The Earth's mantle, a dynamic and heterogeneous layer, is critical to the geodynamic evolution of our planet. Understanding its deformation patterns is essential for unraveling the tectonic history and present-day geodynamics of any region. The tectonic history of the Indian landmass spans the formation of ancient cratons, incorporation into supercontinents such as Rodinia and Gondwana, and eventual collision with Eurasia , resulting in some of the most prominent geological features. Geochronological, geological, and paleomagnetic data suggests that multiple continental blocks came together during the Proterozoic era, leading to the formation of two major supercontinents . The first, Columbia, formed around 1.8 billion years ago , while the second, Rodinia, emerged around 1.0 billion years ago . One significant feature of these ancient tectonic events within the Indian shield is the presence of a Mesoproterozoic orogenic belt that runs in the ENE-WSW direction . This collisional zone, known as the Central Indian Tectonic Zone (CITZ) or Satpura Orogen, is an important suture zone where continental fragments collided and amalgamated during the Mesoproterozoic Era.

The CITZ is a suture zone that demarcates the boundary between the northern and southern Indian cratons (Fig. ). It originated during the Mesoproterozoic orogeny in central India, when the northern Bundelkhand Craton collided with a disordered group of South Indian cratons (Bastar, Dharwar, and Singhbhum cratons) . However, the accretion timing and the kinematics of convergence between the northern and southern Indian cratons are still debated . Some hypotheses propose that the southern Indian cratons underthrust northward beneath the northern Indian craton , while others propose the opposite, with southward-directed underthrusting of the northern Indian craton beneath the southern Indian cratons . Furthermore, recent research has shown the likelihood of double-sided subduction in the central region , which would involve processes such as slab breakoff and subduction polarity reversal. The CITZ has thus been a focal point of several tectonic activities, including continental collision, subduction, and mantle deformation. In addition to these Proterozoic tectonic processes, central India was subsequently modified by the Reunion mantle plume activity during the late Cretaceous (∼66 Ma) . Mantle plumes are commonly interpreted as thermochemical upwellings that originated near the core-mantle boundary that can generate extensive decompression melting, causing significant modification of the lithosphere . The Reunion plume triggered not only the Deccan traps but also major lithospheric thinning and anisotropic fabric alterations . Such plume-lithosphere interaction is likely to result in modifications that potentially overprints pre-existing Proterozoic tectonic fabrics within the CITZ region.

The Central Indian Suture (CIS) and the Southeast Sakoli Shear (SES) form the southern limit of the CITZ, while the Son Narmada North Fault (SNNF) defines the northern boundary (Fig. ). The numerous main faults that exist within these borders are the Son Narmada South Fault (SNSF), Barwani Sukta Fault (BSF), Gavligarh Fault (GF), Tapti Fault (TF), and Tan Shear (TS) (Fig. , ). These fault systems highlight the dynamic character of tectonic interactions and accompanying deformation within the CITZ, contributing to the evolving structural framework of this geologically active zone.

Seismic anisotropy, caused by the directional dependence of seismic wave velocities, can be used to effectively define crustal and upper mantle deformation . Seismic anisotropy can be explained by two main deformation mechanisms: shape preferred orientation (SPO) and lattice preferred orientation (LPO) (e.g., ). Crustal anisotropy is commonly attributed to LPO in mica-rich minerals such as biotite and amphibole, as well as SPO in cracks, fractures, and compositional layering formed during deformation . Upper crustal anisotropy is primarily caused by structural features such as faults, layering, flat mineral alignment, and consistently oriented cracks . The LPO of mica or amphibole minerals can explain anisotropy below the crack closure depth in the mid- to lower crust . Upper mantle seismic anisotropy is primarily attributed to the strain-induced LPO of anisotropic minerals, like olivine, orthopyroxene, and clinopyroxene, arising from the dislocation creep deformation mechanism, which is characterized by the creeping motion of crystal dislocations caused by mantle flow and tectonic stresses at the plate boundaries due to past and ongoing deformations . Anisotropy in subduction zones may reflect fossilized fabrics in the overriding plate or stress-aligned crack systems influenced by slab dehydration and fluid migration, whereas in collisional orogens it is frequently associated with ductile shearing and lithospheric shortening, which produce coherent mineral fabrics parallel to orogenic trends . In contrast, rift settings show anisotropy linked with extensional strain fields, with vertically aligned cracks and dike intrusions producing different anisotropic signatures.

Shear wave splitting (SWS) analysis of SK(K)S core-refracted phases is used globally to quantify seismic anisotropy. SK(K)S splitting measurements at any seismic station provide the integrated seismic anisotropy of the crust and upper mantle. When a shear wave travels across an anisotropic media, it splits into two pseudo “fast” and “slow” S-waves that travel in perpendicular directions with different wave speeds . SWS measurements provide seismic anisotropy parameters, polarization azimuth of the fast wave (fast polarization direction (FPD)), and time difference between fast and slow wave (delay time (δt)), which provides critical information regarding strain patterns, mantle flow, and mechanical behavior of the lithosphere and sub-lithospheric system on the receiver side, allowing for inferences about mantle convection and tectonic history of the region .

A multitude of seismological investigations have been performed to elucidate the evolution and formation mechanisms of various regions in India by examining seismic anisotropy and its direct correlation to mantle deformation patterns, whether arising from present-day plate motion or the cumulative impact of prior and present-day orogenic events. Shear wave splitting has been extensively studied in the Himalayan region (northern India) to comprehend crustal and mantle deformation linked to the India-Eurasia collision . Similarly, numerous research studies on seismic anisotropy and its implications for lithospheric deformation have been performed in southern India to better understand the evolution and formation mechanism of the southern Indian cratons. The SWS study has also highlighted the geological complexities of northeastern and eastern India . A recent study on seismic anisotropy across the whole Indian region found that anisotropy in central India is caused by both frozen-in anisotropy and the shearing of the upper asthenosphere. Despite extensive research in these areas, the CITZ has received relatively less focus, despite its substantial importance. Due to a shortage of seismic stations, the region has minimal seismological data, and no extensive SWS study has been undertaken in this area (Fig. ). Figure  shows the previously conducted splitting measurements for all of India (represented by blue bars) alongside our study region (grey-shaded area), thereby addressing the data deficiency in central India. The CITZ region experienced subduction-accretion-collision tectonics during the Mesoproterozoic era and mantle plume activity during the Cretaceous period, resulting in the formation of Peninsular India . Given its importance as a collisional zone, it is critical to investigate seismic anisotropy in this region to understand the deformation mechanisms. This study examines teleseismic SK(K)S waveform data collected from nine temporarily deployed broadband seismic stations in the CITZ region to assess the seismic anisotropy of the upper mantle at each station. We employed transverse energy minimization methodology to determine the SWS parameters. The resultant SWS parameters provide new insights into the deformational architecture of the crust and upper mantle in the study area. A comprehensive geodynamic model for the CITZ is presented to elucidate the observed SWS patterns, which reveal the tectonic context and mantle deformation mechanisms responsible for the anisotropy beneath the region.

We utilize broadband seismic data collected from nine portable seismic stations that were operational from July 2023 to July 2025 (Fig. ). The seismic stations, equipped with Trillium Compact broadband seismometers, were deployed in three phases, with three stations installed in each phase. The earthquakes, having magnitudes exceeding 5.5, have been chosen for the SKS phase occurring within the epicentral distance ranges of 84–145° and for the SKKS phase, 84–180° (Fig. ). We have 424 teleseismic earthquakes based on magnitude and epicentral distance criteria, which were analyzed for SK(K)S-derived seismic anisotropy for all seismic stations. A band-pass filter having lower and higher cutoff frequencies of 0.05 and 0.2 Hz, respectively, is subsequently employed to enhance the SK(K)S seismic phases.

The SWS measurements were performed using the SplitLab , a software package developed in MATLAB. SplitLab offers multiple splitting algorithms to determine the optimal SWS parameters, FPD and δt. To ensure the reliability of SWS parameters, we employ two distinct and independent methodologies: rotation correlation and transverse energy minimization. The rotation correlation (RC) method determines the parameters FPD and δt by finding the orientation that results in the highest correlation between the calculated fast and slow shear wave components. This method, however, may result in systematic inaccuracies, especially when the initial polarization aligns closely with the polarization of the fast or slow direction of the anisotropic media. Although predictable and well-documented, these errors can be accounted for to retrieve accurate splitting parameters.

The transverse energy minimization (TEM) technique calculates FPD and δt by minimizing the energy on the transverse component after applying the SWS correction. The TEM approach is generally considered more robust for SK(K)S splitting measurements over a wide range of back azimuths. However, systematic errors, similar to those found in the RC approach, may have an impact, especially in circumstances with modest signal-to-noise ratios. This method assumes a single homogenous anisotropic layer and transverse isotropy with a horizontal symmetry axis. The TEM method follows a grid-search approach to ascertain the final optimal splitting parameter, which helps in determining the reliability of the results obtained from the TEM method. It accounts for potential fluctuations, likely induced by data noise, by providing a confidence level for each splitting parameter. Consequently, while the RC method provides an independent estimate of the FPD and δt, the TEM method determines these parameters from the minimum of the transverse-energy surface. The output from the TEM method includes the lower confidence limit, best-fitting value, and upper confidence limit for both the FPD and δt, enabling us to differentiate well-constrained splitting parameters from those with greater uncertainty. The final reported FPD and δt values were consistently taken from the TEM method, ensuring that the measurements used in the subsequent analysis were obtained using the same methodology.

Figure  shows an example of the SKS-derived good splitting measurement recorded at station SG01. The high-quality split event demonstrates a corrected particle motion with a good overlap between the fast and slow components (Fig. ). Before correction, the particle motion exhibits an elliptical trajectory; however, after correction, it takes a linear path that corresponds to the back azimuth (Fig. ). The transverse energy nears zero in the adjusted transverse component plot, and the splitting parameters FPD and δt are consistent, with FPD agreeing within 10° across several methods and δt within 0.2 s (Fig. ). Figure  shows an example of the SKKS derived splitting measurements at station SG01. Figure 6 shows an example of a null measurement at the SG02 seismic station, in which particle motion before and after correction coincides and aligns linearly with the back-azimuth. In such cases, the energy of the transverse component is at a minimum. Null measurements, which show negligible tangential (SH) energy, may occur in three situations. The first occurs when no anisotropy is encountered and the initial SV polarization of the SK(K)S phases is unaffected by anisotropy. It appears practically impossible that any tectonic plate is isotropic in nature. Second, if the region is isotropic in nature due to complex anisotropy (multiple horizontal anisotropic layers with different fast polarization directions). Third, if the initial SV polarization is similar to that of the slow/fast anisotropy axis, the shear wave will not be split into two quasi-S waves .

To ensure accurate SK(K)S splitting measurements, we performed a post-analysis verification of station orientation using the polarization characteristics of SK(K)S phases. For each event, we manually checked the particle motion before and after splitting correction. In particular, we compared the corrected radial component particle motion to the theoretical backazimuth of the incoming wave. After applying the splitting correction, the particle motion is expected to align with the source polarization direction, which coincides with the backazimuth . For all final split datasets, the corrected particle motion projected onto the radial-transverse plane consistently aligns with the theoretical backazimuth, confirming that the stations are properly oriented and that orientation errors are minimal.

After completing the splitting analysis, we have 129 reliable splitting measures, including 82 SKS splitting measurements and 47 SKKS splitting measurements. Table S1 in the Supplement contains detailed information about the earthquake events (event date, time, longitude, depth, and magnitude), station name, phase information, and shear wave splitting parameters with associated error (FPD, FPD error, δt, and δt error) for each splitting measurement. We found six null measurements in our dataset, four at the SG02 seismic station and two at SG09. Figure S1 in the Supplement shows a rose diagram of the backazimuthal coverage for the splits (in black) and the nulls (in red) for the SG02 and SG09 stations, demonstrating that the null and split measurements are obtained from different backazimuths, but their numbers are limited, so detailed modelling is the fundamental limitation of this study.

The majority of the analysed earthquakes are located in the back-azimuth range of 90 to 120° (Fig. 7c), a constraint imposed by India's geographic position relative to the global distribution of large teleseismic earthquakes at appropriate epicentral distances. This narrow window is a well-documented limitation of SK(K)S splitting studies on the Indian subcontinent . Figure  shows a histogram of individual splitting parameters, demonstrating a predominant concentration of FPDs at 40° and δt around 0.85–1.1 s. The backazimuthal variation of individual FPDs observed at all seismic stations (Fig. 7c) indicates the existence of multiple anisotropic strata. However, the number of high-quality SKS measurements per station in our dataset is relatively limited, and backazimuthal coverage remains constrained (Fig. S2). Constructing meaningful splitting parameters versus backazimuth plots requires dense and well-distributed backazimuthal sampling, ideally spanning at least 180° with adequate event density in each bin, which is not uniformly achievable across all stations in our network. Splitting parameters for two layers of anisotropy modeling, whether at individual stations or station groups, can only be interpreted reliably when the variation of splitting parameters with backazimuth demonstrates a clear 90° periodicity . and have explicitly demonstrated that when measurements are concentrated in a narrow backazimuthal range, layered models produce statistically indistinguishable fits, making unique structural determination impossible. Given these limitations, we have chosen not to proceed with two-layer anisotropy modeling in this study.

After calculating individual SK(K)S splitting measurements, the station-averaged splitting parameter for each station is calculated by averaging all splitting parameters derived from numerous events. The von Mises approach was used to calculate the circular mean of FPD, whereas station-averaged δt was calculated using the arithmetic mean. The station-averaged SK(K)S splitting parameter for each seismic station is shown in Table . The average splitting δt at each station vary from 0.74 to 1.4 s (Fig. ). We have the lowest value of δt at the SG02 seismic station. Stations SG01, SG03, SG04, SG05, and SG08 show high anisotropy (>1 s), while stations SG02, SG06, SG07, and SG09 have low δt (<1 s). The FPD values across all stations range from 24.84° (SG06 seismic station) to 70.88° (SG03 seismic station). Most seismic stations have a predominant NE–SW alignment of FPDs, which correlates closely to the Absolute Plate Motion (APM) direction of the Indian Plate with the ITRF2020 model in the no-net-rotation frame of reference (https://www.unavco.org/software/geodetic-utilities/plate-motion-calculator/plate-motion-calculator.html, last access: 20 September 2026).

SK(K)S phases are typically considered to be near-vertical incident at seismic stations, therefore, the measurements are characterized by poor depth resolution . Furthermore, to add depth constraints to the observed SKS splitting parameters, we conducted a ray-piercing-point spatial coherency analysis based on the spatial coherence of the high-quality SK(K)S splitting measurements . In the case of steep incident angles, the ray-path information is used to calculate the ray-piercing point at any assumed depth of the anisotropic layer. The mean splitting information is thus computed within the spatial averaging window, with the ray-piercing point as the center. The search for anisotropy depth was carried out at 5 km intervals within the depth range of 0–400 km. The variation factor is calculated as the weighted sum of the circular mean of the FPDs and the arithmetic mean of the splitting time delays. The depth value with the least variation factor was chosen as the optimal depth of the anisotropy. The variation factor was calculated at each depth for block sizes ranging from 0.05 to 4.0°, with an incremental interval of 0.05°. The optimum number of splitting measurements and spatial consistency in each block affect the variation factor. Both block-size configurations (dx=0.30° and dx=0.35°) show consistent behavior, with the variation factor declining steeply from near-surface values (∼0.95–1.0) down to a broad minimum in the 200–250 km depth range before stabilizing at ∼0.3–0.4 for greater depths. Figure  depicts the variation factor with varying anisotropic depth for block sizes of 0.30 and 0.35. Our study found that the variation factor curve gradually flattens beyond ∼250 km, indicating that the technique's sensitivity decreases as depth increases and ray-piercing point separations become larger than the spatial scale of anisotropy variations. Figure S4 in the revised manuscript demonstrates the piercing points at 50, 100, 200, and 300 km depth. The spatial distribution of ray-piercing points exhibits progressive lateral migration with increasing depth, which is consistent with the expected geometry of steeply-incident SK(K)S raypaths. Individual ray-piercing-point spatial coherency analysis for each seismic station was attempted, but the split measurements were insufficient to produce an acceptable result.

This study focuses on mantle deformation beneath the CITZ, supported by SWS data from the Satpura Gondwana region. The key implications of our study are as follows:

The SK(K)S splitting results indicate the presence of a significant anisotropic fabric beneath the studied area.