1 Introduction

Himalayan arc is one of the most seismically active convergent boundaries, having potential to generate large earthquakes due to the ongoing continental–continental collision between the Indian and Eurasian plates (figure 1). The earthquakes that have caused the most damage in the last two centuries in the Himalayan and contiguous plains are: 1803 Garhwal (MW 7.6), 1897 Assam (MW 8.1), 1905 Kangra (MW 7.8), 1934 Nepal Bihar (MW 8.2), 1950 Assam (MW 8.6) and 2015 Gorkha (MW 7.8) earthquakes. Occurrence of these earthquakes along the Himalayan arc manifests underthrusting of the Indian plate beneath the Eurasian plate (Lavé and Avouac 2000; Stevens and Avouac 2015), with the active seismic slip being mostly limited on the Main Himalayan Thrust (MHT), or on the imbricate thrust faults within the Himalayan wedge (Pandey et al. 1995; Cattin and Avouac 2000). Furthermore, most of the Himalayan earthquakes involve thrust motion with the slip vectors oriented perpendicular to the Himalayan arc. Apart from the conventional Himalayan earthquakes, a few earthquakes have been observed in the subducting Indian slab south of the Himalayan arc and their occurrences are commonly associated with the subducting aseismic ridges of the Indian plate, such as the Munger–Saharsa and Delhi–Haridwar ridges.

Figure 1
figure 1

General seismotectonic setting of the Himalayan arc. Earthquake epicentres (1900–2021 from NEIC, Masse and Needham 1989), focal mechanisms with compressional quadrant shaded in grey (1976–2021, GCMT, Ekström et al. 2012) are shown. Black diamonds represent significant earthquakes. Subsurface ridges in the Indo-Gangetic plains (Sastri et al. 1971; Rao 1973). Black arrow indicates the India plate motion relative to Eurasia (Altamimi et al. 2012). Red transparent patches annotated with year are the ruptures of major and great plate boundary earthquakes (Bilham et al. 2001; Diehl et al. 2017). Study area with the location of the December 2020 Haridwar event (blue star) is marked by a black rectangle. MFT: Main Frontal Thrust; MBT: Main Boundary Thrust; MCT: Main Central Thrust; BNS: Bangong–Nujiang Suture; IYS: Indus–Yangpo Suture.

Full size image

On December 1, 2020, an earthquake with a local magnitude of 4.3 occurred at a depth of 35 km near Haridwar (Uttarakhand), India. The epicentre (77.964°E; 30.057°N) is in the Indo-Gangetic plain and ~3–4 km south of the Himalayan front or Main Frontal Thrust, MFT (figure 2). The earthquake is about 200 km north–northeast of the M 4.5 earthquake that struck Delhi on May 29, 2020. Despite the absence of large earthquakes, the Delhi region has been deemed seismically active with smaller to moderate magnitude events, e.g., the 1956 Bulandshahar earthquake, the 1960 Faridabad–Gurgaon earthquake (M 4.9). Due to its proximity to far-field large earthquakes from the Himalayan arc and the small and moderate magnitude earthquakes in the foreland basin, the Indo-Gangetic Plains and the adjoining Siwalik Himalaya, with a population of over 500 million, have been considered the seismically most vulnerable regions (Singh et al. 2010; Mittal et al. 2016). The deep sediments of Indo-Gangetic plains further add to the seismic hazard due to magnification of seismic waves even from small-magnitude earthquakes (Shukla et al. 2007). The average depth of sediments beneath the Gangetic plain range from 1 to 6 km (Lyon-Caen and Molnar 1985; Borah et al. 2015). At least three aseismic ridges, namely, Aravalli–Delhi fold belt (ADFB), Faizabad ridge and Munger–Saharsa ridge on the Indian plate extend beneath the Indo-Gangetic plains and possibly extend northward beneath the Himalayan wedge. The ADFB is the most prominent amongst them and includes several mapped/inferred faults, e.g., the Great Boundary Fault (GBF), Sohna Fault, Delhi–Haridwar Ridge, Mathura Fault, and Moradabad Fault (Valdiya 1976; Dasgupta et al. 1987). Most earthquakes of the ADFB show reverse or oblique motion on steep dipping faults (Shukla et al. 2007). The northward extension of ADFB beneath the Indo-Gangetic plains appears to be less seismogenic as compared to its southward extension. The 2020 Haridwar earthquake is apparently located on the northward extension of the ADFB. Despite its smaller magnitude (M 4.3), it is critical to understand its source mechanism, occurrence process and its tectonic implications on the Himalayan and foreland seismo-tectonic systems for future hazard assessment.

Figure 2
figure 2

Seismotectonic map of the study area. Significant events (1900–2021, NEIC) from this region are plotted. The focal mechanism of December 1, 2020 (from fociMT) with compressional quadrant shaded in red colour (not scaled with magnitude) is shown and is showing strike-slip mechanism with NNE P-axis. Focal mechanisms with black-coloured compressional quadrant represent the event nearby Delhi–Haridwar ridge from the various study. Focal mechanisms with grey compressional quadrant (1976–2021, GCMT) represent events along the Himalayan arc. Blue triangle represents the local seismic network around the source region operated by CSIR-NGRI. Brown colour triangles denote NCS stations. Major seismotectonic features are shown as SF: Sohna Fault; GBF: Great Boundary Fault; MDF: Mahendragarh–Dehradun Fault. Approximate location of piedmont fault is after Yeats and Thakur (2008).

Full size image

2 Data

CSIR-NGRI is operating a local network in the Uttarakhand and adjoining Indo-Gangetic plains with broadband sensors (figure 2). A total of 45 broadband and five strong motion seismic stations data along with five stations operated by the National Center for Seismology (phase data available at ISC) were used for locating the earthquake by using SEISAN software (ver. 11) (Ottemöller et al. 2013). The nearest and farthest stations used here are located at 45 and 270 km, respectively. The estimated hypocentre location is at 30.057°N and 77.964°E with a depth of 36 km (35.8 km) with an uncertainty of <2 km corresponding to an RMS error in travel time of 0.3 s. The magnitude of the earthquake ML is 4.3. Inclusion of NCS stations helped in adequately covering the hypocentre with no azimuthal gap (figure 2).

3 Source characteristics

3.1 Focal mechanism

Over a wide range of magnitudes, the seismic moment tensor has been used as a standard parameter for understanding earthquake kinematics source processes. Globally, for larger earthquakes (MW > 4.5), source parameters are routinely done using teleseismic waveforms by the Harvard Centroid Moment Tensor (CMT) project (e.g., Ekström et al. 2012) and by the United States Geological Service (USGS) (e.g., Sipkin and Needham 1993). For MW ≤ 4.5 earthquakes, moment tensor solutions require local or regional data (e.g., Ford et al. 2009; Vavryuk and Kim 2014). The MT inversion may be performed using a variety of input data from body-wave amplitudes to full waveforms.

To determine the focal mechanism of the earthquake, we first used a full waveform inversion algorithm as implemented in the ISOLA program package (Sokos and Zahradnik 2008). ISOLA is widely established as a reliable program for estimating focal mechanisms of varying magnitudes at local and regional distances. ISOLA is based on an extension of the method proposed by Kikuchi and Kanamori (1991) to regional and local distances. One of the most important steps in the analysis is the selection of an appropriate crustal model to compute Green’s function which represents the earth’s response as it plays an important role in estimating Green’s function. The inversion is performed in the time domain while the Earth response is represented by Green's function computed based on discrete wavenumber algorithm (Bouchon 1981) using 1-D velocity model of Borah et al. (2015). During inversion, the seismograms were filtered in the frequency range between 0.13 and 0.18 Hz. The MT inversion is retrieved by the least-squares inversion and the best matching between observed and synthetic waveforms can be shown based on variance reduction values (VR) (Sokos and Zahradnik 2008). The best fit double couple model has a VR value of 0.57 and reveals the mechanism of predominantly strike-slip in nature with nodal planes oriented in the ENE–WSW (NP1: strike, φ = 70°, dip, δ = 83°, rake, λ = 33°) and NW–SE (NP2: φ = 336°, δ = 57°, λ = 172°) directions (figure 3). The estimated depth of 36 km is similar to that estimated using phase data (figure 3b). The magnitude is estimated as MW 3.8.

Figure 3
figure 3

Moment tensor inversion result for the 1 December 2021, Haridwar earthquake. (a) Comparison of observed (black) and synthetic (red) waveforms. Station name in each case is indicated in the N–S component. The number at the right top in the waveform panel represents the variance reduction between the waveforms. (b) The best-fitting mechanism for different trial depths is plotted in a lower hemisphere projection. The best fit is for a depth of 36 km (blue colour).

Full size image

The focal mechanism is further confirmed using the polarity of first arrivals and P-wave spectra to constrain double-couple MT solution (Kwiatek et al. 2016). We use fociMT inversion method which is based on the concept of Fitch et al. (1980) and Wiejacz (1992). The input data for the MT inversion are the P-wave spectra and first motion polarities of the vertical components (Kwiatek et al. 2016). We selected stations having clear P-arrival around the event and estimated the focal mechanism, which is predominantly strike-slip in nature with nodal planes oriented in the ENE–WSW (NP1: strike, φ = 83°, dip, δ = 83°, rake, λ = 30°) and NW–SE (NP2: φ = 349°, δ = 60°, λ = 172°) directions and the P-axis is oriented in the NNE direction (figure 2) which is similar to that obtained using the waveform inversion. The estimated MW is 3.9.

3.2 Source parameters

The frequency analysis method is more commonly used for estimating basic source parameters (e.g., Abercrombie and Rice 2005; Mandal and Johnston 2006; Allmann and Shearer 2007) using Brune's (1970, 1971) theoretical spectrum of circular rupture model for small earthquakes, which includes a flat plateau at low frequencies and decreases by ω−2 at high frequencies. Using the SourceSpec Python package (Satriano 2020), we determined the source parameter using S-wave spectra. The combined effect of the source, crustal wave propagation, and site contributions determine the amplitude spectrum of ground motion for a given distance R from the earthquake source, as shown below:

$$v\left(f, R\right)={\Omega }_{0} \left(f\right) \cdot 2\pi f\cdot A\left(f, R\right) \cdot H\left(f\right)$$
(1)

where v(f, R) is the velocity and f is the frequency, A(f, R) is seismic wave attenuation factor along the source-to-station propagation path, and H(f) is a site term accounting for near-surface propagation effects beneath the station.

The source-radiated displacement spectrum is expressed as:

$${\Omega }_{0}\left(f\right)=\frac{{F}_{s}{R}_{\theta ,\varphi }}{4\pi \rho R{\beta }^{3}}.\frac{{M}_{0}}{1+{\left(f/{f}_{c}\right)}^{2}}.$$
(2)

Here M0 and fc are seismic moment and corner frequency, respectively. Rθ,φ is the radiation pattern, which represents the average S wave radiation pattern over azimuth and take-off angles and is taken equal to 0.55 (Boore and Boatwright 1984). R is the hypocentral distance. Fs accounts for the free-surface effect, taken equal to 2. ρ and β are density and shear wave velocity at the source region, respectively.

Following Eshelby (1957), the radius r and uniform stress drop (∆σ) of a circular crack in an infinite Poisson solid are linked to M0 and fc through the equations:

$$\Delta \sigma =\frac{7}{16}\frac{{M}_{0}}{{r}^{3}}$$
(3)
$$r=\frac{2.34\beta }{2\pi {f}_{c}}.$$
(4)

As far as apparent stress (τa) is concerned, we followed the standard method of Wyss and Brune (1968) using the definition

$${\tau }_{a}=\mu \frac{{E}_{r}}{{M}_{0}}$$
(5)

where Er is the radiated energy, which is determined following the approach described in Boatwright et al. (2002) and Lancieri et al. (2012).

The source amplitude spectra of the 2020 event are shown in figure 4, where we used six stations with a good signal-to-noise ratio (S/N > 3). The estimated seismic moment (M0) is 4.9×1014 Nm, corresponding to an MW of 3.73 (±0.07) and a corner frequency (fc) of 5.1 (±0.6). The fracture radius is approximately 269 (±32) m. A value of 9.4 ± 3.7 MPa and 3.03×1010 J are obtained for the stress drop (∆σ) and estimated radiated energy (Er), respectively. The scaled energy which is the ratio of radiated energy and M0 is 6.2×10−5. The calculated apparent stress drop (τa) is 2.5 MPa based on the assumption of a rigidity modulus (μ) of 4.03×1010 Pa from the velocity model of Borah et al. (2015) at the source depth of the event.

Figure 4
figure 4

Best fitting of Brune’s displacement spectra obtained to estimate source parameters. Estimated average M0 is 4.9×1014 Nm which is equivalent to the MW of 3.73 (±0.07) and a corner frequency of 5.1 (±0.6).

Full size image

4 Discussion

The occurrence of an event like December 2020 at a depth of 36 km within the Indian plate and ~20 km south of the mountain front, is rare. It is an intraplate earthquake which occurred on the Delhi–Haridwar ridge beneath the 5–6 km thick sediments of Indo-Gangetic plains. Data from our dense and close seismic network, along with the data from NCS stations, provided us the opportunity to look into the source characteristics of the 2020 Haridwar event to understand the causative reason and its seismotectonic implications. The estimated focal mechanism suggests that predominant sinistral or dextral strike-slip motion was involved on the ENE–WSW or NW–SE oriented nodal plane, respectively, which is consistent with the stress regime inferred from the earthquakes in the Delhi region (figures 2 and 3). The earthquake's depth is comparable with earthquakes in the Delhi region, which occur at depths of up to 30 km (figure 5). The estimated stress drop of 9.4 MPa for the event implies high-stress drop and is consistent with the higher stress drop for the stable continental events. Lower strain rates and longer average fault healing time in the intraplate zone (Scholz 2002; Venkataraman and Kanamori 2004) could explain the larger stress drop and probably related to the strike-slip mechanism of the events (Allmann and Shearer 2007). It was observed that there is no relation between stress drop and magnitude (Venkataraman and Kanamori 2004; Allmann and Shearer 2007). Previous research by Sharma and Wason (1994) and Sivaram et al. (2013) found that the stress drop estimated for the interplate earthquake from the Kumaun Himalayan region is <1 MPa for similar magnitude events. In fact, large value of stress drop of 13 MPa for November 25, 2007 (MW 4.1) (Singh et al. 2010) and 12.4 MPa for March 5, 2012 (MW 4.6) (Mittal et al. 2016) earthquakes in the Delhi region has been reported. Further, similar high-stress drop estimates have been reported for other significant intraplate earthquakes in India, e.g., 21 MPa for the 2001 Bhuj mainshock of MW 7.7 (Negishi et al. 2001; Antolik and Dreger 2001), 7 MPa for the 1993 Latur earthquake of MW 6.3 (Baumbach et al. 1994) and about 20 MPa for the MW 5.8 Jabalpur earthquake (Singh et al. 1999). The scaled energy, expressed as a ratio of Er to M0 is an order of 10−5. In general, the ratio of energy to moment is 10−6–10−5 for smaller events, but for the larger events (>1016 Nm), the ratio increases to between 10−5 and 10−4 (Venkataraman and Kanamori 2004). Our estimated value of 6.2×10−5 is higher than the average for this type of magnitude which could be due to the higher stress drop and fault strength.

Figure 5
figure 5

General depth section of Himalaya. Based on the depth estimation, the epicentre location of December 1, 2020 Haridwar events is marked with focal mechanism and indicates intraplate origin. MFT, MBT, MCT, and MHT are also marked.

Full size image

Farther south of the Haridwar event along the ADFB, near Delhi, many moderate and smaller magnitude earthquakes have occurred. Although the overall trend of ADFB is NE–SW, there are several faults within ADFB which criss-cross ADFB. It is difficult to associate these earthquakes to an identified fault in the region. Their focal mechanisms generally exhibit oblique (strike-slip and reverse motion on steep faults) motion on the faults which are generally transverse to the trend of ADFB. The 2020 Haridwar earthquake is quite similar to the earthquakes of Delhi region in terms of focal mechanism and sense of motion.

One of the critical questions here is the seismogenesis of this earthquake. We qualitatively evaluated all possibilities for the 2020 Haridwar earthquake occurrence based on the estimated source parameters. We examined the possibilities of whether this earthquake occurred (i) in response to the southward propagation of the Himalayan wedge in the form of piedmont fault, (ii) due to flexure in the Indian plate caused by the subduction of the Indian plate beneath Eurasia along the Himalayan arc, (iii) due to strong coupling on the MHT, and (iv) on the northward extension of the Delhi–Haridwar ridge. The first hypothesis can be refuted straight away, and as it occurred at 36 km depth we confirm that the earthquake did not happen on the piedmont fault. In the Indo-Gangetic plains of Pakistan and India, Yeats and Thakur (2008) proposed a southerly progression of thrust front and identified piedmont fault. In the region of earthquake, they mapped piedmont fault at about 20 km south of the Main Frontal Thrust (MFT), the southernmost thrust fault or the deformation front of the Himalayan arc. Although the epicentre of the earthquake is closer to the MFT (~3–4 km south of the MFT), it is unrelated to any of these surface-mapped faults in the Himalayan wedge or on the Indo-Gangetic plains due to its deeper focal depth. Even the second and third hypotheses, i.e., the flexure on the Indian plate due to Indian plate subduction, also caused by the strong locking on the MHT, similar to the subduction zone's outer rise earthquakes (e.g., Chapple and Forsyth 1979; Christensen and Ruff 1988; Craig et al. 2014), can be refuted due to larger focal depth of this earthquake. The outer rise earthquakes are generally shallow and involve normal faulting, which is not the case for the 2020 Haridwar earthquake. We propose that it probably occurred on the northward continuation of the Delhi–Haridwar ridge as similar earthquakes occur on this ridge in and around Delhi region. Similar to the Haridwar earthquake, a strong earthquake in 1988 Udaipur (Nepal) with a magnitude of 6.8, occurred on the northward extension of the Munger–Saharsa ridge, with similar focal depth, location, and focal mechanism (Chen and Kao 1996; Ghimire and Kasahara 2007). However, the Munger–Saharsa ridge is not as active as the ADFB, but that could be due to lack of nearby seismic stations close to Munger–Saharsa ridge. In fact, the occurrence of 1988 and 2020 type of earthquakes pose an additional seismic hazard. So far our perception of hazards in the Himalaya and adjoining regions is limited to that posed by the Himalayan earthquakes on the MHT. But the occurrence of these earthquakes, though at relatively deeper depths, may pose additional hazards if the magnitude of such earthquake is large, as witnessed during the 1988 earthquake of MW 6.8, which took ~1000 lives.

5 Conclusions

The 2020 Haridwar earthquake is unusual in terms of its location, focal depth and source parameters. We estimated its source parameters using a dense and local seismic network around it. The earthquake occurred at 36 km within the Indian plate and south of Himalayan convergent boundary. It involved strike-slip motion on steep planes and exhibited relatively higher stress drop. It appears to have occurred on the northward extension of Aravalli–Delhi fold belt, which in this part is buried under the 5–6 km thick Indo-Gangetic plains. The source parameters of this earthquake are also similar to that of the earthquakes in the Aravalli–Delhi fold belt. We recall that a similar but of larger magnitude earthquake (MW 6.8 1988 Udaipur, Nepal earthquake) occurred on the northward extension of Munger–Saharsa ridge. Occurrence of such earthquakes on the ridges on the Indian plate, which are extending northward and subduction beneath the Himalayan wedge, pose additional seismic hazard which is generally not considered as we tend to focus on the seismic hazard posed by the Himalayan earthquakes occurring on the MHT. We acknowledge that although they are rare events, they can still pose significant seismic hazards.