Combined Multifactor Evidence of a Giant Lower-Mantle Ring Structure below the Eastern Mediterranean

In the Arabian-Northern African region, interaction of the Nubian, Arabian and Eurasian plates and many small tectonic units is conspicuous. In order to better understand this interaction, we use satellite derived gravity data (retracked to the Earth’s surface) recognized now as a powerful tool for tectono-geodynamic zonation. We applied the polynomial approximation to the gravity data which indicated the presence of a large, deep ring structure in the eastern Mediterranean centered below the Island of Cyprus. Quantitative analysis of residual gravity anomaly provides an estimate of the deep anomalous body’s upper edge at a depth of about 1700 km. Computations of the residual gravity anomalies for the lower mantle also indicate presence of anomalous sources. The GPS vector pattern coinciding with the gravity trend implies counter clockwise rotation of this structure. Independent analyses of the geoid isolines map and seismic tomography data support the existence of a deep anomaly. Paleomagnetic data analysis from the surrounding regions confirms a counter clockwise rotation. Numerous petrological, mineralogical, geodynamical and tectonic data suggest a relation between this deep structure and near-surface processes. This anomaly sheds light on a number of phenomena including the Cyprus gravity anomaly, counter clockwise rotation of the Mesozoic terrane belt and asymmetry of basins along continental transform faults.


Introduction
Understanding the relationship between deep geodynamics and subsurface geological processes is one of the prime challenges in Solid Earth Sciences [1] [2]. Here we present a combined analysis of the northern Africa-eastern Mediterranean area, where large tectonic plates (Nubian, Arabian and Eurasian) and a number of comparatively small tectonic plates interact [3]. This ~24 million km 2 region ( Figure 1) includes active rifts and collision belts, a complex pattern of continental and oceanic crusts of different ages, intense seismic activity, as well as several high-amplitude gravity anomalies and seismic velocity anomalies observed at great depths. In this region, zones of the final phases of subduction and the initial stages of rifting (spreading) are comparatively closely located (e.g., [4] [5] [6] [7] [8]). The region is located in the junction zone between East Gondwana and Eurasia, the geological-geophysical instability of which is determined by the intensity of geodynamics-both collisional and rift spreading. Here, different folded belts and cratons are developed and a variety of geological and geophysical processes are manifested (e.g., [3] [7] [9]). The eastern Mediterranean is a tectonically complex region evolving in the midst of the progressive Afro-Eurasian collision. The combination of these geological-geophysical elements requires a structural-geodynamic analysis using modern methodological and numerical methods. Geophysical surveys of the eastern Mediterranean have indicated an extensive zone of development of thinned continental crust [2] and identified a chain of pre-Alpine terranes in this region [4] [10]. Of these geological and geophysical features, the most striking and least well understood is the Cyprus anomaly-one of the highest-amplitude gravity anomalies in the world (about 200 mGals in the Bouguer reduction) [11]. In this paper, we present a combined analysis of a number of significant geophysical and geological factors that shed light on the deep structure of the studied region.
Main outlines of the article's sections are presented below. Section 2 displays application of the polynomial approximation to satellite derived gravity data and quantitative interpretation of the residual anomaly; computation of residual gravity anomalies from the lower mantle (based mainly on seismic data) confirms generally existence of anomalous object.
Brief analysis of geoid isolines map and its relationship with the deep anomalous structure is presented in Section 3.
Combined analysis of the main available geological and geophysical factors underlying presence of the giant deep rotating structure is shown in Section 4.
Section 5 is suggested to integrated examination of the numerous geological and geodynamic (both local and regional) features of the deep ring-like object; separately its influence to some modern geological-geophysical phenomena is discussed.

Examination of Satellite Derived Gravity Data
Regional gravity observations (shipborne and airborne) in the pre-satellite epoch are not uniform, containing large "white spots" resulting from variations in different scales, as well as differences in the accuracies and methodologies of gravity field observations and processing. In addition, most of the available shipborne and airborne gravity data are studied with errors in their gravity field computations exceeding 2 -3 mGals and up to 5 mGals. Modern satellite gravimetric data are a powerful and effective tool for regional tectono-geodynamic zonation including data segmentation, transformation, and comprehensive tectono-structural interpretation [12]- [17]. The satellite gravity data for this study were obtained from the World Gravity DB as retracked from Geosat and ERS missions [18]. For the gravity map construction depicted in Figure 1, more than 9.5 × 10 6 satellite gravity observations were compiled. These observations were made with regular global 1-minute grids, where the error of gravity data computation (retracked to marine/land surface) was estimated at 1 -1.2 mGals. Examination of satellite data within the context of a tectonic regional analysis in many cases enables to detect influences on crustal structure and thickness of lithosphere (e.g., [14] [16]).
Eppelbaum and Katz [15] have shown that for investigation of deep structure in large regions (many million km 2 ) the satellite observed and retracked to the Earth's surface gravity data may be employed without any additional reductions. Initially the studied Arabian-Northern African region was limited by the coordinates of 0 -38˚N and 30 -57˚W [15] [16]. In the both aforementioned works various peculiarities of the earth's crust and lithosphere were analyzed. However, in these works consistent relationships between the regional counterclockwise GPS pattern (e.g., [6] [19]) and the lithosphere's structure were not found. Therefore, it was concluded that the main source causing the GPS outline may occur at large depths.
In the present investigation the investigated area is extended to 0 -55˚N and 22 -62˚W. For the satellite gravity set analysis were applied several reliable mathematical tools including polynomial approximation and some other procedures.

Polynomial Approximation
To better resolve the nature of these anomalies, we implement the polynomial approximation, a powerful instrument for solving a range of problems in mathematics and applied sciences (e.g., [20] [21]). Polynomial computation identifies generalized trends in datasets. Processing of the aforementioned satellite gravity set ("big data") practically eliminates the appearance of random main components that can lead to errors in smaller datasets.
The general trend obtained from the polynomial cubic surface ( ) Figure 2) (residual anomaly) is similar to results that were obtained by distance weighting and nonlinear filtering. The main trend in all these processed gravity maps reflects a deep oval (quasi-ring) structure superficially reflecting a deep source of this anomaly. It should be noted that a comparable trend is reflected in a free air gravity map of long wavelength (800 -3500 km) anomalies [22].

Quantitative Analysis of the Residual Gravity Anomaly
It is conventional practice to interpret residual potential field anomalies quantitatively (e.g., [23] [24]). As it is well known, depth of anomalous source is associated with the wide of anomaly, inclination of its branches, square occupied by anomaly area and some other parameters. A preliminary analysis of the obtained gravity anomaly was carried out using improved tangent, characteristic point and areal methods developed to examine potential geophysical anomalies in conditions where the level of the normal field is unknown [25]. These methods suggested that an upper edge of the giant deep ring structure (GDRS) occurs at a depth of about 1700 km, indicating that the anomalous source is in the lower mantle. We will designate a projection of the GDRS to the upper geological sections (lithosphere and near-surface) as GDRSP.

Computations of Residual Gravity Anomalies for the Lower Mantle
To test the abovementioned hypothesis, computations of residual gravity anomalies for the lower mantle were carried out. The gravity effect of density variations in the lower mantle is hardly visible in the observed gravity field since the last one is dominated by the crustal and upper mantle density heterogeneity (e.g., [13]). Another factor, which also hides the lower mantle effects, is the dynamic topography induced by mantle flow (e.g., [3] [12]). The effect of the dynamic topography is clearly visible in the residual isostatic anomalies computed for the Middle East and surrounding areas [13] [14]. It has been demonstrated above that using the specific data processing provides nevertheless an opportunity, to refine patterns related to the lower mantle ( Figure 2). Another way is to refine the lower mantle gravity field by subtracting gravity effects of the crust and upper mantle based on independent models obtained from various kinds of seismic data constrained by mineral physics [13]. Here we demonstrate the results of this approach for the study area.
The procedure of computation of the residual gravity field consisted of two steps: 1) removing of the crustal gravity effect (including topography/bathymetry) from the observed gravity field, and 2) computation of the residual gravity anomalies for the lower mantle by removing the effect of the upper mantle from the total mantle field (the final map is shown in Figure 3). In the first stage, a 3D density model of the crust has been constructed for the Middle East and surroundings based on nearly all available seismic and a large set of geophysical prospecting data. Based on this model, the gravity effect of the crust has been determined relative to a 1D reference density model and removed from the initial gravity field together with the effect of topography/bathymetry. Accordingly, the effect of the dynamic topography as a part of the observed one is also excluded from the initial field [13].
Reducing the gravity effect of the upper mantle was based on available tomography models. For this purposes we employed S-wave variations from the model SL2013sv [26]. For depths less than 300 km, the density variations have been obtained based on the mineral physics approach [27]. Below, a constant conver- = 0.28 (where ρ is the density, Vs is the velocity of shear waves) [28]. The gravity effect of the upper mantle with the transition zone (to a depth of 700 km) has been calculated relative to a 1D reference model and removed from the whole mantle anomalies [29]; the final results are shown in Figure 3.
The most pronounced negative anomaly corresponds to the Afar triple junction (this tectonic feature is shown in Figure 4), while the northern part related to Eurasia and the Eurasia-Arabia transition zone is characterized mainly by significant positive anomalies.
The lower mantle gravity field was significantly modified compared to the whole mantle anomalies. For example, the effect of the Afar plume is almost reduced, while South Eurasia is now characterized by predominantly negative anomalies ( Figure 3). The residual gravity field is clearly composed of large and Figure 3. Residual gravity anomalies for the lower mantle after removing effect of the upper mantle (methodology is presented in [13]) from the total mantle field. The upper mantle gravity anomalies were computed based on the tomography model SL2013sv [26] as described in [29]. White ellipse shows approximate location of the maximum in Figure 2. Positioning mid-scale anomalies roughly dominated at the intervals > 1000 km and <400 km correspondingly. The mid-scale anomalies cannot be generated by lower mantle heterogeneity due to a large distance to the original density anomalies. Obviously, they are related to insufficient resolution of the initial tomography model and uncertainties of the velocity-to-density conversion as demonstrated in [29]. We can select two groups of the large-scale positive residual anomalies. The first group combines the maxima in the Eastern Mediterranean and around the Red Sea ( Figure 3). This broad anomaly fits well to the residual anomaly revealed by the numerical analysis of the initial field ( Figure 2) as shown by the white ellipse. Therefore, this analysis also confirms the presence of the deep dense structure in the lower mantle in the study area. Another strong positive anomaly (right side of Figure 3) is localized within the collision zone of Arabia and Eurasia.

Brief Analysis of Geoid Data
Generalized geoid isolines (compiled on the basis of the EMG2008) and their comparison with GPS data are shown in Figure 4. It is well known that the geoid map reflects integrated effects from the Earth's crust, mantle and core (e.g., [31]). Behavior of the geoid isolines ( Figure 4) reflects a large quasi-circular anomaly nicely correlating with the GPS vectors orientation (see arrows in Figure 4). Interestingly that numerous paleomagnetic, tectonic-geodynamic data for the region under study also indicate mainly counterclockwise rotation.
In general, elevations of the geoid reflect the arch of the structure, and the lower one-periclinal part. Geodynamically this map agrees with the GPS data pattern. The geoid anomalies also correlate to a large extent with the regional tectonic elements presented in this figure. The geoid isolines show an elongated zone occurring from the NNE to SSW, which separates areas of positive and negative geoid values. These inhomogeneities in the Earth deep mantle have been previously interpreted as the planetary Ural-African Step [32]. At present, based on the GPS, paleomagnetic and seismic data analysis (see below) we propose that such a behavior of geoid isolines is caused by the integrated effect of the earlier recognized Ural-African Step and large anomalous source occurring at a great depth (GDRS).

Integrated Analysis of Geophysical and Geological Factors
Geodynamic analyses (e.g., [33]) indicate that arched segments of geological structures usually are the most non-stable zones (especially in the case of rotation). The Red Sea spreading zone is sharply outlined by an intensive trend of the Bouguer gravity anomalies [34] coinciding with the long axis of the residual satellite observed regional gravity anomaly (GDRSP's arched sector) ( Figure 5) in center of which occurs the high-amplitude Cyprus Bouguer gravity anomaly [11].
The Red Sea zone contains also signatures of the deep geodynamic activity (e.g., [17] [49] [50] [51]). Thus, it provides an additional argument for the deep ring structure influence on the overlying slabs. Figure 5 shows a simplified tectonic scheme of the region overlain on the gravity polynomial cubic surface approximation accompanied by rotational geodynamic elements. Elements of rotational dynamics (mainly obtained from paleomagnetic data analysis ( Figure 5)) largely coincide with the counter clockwise rotation of the GDRSP derived from GPS data. In this context, it should be noted that Muttoni et al. [52] on the basis of a study of paleomagnetic data of the Triassic period suggested that the Neotethys Ocean spreading axis was rotated counterclockwise at that period. Rotation of the deep structure and its connection with the GDRSP can be also confirmed paleobiogeographically. Late Jurassic shallow-water deposits of the Negev, Antilebanon, Northern and Southern Palmyrides terranes include coral biostroms with brachiopod and echinoid banks. The brachiopod fauna isolines of polynomial obtained regional gravity trend (see Figure 2), 11) rotational geodynamic elements derived from: a) paleomagnetic (major) and tectonic (minor) data: 1) Menderes (W. Anatolia) [ (Somalirhynchia-Septirhynchia) in these facies are very similar to those of the Ethiopian paleobiogeographical province in Saudi Arabia, Ethiopia, and Somalia [35]. Thus, the sedimentary deposits of the foreland of Northern Arabia and Eastern Nubia are tectonically discordantly joined with the allochthonic Mesozoic terrain belt ( Figure 5) rotated towards the Gondwana counterclockwise.
The presence of the GDRS below the Eastern Mediterranean is also supported by independent deep seismic tomography [53]- [58] which indicate anomalous compressional and shear wave distributions at depths of 1000 -1800 km. Figure   6 (after [58]) displays a meridional seismic tomography section along the Antalya region (which occurs at the latitude of 40˚). Here seismic tomography results clearly indicate a presence of a mantle occurring inhomogeneous anomalous source within the GDRS contour. It is necessary to note that a rotating circular (elliptical) inhomogeneous structure may produce very complex and diverse effects inside it [33], so here anomalous processes of different signs may simultaneously be observed.
Here A red line marks the location of the modern geological record that authors of [58] interpret to have formed during the subduction of the slab; B is location map of vertical cross-sections C (P-waves) and D (S-waves). Relative amplitude strength, vertical, lateral extent and dip trend are very similar between tomographic models. The center of the GDRSP is also associated with low regional heat flow ano-  (Figure 5 and Figure 7). The initial spreading of the Kiama hyperzone [35] apparently took place in the present Eastern Persian Gulf [39]; this tectonic unit may have been moved to its present location along the transform faults under the influence of the GDRSP. The Mesozoic terrane belt occurring within the GDRSP, has been rotated counter clockwise about 130 m.y. ago [35]. [64] [65]) and major tectonic features.
Outcrops of the various deep magmatic elements were recognized in Cyprus (e.g., [66] [92] [93]). This indicates a high level of tectono-geodynamic activity within the GDRSP. The GPS pattern [6] [19] displays the clear counter clockwise rotation in the central-western part of the region (Figure 7) similar to expected effects of rotation of the so-called geodynamical vortex structure (correlating with the gravity generalized trend shown in Figure 2) in the center on which the high-amplitude Cyprus gravity anomaly [11] is located.
The axis of separation of the Western Caucasus (WC) and Eastern Caucasus (EC) zones-the Main East European Fault (MEEF) (e.g., [94] [95]); at present the MEEF is considered as a system of faults)-continues to the North within the Eastern European platform ( Figure 5 and Figure 7). This fault, however, does not continue to the southern analog of this structure, the Eastern Mediterranean Nubian Fault (EMNB) [16] [35] [65] (shifted relative to MEEF for 500 -600 km to the west), which is an axial fault of the GDRSP of the Mesozoic age ( Figure 5 and Figure 7).
It should be noted that the MEEF divides the regions with predominantly counter clockwise (west) and clockwise (east) paleomagnetic rotation ( Figure 5) as well as similar location of the GPS vectors ( Figure 4).
Summarizing the analysis of the deep mineral-petrological and tectono-geodynamic indicators in the region ( Figure 5 and Figure 7), we conclude: 1) The GDRSP is well marked by a variety of different magmatic features.
2) The largest occurrence of rocks and minerals of deep origin is concentrated in the GDRSP apical zone, in the center of which the Cyprus ophiolites are located; numerous deep mantle minerals were discovered in Cyprus (e.g., [66] [92]).   3) The belt of Cenozoic traps along the line corresponding to the Red Sea strike [96] (Bosworth et al., 2005) agrees to the modern long axis of the GDRSP.
4) The ancient axis of the GDRSP is emphasized by EMNB and, possibly, by its northern extension-MEEF ( Figure 5). The last fault divides the Caucasus Mts. into the Western and Eastern ones.
5) It is significant that both axes: ancient-Mesozoic (along EMNB) and modern-Cenozoic (along the Red Sea strike) were activated in the Late Cenozoic. They locate at an angle of 35˚ -40˚ each to other, and show the contours the western and eastern boundaries of the Sinai Plate originated at that time.
These paleomagnetic data agree in general with the isoline shape of the gravity data polynomial approximation and the GPS global pattern.
The reconstruction of the axial position of the GDRSP at the boundary between Mesozoic and Cenozoic is a natural structural limitation of the Sinai plate, whose geodynamic unity has been amply documented [99]. The development of asymmetric basins along the continental transform faults over the eastern part of the GDRSP was studied by Ben-Avraham [100]. The asymmetric structure of these basins and their left regional counter-clockwise rotation may be interpreted as an obvious influence of the GDRSP.  conclusions about the conjugate deformation of the Earth's ellipsoid around the 35˚ latitude, and 9) tectonic pattern. We have presented nine independent positive indicators (we consider only primary features) and random coincidence of these factors (e.g., [107]) is very small.

Conclusions
This study sheds light on the relationship between earlier unknown deep phenomena and near-surface geological-geophysical features. It appears that the presence of a deep rotating ring structure (superplume?) is reflected by several main independent indicators: satellite derived gravity data polynomial processing and computations of residual gravity anomalies for the lower mantle, results of seismic tomography, GPS data patterns, paleomagnetic data examination, behavior of geoid isolines, as well as evidence from the tectonic-structural, geodynamic, petrological and mineralogical data. Probabilistically, a random coincidence of all these factors is extremely small. Obviously, the deep ring structure influencing many tectonic-geodynamical processes is a zone that may be a geodynamic factor contributing to the Red Sea spreading. We propose that this structure affects, for instance, geological-geophysical phenomena such as the high-amplitude Cyprus gravity anomaly, the Sinai plate configuration, counter clockwise rotation of the Mesozoic terrane belt, the geometry of asymmetrical basins along the Dead Sea Transform and the movement of the tectonic unit of the Kiama paleomagnetic zone.