonbothsidesofthebasinare
complex (Fig. 1) because of a long history of plate coveredbyMiocenedepositsthatoutcropextensively
interactions and associated deformation. The Erzin- west and north of the Erzincan Basin and include
can Basin is bounded by the Kesis Mountains in the limestone, marls, green clay, evaporites and fluvial
north and by the Munzur Mountains in the south. deposits(TU¨YSU¨Z,1993;WESTAWAYandARGER,2001;
These ranges have very different geological struc- KOC¸YIG˘IT, 2003; RICE et al., 2009) (Fig. 2).
tures. The Kesis Mountains are characterised by the
RefahiyeComplex(Fig. 2),whichcontainsophiolites 2.2.1 Geological Structure of the Erzincan Basin
with large amounts of serpentinite and metamorphic
rocks. The Refahiye Complex is overlain by the TheErzincanBasinisfilledwithunconsolidatedPlio-
Sipiko¨r Formation, which contains siliciclastic and Quaternary sediments that contain playa deposits,
carbonate sedimentary rocks (RICE et al., 2009). clastics and basin margin conglomerates. The con-
South of the Erzincan Basin, the basement rocks glomerates are composed of ophiolitic melange
consist of Upper Triassic to Lower Cretaceous clastics and carbonates. The central part of the basin
carbonates of the Munzur Formation (O¨ZGU¨L and isfilledmostlybysilts,sandsandgravels(BARKAand
TURS¸UCU, 1984), which was called the Munzur Dag GU¨LEN, 1989). The thickness of the sedimentary
unitbyRICEetal.(2009).TheKarayaprakMelangeis layers has been debated in various papers. The basin
a 4-km-thick, variably tectonized mixture of blocks depthwasestimatedbyanempiricalrelationbetween
consisting of serpentine, basalt, radiolarite, massive thelengthandthethicknessofthebasinas2.5–3 km
Vol.170,(2013) GeophysicalImagesoftheNorthAnatolianFaultZoneintheErzincanBasin 413
(HEMPTON and DUNNE, 1984). GAUCHER (1994) esti- of acceptable quality; one station on the East Profile
mated the basin depth in the southeastern part of the (EE4) and four stations on the West Profile (EW7,
basin to be between 0.65 and 2.1 km by using the EW8EW11EW12)couldnotbeusedbecauseofthe
SP-converted phase of the aftershocks of the 1992 high levels of cultural noise around the city of
Erzincan earthquake. AKTAR et al. (2004) used the Erzincan (Fig. 1b). The MT impedance tensor
same aftershocks and modelled the Erzincan Basin (Z) and vertical magnetic field transfer functions
with a depth of 9 km. They imaged a low-velocity (T) were calculated using the statistically robust
corridor that extended in a NW–SE direction. They method of EGBERT and BOOKER (1986).
also determined the thickness of unconsolidated
sedimentary units in the basin to be between 3 and
4 km. Following this research, a 1D crustal velocity 4. Dimensionality Analysis and Directionality
model was derived for the Erzincan Basin for both P
wave and S wave velocities, and the thickness of Before MT data can be interpreted, dimensional-
unconsolidated sediments in the basin was estimated ityanalysisisneededtodetermineifa1D,2Dor3D
as 2 km (KAYPAK and EYIDOG˘AN, 2005). After 1D analysis is required. A range of dimensionality anal-
interpretation of the seismic velocity structure of the ysis techniques was applied to the Erzincan data, as
basin, KAYPAK (2008) derived a 3D Vp and Vp/Vs described below. If the data can be shown to be
velocity model and estimated the unconsolidated approximately2D,thenakeypartoftheanalysisisto
basin depth to be in the range of 2–3 km. Recently, determine the strike direction.
GU¨RBU¨Z (2010) studied the geometry of the Erzincan
Basin and estimated the basin depth as *4 km.
4.1. Tensor Decomposition
The electrical resistivity of unconsolidated sedi-
mentsismuchlowerthanthatofcrystallinebasement Estimation of the geoelectric strike can be com-
rocks. The resistivity of sediments is sensitive to plicated by small, near-surface conductivity
variations in porosity. This allows geophysical tech- heterogeneities that alter the direction and amplitude
niques such as magnetotellurics to estimate the of the measured electric fields. These distortions are
thickness of a sediment layer and to map subsurface generally frequency independent and referred to as
porosity. In the following sections, magnetotelluric galvanicdistortions(GROOMandBAILEY,1989;BAHR,
datacollectedintheErzincanBasinaredescribedand 1988). There have been various decomposition
interpreted. approachestoremovethesedistortionsanddetermine
the geoelectric strike direction. One of the most
widely used approaches is the Groom and Bailey
3. Magnetotelluric Data Collection (GB) tensor decomposition. In this method the
distortion is assumed to be due to local three-
The Erzincan magnetotelluric (MT) data descri- dimensional (3D) conductivity structures, whereas
bedinthispaperwerecollectedaspartoftheEastern the regional conductivity structure is assumed to be
Anatolian Magnetotelluric Experiment in 2005 2D.
(TU¨RKOG˘LU,2009).SincetheErzincanBasinandNAF TensordecompositionwasappliedtotheErzincan
are oriented approximately NW–SE, MT soundings MT data with the algorithm of MCNEICE and JONES
were collected at 24 sites on two profiles that were (2001), which is an extended form of the GB
normal to the NAF with a nominal site spacing of decompositionthatcanconsidermultipleMTstations
1–2 km (Fig. 1b). The MT data were recorded with and multiple frequencies. The period-dependent
Phoenix Geophysics V5-2000 magnetotelluric sys- strike direction was computed for both profiles, as
tems.Theinstrumentsweresynchronisedwithtiming shown in the first column of Fig. 3. The dominant
signals from global positioning satellites (GPS) to strike direction at short periods (0.001–0.1 s) is
permit remote reference time series processing between N125(cid:3)–135E(cid:3), which is consistent with the
(GAMBLEetal.,1979).MTdataatmoststationswere strikedirectionofthesurfacetraceoftheNAFinthe
414 U¨.Avs¸aretal. PureAppl.Geophys.
basin.ThisgraduallychangestoN60(cid:3)–70(cid:3)Eatlonger normaltotheregionalstrikedirection.However,ina
periods (1–1,000 s) that penetrate to lower crustal 3D situation the skew angle b will be non-zero, and
depths. The latter seems to be the regional strike induced electric currents will flow in the direction of
direction of the area as also reported by TU¨RKOGLU the major axis (a–b).
etal.(2008).Notethatthereisaninherentambiguity The phase tensor method was applied to the
of 90(cid:3) in these directions, so strike directions Erzincan MT data, and the results are shown in the
orthogonal to those listed above are also consistent second column of Figs. 3 and 4b. The a–b direction
with the data. The frequency-dependent strike direc- was found to be between N90-120(cid:3)E at short period
tion indicates that the subsurface resistivity structure bands (0.001–10 s) and N40–50(cid:3)E at long period
is somewhat 3D. Since the focus of this article is on (10–1,000 s).Thesestrikedirectionsforlongperiods
theuppercrustalstructurecorrespondingtothebasin, are consistent with the other approaches described in
the fault parallel strike direction was chosen. This Sect.4.1.However,there are somedifferencesinthe
may introduce errors in the deeper parts of a short-period data (0.1–10 s); these could be distor-
resistivity model, and this can be investigated by tionsrelatedtothecomplexstructureoftheErzincan
analysisoftheroot-mean-square(r.m.s.)misfiterrors area.
computed by the decomposition. This is a quantita- Figure 4b shows a pseudo section of the phase
tive way to evaluate the validity of the tensor tensor ellipses for the two profiles as a function of
decomposition approach as plotted in Fig. 4a. The period. The colour of the ellipses shows the value of
relativelylowr.m.s.misfitvaluesindicatethattensor the skew angle b. According to the figure, high b
decomposition is valid for both profiles at periods of valuesareobservedatperiodsgreaterthan1 s.Thus,
less than 10 s (with the exception of stations EW1 it perhaps can be concluded that 3D effects are
and EW2). present inthe measured impedances at these periods.
4.2. Phase Tensor 4.3. Induction Vectors
Tensor decomposition has the limitation that it MT strike directions computed with the methods
makesanumberofassumptionsaboutthegeoelectric described above contain an inherent 90(cid:3) ambiguity.
structureinthestudyarea,e.g.,theregional geoelec- This ambiguity can be removed by reference to
tric structure is 2D (BIBBY et al., 2005). CALDWELL geological information or by using the vertical
et al. (2004) introduced a method that provides a magnetic field, provided the structure is isotropic.
partial solution for the undistorted impedance tensor The magnetic field transfer function T = [T , T ]
x y
directly from the observed (distorted) impedance relates the vertical and horizontal magnetic fields
tensor where near-surface heterogeneity and regional through H = T H ? T H . This function can be
z x x y y
conductivity structures are 3D. The method is based graphically represented by plotting as an induction
on the phase of the impedance tensor, which is not vector. The real induction vector has components
affected by galvanic distortions (CALDWELL et al., [Re(Tx,), Re (Ty)] (PARKINSON, 1962; WIESE, 1962)
2004; BIBBY et al., 2005). The phase information is and is widely used for dimensionality and direction-
defined by the phase tensor U = X21 9 Y where ality analysis. The real induction vector points at
X and Y are the real and imaginary parts of the conductive discontinuities when plotted in the Par-
impedance tensor. The non-symmetric phase tensor kinson convention. In an ideal 2D geometry, the
can be represented graphically as an ellipse, defined induction vectors will be perpendicular to the
by the major axis U , the minor axis U and the geoelectric strike direction and provide a way to
max min
skew angle b. The parameter a expresses the tensor overcome the inherent ambiguity in strike directions
dependence on the chosen coordinate frame. If the estimated from the impedance tensor. Induction
conductivity distribution is 1D, then the phase tensor vectors for the Erzincan data are plotted in the
willbeaunitcircleandtheskewangleb = 0.Inthe Parkinson convention inFig. 5. Thedirections of the
2D case, b = 0 and a is the direction parallel or induction vectors show significant scatter and are
Vol.170,(2013) GeophysicalImagesoftheNorthAnatolianFaultZoneintheErzincanBasin 415
Figure3
Geoelectricstrikedirectionsobtainedforfiveperiodbands.Blackandwhitehistogramsrepresentthe90(cid:3)ambiguityintheMTgeoelectric
strikedirection.Theco-centredcircles(dashed)indicatethetotalnumberofthefrequenciesthathasthesamestrikeanglesinchosenperiod
bandsatallstations.GBGroomBailey,MJMCNEICEandJONES(2001),CBBCALDWELLetal.(2004)
416 U¨.Avs¸aretal. PureAppl.Geophys.
Figure4
aThe RMSmisfitof tensordecompositionfor the wholeperiod band. bPhase tensorellipses attwo profiles.Thecolourofthe ellipses
indicatestheskewb
inconsistent with an ideal 2D scenario. This scatter oftheregionis3Dratherthan2D,consistentwiththe
could be due to either noise in the data or a 3D complex pattern of faults shown in Fig. 1b.
geoelectric structure. Short induction vectors in the In summary, MT data indicate a complex 3D
middle of the basin may indicate a 1D geoelectric geometryfortheErzincanBasin,anda2Danalysisof
structure. Stations at the edge of the Erzincan Basin thesedatashouldbeundertakenwithcare.Overall,the
have longer induction vectors, indicating a strong geoelectric strike direction of the Erzincan Basin is
horizontal change in conductivity. These induction about N120(cid:3)E at short periods (0.001–10 s) and is
vectordirectionssuggestthatthegeoelectricstructure consistentwiththestrikedirectionofthesurfacetraceof
Vol.170,(2013) GeophysicalImagesoftheNorthAnatolianFaultZoneintheErzincanBasin 417
Figure5
RealinductionvectorsfordifferentperiodbandsplottedintheParkinsonconvention
theNAF(BARKAandGU¨LEN,1989;AKTARetal.,2004). analysis,withthelowestfrequenciessamplingdeepest
Thegeologicalconstraintwasessentialinovercoming into the Earth. To obtain a resistivity model as a
the90(cid:3)ambiguity,owingtothescatterintheinduction functionoftruedepth,forwardmodellingorinversion
vectors.Atlongerperiodsthestrikedirectionisbetween methods need to be applied to the MT data. The Erz-
N60(cid:3)–70(cid:3)EintheErzincanregion. incanMTdatahavebeenanalysedwith1D,2Dand3D
approaches,asdescribedinthefollowingsections.
5. Magnetotelluric Modeling of the Erzincan Basin
5.1. Magnetotelluric Data
Magnetotelluric data are recorded in the time The impedance tensor was rotated in the co-
domain and transformed to the frequency domain for ordinate system of N120(cid:3)E that was derived in Sect.
418 U¨.Avs¸aretal. PureAppl.Geophys.
4,andapparentresistivityandphaseswerecomputed. MTstationsonbothprofiles.Thisisprimarilydueto
Theverticalmagneticfieldtransferfunctionwasalso the presence of static shifts in the MT data. In
projected into this co-ordinate system. Figures 6 and addition non-uniqueness arises in the MT inverse
7illustratetheapparentresistivity,phaseandvertical problem because only the conductance of a buried
magneticfieldtransferfunctiondatainpseudosection layer is well defined by MT data. Various combina-
format and fitting curves format, respectively. Three tions of layer conductivity and thickness with the
main features can be observed in the data: sameproductwillallbeabletofitthemeasureddata.
Therefore,athick,lower conductivitylayer willgive
• Nearlyconstantphasevaluesareobservedoverthe
exactly the same MT response as a thin, more
period range 0.01–1 s for both polarizations of the
conductive layer. However, in this study, the con-
MT impedance data. This implies that the shallow
ductive layer is located at the surface, and the non-
resistivity structure does not vary strongly verti-
uniqueness problem is less serious. This is because
cally or horizontally in the Erzincan Basin.
the highest frequency data directly sample the
• Increasing apparent resistivity in the period range
conductivity of the upper part of the layer, partially
1–30 s.
overcoming the non-uniqueness.
• Decreasing apparent resistivity at periods longer
Theseismicvelocitiesinthisareaarebetween4.5
than 30 s. This implies the presence of a low
and 5.5 km/s, and according to MEJU et al. (2003)
resistivity (conductive) layer at depth.
these velocities correspond to 500–900 Xm resistiv-
ities. Thus, in our analysis, the resistivity of the
second layer (a resistive layer between two conduc-
5.2. One-Dimensional(1D)InversionoftheErzincan
tors)wasfixedto500Xm(Fig. 8andSupplementary
MT Data
Fig.A1)withtheassumptionthattheresistivityofthe
One-dimensional inversion was performed using basement rock does not change significantly over
WinGlink1 software package, which allows the user such short distances. In this case, the 10–30 Xm top
to fix the resistivity or thickness of a layer. MT layer that corresponds to the sedimentary fill of the
stationsinsidetheErzincanBasinwheretheapparent Erzincan Basin has a thickness in the range of
resistivity and phase curves suggest a 1D geoelectric 2.9–4 km and 2.5–3.6 km in the West and East,
structure (Fig. 7 and Supplementary FigA2) were respectively. Thus, the mean thickness of the Erzin-
used in 1D inversion to estimate the basin thickness. can Basin can be estimated as 3.25 ± 0.67 km.
Since the boundary between the low resistivity basin
sediments and the upper crustal crystalline rocks is
5.3. Two-Dimensional (2D) Inversion
expected to be sharp, this approach is useful because
of the Erzincan MT Data
it permits sharp changes in the resistivity. MT data
from four stations were inverted using this approach. In the previous section, it was shown that the
Two stations were chosen from the central West ErzincanMTdatacanbeconsidered2Dintheperiod
Profile (EW-5 and EW-6), and the other two stations band 0.01–100 s with a strike-direction of N120oE.
were chosen from the central East Profile (EE-5 and Whentheearthisassumedtobetwo-dimensional,the
EE-6). Initial inversions revealed a three-layer resis- data can be divided into two independent modes: (1)
tivity structure with the middlelayerhavinga higher thetransverse-electric(TE)modehaselectriccurrent
resistivity than those above and below. The upper flowingparalleltothegeoelectricstrikedirectionand
conductivelayercan beidentifiedasthesedimentary (2) the transverse-magnetic (TM) mode with electric
basin. The estimated thickness of the basin was currentflowingperpendiculartothegeoelectricstrike
observedtovarysignificantlybetweencloselyspaced direction. These two modes are sensitive to different
aspects of the subsurface resistivity structure. The
TE mode is particularly sensitive to along-strike
1 WinGLinkis a multidisciplinary softwareprogramdevel-
conductors, whereas the TM mode is more sensitive
oped by WesternGeco to process, interpret and integrate
geophysicaldata. to resistive features (WANNAMAKER et al., 1989;