sign in sign up
<<
Home , Anatolia, Anatolian Plate, Bitlis, Eastern Anatolia Region, Seismic tomography

geosciences

Article Crustal Structure of the Eastern () Based on Seismic Tomography

Irina Medved 1,2,* , Gulten Polat 3 and Ivan Koulakov 1

1 Trofimuk Institute of Petroleum and SB RAS, Prospekt Koptyuga, 3, 630090 Novosibirsk, Russia; [email protected] 2 Sobolev Institute of Geology and Mineralogy SB RAS, Prospekt Koptyuga, 3, 630090 Novosibirsk, Russia 3 Department of Civil Engineering, Yeditepe University, 26 Agustos Yerleskesi, 34755 , Turkey; [email protected] * Correspondence: [email protected]; Tel.: +7-952-922-49-67

Abstract: Here, we investigated the crustal structure beneath eastern Anatolia, an area of high seismicity and critical significance for earthquake hazards in Turkey. The study was based on the local tomography method using data from earthquakes that occurred in the study area provided by the Turkiye Cumhuriyeti Ministry of Interior Disaster and Emergency Management Directorate Earthquake Department Directorate of Turkey. The dataset used for tomography included the travel times of 54,713 P-waves and 38,863 S-waves from 6355 seismic events. The distributions of the resulting seismic velocities (Vp, Vs) down to a depth of 60 km demonstrate significant anomalies associated with the major geologic and tectonic features of the region. The was revealed as a high-velocity anomaly, and the low-velocity patterns north of the suture are mostly associated with eastern Anatolia. The upper crust of eastern Anatolia was associated with a ~10 km  thick high-velocity anomaly; the lower crust is revealed as a wedge-shaped low-velocity anomaly.  This kind of seismic structure under eastern Anatolia corresponded to the hypothesized existence of Citation: Medved, I.; Polat, G.; a lithospheric window beneath this collision zone, through which hot material of the asthenosphere Koulakov, I. Crustal Structure of the rises. Thus, the presented results help to clarify the deep structure under eastern Anatolia. (Turkey) Based on Seismic Tomography. Keywords: seismic tomography; Eastern Anatolia; Arabian plate; collision zone; crustal structure Geosciences 2021, 11, 91. https:// doi.org/10.3390/geosciences11020091

Academic Editors: Pier Paolo 1. Introduction G. Bruno and Jesus Martinez-Frias Active tectonic forces at the boundaries of lithospheric plates to strong seismic Received: 27 January 2021 activity in tectonically active that affects the infrastructure and lives of people Accepted: 12 February 2021 Published: 15 February 2021 therein. In many cases, the plate interaction mechanisms, which are important for assessing long-term seismic hazards, are not completely clear. Promptly addressing problems related

Publisher’s Note: MDPI stays neutral to collisional processes requires robust information about the main structural elements with regard to jurisdictional claims in in the crust and . For this reason, geophysical investigations, particularly published maps and institutional affil- seismic tomography studies, are in high demand for establishing geodynamic scenarios of iations. lithospheric evolution in many regions of the world. We implemented seismic tomography to study the crustal structure beneath eastern Anatolia, including the eastern Anatolian accretionary complex and the surrounding areas. Figure1 shows the main tectonic features of the study region. The is bounded by two Neo-Tethyan sutures: the Izmir--Erzincan˙ suture (IAES, shown Copyright: © 2021 by the authors. in Figure1) in the north and the Bitlis suture in the south [ 1]. Moreover, the IAES separates Licensee MDPI, Basel, Switzerland. This article is an open access article the Pontides Unit (PU) in the north from the East Anatolian Accretionary Complex (EAAC) distributed under the terms and to the south [2] (Figure1). The eastern part of Turkey is characterized by the junction of conditions of the Creative Commons three large plates: the in the west, the Arabian plate in the south, and the Attribution (CC BY) license (https:// in the east. This system is divided by tectonic faults, including the North creativecommons.org/licenses/by/ Anatolian fault (NAF) and the East Anatolian fault (EAF), which intersect in the study 4.0/). area and form the Karliova triple junction (KTJ) (Figure1). This system of plates is pushed

Geosciences 2021, 11, 91. https://doi.org/10.3390/geosciences11020091 https://www.mdpi.com/journal/geosciences Geosciences 2021, 11, x FOR PEER REVIEW 2 of 12

Geosciences 2021, 11, 91 2 of 12 cluding the (NAF) and the East Anatolian fault (EAF), which inter- sect in the study area and form the Karliova triple junction (KTJ) (Figure 1). This system ofnorthward plates is pushed by the movementnorthward ofby the the Arabian movement plate of atthe a Arabian rate of ~18 plate mm/year at a rate with of ~18 respect mm/yr to withthe Eurasian respect to plate the Eurasian [3,4]. This plate movement [3,4]. This causes movement westward causes displacement westward ofdisplacement the Anatolian of theblock Anatolian along the block right-lateral along the strike-slipright-lateral NAF strike at- aslip rate NAF of ~24 at a mm/year.rate of ~24 Tomm/year. the south, To the the south,Anatolian the Anatolian plate is bounded plate is bybounded the EAF, by which the EAF, exhibits which left-lateral exhibits left displacement-lateral displacement at a rate of at~9 a mm/yearrate of ~9 mm/yr [4]. To the[4]. westTo the of west the NAFof the is NAF the central is the central Anatolian Anatolian fault zone fault (CAFZ), zone (CAFZ), which whichis a transtensional is a transtensional left-lateral left-lateral strike-slip strike fault-slip zone fault extendingzone extending from centralfrom central Anatolia Anatolia to the toMediterranean the Mediterranean [5 Se]. a [5].

FigureFigure 1. 1. TheThe simplified simplified regional regional tectonic tectonic map map of of Eastern Eastern Turkey Turkey with with topographic topographic relief relief (modified (modified from from [ [66]).]). Pink Pink areas areas areare Neogene Neogene–Quaternary–Quaternary volcanism. volcanism. Arrows Arrows indicate indicate the the direction direction of of plate plate and and fault fault motions. motions. BPM BPM–Bitlis-Poturge–Bitlis-Poturge ; Massif; CAFZ–Central Anatolia Fault Zone; EAAC—East Anatolian Accretionary Complex; EAF—East Anatolian Fault; IAES– CAFZ–Central Anatolia Fault Zone; EAAC—East Anatolian Accretionary Complex; EAF—East Anatolian Fault; IAES– Izmir-Ankara-Erzincan Suture; KTJ—Karlova Triple Junction; MOFZ—-Ovacik Fault Zone; NAF—North Anato- Izmir-Ankara-Erzincan Suture; KTJ—Karlova Triple Junction; MOFZ—Malatya-Ovacik Fault Zone; NAF—North Anatolian lian Fault; PU—Pontide Unit. Fault; PU—Pontide Unit. The region of eastern Anatolia features complex tectonic structures, which formed The region of eastern Anatolia features complex tectonic structures, which formed duringduring its its multistage multistage evolutionary evolutionary history. history. The The general general shortening shortening regime regime of eastern of eastern An- atoliaAnatolia is mostly is mostly controlled controlled by the by Arabian the Arabian-Eurasian-Eurasian collision, collision, which which began began in the in mid the- mid-Mi- oceneMiocene [1,7[–19,7].– 9The]. The continental collision between between the the Arabian Arabian and and Eurasian Eurasian plates plates began began afterafter the the closure closure of of the the Tethys Tethys , Ocean, which which separated separated Gondwana Gondwana and and Laurasia. Laurasia. The for- The mationformation of marginal of marginal basins basins and and the the accretion accretion of of island island arcs arcs and and other other continental continental fr frag-ag- ments,ments, as as well well as as the the subduction of of intervening intervening portions portions of of the the oceanic oceanic lithosphere, lithosphere, took took placeplace during during the the closure closure of of the the Neo Neo-Tethys.-Tethys. The The East East Anatolian Anatolian Accretionary Accretionary Complex Complex (EAAC(EAAC,, shown shown in in Figure Figure 11)) waswas formedformed duringduring subductionsubduction from the Late CretaceousCretaceous toto thethe earliest earliest Oligocene Oligocene and and included included Neo Neo-Tethyan-Tethyan material. material. The The EAAC, EAAC, having having formed formed on on thethe northward northward-subducting-subducting oceanic oceanic lithosphere, lithosphere, can can be be considered considered as as a a remnant remnant of of a a large large accretionaryaccretionary prism prism situated situated between between the the Pontides Pontides and and the the Bitlis Bitlis-Portuge-Portuge massif. massif. There There is nois nosubducting subducting slab slab beneath beneath the EAAC. the EAAC.;; This Thiscould could be due be to due the to detachment the detachment of the ofplate the underplate underthe collision the collision zone. zone.[10] argues [10] argues that the that slab, the slab,whose whose subduction subduction was wasgenerating generating the the Pontide arc in the north, was not attached to the Arabian plate. Instead, the slab was possibly attached to the Bitlis-Poturge block before the break-off. The Bitlis-Portuge massif (BPM shown in Figure1) consists of medium-to-highly metamorphosed units [11].

Geosciences 2021, 11, 91 3 of 12

Several petrological studies suggest that subduction was the likely cause of volcanic activity in the region [10,12]. Although this issue is still actively discussed in the scientific community, most argue that the large volume of melt generated under eastern Anatolia was caused by hot asthenospheric upwelling. Most of the area of eastern Anatolia is covered by the products of recent volcanism (with ages from 20 Ma to the present day). According to the results of isotopic dating, the younger magmatism (6–4 Ma) occurred around Van, whereas the older magmatism is mostly concentrated in the northern part of the plateau (eastward part of the study area). This suggests that the heating and melting zones migrated southward during the past 20 Ma [13,14]. In the high-elevation East Anatolian Plateau, an uplift of ~2 km is observed relative to the adjacent parts of the Arabian plate to the south (e.g., [9,15]). According to the results of receiver function analysis and surface wave tomography studies, the crust of the Anatolian plate gradually thickens to the east. The deepest Moho boundary is observed beneath eastern Anatolia (~50 km) [16]; beneath central Anatolia, the crustal thickness is ~37–47 km, and beneath western Anatolia, the crust is ~30 km thick [16,17]. According to the results of [18], in the latitudinal direction, the crust thickens northward from 42 km near the southern part of the Bitlis suture zone to 50 km along the northern edge of the Anatolian plateau in the vicinity of the North Anatolian fault. The thickness of the crust on the Arabian plate is ~40 km [19]. Some geodynamic models (see, for example, [11,20–22]) suggest that the thickening of the eastern Anatolia crust and its uplift are associated with the delamination of the mantle lithosphere within the Arabian plate. Several regional seismic tomography models consis- tently show traces of slab detachment/break-off and asthenospheric upwelling beneath the Arabia- convergence zone (see, for example, [23–30]). In all of the above studies, high-velocity anomalies beneath the margins of the Arabian and Eurasian plates indicate the subsidence of delaminated material, whereas low-velocity anomalies are observed directly under the collision zone, which can be explained by the absence of strong thinning of the mantle lithosphere and the filling of this gap with hot asthenospheric material. Over the last decade, many regional- and local-scale tomography studies were per- formed for different parts of Turkey. Seismic velocity models of the lithosphere were - tained for the areas of central Anatolia [31,32], western Turkey [33], eastern Turkey [34,35], the North Anatolian fault [36–38], the Marmara Basin [39,40], and the Erzincan Basin [41]. As shown above, a large number of studies have been performed on the regional structures of Turkey, and the geodynamic processes in these areas were successfully inves- tigated. However, despite the abundance of publications, the crustal structure of eastern Turkey is far from being perfectly understood, as many of the derived models demonstrate considerable inconsistencies with each other. In this study, we present new seismic tomog- raphy models of the crust and uppermost mantle (down to ~60 km), which were built using a new large set of travel time data. These results can help determine the links between the surface geology and deep structures controlled by complex collisional processes.

2. Data and Method The models of seismic velocities were obtained using the arrival times of P- and S-body waves from local events in the eastern Anatolia region recorded by stations in Turkey and the surrounding countries. The dataset was provided by the Turkiye Cumhuriyeti Ministry of Interior Disaster and Emergency Management Directorate Earthquake Department Di- rectorate of Turkey. The magnitudes of most of the earthquakes in the study were generally between 1 and 7, and their focal depths were mostly between 5 and 35 km. The dataset included the travel times of 114,800 P-waves and 88,962 S-waves from 13,029 earthquakes were recorded by 187 seismic stations. To perform the tomographic inversion, we discarded data with fewer than a total of eight P and S picks per event. At the stage of determining the preliminary source locations in the initial 1D velocity model, we removed the data with residuals exceeding 1 s. After applying these two criteria, for the tomographic inversion, we obtained a dataset with Geosciences 2021, 11, x FOR PEER REVIEW 4 of 12

The dataset included the travel times of 114,800 P-waves and 88,962 S-waves from 13,029 earthquakes were recorded by 187 seismic stations. To perform the tomographic inversion, we discarded data with fewer than a total of Geosciences 2021, 11, 91 4 of 12 eight P and S picks per event. At the stage of determining the preliminary source locations in the initial 1D velocity model, we removed the data with residuals exceeding 1 s. After applying these two criteria, for the tomographic inversion, we obtained a dataset with 54,71354,713 P P-- and and 38,863 38,863 S S-wave-wave picks picks from from 6355 6355 events events (almost (almost 15 15 picks picks per per event event on on average). average). FigureFigure 2 demonstrates the coverage of the rays from this dataset.

FigureFigure 2. 2. DistributionDistribution of of data data used used for for seismic seismic tomography tomography in in the the study study region. region. Blue Blue triangles triangles denote denote seismic seismic stations. stations. Red Red dotsdots are are the the events. events. Grey Grey lines lines indicate indicate the the horizontal horizontal projection projection of of the the ray ray-path.-path.

TheThe inversion inversion of thethe P-P-and and S-waves S-waves arrival arrival times times were were performed performed to obtainto obtain 3D velocity3D ve- locitymodels models and the and locations the locations of the eventsof the byevents using by the using nonlinear the nonlinear local earthquake local earthquake tomography to- mographycode LOTOS code (Local LOTOS Tomography (Local Tomography Software) [42 Software)]. This code [42 has]. This been code used has to studybeen used various to studycollision various zones collision with similar zones data with distribution similar data geometries, distribution such geometries, as the Tien such Shan as [43 the,44 Tien], the ShanCaucasus [43,44 [],24 the], [45], Calabria [24], India [46 [],45 and], Calabria others. [46], and others. TheThe calculations calculations began began with with locating locating the the sources sources in in the the reference reference 1D 1D velocity velocity model model usingusing the the grid grid se searcharch method. method. At At this this step, step, to to make make the the calculations calculations faster, faster, the the travel travel times times ofof seismic seismic waves waves were were computed computed using using a a table table of of travel travel times times calculated calculated at at a a preliminary preliminary stagestage for for all all possible possible depths depths of of events events and and epicentral epicentral distances. distances. Then,Then, we performed performed several several iterations iterations of the of thetomographic tomographic inversion inversion procedure, procedure, which whichincluded included the relocation the relocation of sources, of sources, the calculation the calculation of the sensitivity of the sensitivity matrix, and matrix, the inversion. and the inversion.The source The relocation source procedurerelocation inprocedure the first iterationin the first was iteration performed was withperformed the reference with the 1D referencevelocity model1D velocit andy then model with and the then updated with 3Dthe velocityupdated models 3D velocity calculated models in thecalculated preceding in theiterations. preceding The iterations. search forThe the search optimal for the source optimal coordinates source coordinates was based was on based the gradient on the gradientoptimization optimization method method (see [42 ]),(see which [42]), morewhich rapidly more rapidly calculates calculates the most the likely most sourcelikely sourcelocations. locati Theons. algorithmThe algorithm of a of 3D a 3D seismic seismic ray ray tracing, tracing, based based on on thethe bending method method proposedproposed by by [ [4747]],, was was used used to to calculate calculate the the travel travel times times within within the the iteration. iteration. TheThe velocity velocity models models were were parameterized parameterized using using a a set set of of nodes nodes deployed deployed according according to to the ray density. In map view, the nodes were installed regularly with a spacing of 25 km. In the ray density. In map view, the nodes were installed regularly with a spacing of 25 km. the vertical direction, the grid spacing was inversely proportional to the ray density, but the In the vertical direction, the grid spacing was inversely proportional to the ray density, spacing could not be smaller than a predefined value of 10 km. The results were calculated but the spacing could not be smaller than a predefined value of 10 km. The results were using four grids (with orientations of 0◦, 22◦, 45◦, and 67◦) and then averaged them in one calculated using four grids (with orientations of 0°, 22°, 45°, and 67°) and then averaged 3D model. The inversion was performed simultaneously for the P- and S-wave velocities, as well as for the source parameters (dx, dy, dz, and dt) and for the station corrections. For the inversion, we used the LSQR algorithm by [48] and [49]. The inversion stability

was controlled by damping and flattening. This procedure minimized the differences in the solutions from neighboring nodes. To derive the final models, we used five iterations in total. Geosciences 2021, 11, x FOR PEER REVIEW 5 of 12

Geosciences 2021, 11, 91 them in one 3D model. The inversion was performed simultaneously for the P- and5 of S 12- wave velocities, as well as for the source parameters (dx, dy, dz, and dt) and for the station corrections. For the inversion, we used the LSQR algorithm by [48] and [49]. The inversion stability was controlled by damping and flattening. This procedure minimized the differ- The optimal initial 1D model was estimated after several runs of the full inversion. In ences in the solutions from neighboring nodes. To derive the final models, we used five the first run, we used a starting model defined as an average model derived from previous iterations in total. studies in eastern Turkey (e.g., [50]). Then, to determine the optimal reference model, The optimal initial 1D model was estimated after several runs of the full inversion. we computed the average values of the P- and S-wave velocities at certain depths were In the first run, we used a starting model defined as an average model derived from pre- computed after the full inversion procedure. These values were used as a new reference vious studies in eastern Turkey (e.g., [50]). Then, to determine the optimal reference model for the next run. Thus, after several runs, we obtained the optimal reference model model, we computed the average values of the P- and S-wave velocities at certain depths presented in Table1. The P and S velocity values in the 1D model were defined at few were computed after the full inversion procedure. These values were used as a new refer- depth levels. The velocity changes were linear between those depths. We avoided sharp ence model for the next run. Thus, after several runs, we obtained the optimal reference interfaces, such as the Moho, to define artificial horizontal structures to obtain the final modelvelocity presented model. in Table 1. The P and S velocity values in the 1D model were defined at few depth levels. The velocity changes were linear between those depths. We avoided sharpTable 1.interfaces,The 1D Reference such as Model. the Moho, to define artificial horizontal structures to obtain the final velocity model. Depth, km Vp, km/s Vs, km/s Vp, Vs Table 1. The− 51D Reference Model. 5.5 3.3 1.66 Depth,10 km Vp, 5.89 km/s Vs, 3.39 km/s Vp 1.73, Vs −5 5.5 3.3 1.66 20 6.12 3.61 1.69 10 5.89 3.39 1.73 2030 6.12 6.51 3.61 3.8 1 1.71.69 3040 6.51 7.2 4.223.8 1 1.70.71 4060 7.717.2 4.22 4.51 1 1.70.70 60 7.71 4.51 1.70 120120 8.2 8.2 4.6 4.6 1 1.78.78

3.3. Seismic Seismic T Tomographyomography M Modelsodels and T Testsests TheThe main main local local tomography tomography results results are are the the 3D models of PP-- and S-waveS-wave velocity anomalies,anomalies, which which are are presented presented in inhorizontal horizontal slices slices at depths at depths of 10, of 30, 10, and 30, 50 and km 50in Figure km in 3Figure and in3 fourand vertical in four verticalsections sectionspassing through passing throughthe whole the study whole area study and crossing area and the crossing Bitlis suturethe Bitlis zone suture in Figure zone in4. Figure4.

Figure 3. PP-- and S S-velocity-velocity anomalies in horizontal sections.

Geosciences 2021, 11, 91 6 of 12 Geosciences 2021, 11, x FOR PEER REVIEW 6 of 12

FigureFigure 4.4. P-velocityP-velocity anomaliesanomalies inin verticalvertical sectionssections (at(at left),left), absoluteabsolute P-velocitiesP-velocities(in (inthe themiddle), middle), the the Vp/Vs Vp/Vs ratioratio (at(at right).right). HorizontalHorizontal sectionsection 1010km km withwithmain main geologicalgeological structurestructure elements elements and and profiles profiles locations locations on on the the top. top.

FigureFigure3 3 shows shows that that the the derived derived P- P- and and S-wave S-wave velocity velocity structures structures correlated correlated fairly fairly wellwell withwith eacheach other, which may serve as an informal informal veri verificationfication criterion. criterion. Indeed, Indeed, at at a aregional regional scale, scale, most most of of the the geological geological structures structures (for (for example, example, sedimentary sedimentary basins basins and and ig- igneousneous provinces) provinces) were were similarly similarly expressed expressed by by the the P P-- and S-waveS-wave velocities.velocities. TheThe verticalvertical sectionssections ofof Vp/VsVp/Vs ratio models, shown in FigureFigure4 4,, demonstrated demonstrated similar similar structures structures as as in in thethe verticalvertical sectionssections ofof thethe mainmain resultingresulting dVpdVp models.models. TheThe factfact thatthat thethe independentlyindependently derivedderived dVpdVp andand dVsdVs highlightedhighlighted thethe samesame structures,structures, thethe existingexisting datadata provideprovide sufficientsufficient coveragecoverage toto reliablyreliably recoverrecover thethe structuresstructures discussed.discussed. ToTo assessassess thethe rolerole ofof randomrandom noise noise in in the the data, data, we we performed performed the the so-called so-called odd/even odd/even test,test, which consists consists of of independent independent inversions inversions of oftwo two data data subsets subsets distinguished distinguished by a by ran- a randomdom criterion, criterion, for for example, example, using using odd odd-- and and even even-numbered-numbered events events.. Then, Then, we comparedcompared thethe inversioninversion resultsresults obtainedobtained usingusing thethe samesame parametersparameters andand thethe samesame algorithms.algorithms. IfIf thethe imagesimages werewere stronglystrongly different,different, itit meansmeans thatthat thethe randomrandom factorfactor inin thethe datadata waswas essential.essential. TheThe results of the the odd/even odd/even test test in inour our case case were were demons demonstratedtrated for forboth both the theP-wave P-wave and andthe S the-wave S-wave velocity velocity anomalies anomalies at a depth at a depth of 30 km of 30 in kmFigure in Figure 5. We5 could. We couldobserve observe a good acorrelation good correlation between between the major the recovered major recovered structures. structures. The main features The main of the features velocity of het- the velocityerogeneities, heterogeneities, which are important which are for important our interpr foretation, our interpretation, could be traced could in be all traced the models. in all the models. Synthetic modeling is an important step in tomography, as it provides estimates of the spatial resolution and reliability of the recovered model and allows the optimal values of free parameters (grid spacing, damping, and smoothing) to be determined. When performing synthetic modeling, we simulated the conditions of actual tomographic studies. Synthetic travel times were calculated for the existing source-receiver pairs through a predefined 3D synthetic model using the bending ray tracing algorithm and were then perturbed with random noise. To start recovering the model, we started processing with

Geosciences 2021, 11, 91 7 of 12

the absolute source determination results with the initial 1D velocity model, similar to the experimental data inversion. All the steps and controlling parameters during the recovery Geosciences 2021, 11, x FOR PEER REVIEW 7 of 12 procedure were identical to the case of computing the main tomographic model based on the experimental data.

Figure 5. Odd/Even events events test. test.

SyntheticTo assess themodeling horizontal is an resolution important of thestep model, in tomography, we performed as it a provides series of checkerboardestimates of thetests spatial with resolution anomalies and of different reliability sizes, of the as recovered presented model in Figure and6 allows. We preset the optimal a horizontal values ofcheckerboard free parameters with (grid rectangular spacing, positive damping and, and negative smoothing) anomalies to be with determined. lateral dimensions When per- of forming40 × 40 km,synthetic 60 × 60modeling, km, and we 80 ×simulated80 km, and the withconditions amplitudes of actual of ± tomographic6% (Figure6). studies. In this Syntheticparticular travel case, thetimes anomalies were calculated remained for unchanged the existing with source depth.-receiver As shown pairs inthroughFigure a6 pre-, the definedcheckerboard 3D synthetic structure model was robustlyusing the resolved bending for ray both tracing the P-wave algorithm and and the S-wavewere then velocity per- turbedmodels. with For random shallower noise. depths To (upstart to recover ~10–15ing km), the even model, the 40we km started size patternprocessing was with robustly the absoluteresolved source in most determination parts of the study results area. with At a the depth initial of ~30 1D km, velocity the anomalies model, similar at 60 × to60 the km experimentalwere more resolvable data inversion. than those All the at steps a depth and ofcontrolling 40 km. Additionally, parameters during in the 60the km recovery depth proceduresection, the were resolvable identical pattern to the sizecase was of computing approximately the main 60 km. tomographic However, themodel models based with on thesmaller experimental anomalies data. were strongly smeared at this depth. ToBased assess on the this horizontal test, we could resolution conclude of the that model, for most we partsperformed of the a study series area, of checker- we had boardsufficient tests resolution with anomalies to reveal of thedifferent main structures,sizes, as presented which will in beFigure discussed 6. We later. preset The a hori- parts zontalof the checkerboard model where thewith checkerboard rectangular waspositive not correctlyand negative restored anomalies should with be considered lateral dimen- with sionsprudence of 40in × the40 km, interpretation. 60 × 60 km, and 80 × 80 km, and with amplitudes of ± 6% (Figure 6). In this particular case, the anomalies remained unchanged with depth. As shown in Figure 6, the checkerboard structure was robustly resolved for both the P-wave and the S-wave velocity models. For shallower depths (up to ~10–15 km), even the 40 km size pattern was robustly resolved in most parts of the study area. At a depth of ~30 km, the anomalies at 60 × 60 km were more resolvable than those at a depth of 40 km. Additionally, in the 60 km depth section, the resolvable pattern size was approximately 60 km. However, the models with smaller anomalies were strongly smeared at this depth.

Geosciences 2021, 11, x FOR PEER REVIEW 8 of 12

Geosciences 2021, 11, 91 8 of 12 Geosciences 2021, 11, x FOR PEER REVIEW 8 of 12

Figure 6. Checkerboard tests for three P- and S-velocity models. The sizes of anomalies are 40, 60, 80 km. Gray lines mark the boundaries of the synthetic anomalies.

Based on this test, we could conclude that for most parts of the study area, we had FigureFigure 6. 6. CheckerboardCheckerboard tests testssufficient for for three three P P-resolution- and and S S-velocity-velocity to reveal models. models. the The Themain sizes sizes structures, of of anomalies anomalies which are are will40, 40, 60, 60, be 80 80discussed km. km. Gray Gray later.lines lines ma markTherk parts thethe boundaries boundaries of of the the synthetic syntheticof theanomalies. anomalies. model where the checkerboard was not correctly restored should be considered with prudence in the interpretation. Based To estimate on this the the test, vertical vertical we could resolution, resolution, conclude we we thatperformed performed for most a synthetic aparts synthetic of thetest teststudy with with anomaliesarea, anomalies we had de- sufficientfineddefined along along resolution vertical vertical profiles.to reveal profiles. Because the Becausemain of structures, the of trade the trade-off- offwhich between will between be the discussed source the source depth later. deptha Thend veloc- parts and ofityvelocity the distribution, model distribution, where the verticalthe the checkerbo vertical resolution resolutionard iswas typically not is typically correctly much much poorerrestored poorer than should the than horizontal be the considered horizontal reso- withlutionresolution prudence in cases in cases in of the passive of interpretation. passive source source tomography tomography studies. studies. In In Figure Figure 7,7, we present vertical vertical checkerboard To estimate tests tests the associated associated vertical resolution, with with the the verticalwe vertical performed profiles profiles a usedsynthetic used to to prese present test ntwith the the anomalies main main results. results. de- finedAcross along these these vertical sections, sections, profiles. the the anomalies anomalies Because wereof were the defined trade defined-off within between within a 100 a the 100 km source kmthick thick depthband. band. aInnd the veloc- In hor- the ityizontalhorizontal distribution, direction, direction, the the vertical theanomalies anomalies resolution had had a sizeis a typically size of 90 of km. 90 much km. In the Inpoorer thevertical vertical than direction, the direction, horizontal the the signs reso- signs of lutiontheof the anomalies anomalies in cases change of changed passived at atdepths source depths of tomography of20 20 km km and and studies.70 70 km. km. F ForInor theFigure the shallow shallow 7, we part part present of of the the vertical model, model, checkerboardthethe recovery recovery result resulttests associatedof this this test test appearwith appeared theed vertical to be robust robust; profiles; the used vertical to prese resolution nt the was main sufficient sufficient results. Acrosstoto detect detect these the the sections,change change in in the the the anomalies anomaly anomaly sign were at defined a depth within of 20 km a 100 depth, km thick but the band. model In the did hor- not izontalhave a direction,sufficient sufficient theresolution anomalies to recognize had a size anomaly of 90 km. sign In thechanges vertical at direction, 70 km. It may the signs also be of theobserved anomalies that thechange anomalies d at depths at the of ends 20 km of theand profiles profiles70 km. were For the smeared shallow due part to of the the poor model, data thecoverage recovery along result the ofmargins this test of appear the study ed to area. be robust ; the vertical resolution was sufficient to detect the change in the anomaly sign at a depth of 20 km depth, but the model did not have a sufficient resolution to recognize anomaly sign changes at 70 km. It may also be observed that the anomalies at the ends of the profiles were smeared due to the poor data coverage along the margins of the study area.

Figure 7. Vertical checkerboardcheckerboard test.

Figure 7. Vertical checkerboard test.

Geosciences 2021, 11, 91 9 of 12

4. Discussion and Interpretation The horizontal sections of the obtained seismic tomography model demonstrated a strong predominance of high-velocity anomalies in the southern part of the studied region beneath the Arabian plate. To the north of the Bitlis suture zone, we observed low-velocity anomalies that mark a boundary between the Arabian plate and the East Anatolian plate. Horizontal sections at a depth of 10 km (Figure3) showed a clear correlation between the low-velocity anomalies and known tectonic features, such as the NAF, the EAF, and the MAFZ, which are highlighted in the Figures with black dotted lines. Deeper horizontal sections at depths of 30 and 60 km show that the entire study area to the north of the Bitlis suture zone is associated with low-velocity anomalies. The four vertical profiles shown in Figure4 are oriented north-south and are nearly perpendicular to the Bitlis suture zone. The first section crosses the Anatolian block, the second section crosses the NAF and EAF, the third section passes through the KTJ, and the fourth section passes through the EAAC and crosses the volcanic province around . In all vertical sections, the Bitlis suture zone was associated with a low-velocity anomaly at the surface, which is most likely due to the accretion of sedimentary and upper crustal rocks at the boundary between the Arabian and East Anatolian plates. Moreover, the vertical sections demonstrate some clear structural differences between the high-velocity Arabian plate to the south and the low-velocity East Anatolian plate to the north of the Bitlis border. Compared with the crust of eastern Anatolia, the crust of the Arabian plate was denser and more consolidated; therefore, high P- and S-wave velocities prevail. In addition, according to the results shown in the vertical sections, a shallow high- velocity anomaly with a thickness of ~10–12 km was observed to the north of the Bitlis suture boundary, which was underlain by a thick low-velocity anomaly. Based on our model, it was difficult to evaluate the actual thickness of this deeper layer, as, according to the results of the synthetic test shown in Figure7, the anomalies were smeared at great depth in the obtained models. This low-velocity anomaly has an inclined shape that dips northward from the Bitlis suture zone underneath the East Anatolian plate. This feature seems to be consistent with available receiver function results showing the gradual northward thickening of the crust [16,17]. According to [18], in the Bitlis suture zone, the Moho depth is 42 km, while in the northern part, for example, under the NAF, the Moho reaches 50 km in depth [19]. Beneath eastern Anatolia, the crustal structure appears to be atypical. In the upper part of the velocity sections, the velocities appear to be higher than those of regular continental crustal structures, whereas, in the lower crust, we observe prominent low- velocity anomalies. The higher-velocity structures in the upper crust might be associated with the widespread distribution of Neogene-Quaternary volcanism. Strongly consolidated intrusions and massive effusive flows have an integral effect of increasing the velocity in the upper crust. The low velocities of seismic waves in the lower mafic crust can be explained by active shortening, which led to the entrainment of highly weakened material in the lower crust. This material was immersed to great depth, where it underwent phase changes and melting. An important additional factor affecting the lower crust is the absence of strong thinning of the mantle lithosphere beneath eastern Anatolia (see, for example, [20,21]). As a result, the lithospheric window is filled with overheated asthenospheric material that reaches the bottom of the crust. In this case, the low velocities in the lower crust in our tomography models may represent anomalous heating of the crust and possible melting due to elevated heat flow from the asthenosphere. Our findings are strongly consistent with the modeling results by [51], who presented a rheological model of eastern Anatolia and showed that the crust consists of a strong brittle upper crust (down to depths of 10–15 km) and a weak ductile lower crust (depths from 10–15 km to 40 km). In the southern part of vertical profiles 1, 2, and 3 (Figure4), there was a high-velocity anomaly related to the Arabian plate, which has north immersion trending underneath the Geosciences 2021, 11, x FOR PEER REVIEW 10 of 12 Geosciences 2021, 11, 91 10 of 12

the observed low-velocity anomaly. This positive anomaly was divided into two parts: oneobserved has an low-velocity isometric shape anomaly. and was This located positive directly anomaly below was dividedthe Bitlis into suture two zone, parts: and one anotherhas an isometric larger anomaly shape and was was located located to the directly south. below We thepropose Bitlis that suture the zone, former and anomaly another representslarger anomaly a remnant was located of the to Neo the- south.Tethys Weoceanic propose lithosphere that the formerremaining anomaly under represents the Bitlis a sutureremnant zone, of the the Neo-Tethys main part oceanicof which lithosphere broke off remainingand plunged under into the the Bitlis mantle suture during zone, sub- the duction.main part The of whichlarger part broke belongs off and to plunged the crust into of thethe mantleArabian during plate. subduction.A schematic Thestructure larger ofpart the belongs crust beneath to the crust eastern of the Anatolia Arabian according plate. A schematic to the results structure obtained of the in crust this workbeneath is showneastern in Anatolia Figure 8. according to the results obtained in this work is shown in Figure8.

FigureFigure 8. 8. SchemeScheme of of the the Crustal Crustal Struc Structureture along along profile profile 1 under Eastern Anatolia.

InIn the the southern southern part part of of the the vertical vertical sections, sections, we we can can see see that that the the crust crust of of the the Arabian Arabian plateplate is is associated associated with with a ahigh high-velocity-velocity anomaly anomaly reaching reaching depths depths of more of more than than 60 km, 60 km,alt- houghalthough the theMoho Moho boundary boundary reaches reaches onl onlyy 40 km 40 km,, according according to [ to19 [].19 This]. This structure structure can can be explainedbe explained by the by theongoing ongoing northward northward dipping dipping of the of theArabian Arabian plate. plate. Since Since the crust the crust of the of Arabianthe Arabian plate plate does does not differ not differ structurally structurally (thick (thick sedimentary sedimentary layer, layer, many many faults, faults, etc.), etc.), the velocitythe velocity anomaly anomaly througho throughoutut all of allthe of investigated the investigated volume volume (up to (up a depth to a depthof 60 km) of 60 looks km) uniformlylooks uniformly positive. positive. In the In northern the northern part part of vertical of vertical profile profile 1, 1, a a high high-velocity-velocity anomaly anomaly standsstands out, out, which which most most likely likely belongs belongs to to a a denser denser crust crust under under the the Pontides. Pontides.

5.5. Conclusions Conclusions UsingUsing the LOTOSLOTOS algorithm,algorithm, we we obtained obtained a 3Da 3D model model of heterogeneousof heterogeneous crustal crustal P- and P- S-wave velocities beneath eastern Anatolia down to a depth of ~60 km. We used a large and S-wave velocities beneath eastern Anatolia down to a depth of ~60 km. We used a amount of data provided by the T. C. Ministry of Interior Disaster and Emergency Manage- large amount of data provided by the T. C. Ministry of Interior Disaster and Emergency ment Directorate Earthquake Department Directorate of Turkey. The obtained tomographic Management Directorate Earthquake Department Directorate of Turkey. The obtained models were verified using different tests showing high reliability and fair resolution. tomographic models were verified using different tests showing high reliability and fair The tomographic models demonstrate a clear separation of the study region into two resolution. tectonic units. To the south of the Bitlis suture zone, the crust of the consolidated Arabian The tomographic models demonstrate a clear separation of the study region into two plate is marked as a high-velocity anomaly. The low-velocity patterns to the north of tectonic units. To the south of the Bitlis suture zone, the crust of the consolidated Arabian the Bitlis suture are mostly associated with eastern Anatolia. According to the presented plate is marked as a high-velocity anomaly. The low-velocity patterns to the north of the vertical sections, beneath the East Anatolian plate, we observed a thin (~10–12 km) high- Bitlis suture are mostly associated with eastern Anatolia. According to the presented ver- velocity anomaly in the upper crust, which might be associated with massive Neogene- tical sections, beneath the East Anatolian plate, we observed a thin (~10–12 km) high-ve- Quaternary magmatic activity in this area. In the lower crust, we observed an inclined locitylow-velocity anomaly anomaly in the upper having crust, the smallestwhich might thickness be associated beneath thewith Bitlis massive zone Neogene and the largest-Qua- ternarythickness magmatic beneath activity the northern in this partarea. of In the the section lower crust, (>50 km).we observe This featured an inclined seems tolow be- velocityconsistent anomaly with the having general the northwardsmallest thickness thickening beneath trend the of theBitlis crust, zone as and revealed the largest from thicknessexisting receiver beneath function the northern studies. part of the section (>50 km). This feature seems to be con- sistentIt with is most the likely general that northward the upper thickening crustal high-velocity trend of the anomaly crust, as corresponds revealed from to the existing strong receiverfelsic layer function of the studies. crust, while the low-velocity anomaly is associated with the weak lower maficIt crust. is most Thus, likely the that presented the upper results crustal confirm high the-velocity hypothesis anomaly of the corresponds existence of a to litho- the strongspheric felsic window layer beneath of the crust, eastern while Anatolia, the low through-velocity which anomaly hot material is associated of the with asthenosphere the weak lowerrises, andmafic these crust. findings Thus, the help presented clarify the results deep confirm structure the under hypothesis eastern of Anatolia. the existence of a lithospheric window beneath eastern Anatolia, through which hot material of the asthen- osphere rises, and these findings help clarify the deep structure under eastern Anatolia.

Geosciences 2021, 11, 91 11 of 12

Author Contributions: G.P. analyzed the data; I.M. and I.K. obtained seismic tomography models and contributed to preparing figures; All authors participated in discussions of the results and writing the manuscript. All authors have read and agreed to the published version of the manuscript. Funding: The reported study was funded by Russian Foundation for Basic Research, project number 19-35-60002. Data Availability Statement: The full directory of LOTOS code with data corresponding to this study is available at https://doi.org/10.5281/zenodo.3944178 (accessed on 15 February 2021). Conflicts of Interest: The authors declare no conflict of interest.

References 1. Okay, A.I.; Zattin, M.; Cavazza, W. Apatite fission-track data for the Miocene Arabia-Eurasia collision. Geology 2010, 38, 35–38. [CrossRef] 2. Mueller, M.A.; Licht, A.; Campbell, C.; Ocako˘glu,F.; Taylor, M.H.; Burch, L.; Ugrai, T.; Kaya, M.; Kurto˘glu,B.; Coster, P.M.C.; et al. Collision Chronology Along the Izmir-Ankara-Erzincan˙ Suture Zone: Insights From the Sarıcakaya Basin, Western Anatolia. 2019, 38, 3652–3674. [CrossRef] 3. Tan, O.; Taymaz, T. Active tectonics of the Caucasus: Earthquake source mechanisms and rupture obtained from inversion of teleseismic body waveforms. Spec. Pap. Geol. Soc. Am. 2006, 409, 531. 4. McClusky, S.; Balassanian, S.; Barka, A.; Demir, C.; Ergintav, S.; Georgiev, I.; Gurkan, O.; Hamburger, M.; Hurst, K.; Kahle, H.; et al. Global Positioning System constraints on plate kinematics and dynamics in the and Caucasus. J. Geophys. Res. Solid 2000, 105, 5695–5719. 5. Yıldırım, C.; Sarıkaya, M.A.; Çiner, A. Late Pleistocene intraplate extension of the Central Anatolian Plateau, Turkey: Inferences from cosmogenic exposure dating of alluvial fan, landslide, and moraine surfaces along the Ecemi¸sFault Zone. Tectonics 2016, 35, 1446–1464. [CrossRef] 6. Özacar, A.A.; Zandt, G.; Gilbert, H.; Beck, S.L. Seismic images of crustal variations beneath the East Anatolian Plateau (Turkey) from teleseismic receiver functions. Geol. Soc. Spec. Publ. 2010, 340, 485–496. [CrossRef] 7. ¸Sengör, A.M.C.; Yilmaz, Y. Tethyan evolution of Turkey: A plate tectonic approach. Tectonophysics 1981, 75, 181–241. [CrossRef] 8. Allen, M.; Jackson, J.; Walker, R. Late Cenozoic reorganization of the Arabia-Eurasia collision and the comparison of short-term and long-term deformation rates. Tectonics 2004, 23.[CrossRef] 9. Dewey, J.F.; Hempton, M.R.; Kidd, W.S.F.; Saroglu, F.; ¸Sengör, A.M.C. Shortening of continental lithosphere: The neotectonics of Eastern Anatolia - A young collision zone. Geol. Soc. Spec. Publ. 1986, 19, 1–36. [CrossRef] 10. Keskin, M. Magma generation by slab steepening and break-off beneath a subduction-accretion complex: An alternative model for collision-related volcanism in Eastern Anatolia, Turkey. Geophys. Res. Lett. 2003, 30, 7–10. [CrossRef] 11. ¸Sengör, A.M.C.; Özeren, S.; Genç, T.; Zor, E. East Anatolian High Plateau as a mantle-supported, north-south shortened domal structure. Geophys. Res. Lett. 2003, 30, 2–5. [CrossRef] 12. Yilmaz, Y. Comparison of young volcanic associations of western and eastern Anatolia formed under a compressional regime: A review. J. Volcanol. Geotherm. Res. 1990, 44, 69–87. [CrossRef] 13. Schleiffarth, W.K.; Darin, M.H.; Reid, M.R.; Umhoefer, P.J. Dynamics of episodic Late Cretaceous-Cenozoic magmatism across Central to Eastern Anatolia: New insights from an extensive geochronology compilation. Geosphere 2018, 14, 1990–2008. [CrossRef] 14. Rabayrol, F.; Hart, C.J.R.; Thorkelson, D.J. Temporal, spatial and geochemical evolution of late Cenozoic post-subduction magmatism in central and eastern Anatolia, Turkey. Lithos 2019, 336, 67–96. 15. ¸Sengör, A.M.C.; Kidd, W.S.F. Post-collisional tectonics of the Turkish- and a comparison with . Tectonophysics 1979, 55, 361–376. [CrossRef] 16. Vanacore, E.A.; Taymaz, T.; Saygin, E. Moho structure of the anatolian plate from receiver function analysis. Geophys. J. Int. 2013, 193, 329–337. [CrossRef] 17. Tezel, T.; Shibutani, T.; Kaypak, B. Crustal structure variation in western Turkey inferred from the receiver function analysis. Tectonophysics 2010, 492, 240–252. 18. Zor, E.; Sandvol, E.; Gürbüz, C.; Türkelli, N.; Seber, D.; Barazangi, M. The crustal structure of the East Anatolian plateau (Turkey) from receiver functions. Geophys. Res. Lett. 2003, 30.[CrossRef] 19. Gök, R.; Pasyanos, M.E.; Zor, E. Lithospheric structure of the -continent collision zone: Eastern Turkey. Geophys. J. Int. 2007, 169, 1079–1088. [CrossRef] 20. Gans, C.R.; Beck, S.L.; Zandt, G.; Biryol, C.B.; Ozacar, A.A. Detecting the limit of slab break-off in central Turkey: New high-resolution Pn tomography results. Geophys. J. Int. 2009, 179, 1566–1572. 21. Memi¸s,C.; Gö˘gü¸s,O.H.; Uluocak, E.¸S.;Pysklywec, R.; Keskin, M.; ¸Sengör, A.M.C.; Topuz, G. Long Wavelength Progressive Plateau Uplift in Eastern Anatolia Since 20 Ma: Implications for the Role of Slab -Back and Break-Off. Geochem. Geophy. Geosys. 2020, 21.[CrossRef] 22. Pearce, J.A.; Bender, J.F.; De Long, S.E.; Kidd, W.S.F.; Low, P.J.; Güner, Y.; Saroglu, F.; Yilmaz, Y.; Moorbath, S.; Mitchell, J.G. Genesis of collision volcanism in Eastern Anatolia, Turkey. J. Volcanol. Geotherm. Res. 1990, 44, 189–229. [CrossRef] Geosciences 2021, 11, 91 12 of 12

23. Koulakov, I.; Zabelina, I.; Amanatashvili, I.; Meskhia, V. Nature of orogenesis and volcanism in the Caucasus region based on results of regional tomography. Solid Earth 2012, 3, 327–337. [CrossRef] 24. Zabelina, I.; Koulakov, I.; Amanatashvili, I.; El Khrepy, S.; Al-Arifi, N. Seismic structure of the crust and uppermost mantle beneath Caucasus based on regional earthquake tomography. J. Asian Earth Sci. 2016, 119, 87–99. [CrossRef] 25. Zor, E. Tomographic evidence of slab detachment beneath eastern Turkey and the Caucasus. Geophys. J. Int. 2008, 175, 1273–1282. [CrossRef] 26. Lei, J.; Zhao, D. Teleseismic evidence for a break-off subducting slab under Eastern Turkey. Earth Planet. Sci. Lett. 2007, 257, 14–28. [CrossRef] 27. Portner, D.E.; Delph, J.R.; Berk Biryol, C.; Beck, S.L.; Zandt, G.; Özacar, A.A.; Sandvol, E.; Türkelli, N. Subduction termination through progressive slab deformation across Eastern Mediterranean subduction zones from updated P-wave tomography beneath Anatolia. Geosphere 2018, 14, 907–925. [CrossRef] 28. Kaban, M.K.; Petrunin, A.G.; El Khrepy, S.; Al-Arifi, N. Diverse Continental Subduction Scenarios Along the Arabia-Eurasia Collision Zone. Geophys. Res. Lett. 2018, 45, 6898–6906. [CrossRef] 29. Al-Lazki, A.I.; Seber, D.; Sandvol, E.; Turkelli, N.; Mohamad, R.; Barazangi, M. Tomographic Pn velocity and anisotropy structure beneath the Anatolian plateau (eastern Turkey) and the surrounding regions. Geophys. Res. Lett. 2003, 30, 4–7. [CrossRef] 30. Maggi, A.; Priestley, K. Surface waveform tomography of the Turkish-Iranian plateau. Geophys. J. Int. 2005, 160, 1068–1080. [CrossRef] 31. Salah, M.K.; ¸Sahin,¸S.;Topatan, U. Crustal velocity and Vp/Vs structures beneath central Anatolia from local seismic tomography. Arab. J. Geosci. 2014, 7, 4101–4118. [CrossRef] 32. Warren, L.M.; Beck, S.L.; Biryol, C.B.; Ozacar, A.A.; Yang, Y. Crustal velocity structure of Central and Eastern Turkey from ambient noise tomography. Geophys. J. Int. 2013, 194, 1941–1954. [CrossRef] 33. Salaün, G.; Pedersen, H.A.; Paul, A.; Farra, V.; Karabulut, H.; Hatzfeld, D.; Papazachos, C.; Childs, D.M.; Pequegnat, C.; Afacan, T.; et al. High-resolution surface wave tomography beneath the Aegean-Anatolia region: Constraints on upper-mantle structure. Geophys. J. Int. 2012, 190, 406–420. [CrossRef] 34. Angus, D.A.; Wilson, D.C.; Sandvol, E.; Ni, J.F. Lithospheric structure of the Arabian and Eurasian collision zone in eastern Turkey from S-wave receiver functions. Geophys. J. Int. 2006, 166, 1335–1346. [CrossRef] 35. Salah, M.K.; ¸Sahin,¸S.;Aydin, U. Seismic velocity and Poisson’s ratio tomography of the crust beneath East Anatolia. J. Asian Earth Sci. 2011, 40, 746–761. [CrossRef] 36. Koulakov, I.; Bindi, D.; Parolai, S.; Grosser, H.; Milkereit, C. Distribution of seismic velocities and attenuation in the crust beneath the North Anatolian fault (Turkey) from local earthquake tomography. Bull. Seismol. Soc. Am. 2010, 100, 207–224. [CrossRef] 37. Kaypak, B. Three-dimensional VP and VP/VS structure of the upper crust in the Erzincan basin (eastern Turkey). J. Geophys. Res. Solid Earth 2008, 113.[CrossRef] 38. Yolsal-Çevikbilen, S.; Biryol, C.B.; Beck, S.; Zandt, G.; Taymaz, T.; Adiyaman, H.E.; Özacar, A.A. 3-D crustal structure along the North Anatolian Fault Zone in north-central Anatolia revealed by local earthquake tomography. Geophys. J. Int. 2012, 188, 819–849. [CrossRef] 39. Karabulut, H.; Özalaybey, S.; Taymaz, T.; Aktar, M.; Selvi, O.; Kocaoglu,ˇ A. A tomographic image of the shallow crustal structure in the Eastern Marmara. Geophys. Res. Lett. 2003, 30, 1–4. [CrossRef] 40. Polat, G.; Özel, N.M.; Koulakov, I. Investigating P-and S-wave velocity structure beneath the (Turkey) and the surrounding area from local earthquake tomography. Earth Planets Sp. 2016, 68, 1–14. [CrossRef] 41. Gökalp, H. Tomographic Imaging of the Seismic Structure Beneath the East Anatolian Plateau, Eastern Turkey. Pure Appl. Geophys. 2012, 169, 1749–1776. [CrossRef] 42. Koulakov, I. LOTOS code for local earthquake tomographic inversion: Benchmarks for testing tomographic algorithms. Bull. Seismol. Soc. Am. 2009, 99, 194–214. [CrossRef] 43. Zabelina, I.V.; Koulakov, I.Y.; Buslov, M.M. Deep mechanisms in the Kyrgyz Tien Shan orogen (from results of seismic tomography). Russ. Geol. Geophys. 2013, 54, 695–706. [CrossRef] 44. Sychev, I.V.; Koulakov, I.; Sycheva, N.A.; Koptev, A.; Medved, I.; El Khrepy, S.; Al-Arifi, N. Collisional Processes in the Crust of the Northern Tien Shan Inferred From Velocity and Attenuation Tomography Studies. J. Geophys. Res. Solid Earth 2018, 123, 1752–1769. [CrossRef] 45. Koulakov, I.; Gerya, T.; Rastogi, B.K.; Jakovlev, A.; Medved, I.; Kayal, J.R.; El Khrepy, S.; Al-Arifi, N. Growth of mountain belts in central triggers a new collision zone in central India. Sci. Rep. 2018, 8, 1–8. [CrossRef][PubMed] 46. Koulakov, I.; Jakovlev, A.; Zabelina, I.; Roure, F.; Cloetingh, S.; El Khrepy, S.; Al-Arifi, N. Subduction or delamination beneath the Apennines? Evidence from regional tomography. Solid Earth 2015, 6, 669–679. [CrossRef] 47. Um, J.; Thurber, C.A. Fast algorithm for two-point seismic ray tracing. Bull. Seismol. Soc. Am. 1987, 77, 972–986. 48. Paige, C.C.; Saunders, M.A. LSQR: An algorithm for sparse linear equations and sparse least squares. ACM Trans. Math. Softw. 1982, 8, 43–71. [CrossRef] 49. Nolet, G. propagation and seismic tomography. Seism. Tomogr. 1987, 1–23. Available online: https://link.springer. com/chapter/10.1007%2F978-94-009-3899-1_1 (accessed on 14 February 2021). 50. Çivgin, B.; Kaypak, B. Estimation of the crustal structure in Central Anatolia (Turkey) using receiver functions. Turk. J. Earth Sci. 2017, 26, 314–330. [CrossRef] 51. Oruç, B.; Gomez-Ortiz, D.; Petit, C. Lithospheric flexural strength and effective elastic thicknesses of the Eastern Anatolia (Turkey) and surrounding region. J. Asian Earth Sci. 2017, 150, 1–13. [CrossRef]