Delineation of partial melts and crustal heterogeneities within the crust beneath Kumaon Himalaya, India from Lg wave attenuation

The crustal seismic attenuation or the Q structure is studied by using the Fourier spectra of Lg-wave along the Tanakpur- Dharchula- Dharma transect in the Kumaon Himalaya. The 1 Hz Lg Q (Q0) values are computed between different pairs of two stations and the observed values are later utilized to calculate the lateral variation in the Q0 values by following a back projection algorithm. This computation of Q0 values utilizes five regional distance earthquakes having moment magnitude (Mw) ≥ 4.0, which lie along the great circle path of the transect. Three of the five earthquakes occurred in the Tibetan plateau and the and the others occurred to the southwest on the Indian shield and are well recorded at all the 32 broadband seismographs operated between September 2018 and March 2022. The estimate Qo values range from 63 ± 2 and 203 ± 25, with the lowest value in the Lesser Himalaya and the highest across part of the Indo Gangetic Plain and Siwalik Himalaya. The Q0 model has low values ∼200 along the profile in the Indo Gangetic Plain and the Siwalik Himalaya, and are correlated with 2–5 km thick sedimentary layers below the Himalaya and the adjoining Indo-Gangetic Plain. We observe two distinctly different Q0 values to the northeast in the Lesser Himalaya tectonic unit. The region lying between the South Almora Thrust (SAT) and the Berinag Thrust (BT) shows extremely low Q0 values (∼60) but increases further north towards the Vaikrita Thrust (VT) to ∼200. The possible explanation for observing such huge variation of the Q0 values within a single tectonic unit may be the presence of fluid rich ramp structures, which introduces crustal heterogeneities and traps the aqueous fluids or partial melts lying within the crust. The Lg Q0 values decrease to the North and become ∼166 for station pairs in the Higher Himalaya and Tethys Himalaya tectonic units. The low Q0 values observed in this region may be correlated with low viscous partial melts in the form of Miocene leucogranite plutons, which resulted out of the Indo-Asian collision. The attenuation structure along the profile in the Kumaon Himalaya can be used to estimate ground motions of future earthquakes in the area and can contribute to seismic hazard assessment in the Himalaya and neighbouring regions.

www.nature.com/scientificreports/ studies worldwide to characterize the attenuation property of a medium through Quality factor (Q) estimation utilizing different types of seismic waves for e.g. coda-wave, Lg-wave, P-wave and S-wave, etc. recorded at local and regional distances from an earthquake location 4,[11][12][13][14][15][16][17][18][19] . The effect of fluid or melt fraction is more dominant in attenuation property (1/Q) compared to the S-wave velocity. While the S-wave velocity in crust varies by less than 20%, Q varies by a factor of 3 on major continents 20 . The present profile, which lies in the Kumaon sector of the northwest Himalaya stretches to about 200 km in length along the northeast-southwest direction. The transect cuts perpendicular across the major fault structures in the region that have a strike directing northwest-southeast.
The present seismic profile consists of stations that are fit by the great-circle path under a least square criterion 3 and thus provide us with the opportunity to invert for the lateral variation of Lg Q along this best-fit profile in the Kumaon section of the northwest Himalaya (Fig. 1). We have utilized the interstation Q 0 computation approach 3 to image the attenuation across the major structural features of the Kumaon Himalaya that extends from the Indo-Gangetic plain in the south across the Main Frontal Thrust (MFT) to the north of the South Tibetan Detachment (STD) fault in the Tethys Himalaya. The first step involves calculation of the interstation Q 0 values from the selected pair of stations. The interstation Q 0 values are then used to map the lateral variations of Lg Q 0 using a back projection method. We find low Lg Q 0 along the entire length of the profile, which we interpret in terms . Inset 1 shows the Indian map with the study area marked as a red square. Inset II shows the geographical location of the five regional earthquakes (solid star) used in analysis and the recording stations (filled triangle). Three earthquakes in the Tibet are represented in the form of solid stars with different sizes and colours. Rest two earthquakes from Indian shield are represented as solid blue star. The figure has been generated using the Generic Mapping Tool (GMT), version 6.0 (https:// docs. gener ic-mappi ngtools. org/6. 0/).

Seismotectonics and previous geophysical studies
The spectacular Himalaya, mountain chain resulted from the northwards underthrusting of the Indian plate with the Eurasia since 50 Ma 21,22 . The subduction of the Indian plate beneath the Himalaya caused lateral deformations giving rise to fault structures from north to south. These crustal scale faults have a NW-SE strike and are designated as the Indus Tsangpo Suture Zone (ITSZ), the Main Central thrust (MCT), the Main boundary thrust (MBT) and the Main frontal thrust (MFT) 23,24 . The MFT bounds the Himalaya from the adjoining Indo-Gangetic Plain (IGP) to its south. Four different tectonic units referred to as the Siwalik Himalaya; the Lesser Himalaya; the Higher Himalaya and the Tethys Himalaya are separated from each other by thrust faults at the respective southern and northern boundaries by the MFT, MBT, MCT, STD and ITSZ respectively (Fig. 1). The other major thrust faults between the MBT and the MCT are the Almora thrust which is further divided into a north Almora thrust (NAT) and south Amotha thrust (SAT) and the Barinag Thrust (BT). These surface faults penetrate in depth to about 5 km 25 .
The western Kumaon region lies in the central seismic gap (CSG) and several geological and geophysical studies suggest a ramp structure involving the Main Himlayan Thrust (MHT) at mid-crustal depth that is responsible for major and great earthquakes in the Himalaya 26 . Recently, crustal investigations using receiver function analysis along the Kali River of the Kumaon Himalaya have shown variations in the crustal thickness of ~ 38 km to 41 km from the Indo-Gangetic Plain (IGP) to the Higher Himalaya and have also detected partial melts at mid-crustal depths 27 . A study to the west of the Kali river region 28 found a flat-ramp-flat structure of the MHT and a low shear wave velocity zone (LVZ) beneath the Lesser Himalaya between the Main Frontal Thrust (MFT) and the Main Central Thrust (MCT). This LVZ is well corroborated by low resistivity found in magnetotelluric studies 29,30 . Receiver function analysis on the eastern side of the Kali river in Nepal reported that the Moho beneath western Nepal is gently dipping northward, from a depth of ~ 40 km beneath the foothills to ~ 58 km beneath the Higher Himalaya and then even deeper beneath southern Tibet 31 . A mid-crustal lowvelocity zone at ~ 12-18 km depth, beneath the Lesser Himalaya in western Nepal was also reported 31 . This LVZ lies in between the MBT and the MCT and is likely caused by fluids expelled from underthrust sedimentary rocks trapped at the MHT 31 . The spatial distribution of simulated crustal Vp/Vs ratios reveals a large lateral variation in the crustal composition of the Uttarakhand Himalayan region 32 . Large variations in the crust's thickness (28.3 to 52.9 km) and Poisson's ratio (0.17 to 0.36) indicate the presence of serpentinization and high pressure metamorphic fluids in the crust 32 .
In a Lg Q o attenuation study 33 conducted along a profile in the Garhwal Himalaya area of the northwest Himalaya found Indian shield-like Q o values of 742 ± 235 in the Lower Himalaya, whereas the High Himalaya is distinguished by an exceptionally low Q o value of 30-60. Their results suggest the presence of a low viscosity channel in the Higher Himalaya. A similar study 34 for the Ladakh-Karakoram and southern Tibet suggests high crustal seismic attenuation (Q o ∼70) beneath the region with possible presence of aqueous fluid/partial melt in the Ladakh crust.
All of the above mentioned geophysical studies in the Garhwal and Kumaon region of Uttarakhand, India highlights the fact that this region has been seismically very active in the recent decades and involves a structurally complex and heterogenous crust. According to the 2011 census this region has a population density of 69.77% residing in rural villages with poor living conditions and with bare minimum services. There is also very poor enforcement of earthquake resistant building practices, which makes the region highly vulnerable to future seismic hazards of ground motion. An ability of accurately estimate ground motions from future earthquakes is needed to address the seismic hazard of a region requires estimation of ground motion from future earthquakes. This ability requires knowledge of the crustal velocity structure, characterization of earthquake source zones, and quantification of seismic wave attenuation from the source to any point on the surface of the Earth. The first two requirements have been extensively studied for the Garhwal-Kumaon Himalaya, over the past decade 26,27,[35][36][37][38][39][40] but gaps exist in our knowledge of the attenuation structure. The present study focuses on estimation of seismic wave attenuation using the Lg wave spectra from earthquakes recorded at regional distances to ascertain the quality factor (Q) for the crust beneath the Kumaon Himalaya, India. This should provide more constraints on assessment of future seismic hazard scenarios in the Kumaon sector of the northwest Himalayas.

Data
CSIR-NGRI initiated a project to study the seismic hazard aspect of the Kumaon Himalaya referred to as "Synthesis of Earthquake Hazard scenario in NW Himalaya by Investigating the multi-scale Variations in structural and seismotectonic Assemblages" (SHIVA). The seismic attenuation study in the Kumaon Himalaya focused on the central seismic gap zone along the SW-NE transect covering MFT, MBT, MCT and STD and used 32 broadband seismic stations that operated September-2018 to March-2022 41,42 . All the stations were equipped with CMG-3 T (120 s period) sensor and REFTEK (RT 130-01) data loggers with 4 GB swappable hard disk and GPS. Details of individual stations are presented in Table 1.
The Lg wave recorded on the seismograms of regional earthquakes (Table 2) are extracted with corner at 0.5 Hz using in the time window appropriate for a wave travelling with a group velocity between 3.0 and 3.6 km s −13 (Fig. 2).
The Lg wave was then cut out and a cosine taper of width 0.05 was applied to avoid spectral leakage. The instrument response was removed after transforming to the frequency domain using the fast Fourier transform To stabilize the measurement of Q o , we applied a 15-point smoothing to the ground displacement spectra 3 and computed the spectral ratio in the frequency band 0.4-2 Hz. Here we have utilized a total five regional events of which three events were in the Tibet region and two were from the Indian shield region. Figure 3a,b shows the plot of example of Lg amplitude spectra of selected stations.

Methodology
Lg is a regional seismic wave comprised of multiple shear wave reverberations trapped in the crustal waveguide, and is important for studying the crustal structure. The Lg Q at frequencies greater than 1 Hz is observed to obey a frequency dependence power law of the form  3,20 . The quality factor (Q o ) generally increases with frequency. We analysed the Lg Q at 1.0 Hz to maintain the uniformity within the analysis and to accurately determine the relative variation within different lithotectonic units.

Estimation of Lg Q o .
We have utilized the two-station technique 3 to compute the Lg Q attenuation structure beneath the Kumaon Himalaya. If two-stations denoted as station i and station j then the method collects the Lg amplitude spectra, which is denoted as A i (f) and A j (f) at a particular frequency, f respectively. Then we calculate the scaled amplitude ratio following the methodology 3 , which states that  Equation (2) can be written as. where V Lg is the typical Lg group velocity. The Q o and η values can be calculated by fitting a line using the leastsquares criterion. While the computation of Q o is stable, the measurement of stable interstation η is difficult. Further, due to a limited availability of earthquake-station pair measurements, we restricted this study to Q o variation only. Accordingly, we used η = 0 in Eq. (3) and computed the Q o along with the associated error in individual two-station data.
Initially, we selected all possible two-station pairs for the analysis to estimate the interstation spectral ratios but later on we applied the suggested criterion 3 to examine the final two-station pairs for the Lg Q analysis.
There is another important limitation to this method, which assumes that the recording stations are aligned exactly with the source, but in a real scenario, a perfect alignment of stations and source is usually not possible 3 . To define the alignment, an angle δθ, which is the difference between the azimuths from the source and the two stations, is used. A previous study 43 allowed δθ up to 10°, whereas another study 3 used 15°. The amplitude spectra of two events one from the north-eastern and other from the south-western side of the profile are shown in Fig. 3a and b, respectively. Inversion for regional variation of Q o . The inversions are composed of two steps. In the primary step, interstation Q o is measured from selected pairs of two stations (Table 3). In the second step, the interstation Q o values are mapped using inversion. We have applied the singular value decomposition method 44 to compute the variations of Lg Q o attenuation along the profile in the Kumaon Himalaya. We have divided our NE-SW profile in to four different cells with the M denoting the discrete Lg Q o values in different grids, and where N is the number of observations (Q value along a two-station path, so N = number of paths). If there are N interstation Q o measurements and the region is divided into M cells, the quality factor value for individual cells (Q m ) is related to the two-station measurements of quality factor (Q n ) as: where Δ mn is the length of the nth ray in the mth cell. εn is the error for each Q n measurement.

Results and discussion
Assessment of seismic hazard in an earthquake prone area involves quantification of its crustal velocity structure, earthquake source properties and the seismic wave attenuation in the area. The latter requires estimates of the seismic quality Q. Estimation of Q at 1 Hz (Q 0) of the Lg waves recorded from regional earthquakes is useful because of Q 0Lg is sensitive to crustal properties 3,6,19 . Lg attenuation is also an important factor in determining seismic hazard because Lg of the large amplitude at regional distances 45 . The seismic risk in the Himalaya is high as it is expected to host a Mw 8.0 or greater anytime in future 46 . The present study area, which forms the eastern part of the CSG in the Himalaya has hosted many moderate to strong earthquakes in the past with the two most significant earthquakes being the 1991 Mw 6.8 Uttarkashi earthquake and the 1999 Mw 6.6 Chamoli earthquakes. Quantifying the obvious hazard is an essential to prerequisite to reducing the seismic risk to the   www.nature.com/scientificreports/ population in this region. We computed the Lg attenuation characteristics using five regional events with M w 4.0 to 5.5 (Table 2) along the NE-SW profile of ∼200 km along the Tanakpur-Dharchula-Dharma transect in eastern Kumaon Himalaya. We utilized the interstation Q 0 computation approach 3 to quantify the attenuation at different places along the 200 km seismic profile. We find significant differences of Lg Q o along the profile. The individual Lg Q o values resulting from the various station pairs are represented in Table 3 and some examples for linear regression fit for two station pairs are shown in Fig. 5.
The value of Lg Q o from different events and different pairs of stations are plotted in Fig. 6. The Lg Q o values vary 20 to 203 for 64 combinations of two stations along the profile.
To better understand the lateral variability of the Q o along the profile, we divide the profile into four blocks. From north to south, first block covers the IGP and Siwalik Himalaya and the second and third blocks cover the lesser Himalaya and the fourth block consists of the higher Himalaya and the Tethys Himalaya. The linear inversion of Q o values computed using Eqs. (5) and (6), utilizing all the individual interstation Lg Q o values (Fig. 6) resulted in a minimum value of 63 ± 2 and a maximum of 203 ± 25 for four individual blocks along the profile (Fig. 7). The Lg Q o values along the profile are nearly equal to 200 for all the blocks except for a part of the second block lying between the South Almora Thrust (SAT) and the Berinag Thrust (BT), which shows characteristically low Q o value of 63.
Our maximum Lg Q o values resulting from the inversion of interstation Lg Q o values is 203 ± 25, observed for the first block of the profile that includes some stations from Indo-Gangetic Plain, Siwalik Himalaya and some part of the Lesser Himalaya to the south of the SAT (South Almora Thrust). The reason for such low Lg Q o values in the region can be attributed to the presence of younger tertiary sediments, which has characteristically low Vp/Vs ratios and cause decay of Lg wave amplitude below the paired stations. Another possibility is the lateral variation of crustal structure and sedimentary thickness (2-5 km) 47 .
This second block in the profile shows the low value of Lg Q o 63 which reflects strong attenuations which consists of major secondary fault structures of the lesser Himalayan tectonic units and is mainly bounded between the South Almora Thrust (SAT) in the south and the Berinag Thrust (BT) in north. The low value of Q o correlates with low resistivity and a low seismic velocity zone 28,29,48 detected to the west.     53,54 , which report a low Q 0 ≈ 44 and Q 0 ≈ 74 respectively. The study 38 investigated the seismotectonic perspective for the Kumaon Himalaya and reported a complex faulting pattern, which is correlated with our reported low Q0 values to the south of the VT in the MCT zone. The reason for observing such structural or crustal scale heterogeneities can be ascribed to the presence of fluidrich zone as well as strain localization and large stress build-up due to the locking in the ramp structure on the MHT in the Dharchula or the Chipalkot crystalline belt. We believe that the ramp structure extends further south into the region between BT and SAT in the Inner lesser Himalayas and leads to lowering of the Q 0 values in this part of the lesser Himalayas.
Another significant study 39 utilizing the P-wave receiver functions reported the presence of a ramp in the crust, which dips at about ∼20°. This ramp or duplex structure is responsible for introducing crustal heterogeneities within the crust. The study also reports the presence of a low velocity layer in the lesser Himalaya with aqueous fluids that can result in high attenuation of the Lg spectra below the lesser Himalayan crust. The presence of aqueous fluid is well correlated with the mechanism of decompression melting 55 , which can be associated with the melting of the lower crust in the region.
High Q o were reported for the southern part of the Lesser Himalaya zone between MBT and MCT in the Garhwal section, to the northwest of the present profile 33 .
For our profile, we observed a characteristically low Q o values and high attenuation in the range from 63 to 200 for the Lesser Himalayan tectonic unit, which lies between the two crustal scale faults MBT and the MCT. Our observations in the Higher Himalaya tectonic unit are similar to those from the previous study 33 , which has also reported low Lg Q 0 values between 30 and 60 in the actively deforming Higher Himalaya. To the south of the MCT zone in the Lesser Himalaya they have reported high Lg Q 0 values of 742 ± 235, which also shows a large error associated with the value and this is not an acceptable value for the actively deforming Himalayas. This large error reported for the Lesser Himalaya section in Garhwal Himalaya 33 does not support the underthrusting mechanism of the Indian plate beneath Himalayas, which makes the region seismically very active and contradicts the presence of a stable lithosphere like the Indian Shield beneath the Himalayas.
Our findings show low Q o values and high attenuation for the Lesser Himalayan tectonic unit in accordance with LVL and low resistivity in between MBT and MCT 29 . A previous study of this region also reported high Poisson's ratio due to the presence of fluids/partial melts at mid-crustal depths, which in turn controls the rheological strength of the crust leading to high crustal attenuation in the region 32 . The presence of fluids beneath the Higher Himalaya and the Tethys Himalaya tectonic unit has led to the long-term structural and compositional evolution of the active fault zones through decrease in fault friction 28 . This decrease in frictional strength of the fault causes a more heterogeneous fault behaviour and leads to high seismic attenuation of the L g waves [56][57][58] . The Tethys Himalaya north of the STD is reported to be actively deformed and consists of abundant normal faults with east-west extension 59 , which can lead to the presence of strong small-scale structural heterogeneities within the crust that might have also contributed observed low Lg Q o values in the region.
There have been other studies available from the investigation of seismic attenuation utilizing different types of waves for the Himalaya-Tibet region (Fig. 8). Similar results of lower Q p (≈ 44) and Q s (≈ 87) at 1 Hz utilizing body waves were reported for the northeast Himalaya 60 . The results from most of the seismic attenuation studies reports low Q o values for the Himalaya-Tibet orogenic belt and the possible reasons include presence of crustal heterogeneities and trapped partial melts in the mid-crust 61 . Another significant reason controlling the seismic Q factor beneath Himalaya can be due to the crustal thickening and presence of thick sedimentary

Conclusions
Lg spectra recorded along the Tanakpur-Dharchula-Dhrama transect is utilized to study the seismic attenuation or the Q structure at 1 Hz beneath the western Kumaon sector of the NW Himalaya. A two-station method and a back projection algorithm are used to obtain models of laterally varying Lg Q 0 (Lg Q at 1 Hz). This resulted in Q 0 values to be low (∼ 63 -203) for the entire profile. We interpret these extremely low Lg Q 0 values in the Kumaon Himalaya region being caused by the presence of crustal heterogeneities due to ramp structures and/or trapped fluids/partial melts caused due to high temperature in the crust.
Our results are consistent with the previous findings of strong attenuation beneath Higher Himalaya and Tethys Himalaya in Tibet as well as other sections of the northwest and central Himalayas but somehow differs for the Lesser Himalaya, which earlier reported low attenuation. Though this study has been conducted with a good quality of seismic datasets but there is a scope of obtaining Lg Q 0 tomographic images along the profile, which would be more useful in having better constraints on the crustal velocity structure variations in the actively deforming Himalaya collision zone.