Evidence for anomalous mantle upwelling beneath the Arabian Platform from travel time tomography inversion

    Evidence for anomalous mantle upwelling beneath the Arabian Platform from travel time tomography inversion Ivan Koulakov, Evgeniy Burov, Sierd Cloetingh, Sami El Khrepy, Nassir Al-Arifi, Natalia Bushenkova PII: DOI: Reference:

S0040-1951(15)00651-4 doi: 10.1016/j.tecto.2015.11.022 TECTO 126853

To appear in:

Tectonophysics

Received date: Revised date: Accepted date:

17 August 2015 12 November 2015 20 November 2015

Please cite this article as: Koulakov, Ivan, Burov, Evgeniy, Cloetingh, Sierd, El Khrepy, Sami, Al-Arifi, Nassir, Bushenkova, Natalia, Evidence for anomalous mantle upwelling beneath the Arabian Platform from travel time tomography inversion, Tectonophysics (2015), doi: 10.1016/j.tecto.2015.11.022

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Evidence for anomalous mantle upwelling beneath the

T

Arabian Platform from travel time tomography

by

SC R

IP

inversion

NU

Ivan Koulakov1,2 ([email protected], corresponding author), Evgeniy Burov3,4 ([email protected]),

MA

Sierd Cloetingh5 ([email protected])

D

Sami El Khrepy6,7 ([email protected]),

TE

Nassir Al-Arifi7 ([email protected])

1.

CE P

Natalia Bushenkova1,2 ([email protected])

Trofimuk Institute of Petroleum Geology and Geophysics SB RAS, Prospekt Koptyuga, 3,

AC

630090, Novosibirsk, Russian Federation, 2.

Novosibirsk State University, Novosibirsk, Russia, Pirogova 2, 630090, Novosibirsk, Russia

3.

Sorbonne Universités, UPMC University Paris VI, F-75005 Paris, France

4.

CNRS, UMR 7193, Institut des Sciences de la Terre Paris (iSTeP), F-75005 Paris, France

5.

Faculty of Geosciences, Utrecht University, Budapestlaan 4, 3584, Utrecht, The Netherlands

6.

King Saud University, Riyadh, Saudi Arabia, P.O. Box 2455, Riyadh 11451, Saudi Arabia.

7.

National Research Institute of Astronomy and Geophysics, NRIAG, 11421, Helwan, Egypt. Revision of the paper TECTO10593 submitted to Tectonophysics November, 2015 1

ACCEPTED MANUSCRIPT Abstract We present a new model of P-velocity anomalies in the upper mantle beneath the Arabian Peninsula, Red Sea, and surrounding regions. This model was computed with the use of travel time data from

IP

T

the global catalogue of the International Seismological Center (ISC) for the years of 1980-2011. The

SC R

reliability of the model was tested with several synthetic tests. In the resulting seismic model, the Red Sea is clearly associated with a higher P-velocity anomaly in the upper mantle at least down to 300 km depth. This anomaly might be caused by upward deviation of the main mantle interfaces caused

NU

by extension and thinning of the lithosphere due to passive rifting. Thick lithosphere of the Arabian Platform is imaged as a high-velocity anomaly down to 200-250 km depth. Below this plate, we

MA

observe a low-velocity structure that is interpreted as a hot mantle upwelling. Based on the tomography results, we propose that this upper mantle anomaly may represent hot material that

D

migrates westward and play a major role in the formation of Cenozoic basaltic lava fields in western

TE

Arabia. On the northeastern side of the Arabian Plate, we clearly observe a dipping high-velocity

CE P

zone beneath Zagros and Makran, which is interpreted as a trace of subduction or delamination of the

AC

Arabian Plate lithosphere.

Key words: seismic tomography, Arabian Plate, Red Sea, Cenozoic volcanism

Highlights: 1. A new upper mantle tomographic model is constructed for the Arabian Plate region 2. The derived anomalies fit the major structural units of the region 3. A low-velocity anomaly beneath the Arabian Plate traces a mantle upwelling 4. This hot area was responsible for the origin of recent volcanism in Western Arabia 5. A high velocity beneath the Red Sea may indicate a passive rifting mechanism

2

ACCEPTED MANUSCRIPT Introduction The western part of the Arabian Peninsula along the Red Sea Rift is characterized by widespread Cenozoic basaltic lava fields (Figure 1), also called harrats, which cover large areas of

T

more than 180,000 square km in total. Some of these volcanic centers appear to have been active in

IP

recent past and even currently. Camp et al. (1987) identified more than twenty large eruptions in

SC R

Saudi Arabia and Yemen during the last 1,500 years. For example, the existence of ongoing magmatic activity was supported by recent unrest in the Harrat Lunayyir in April-June 2009 (e.g.,

NU

Pallister et al., 2010). More than 30,000 earthquakes, with some of them reaching a magnitude of M5.4, were recorded during this period. Many geophysical observations (e.g., Pallister et al., 2010;

MA

Baer and Hamiel, 2010; Al-Amri et al., 2012; Hansen et al., 2013) indicated that this crisis was caused by the ongoing activation of magmatic activity, even though the magma did not reach the

D

surface and a volcanic eruption did not occur. The structural cause of this ―missed‖ eruption in the

TE

Lunayyir basaltic field is discussed by Koulakov et al. (2015).

CE P

The lava composition from recent eruptions in western Arabia appears to indicate deep origins, probably arising from a mantle plume (Duncan and Al-Amri, 2013). Duncan and Al-Amri (2013) provide an overview of geochemical analyses of lavas from different harrats in western

AC

Arabia. They point out that the lava composition includes basanite to alkali olivine basalts of Hawaiian type, which mostly characterize plume-related hotspots, considerably different from the tholeiitic basalts observed in the Red Sea spreading center. Camp and Roobol (1992) found that the basalt properties in the western Arabian harrats were closer to the spreading-type in the older volcanic fields but became more alkalic in younger volcanoes. The distribution of volcanic activity appears to be asymmetric with respect to the Red Sea. Most of the basaltic fields are located along the eastern side of the rift. Along the African Red Sea coast, there are no recent volcanic manifestations, except for the Afar region, which is located at the southernmost edge of the Red Sea and is presumed to be associated with a mantle plume (e.g., 3

ACCEPTED MANUSCRIPT Ebinger and Sleep, 1998). The origin of Cenozoic volcanism in western Arabia and its possible links with the Red Sea rifting and Afar plume are subject to active debate. Some authors (e.g., Altherr et al., 1990) suggest that the opening of the Red Sea is caused by ascending hot mantle material (i.e., an

T

active rifting mechanism), which is spread over a large region surrounding the rift leading to

IP

anomalous heating of the overlying crust and volcanic activity. Another model invokes passive

SC R

extension of the Red Sea due to relative displacement of the lithospheric plates. Thinning of the lithosphere during the rifting causes an asymmetric mantle upwelling (e.g., Bohannon et al., 1989;

NU

Watremez et al., 2013) that mostly affects the eastern side of the rift and results in volcanic activity. An alternative scenario links the volcanic manifestations in western Arabia with the Afar plume (e.g.,

MA

Courtillot et al., 1984; Debayle et al., 2001). This concept is supported by geochemical similarity of lavas in the East African Rift and western Arabia (e.g., Bertrand et al., 2003) as well as by

D

geophysical observations of seismic anisotropy described in the next paragraph, which might indicate

TE

a path of northward plume migration (e.g. Chang et al., 2011, Lazar et al., 2012). Another mechanism

CE P

for explaining the volcanism in western Arabia invokes the presence of a mantle plume directly beneath the Arabian Platform (e.g., Chang and van der Lee, 2011). This hypothesis is supported by receiver function studies (Vinnik et al., 2003), detecting anomalously low S-velocities below the

AC

cratonic lithosphere of the Arabian Plate. Many geophysical studies of the deep structure beneath the Red Sea, Arabian Plate, and surrounding regions have been recently conducted to further examine the source of volcanic activity in western Arabia. For example, the analysis of SKS splitting (Hansen et al., 2006; Elsheikh et al., 2014) shows E-W oriented fast directions to the south of the Arabian Plate in Yemen. However, in the western part of Arabia, the anisotropy is generally oriented S-N which is interpreted as northward migration of the material from the Afar plume. A receiver function study by Hansen et al. (2007) highlights the variations of crustal thickness and depth to the lithosphere-asthenosphere boundary in

4

ACCEPTED MANUSCRIPT the Arabian Peninsula. Another receiver function study by Vinnik et al. (2003) provides evidence for anomalously low velocities below the thick lithosphere in central Arabia. Most of the information on mantle structure comes from several regional and global surface

T

and body wave tomography models. For example, a number of global and regional studies covering

IP

large regions of Africa and Arabia e.g. (Ritsema et al., 1999; Ritsema and van Heijst, 2000; Sebai et

SC R

al., 2006; Montagner et al., 2007: Ekström, 2011) provide evidence for mantle origins of volcanism in Afar and Arabia. Another model by Pasyanos and Nyblade (2007) revealed consistent low-velocity

NU

patterns below the volcanic fields in western Arabia but also detected a neutral velocity anomaly beneath the Red Sea. Similar patterns were found by higher resolution surface wave tomography

MA

models by Park et al. (2008) and Chang and Van der Lee (2011). Body wave tomography is also used for studying the Arabian region, but it faces the problem

D

of sparsely distributed stations and very uneven distribution of seismicity that causes lower resolution

TE

in the crust and uppermost mantle. Usually most details on mantle structure come from global

CE P

tomography models (e.g. Simmons et al., 2011). Some of the results were obtained from teleseismic data recorded by regional and local temporal networks (e.g., Park et al. 2007). On a local scale, teleseismic tomography was implemented for the Afar region by Stork et al. (2013). Over the last 5

AC

years, many studies have been conducted on the northernmost East-African rift and in Yemen, including tomographic studies (e.g., Hammond et al., 2013; Korostelev et al., 2014). These have demonstrated that low velocities underlie the recent volcanic zones in Yemen. Teleseismic tomography by Benoit et al., (2003) reveals a low-velocity anomaly underlying the high-velocity Arabian Plate. The distribution of seismic attenuation based on the analysis of waveform data recorded by a temporal network in Saudi Arabia (Al-Damegh et al., 2004) identified areas of low attenuation beneath the Arabian Plate and higher attenuation beneath the western Arabian coast. Along the axis of the Red Sea, a narrow zone of low attenuation was observed.

5

ACCEPTED MANUSCRIPT A travel time tomography study was performed by Hansen et al. (2012) for the entire African realm based on the data of the International Seismological Center (ISC) and available data of regional networks. Uppermost mantle structures were also studied using Pn and Sn travel times over all Asia

T

(e.g., Ritzwoller et al., 2002) and the Arabian region (Al-Lazki et al., 2004). The results of these body

IP

wave tomography studies are generally consistent with those provided by surface wave tomography.

SC R

The Arabian region was also covered by a series of tomography models obtained through the analysis of the ISC data across Asia (Koulakov, 2011), the East African Rift (Koulakov, 2007) and Iran

NU

(Alinaghi et al. 2007). In this study, we revise these models with a focus on the Arabian Plate and surrounding areas using similar methodology but with new datasets collected by the ISC over the last

MA

few years. It is important to note that the most recent observations include data from stations that were not available for previous studies. As such, these stations improve coverage and help to better

CE P

Data and Algorithm

TE

D

illuminate intraplate regions.

In this study, we use travel times of P and S body waves from the revised ISC catalogue,

AC

which spans the time period from 1980 to 2011 (ISC, 2011). The distributions of seismic stations and events used are shown in Figure 2. In this study, we used two groups of data. In the first group, we considered events in the study region recorded by worldwide stations at all possible epicentral distances. We selected events with no less than 30 recorded P and S arrival times surrounded by stations with an azimuthal gap of less than 180º. All these events were re-located using the algorithm developed by Koulakov and Sobolev (2006a). At this step, the data with residuals of more than 4 s were rejected, reducing the data catalogue to approximately 30%. Travel times were computed using the 1D velocity model AK135 (Kennett et al., 1995) and were corrected for crustal heterogeneities using the updated global model CRUST2.0 (Bassin et al., 2000), which has a resolution of 2 degrees in most parts of the world and 1 degree over Eurasia. In total, in the first data group, we used travel 6

ACCEPTED MANUSCRIPT time data from more than 68,000 events with approximately 900,000 P-phases and 100,000 S-phases recorded by the worldwide seismic stations at all available epicentral distances. When possible, we used depth phases (mostly pP and sS) to calculate depths of remote sources.

T

The second group of data uses travel times of P and S waves from remote events at

IP

teleseismic distances recorded by stations within the study area. For this part of the data, we selected

SC R

events with no less than 10 recorded P and S arrival times by the stations in the study region. This approach provided an additional 150,000 P-wave arrival times and 10,000 S-wave arrival times.

NU

The tomography inversion was based on the algorithm developed by Koulakov and Sobolev (2006b), which has been used previously with older datasets to examine different portions of the

MA

study area (Koulakov, 2007, 2011; Alinaghi et al., 2007). Inversions were performed for a series of circular overlapping areas with radii of 8º and depths down to 1000 km (Figure 2A). For each area,

D

the inversion parameters were set separately based on the results of synthetic tests. The

TE

parameterization nodes were installed at 15 horizontal levels located at depths from 25 to 900 km.

CE P

The grid spacing used was based on the density of rays. The minimum grid spacing was 40 km; for areas with sparse ray coverage, the grid spacing increased correspondingly. Examples of grids for two areas are presented in Figure 3. To avoid artifacts related to basic orientation of the grid, we

AC

performed independent inversions for two different grids with different basic orientations (0 and 45º). After performing inversions for two grid orientations in each of nine overlapping circular areas, we merged the results into the three-dimensional model using the algorithm described by Koulakov and Sobolev (2006b). The inversion was performed simultaneously for both P- and S- velocity models, source and station corrections. When using teleseismic data, the source corrections included one parameter per event (dt). For regional events, four parameters per event (dx, dy, dz, and dt) were determined. The inversion was performed using the LSQR algorithm (Paige and Saunders, 1982; Nolet, 1987) with smoothing regularization applied. Damping parameters and weights for station and source corrections 7

ACCEPTED MANUSCRIPT were determined separately for each circular window based on several trials of synthetic model reconstructions. In each case, we tuned the inversion parameters to achieve the best recovery of anomalies within the input synthetic model. These parameters were then used to invert the

T

experimental data. Variance reduction of the residuals following the inversions ranged from 40% in

IP

poorly sampled regions to 55% in densely sampled areas. The final average deviations of the P

SC R

residuals ranged from 0.4 to 0.6 s and those of the S-residuals were approximately equal to 0.6-0.9 s.

NU

Results and Verification

MA

The main result of this study is a 3D distribution of P velocity anomalies which are presented in four horizontal sections in Figure 4 and in four vertical sections in Figure 6. We also present the S-

D

velocity anomalies in the same sections in Figures 5 and 6. The major patterns of the S model are

TE

generally consistent with the P-velocity model. However, due to much less data and their poorer

interpretation.

CE P

quality, the resolution of the S-model was considerably poorer, and of less importance for our

The resulting anomalies are shown only in areas with sufficient ray coverage (with more than

AC

10% of average ray density for the current area) where the parameterization nodes were installed. The velocity anomalies are given in percent in respect to the 1D spherical velocity model AK135 (Kennett et al., 1995).

At shallower sections (100 and 200 km depth, Figure 4), the P-velocity clearly delineates the boundary between the low-velocity Arabian Platform to the east and high-velocity Precambrian shield to the west. The low-velocity anomalies beneath western Arabia correlate with the distributions of Cenozoic volcanic fields. In deeper sections, this low-velocity anomaly seems to migrate to the northeast. An unexpected feature is a linear high-velocity anomaly located beneath the Red Sea in the upper mantle. We therefore carefully checked its existence using various tests that will be presented below and proved its robustness. In the vertical sections of P-wave velocity anomalies 8

ACCEPTED MANUSCRIPT in Figure 6, we observe a high-velocity anomaly beneath the Arabian Plate down to ~200 km depth that deepens below the Zagros belt in the northeastern parts of the profiles. Prior starting the detailed interpretation of the derived results, it is important to present several tests to assess their reliability.

T

It pointed out above, for the location of sources, we used the heterogeneous crustal model,

IP

CRUST2.0 (Bassin et al., 2000). The lateral resolution of 2 degrees might be not sufficient for

SC R

adequate representation of crustal heterogeneities in some areas of strongly variable structures, such as those observed in the Red Sea. In addition, in some areas, robust data on crustal structures are not

NU

available, and for them, this model might be not adequate. In this context, it is important to estimate the effect of Moho depth variations upon mantle structures. In Figure 8, we present a section resulted

MA

from inversion of observed data with the use of a constant Moho at 30 km depth. The distributions of velocity anomalies appears to be almost identical to those in the main model in section 2 in Figure 6.

D

The effect of the crustal variations in this case appears to be weaker than in the study of Europe by

TE

Koulakov et al. (2009). This might be explained by relatively sparse distribution of stations and a

CE P

strong role of the trade-off between shallow velocity distribution, station corrections and source parameters that are determined simultaneously during the inversion. Another argument for a weak effect of the crustal anomalies is based on the synthetic modeling with realistic structures, which will

AC

be presented below.

Synthetic modeling allows the spatial resolution and the optimal values of the inversion parameters to be assessed. Synthetic travel times for the tests were computed for all source-receiver pairs used for constructing the main model including both types of datasets with regional and teleseismic rays. The synthetic anomalies affected the ray along the whole path, even if a part of the ray was located outside the current window. In all cases presented here, the synthetic residuals were perturbed with random noise with 0.3 s average deviation. With this value of noise, after inversion of synthetic data, we observed approximately the same variance reductions as in the cases of observed data inversion. A lower variance reduction in poorly sampled areas indicates that this parameter 9

ACCEPTED MANUSCRIPT represents not only noise in the data, but also the limited resolution capacity. After computing the synthetic travel times, the sources were randomly shifted by an average of 20 km in both the horizontal and vertical directions. Similar to the inversion performed with the observed data, the

IP

trade-off between source parameters and velocity anomalies.

T

inversions with the synthetic travel times start with the step of source locations. This simulates the

SC R

Results from synthetic, checkerboard resolution tests are shown in Figure 9 for the P and S velocity anomalies. We define the input model with alternating positive and negative anomalies with

NU

amplitudes of ±3% and a spacing of 3ºx3.5º (approximately 330x330 km) for the P-model and 6ºx5º for the S-model. The synthetic checkers change sign every 400 km in depth. The checkerboard

MA

patterns are defined throughout the whole earth to model the influence anomalies located outside the study region. The results show that our model is well resolved across Iran, Asia Minor, and the

D

southern part of the Red Sea. In other areas, such as in central Arabia, the resolution is poorer; the

TE

recovered anomalies are strongly smeared and are reduced in amplitude. That being said, the general

CE P

locations of the anomalies are correctly recovered; thus even in poorly resolved regions, the tomography results can be interpreted qualitatively. It is also worth noting that the depth resolution is reasonable; the change of the anomaly sign at 400 km depth is clearly seen in the presented depth

AC

sections. For the S-model, the resolution appears to be much poorer. At shallower depths, the structures beneath the central part of Arabia are not resolved due to lack of rays. At 640 km depth, the anomaly locations are generally recovered, but their amplitudes are strongly reduced. Based on this test, we conclude that the S-model suffers of poor resolution and should be interpreted with prudence. The checkerboard tests in the current study clearly demonstrate how adding the new data improves the resolution of the model. For example, in the checkerboard tests presented by Koulakov (2011), for most parts of the Arabian Peninsula, no structures can be restored, whereas in the current study some information for the same areas can be retrieved. 10

ACCEPTED MANUSCRIPT In Figure 10, we present another set of synthetic tests with two models having slightly different configurations. The synthetic patterns are defined as prisms with polygonal shapes defined in vertical section 2A-2B (same section as used for presenting results in Figure 6). Across the section,

T

the shape of the prism remains unchanged and its thickness is equal to 300 km. These synthetic

IP

models represent the same features as observed in the resulting model after inversion of the

SC R

experimental data. The corresponding recovered models provide the approach to check the robustness of some key results. In particular, our synthetic results show that the high-velocity cratonic

NU

lithosphere of the Arabian Plate and the low-velocity anomaly located beneath the thick lithosphere can be robustly resolved using the available data. We conclude that this provides evidence for the

MA

existence of a mantle upwelling beneath the Arabian Plate. Another feature that is examined by our synthetic test is an interrupted, high-velocity anomaly beneath the Zagros belt on the northeast side of

D

the study area. We show that both continuous and interrupted anomalies can be resolved by our

TE

model.

CE P

In vertical sections of the resulting P-velocity model (Figure 6), beneath the central and northern parts of the Red Sea, we observe an elongated vertical high-velocity anomaly that can be traced down to ~400 km depth. To assess whether this feature may result from downward smearing

AC

of strong crustal anomalies or represents real mantle structures, we consider two synthetic models with a stronger shallower anomaly (8%) in Figure 10A and with weaker deeper anomaly (4%) in Figure 10C. For the case of Model 1 in Figure 10B, we observe that the crustal anomaly beneath the Red Sea is almost not recovered. The poor resolution for the crustal structures beneath the Red Sea is possibly caused by the trade-off effect between the velocity and source parameters at shallow depths. In Model 2 (Figure 10D), the deeper high-velocity anomaly beneath the Red Sea is recovered at the right place, though its amplitude is reduced from 4% to 1.5%. From these two synthetic tests, we conclude that crustal anomalies do not strongly affect the mantle structures in our model. For the Red Sea, it appears that the observed high-velocity anomaly is not a result of downward smearing of 11

ACCEPTED MANUSCRIPT crustal anomalies, as might be suspected, but represents mantle structures, as inferred by the test with Model 2. All synthetic tests show that because of implementing the inversion damping, the recovered

T

amplitudes of anomalies are usually lower than those defined in the synthetic model. In different

IP

parts of the study area, the reductions of amplitudes might be different depending on the distribution

SC R

of data. This amplitude reduction it is not always under our control. Thus, we should point out that amplitudes of anomalies remain not reliably determined. At the same time, the shapes of the

MA

NU

anomalies appear to be robustly reconstructed. Therefore, we interpret all results only qualitatively.

Discussion

D

Comparison with previous results

TE

The P-wave velocity structures presented in this study can be compared with key findings

CE P

from previous seismic studies throughout the Arabian Peninsula and surrounding regions. In respect to the previous model constructed by Koulakov (2011), who used a similar algorithm as in the current study, we have achieved considerable improvement of the resolution for the Arabian region, as

AC

follows from comparison of the synthetic checkerboard tests. Better ray coverage in the current model is realized mostly by using new data collected by the stations in the central part of the Arabian Plate.

The velocity anomalies derived in this study are consistent with most previous tomographic studies for the same region. For example, higher velocities observed beneath the Arabian Plate and lower velocities beneath the western Arabian margin agree well with results from Ritsema and van Heijst (2000); Sebai et al. (2006); Pasyanos and Nyblade (2007) and other studies mentioned above. The resolution of the S-wave velocity in the model from Chang and Van der Lee (2011) is similar to that of our P-wave velocity model. Besides the general trend of regional scale velocity variations 12

ACCEPTED MANUSCRIPT identified by other studies, Chang and Van der Lee (2011) reveal a subtle velocity increase beneath the Red Sea, in the same location as in our P-velocity model. A similar pattern, reflected by lower seismic attenuation, was identified by Al-Damegh et al. (2004). Similar to our model, the presence of

T

a low-velocity anomaly right beneath the higher-velocity Arabian Plate was detected using

IP

teleseismic tomography by Benoit et al. (2003). However, these authors did not provide any

SC R

interpretation for this feature, although it may indicate the presence of hot mantle material beneath the Arabian Plate. The consistency of many different models created using different algorithms

MA

Thick lithosphere beneath the Arabian Plate

NU

provides confidence on the mantle structures discussed below.

Most of the observed anomalies in Figures 4 and 6 are consistent with major structural units

D

throughout the study region. A prominent feature is a high velocity anomaly associated with the

TE

Arabian plate, which likely represents an area of thick lithosphere. At 100 km depth (Figure 4), the

CE P

high velocity anomaly is almost perfectly bounded by belts of Mesozoic folded rocks that define the western border of the Arabian Platform. Based on considering the shape of this high-velocity anomaly in vertical sections (Figure 6), we estimate that beneath this area, the thickest Arabian

AC

cratonic lithosphere may reach 200-250 km. This lithospheric thickness value beneath the Platform is consistent with estimates by Hansen et al. (2007).

Hot mantle beneath thick Arabian Plate? At 400 and 600 km depth (Figure 4), we observe a maximum negative velocity anomaly located below the Arabian Platform, which is interpreted as the center of an anomalously hot mantle zone. As the width of this zone is comparable to its height, it cannot be identified as a classic mantle plume, such as inferred beneath Iceland or Hawaii. With the available data, we cannot confirm the existence of a thin (~100 km diameter) contrasted, column-shaped, plume-related anomaly. 13

ACCEPTED MANUSCRIPT Additionally, elongation of this anomaly into the lower mantle is not observed in the existing global tomography models (e.g., Ritsema et al., 2011) as would be expected for a whole mantle plume. For these reasons, we avoid the term ―plume‖ and instead interpret this anomaly as a ―mantle upwelling‖

T

or ―hot mantle zone‖. On the other hand, many recent plume-related studies (e.g., Burov and Gerya,

IP

2014; Koptev et al., 2015) have suggested that a thermo-chemically buoyant mantle plume would not

SC R

necessary have a persistent tail or that such a tail may be narrow (< 100 km in radius) and hence not resolvable by seismic imaging. As argued by Dannberg and Sobolev (2015), only low-buoyancy

NU

mantle plumes would have thick, seismically detectable tails with radii greater than 200 km. We propose that this slow-velocity and presumably hot mantle material reaches the bottom of

MA

the Arabian Plate lithosphere, as schematically shown in Figure 11. Since this material is unable to penetrate through the rigid cratonic lithosphere, it flows along the bottom of the platform, similar to

D

inferences by recent thermo-mechanical models of plume-lithosphere interactions (Burov et al., 2007;

TE

Burov and Cloetingh, 2010; Guillou-Frottier et al., 2011; Burov and Gerya, 2014; Koptev et al.,

CE P

2015). Beneath western Arabia, it reaches the border of the Arabian Shield and is then entrained toward the rift by return convection flows initiated by the opening of the Red Sea, as shown in Figure 11. As a result, the hot mantle material is directed toward a segment of relatively weak lithosphere

AC

close to the eastern shore of the Red Sea and ascend to the surface forming large basaltic extrusions (harrats) along the western border of the Arabian Peninsula. A similar mechanism was recently suggested by Koptev et al. (2015) for the Tanzanian craton. Note, however, that the strongest lowvelocity anomaly is observed not directly beneath volcanic areas, but more to the east, beneath the middle part of the Arabian Shield, where a regional upwelling is observed (see Figure 1 with topography). We hypothesize that this low-velocity anomaly beneath the Arabian Shield represents an accumulation of hot asthenosphere material, which cannot reach the surface because the overlying lithosphere is too strong. It migrates laterally along the bottom of the lithosphere and can reach the surface only along the coast of the Red Sea as here the lithosphere is thinned due to passive rifting. 14

ACCEPTED MANUSCRIPT Although the total volume of hot material beneath the western margin of Arabia is not as large as beneath the Arabian Shield, it is probably enough to pierce the weakened lithosphere and cause volcanic activity.

T

The hypothesis of feeding volcanoes in western Arabia from hot mantle upwelling beneath

IP

the Arabian Plate seems to contradict a mechanism previously proposed by Chang et al. (2011). They

SC R

explored the anisotropy orientations and claimed that there might be a northward mantle flow from the Afar plume that brings the hot material to initiate the volcanic activity. However, an open

NU

question remains which forces would drive these lateral displacement of mantle material to a distance of more than 1500 km. In addition, the anisotropy used for making this proposal is based on an

MA

analysis of the SKS splitting data. It is known that this method is not always capable identifying the depth of the anisotropic layer. It is therefore possible that the latitudinally oriented anisotropy might

D

be caused by lithospheric structures. Regarding the passive extension of the lithosphere along the Red

CE P

TE

Sea, such lineaments cannot be excluded.

Volcanism in cratonic margins

Based on the interpretation of our seismic model, we conclude that the major cause of

AC

Cenozoic volcanism in Western Arabia may be related to a mantle upwelling located beneath the Arabian Platform. Geochemical analyses of Cenozoic basalts in western Arabia generally indicate a deep origin (e.g., Duncan and Al-Amri, 2013) that may support our hypothesis. It is interesting that many of the world’s Cenozoic intracontinental basaltic fields are associated with large Precambrian platforms. For example, Koulakov and Bushenkova (2010) found that the locations of recent basalts in southern Siberia and Mongolia overly low-velocity anomalies that have deep roots beneath the Siberian Craton. Similar patterns are observed in the Arctic region of Yakutia, where recent volcanoes align with the northeastern border of the Siberian Craton and coincide with low velocities in the tomography model by Jakovlev et al. (2012). In the central East-African Rift system, abundant 15

ACCEPTED MANUSCRIPT volcanic activity in the Eastern Rift branch (and the absence of such in the Western Rift branch) also correlates with seismic tomography images and thermo-mechanical models that suggest asymmetric eastward upwelling of the mantle plume material deviated by the mantle lithosphere keel of the

T

Tanzanian Craton (e.g., Weeraratne et al., 2003; Koulakov, 2007; Adams et al., 2012; Koptev et al.,

IP

2015). From all these studies, we can conclude that many intracontinental volcanic areas near thick

SC R

Precambrian platforms might be associated with mantle ascending flows below these cratons. The most plausible explanation for this correlation is that thick lithosphere acts as a thermal isolator

NU

which blocks vertical heat flow and results in overheating of the mantle, thereby causing the development of plume instability below the continent (Guillou-Frottier and Jaupart, 1995). The

MA

physical plausibility of this mechanism has been tested by a number of analogue (e.g., GuillouFrottier and Jaupart, 1995) and numerical modelling studies (e.g., Burov et al., 2007; Guillou-Frottier

TE

D

et al., 2011; Koptev et al., 2015).

CE P

Subduction or delamination beneath Zagros and Makran? Along profiles 1 and 3 in Figure 6, the high-velocity anomaly beneath the Arabian Plate appears to be connected to slab-shaped anomalies dipping to the northeast beneath the Zagros belt

AC

and Makran. However, along profile 2, this connection seems to disappear. As illustrated in the synthetic tests in Figure 5, this interruption of the lithospheric continuity appears to be robust. The subduction of the Arabian Plate and/or delamination of the mantle lithosphere beneath the Zagros belt and Makran is confirmed by multiple data and therefore is widely accepted by most researchers (e.g., Agard et al., 2009; Regard et al., 2010; Francois et al., 2014a). Such a structure has also been supported by several tomographic models (e.g. Hafkenscheid et al., 2006; Alinaghi et al., 2007). However, precise earthquake locations in the Zagros area (Engdahl et al., 2006) suggest that the lack of mantle seismicity may indicate that present-day subduction is not occurring. Based on a previous tomographic model, Koulakov (2011) proposed delamination of the mantle continental lithosphere 16

ACCEPTED MANUSCRIPT beneath the Zagros belt and subduction of oceanic lithosphere beneath Makran. Koulakov (2011) also showed a gap between the sinking lithospheric segments and suggested that the delamination is independent of the subduction. Based on the current seismic model, we propose an alternative

T

hypothesis, which might explain this gap by the thermal effect of the hot mantle flow coming from

IP

below the Arabian Plate.

SC R

Our results suggest that the center of the upwelling is located closer to the western margin of the Arabian Platform; thus, most of the hot material migrates westward. However, part of the hot

NU

material may also follow the bottom of the craton in a northeastward direction (toward the right side of Figure 11), similar to that shown by Koptev et al., (2015) for the central East African Rift system.

MA

Beneath Arabia, the hot mantle flow may reach a zone of subduction or delamination below the Zagros belt. Since the subduction rate is presumed to be quite slow in this area (<2-3 cm/y, e.g.

D

Francois et al., 2014a), the corresponding Péclet numbers, which defines the ratio of heat advection

TE

to the diffusion rate, are also small (Burov et al., 2014). Hence the hot material may warm the upper

CE P

part of the sinking lithosphere, thereby weakening it, resulting in its boudinage and detachment in some segments of the Zagros-Makran belt. We propose that the gap seen along profile 2 (Figure 11), may correspond to the detachment of the lithosphere resulting from this mechanism. At the same

AC

time, since only a minor portion of the mantle material would be directed to the northeast, the amount of hot material in this area is probably not sufficient to produce similarly strong volcanic manifestations, such as those observed in western Arabia. Note that in Section 2 (Figures 6 and 11), beneath the Arabian Gulf, we observe another weak vertically oriented positive anomaly, which seems to separate the continuity of the low-velocity anomaly beneath the Arabian Plate and the Zagros belt. If this anomaly is robust, it might be speculated that it represents drips of delaminated lithosphere detached from the northeastern margin of the Arabian Plate. Unfortunately, the amplitude of this anomaly is too small, and the resolution in this part is too low to yield a definitive interpretation for this feature. 17

ACCEPTED MANUSCRIPT Deviations of mantle interfaces beneath the Red Sea? A striking feature observed beneath the Red Sea appeared unexpected for us. In our model,

T

the Red Sea is associated with high-velocity anomaly which is observable down to ~400 km depth.

IP

The synthetic modeling presented in Figure 10 shows that this anomaly does really represents mantle

SC R

structure and is not a result of downward smearing of crustal anomalies. It is known that P-velocity, in contrast to S-velocity, is more sensitive to composition than temperature and liquid content [e.g.,

NU

Mechie et al., 1994]. Extension and thinning of the lithosphere beneath the Red Sea lead to passive ascent of mantle material, which temperature does not significantly differ from the temperature of the

MA

asthenosphere (mantle temperature gradients are on the order of 0.3-05°C/km (Schubert et al., 2001). As a result, rocks with deeper composition become shallower beneath the rift and thus appear to have

D

higher P-velocity compared to the surrounding area (in the absence of fluids, retrograde phase

TE

changes do not occur in these rocks at least up to the LAB depth). To illustrate the ascending flow

CE P

beneath the Red Sea, in Figure 11, we show contour lines of absolute velocities which may represent upward deflection of the mantle layers below the Red Sea. Note that the damping during the tomographic inversion causes considerable reduction of the anomaly magnitude, Thus, in reality, the

AC

anomaly beneath the Red Sea may appear to be stronger, and deviations of the contour lines might be stronger than indicated in Figure 11.

Conclusions We have generated new models of P- and S-wave velocity anomalies in the upper mantle beneath the Arabian Peninsula and surrounding regions. Synthetic tests and comparisons with independent models provide confidence in our results. Using new data that have become available from the ISC in recent years, we obtain a fair resolution in intracontinental areas which were not previously illuminated by body wave seismic data. 18

ACCEPTED MANUSCRIPT A distinct positive P-velocity anomaly in the depth interval of the first 200-250 km beneath the Red Sea is a strong indicator of a passive character of rifting. The P-velocity, which is more sensitive to composition than to temperature, shows the ascent of rocks with deeper composition due

T

to extension and thinning of the lithosphere in the rift zone. In case of active rifting, the high P-

IP

velocity anomaly would not be observed. It might even be negative since the compositional effect in

SC R

that case would be negated by temperature while both P and S anomalies - rooted at more important depths than 250 km.

NU

Our results suggest that the cratonic Arabian Platform is associated with a high-velocity anomaly that extends down to 200-250 km depth. Below the craton, we observe low-velocity

MA

anomalies that might represent the upper part of a mantle plume or a hot mantle upwelling. We propose that the hot mantle material has been laterally spreading below the thick and strong Arabian

D

lithosphere and reaches a weak zone in the vicinity of the Red Sea Rift where it has propagated

TE

upward, leading to partial melting and volcanic activity at the surface.

CE P

On the opposite side of the Arabian Plate, we suggest that the plume is connected to subducting or delaminating lithosphere below the Zagros belt and Makran. However, in the central part of the collision zone, there is a ―gap‖ in the seismic velocity anomaly, suggesting detachment of

AC

the lithosphere, in accordance with inferences from geological data and thermo-mechanical models of the Zagros collision which pointed out to the necessity of mantle upwelling associated with the detachment of the subducting lithosphere [Francois et al., 2014a,b]. We hypothesize that this detachment might be caused by hot mantle material that has reached the northeastern part of the Arabian Plate and has weakened the top of the subducting lithosphere.

Acknowledgments Travel time data for this study were acquired from the online catalogue of the International Seismological Center (http://www.isc.ac.uk). IK and NB are supported by the Russian Scientific 19

ACCEPTED MANUSCRIPT Foundation (grant #14-17-00430). The authors extend their appreciation to the Deanship of Scientific Research at King Saud University for funding the study through the international research group project No IRG14-21. This study has been also supported the ERC Advanced Grant 290864

T

RHEOLITH (to EB).

IP

Figure captions:

SC R

Figure 1. Topography/bathymetry and major geographic units across the study area (downloaded from www.marine-geo.org). Red areas indicate the locations of Cenozoic basaltic fields

NU

(harrats). Green dotted line highlights Mesozoic rocks lineating the boundary between the Arabian Shield and Platform.

MA

Figure 2. Earthquake epicenters (red dots) and stations (triangles) used in this study. (A) Zoom to the area of interest. Dotted lines mark circular areas used for inversions. (B) Distributions of global

D

stations and events used in this study in polar coordinates.

TE

Figure 3. Ray density and distributions of parameterization nodes (black dots) at 500 km depth for

CE P

two selected circular areas indicated by bold lines. Other areas are depicted with thinner lines. Figure 4. P-wave velocity anomalies at four different depths in the new tomography model given in percent in respect to the model AK 135. Red and green lines outline harrats and Mesozoic fold

AC

belts. The locations of four cross-sectional profiles are indicated on the 100 km depth panel. Pink dotted ovals marked on the 400 and 600 depth panels with AP indicate the interpreted location of the hot mantle upwelling below the Arabian Plate. Locations of the profiles with distance indications are shown in slice at 100 km depth. Figure 5. Same as Figure 4, but for the S-wave velocity anomalies. Figure 6. P-wave velocity anomalies given in percent in respect to the model AK 135 along four cross-sectional profiles shown in Figure 4. Above each section, the exaggerated topographic relief is shown (from www.marine-geo.org). Figure 7. Same as Figure 6, but for the S-wave velocity anomalies. 20

ACCEPTED MANUSCRIPT Figure 8. P-wave velocity anomalies along profile 2 resulted from inversion with the use of constant Moho depth model. Above the section, the exaggerated topographic relief is shown. Figure 9. Checkerboard resolution tests at 150 km (left) and 640 km (right) depth for the P- and S-

IP

that the signs of the anomalies change at 400 km depth.

T

wave velocity anomalies. The initial synthetic pattern is indicated with thin black lines. Note

SC R

Figure 10. Synthetic models that examine two different realistic anomaly configurations along profile 2. Panels A and C show the input anomalies for Models 1 and 2, respectively, while panels B

NU

and D show the recovered models.

Figure 11. Interpretation of the tomography results. The background shows the P-wave velocity

MA

anomalies along profile 2 (Figure 6). Dotted lines indicate absolute velocity contours in the area around the Red Sea. The bold blue line mark the bottom of the lithosphere and the location of

D

the detached block. The violet arrows indicate flow direction of hot material from the inferred

TE

Arabian Plume. Red triangles denote the locations of Cenozoic volcanoes. Black arrows show

AC

References:

CE P

rifting directions associated with the Red Sea.

Adams, A., Nyblade, A., Weeraratne, D., 2012. Upper mantle shear wave velocity structure beneath the East African plateau: Evidence for a deep, plateau-wide low velocity anomaly. Geophysical Journal International, 189, 123–142. Agard, P., Yamato, P., Jolivet, L., Burov, E., 2009. Exhumation of oceanic blueschists and eclogites in subduction zones: timing and mechanisms, Earth-Science Reviews, 92(1-2), 53-79, doi: 10.1016/j.earscirev.2008.11.002.

21

ACCEPTED MANUSCRIPT Al-Amri, A.M.S., 1995. Recent seismic activity in the northern Red Sea, Journal of Geodynamics, 20, 243-253. Al-Amri, A.M., Fnais, M.S., Abdel-Rahman, K., Mogren, S., Al-Dabbagh, M., 2012.

IP

International Journal of Physical Sciences, 7, 2791-2805.

T

Geochronological dating and stratigraphic sequences of Harrat Lunayyir, NW Saudi Arabia,

SC R

Al-Damegh, K., Sandvol, E., Al-Lazki, A., Barazangi, M., 2004. Regional seismic wave propagation (Lg and Sn) and Pn attenuation in the Arabian Plate and surrounding regions, Geophysical

NU

Journal International, 157, 775-795.

Alinaghi A., Koulakov, I., Thybo, H., 2007. Seismic tomographic imaging of P- and S-waves

MA

velocity perturbations in the upper mantle beneath Iran, Geophysical Journal International, 169 , 1089–1102, doi:10.1111/j.1365-246X.2007.03317.x.

D

Al-Lazki, A. I., Sandvol, E., Seber, D., Barazangi, M., Turkelli, N., Mohamad, R., 2004. Pn

TE

tomographic imaging of mantle lid velocity and anisotropy at the junction of the Arabian,

CE P

Eurasian and African plates, Geophysical Journal International, 158, 1024-1040. Altherr, R., Henjes-Kunst, F., Puchelt, H., Baumann, A., 1990. Volcanic activity in the Red Sea axial trough – Evidence for a large mantle diapir, Tectonophysics, 150, 121-133.

AC

Baer, G., Hamiel Y., 2010. Form and growth of an embryonic continental rift: InSAR observations and modeling of the 2009 western Arabia rifting episode, Geophysical Journal International, 182, 155-167.

Bassin, C., Laske, G., Masters, G., 2000. The current limits of resolution for surface wave tomography in North America, EOS, Transactions - American Geophysical Union, 81, F897. Benoit, M.H., Nyblade, A.A., VanDecar, J.C., Gurrola, H., 2003. Upper mantle P wave velocity structure and transition zone thickness beneath the Arabian Shield, Geophysical Research Letters, 30, doi: 10.1029/2002GL016436.

22

ACCEPTED MANUSCRIPT Bertrand, H., Chazot, G., Blichert-Toft, J., Thoral, S., 2003. Implications of widespread high-μ volcanism on the Arabian Plate for Afar mantle plume and lithosphere composition, Chemical Geology, 198, 47-61.

T

Bohannon, R.G., Naeser, C.W., Schmidt, D.L., Zimmermann, R.A., 1989. The timing of uplift,

IP

volcanism and rifting peripheral to the Red Sea: a case for passive rifting, Journal of

SC R

Geophysical Research, 94, 1683–1701.

Burov E., Francois, T., Agard, P., Le Pourhiet, L., Meyer, B., Tirel, C., Lebedev, S., Yamato, P.,

NU

Brun, J.-P., 2014. Mechanisms of subduction and HP/UHP exhumation of continental crust: rheological and geodynamic controls, Tectonophysics, doi: 10.1016/j.tecto.2014.04.033.

MA

Burov, E., Guillou-Frottier, L., D’Acremont, E., Le Pourhiet, L., Cloetingh, S., 2007. Plume head – lithosphere interactions near intra-continental plate boundaries, Tectonophysics, 434, 15-38,

D

doi: 10.1016/j.tecto.2007.01.002.

TE

Burov, E., Cloetingh, S., 2010. Plume-like upper mantle instabilities drive subduction initiation,

CE P

Geophysical Research Letters, 37, doi: 10.1029/2009GL041535. Burov E., Gerya, T., 2014. Asymmetric three-dimensional topography over mantle plumes, Nature, doi: 10.1038/nature13703.

AC

Camp, V. E., Hooper, P. R., Roobol, M. J., White, D. L., 1987. The Madinah eruption, Saudi Arabia: Magma mixing and simultaneous extrusion of three basaltic chemical types, Bulletin of Volcanology, 49, 489-508. Camp, V.E., Roobol, M.J., 1992. Upwelling asthenosphere beneath western Arabia and its regional implications, Journal of Geophysical Research, 97, 15,255-15,271. Chang, S. J., Van der Lee, S., 2011. Mantle plumes and associated flow beneath Arabia and East Africa, Earth and Planetary Science Letters, 302, 448-454.

23

ACCEPTED MANUSCRIPT Chang, S.‐J., Merino, M., Van der Lee, S., Stein, S., Stein, C. A., 2011. Mantle flow beneath Arabia offset from the opening Red Sea, Geophysical Research Letters, 38, L04301, doi: 10.1029/2010GL045852.

IP

Geophysics, Oxford University Press, New York, 186.

T

Coleman, R. G., 1993. Geological evolution of the Red Sea, Oxford Monographs on Geology and

SC R

Courtillot, V., Achache, J., Landre, F., Bonhommet, N., Galibert, P.Y., Montigny, R., Feraud, G., 1984. Episodic spreading and rift propagation: new paleomagnetic and geochronologic data

NU

from the Afar nascent passive margin, Journal of Geophysical Research, 88, 3315–3333. Debayle, E., Leveque, J.J., Cara, M., 2001. Seismic evidence for a deeply rooted low-velocity

MA

anomaly in the uppermantle beneath the northeastern Afro/Arabian continent, Earth and Planetary Science Letters, 193, 423–436.

D

Dannberg, J., Sobolev, S., 2015. Low-buoyancy thermochemical plumes resolve controversy of

TE

classical mantle plume concept, Nature Communications 6, doi:10.1038/ncomms7960/

CE P

Duncan, R. A., Al-Amri, A. M., 2013. Timing and composition of volcanic activity at Harrat Lunayyir, western Saudi Arabia, Journal of Volcanology and Geothermal Research, 260, 103116.

AC

Ebinger, C. J., Sleep, N. H., 1998. Cenozoic magmatism throughout east Africa resulting from impact of a single plume. Nature, 395, 788-791. Ekström, G., 2011. A global model of Love and Rayleigh surface wave dispersion and anisotropy, 25-250 s. Geophysical Journal International, 187, 1668-1686. doi:10.1111/j.1365246X.2011.05225.x Elsheikh, A. A., Gao, S. S., Liu, K. H., Mohamed, A. A., Yu, Y., Fat-Helbary, R. E., 2014. Seismic anisotropy and subduction‐ induced mantle fabrics beneath the Arabian and Nubian Plates adjacent to the Red Sea, Geophysical Research Letters, 41, 2376-2381.

24

ACCEPTED MANUSCRIPT Engdahl, E. R., Jackson, J. A., Myers, S. C., Bergman, E. A., Priestley, K., 2006. Relocation and assessment of seismicity in the Iran region. Geophysical Journal International, 167, 761-778. Hafkenscheid, E., Wortel, M. J. R., Spakman, W., 2006. Subduction history of the Tethyan region

T

derived from seismic tomography and tectonic reconstructions, Journal of Geophysical

IP

Research: Solid Earth (1978–2012), 111(B8).

SC R

Hammond, J., Kendall, J.-M., Stuart, G., Ebinger, C., Bastow, I., Keir, D., Ayele A., Belachew, M., Goitom, B., Ogubazghi, G., Wright, T., 2013. Mantle upwelling and initiation of rift

NU

segmentation beneath the Afar Depression, Geology, 41, 635–638. Hansen, S., Schwartz, S., Al-Amri, A., Rodgers, A., 2006. Combined plate motion and density-driven

MA

flow in the asthenosphere beneath Saudi Arabia: Evidence from shear-wave splitting and seismic anisotropy, Geology, 34, 869-872.

D

Hansen, S. E., Rodgers, A. J., Schwartz, S. Y., Al-Amri, A., 2007. Imaging ruptured lithosphere

TE

beneath the Red Sea and Arabian Peninsula, Earth and Planetary Science Letters, 259, 256-265.

CE P

Hansen, S. E., Nyblade, A. A., Benoit, M. H., 2012. Mantle structure beneath Africa and Arabia from adaptively parameterized P-wave tomography: Implications for the origin of Cenozoic AfroArabian tectonism, Earth and Planetary Science Letters, 319, 23-34.

AC

Hansen, S. E., DeShon, H. R., Moore-Driskell, M. M., Al-Amri, A., 2013. Investigating the P wave velocity structure beneath Harrat Lunayyir, northwestern Saudi Arabia, using double-difference tomography and earthquakes from the 2009 seismic swarm. Journal of Geophysical Research: Solid Earth, 118, 4814-4826. International Seismological Centre, On-line Bulletin, 2011. http://www.isc.ac.uk, International Seismological Centre, Thatcham, United Kingdom. Guillou-Frottier, L., Jaupart, C., 1995. On the effect of continents on mantle convection, Journal of Geophysical Research, 100 (B12), 24217-24238.

25

ACCEPTED MANUSCRIPT Guillou-Frottier, L., Burov, E., Cloetingh, S., Le Goff, I., Deschamps, Y., Huet, B., Bouchot, V., 2011. Plume-induced dynamic instabilities near cratonic blocks: Implications for P-T-t paths and metallogeny, Global and Planetary Change, 90, 37-50, doi:

T

10.1016/j.gloplacha.2011.10.007.

IP

François, T., Burov, E., Agard, P., Meyer, B., 2014a. Build-up of a dynamically supported orogenic

SC R

plateau: numerical modelling of the Zagros / Central Iran case study, Geochemistry, Geophysics, Geosystems, 15, 2632-2654. doi: 10.1002/2013GC005223.

NU

François, T., Agard, P., Bernet, M., Meyer, B., Chung, S.-L., Zarrinkoub, M.H., Burov, E., Monié, P., 2014b. Cenozoic exhumation of the internal Zagros: First constraints from low-temperature

MA

thermochronology and implications for the build-up of the Iranian plateau, Lithos, 206/207, 100 – 112.

D

Jakovlev, A., Bushenkova, N.A., Koulakov, I.Yu., Dobretsov, N.L., 2012. Structure of the upper

TE

mantle in the Circum-Arctic region from regional seismic tomography, Russian Geology and Geophysics, 53, 963-971, doi: 10.1016/j.rgg.2012.08.001.

CE P

Kennett, B.L.N., Engdahl, E.R., Buland, B., 1995. Constraints on seismic velocities in the Earth from travel times, Geophysical Journal International, 122, 108–124.

AC

Koptev, A., Calais, E., Burov, E., Leroy, S., 2015. Dual continental rift systems generated by plumelithosphere interaction, Nature Geoscience, 8, 388-392. DOI: 10.1016/j.gr.2015.02.002. Korostelev, F., Basuyau, C., Leroy, S., Tiberi, C., Ahmed, A., Stuart, G. W., Boschi, L., 2014. Crustal and upper mantle structure beneath south‐western margin of the Arabian Peninsula from teleseismic tomography, Geochemistry, Geophysics, Geosystems, 15, 2850-2864. Koulakov, I., 2011. High-frequency P and S velocity anomalies in the upper mantle beneath Asia from inversion of worldwide traveltime data, Journal of Geophysical Research, 116, B04301, doi: 10.1029/2010JB007938.

26

ACCEPTED MANUSCRIPT Koulakov, I., 2007. Structure of the Afar and Tanzania plumes based on the regional tomography using ISC data, Doklady Earth Sciences, 417, 1287–1292. Koulakov, I., Sobolev, S. V., 2006a. Moho depth and three-dimensional P and S structure of the crust

T

and uppermost mantle in the Eastern Mediterranean and Middle East derived from tomograpic

IP

inversion of local ISC data, Geophysical Journal International, 164, 218-235.

SC R

Koulakov I., Sobolev, S.V., 2006b. A Tomographic Image of Indian Lithosphere Break-off beneath the Pamir Hindukush Region, Geophysical Journal International, 164, 425-440.

NU

Koulakov I., Bushenkova, N., 2010. Upper mantle structure beneath the Siberian craton and surrounding areas based on regional tomographic inversion of P and PP travel times,

MA

Tectonophysics, 486, 81-100.

Koulakov I., El Khrepy, S., Al-Arifi, N., Kuznetsov, P., Kasatkina, E., 2015. Structural cause of a

D

missed eruption in the Lunayyir basaltic field (Saudi Arabia) in 2009, Geology, v.43, no 6,

TE

G36271, doi: 10.1130/G36271.1.

CE P

Lazar, M., Ben-Avraham, Z., Garfunkel, Z., 2012. The Red Sea–New insights from recent geophysical studies and the connection to the Dead Sea fault, Journal of African Earth Sciences, 68, 96-110.

AC

Mechie, J., Fuchs, K., Altherr, R., 1994. The relationship between seismic velocity, mineral composition and temperature and pressure in the upper mantle—with an application to the Kenya Rift and its eastern flank, Tectonophysics, 236, 453-464. Montagner, J., Marty, B., Stutzmann, E., Sicilia, D., Cara, M., Pik, R., Lévêque, J., Roult, G., Beucler, E., Debayle, E., 2007. Mantle upwellings and convective instabilities revealed by seismic tomography and helium isotope geochemistry beneath eastern Africa, Geophysical Research Letters, 34, 21303, doi: 10.1029/2007GL031098. Nolet, G., 1987. Seismic wave propagation and seismic tomography, in Anonymous Seismic tomography, Springer, pp. 1-23. 27

ACCEPTED MANUSCRIPT Paige, C. C., Saunders, M. A., 1982. LSQR: An algorithm for sparse linear equations and sparse least squares, ACM Transactions on Mathematical Software (TOMS), 43-71. Pallister, J.S., McCausl, W.A., Jónsson, S., Lu, Z.H.M., El Hadidy, S., Aburukbah, A., Stewart,

T

I.C.F., Lundgren, P.R., White, R.A., Moufti, M.R.H., 2010. Broad accommodation of rift-

IP

related extension recorded by dyke intrusion in Saudi Arabia, Nature Geosciences 3: 705–712.

SC R

Park, Y., Nyblade, A. A., Rodgers, A. J., Al Amri, A., 2008. S wave velocity structure of the Arabian Shield upper mantle from Rayleigh wave tomography, Geochemistry, Geophysics, Geosystems,

NU

9, Q07020, doi: 10.1029/2007GC001895.

Pasyanos, M. E., Nyblade, A. A., 2007. A top to bottom lithospheric study of Africa and Arabia,

MA

Tectonophysics, 444, 27–44, doi: 10.1016/j. tecto.2007.07.008. Regard, V., Hatzfeld, D., Molinaro, M., Aubourg, C., Bayer, R., Bellier, O., Yamini-Fard, F., Peyret,

D

M., Abbassi, M., 2010. The transition between Makran subduction and the Zagros collision:

TE

recent advances in its structure and active deformation, Special Publication - Geological

CE P

Society of London, 330 41-64, DOI :10.1144/SP330.4. Ritsema, J., van Heijst, H. J., Woodhouse, J. H., 1999. Complex shear wave velocity structure imaged beneath Africa and Iceland. Science, 286, 1925-1928.

AC

Ritsema, J., van Heijst, H., 2000. New seismic model of the upper mantle beneath Africa, Geology, 28, 63-66.

Ritsema, J., Deuss, A., Van Heijst, H. J., Woodhouse, J. H., 2011. S40RTS: a degree-40 shearvelocity model for the mantle from new Rayleigh wave dispersion, teleseismic traveltime and normal-mode splitting function measurements. Geophysical Journal International, 184, 12231236. Ritzwoller, M. H., Barmin, M. P., Villasenor, A., Levshin, A. L., Engdahl, E. R., 2002. Pn and Sn tomography across Eurasia to improve regional seismic event locations, Tectonophysics, 358, 39-55. 28

ACCEPTED MANUSCRIPT Schubert, G., Turcotte, D.L., Olsen, P., 2001. Mantle Convection in the Earth and Planets, Cambridge, 2001. Sebai, A., Stutzmann, E., Montagner, J., Sicilia, D., Beucler, E., 2006. Anisotropic structure of the

T

African upper mantle from Rayleigh and Love wave tomography, Physics of the Earth and

IP

Planetary Interiors, 155, 48–62.

SC R

Simmons, N. A., Myers, S. C., Johannesson, G., 2011. Global‐scale P wave tomography optimized for prediction of teleseismic and regional travel times for Middle East events: 2. Tomographic

NU

inversion. Journal of Geophysical Research: Solid Earth (1978–2012), 116(B4). Stork, A. L., Stuart, G. W., Henderson, C. M., Keir, D., Hammond, J. O. S., 2013. Uppermost mantle

MA

(Pn) velocity model for the Afar region, Ethiopia: an insight into rifting processes, Geophysical Journal International, 193, 321-328.

D

Vinnik, L., Ravi Kumar, M., Kind, R., Farra, V., 2003. Super-deep low-velocity layer beneath the

TE

Arabian plate, Geophysical Research Letters, 30, doi: 10.1029/2002GL016590.

CE P

Watremez, L., Burov, E., d’Acremont, E., Leroy, S., Huet, B., Le Pourhiet, L., Bellahsen, N., 2013. Buoyancy and localizing properties of continental mantle lithosphere: Insights from thermomechanical models of the eastern Gulf of Aden, Geochemistry, Geophysics, Geosystems,

AC

14, 2800-2817, doi: 10.1002/ggge.20179, ISSN: 15252027. Weeraratne, D.S., Forsyth, D.W., Fischer, K.M., Nyblade, A.A., 2003. Evidence for an upper mantle plume beneath the Tanzanian craton from Rayleigh wave tomography, Journal of Geophysical Research: Solid Earth (1978–2012), 108(B9).

29

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 1. Topography/bathymetry and major geographic units across the study area (downloaded from www.marine-geo.org). Red areas indicate the locations of Cenozoic basaltic fields (harrats). Green dotted line highlights Mesozoic rocks lineating the boundary between the Arabian Shield and Platform.

30

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

B.

AC

CE P

TE

D

A.

Figure 2. Earthquake epicenters (red dots) and stations (triangles) used in this study. (A) Zoom to the area of interest. Dotted lines mark circular areas used for inversions. (B) Distributions of global stations and events used in this study in polar coordinates. 31

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

Figure 3. Ray density and distributions of parameterization nodes (black dots) at 500 km depth for two selected circular areas indicated by bold lines. Other areas are depicted with thinner lines.

32

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 4. P-wave velocity anomalies at four different depths in the new tomography model given in percent in respect to the model AK 135. Red and green lines outline harrats and Mesozoic fold belts. The locations of four cross-sectional profiles are indicated on the 100 km depth panel. Pink dotted ovals marked on the 400 and 600 depth panels with AP indicate the interpreted location of the hot mantle upwelling below the Arabian Plate. Locations of the profiles with distance indications are shown in slice at 100 km depth.

33

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 5. Same as Figure 4, but for the S-wave velocity anomalies.

34

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 6. P-wave velocity anomalies given in percent in respect to the model AK 135 along four cross-sectional profiles shown in Figure 4. Above each section, the exaggerated topographic relief is shown (from www.marine-geo.org).

35

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 7. Same as Figure 6, but for the S-wave velocity anomalies.

36

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

NU

Figure 8. P-wave velocity anomalies along profile 2 resulted from inversion with the use of constant Moho depth model. Above the section, the exaggerated topographic relief is shown.

37

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 9. Checkerboard resolution tests at 150 km (left) and 640 km (right) depth for the P- and S-wave velocity anomalies. The initial synthetic pattern is indicated with thin black lines. Note that the signs of the anomalies change at 400 km depth.

38

AC

CE P

TE

D

MA

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

Figure 10. Synthetic models that examine two different realistic anomaly configurations along profile 2. Panels A and C show the input anomalies for Models 1 and 2, respectively, while panels B and D show the recovered models.

39

NU

SC R

IP

T

ACCEPTED MANUSCRIPT

AC

CE P

TE

D

MA

Figure 11. Interpretation of the tomography results. The background shows the P-wave velocity anomalies along profile 2 (Figure 6). Dotted lines indicate absolute velocity contours in the area around the Red Sea. The bold blue line mark the bottom of the lithosphere and the location of the detached block. The violet arrows indicate flow direction of hot material from the inferred Arabian Plume. Red triangles denote the locations of Cenozoic volcanoes. Black arrows show rifting directions associated with the Red Sea.

40

ACCEPTED MANUSCRIPT Highlights: 6. A new upper mantle tomographic model is constructed for the Arabian Plate region 7. The derived anomalies fit the major structural units of the region

T

8. A low-velocity anomaly beneath the Arabian Plate traces a mantle upwelling

IP

9. This hot area was responsible for the origin of recent volcanism in Western Arabia

AC

CE P

TE

D

MA

NU

SC R

10. A high velocity beneath the Red Sea may indicate a passive rifting mechanism

41