1 Introduction

The state of Sikkim in India lies in the Himalayan region and is bordered on three sides by Bhutan, Tibet, and Nepal. According to the Indian seismic zoning, Sikkim lies in the high-risk zone IV. Sikkim lies along the tectonic plate boundary where the Indian and the Eurasian plate are under collision. The Main Boundary Thrust (MBT) and Main Central Thrust (MCT) are the two major thrust faults passing through the state as shown in Fig. 1a. The region has recorded seismic activity in the past, and the distribution and damage pattern in the 2011 earthquake demonstrate attenuation features of the area [1,2,3,4,5,6,7,8,9,10,11] between the Main Boundary Thrust (MBT) and the Main Central Thrust (MCT). On 18th September 2011, Sikkim was struck by a Mw 6.9 earthquake which caused enormous structural damage to the built environment in the region. The earthquake exposing several structural defects and also highlighted the role of construction malpractices associated with non-engineered construction in the region. This paper analyses the ground motions of the 2011 Sikkim earthquake and describes the observed structural damage caused due to the severe ground shaking associated with this earthquake. Time–frequency analyses are performed on strong ground motion data recorded at six stations (Chungthang, Gangtok, Gezing, Melli, Mangan, and Singtam) as shown in Fig. 1b.

Fig. 1
Two maps of Sikkim. Map a. M C T is a curve going north through West Sikkim and the northern part of South Sikkim, through Mangan in the south of North Sikkim, and south through East Sikkim. M B T is a line that goes west to east in the south of Sikkim. Map b. The six recording stations are located on the east of the epicentre.

a Location of MCT and MBT on the map of Sikkim, b Earthquake epicentre and recording stations. (1) Chungthang, (2) Gangtok, (3) Gezing, (4) Mangan, (5) Melli, (6) Singtam, *Epicentre

2 Observed Damage to Buildings

Several reconnaissance surveys by various researchers [12,13,14,15,16] were undertaken in the aftermath of the earthquake to assess structural damage. The surveys recorded the observed damage to the variety of structures prevalent in Sikkim. It was observed that damage was severe due to the non-engineered structures prevalent in the region (RC, unreinforced masonry (URM), and bamboo structures). In some towns, structural damage in engineered buildings was observed but was concentrated in low-rise buildings and the lower storeys of taller buildings. Table 1 describes the details of the wide variety in the observed structural damage due to the earthquake, as recorded in six towns of Sikkim from where strong ground motions are also available. Based on the severe damage observed in the town of Mangan, a damage intensity level of VIII+ (MMI scale) has been assigned [17]. Similarly, based on the high damage to structures in the towns of Chungthang, Gangtok, Mangan, and Singtam, the four towns were assigned a damage intensity level of VIII [17].

Table 1 Observed damage to buildings

3 Features of Recorded Accelerograms

Table 2 shows the important characteristics of the seismic waves, namely peak ground acceleration (PGA), peak ground velocity (PGV), arias intensity (AI), mean period (Tm), and predominant time (Tp). The maximum PGA of 0.45 g was recorded at Gezing town (in the radial direction): the town was closest to the epicentre, at 50 km. The next highest PGA value of 0.39 g (in the radial direction) was recorded at Mangan town: the town is also situated at an epicentral distance of 50 km. Despite the distance being similar, the difference in peak accelerations is due to the soil properties and the propagation effects of the earthquake.

Table 2 Characteristics of the accelerograms

Chungthang town, situated at a distance of 55 km, experienced a PGA of 0.36 g. Gangtok and Melli which are located at distances of 68 km and 78 km, respectively, recorded the same PGA of 0.28 g. The town of Singtam, which is situated at a distance of 67 km, recorded a slightly lower horizontal PGA of 0.21 g. Velocity time series are derived from the accelerograms, after applying baseline filtering (Butterworth, Bandpass) of cut frequency 0.2–80 Hz and order 4 along with second-order baseline correction. The maximum PGV of 8.42 cm/s was observed at the Gezing station in the radial direction. PGVs (in the radial direction) for Chungthang, Gangtok, Mangan, Melli, and Singtam were 5.69 cm/s, 5.47 cm/s, 4.81 cm/s, 4.68 cm/s and 3.31 cm/s, respectively. Similarly, in the transverse direction, the maximum PGV of 6.44 cm/s was observed at Gangtok, followed by Mangan at 5.41 cm/s, Gezing at 5.35 cm/s, Chungthang at 4.67 cm/s, Melli at 3.21 cm/s, and Singtam at 2.66 cm/s.

4 Fast Fourier Transforms

Fast Fourier Transforms show the frequency content in a seismic wave. Figures 2, 3, 4, 5, 6, and 7 show the FFTs obtained for the accelerograms from the six stations. It is observed that the frequencies were primarily in the range of 0–20 Hz. The radial and transverse components are designated using the symbols, R and T, respectively.

Fig. 2
Two graphs of Fourier amplitude versus frequency in hertz. The frequency is mainly between 0 and 20. In R, the graph is skewed to the right with a peak at 10, 0.11. In T, the graph is more symmetrical with the peak at 10, 0.08. All values approximated.

FFT for Chungthang: (i) R, (ii) T

Fig. 3
Two graphs of Fourier amplitude versus frequency in hertz. The frequency is mainly between 0 and 20. In R, the graph is skewed to the right with local maxima rising to the peak at 20, 0.18. In T, the graph is more uniform with a peak at 5, 0.17. All values approximated.

FFT for Gangtok: (i) R, (ii) T

Fig. 4
Two graphs of Fourier amplitude versus frequency in hertz. The frequency is mainly between 0 and 20. In R, the graph is quite symmetrical with the peak at 10, 0.1. In T, the graph is more uniform with an amplitude of 0.1. All values approximated.

FFT for Gezing: (i) R, (ii) T

Fig. 5
Two graphs of Fourier amplitude versus frequency in hertz. The frequency is mainly between 0 and 20. In R, the graph is slightly skewed to the left with a peak at 20, 0.13. In T, the graph is more symmetrical with a peak at 15, 0.1. All values approximated.

FFT for Mangan: (i) R, (ii) T

Fig. 6
Two graphs of Fourier amplitude versus frequency in hertz. The frequency is mainly between 0 and 20. In R, the graph is quite symmetrical with a peak at 15, 0.2. In T, the graph is also quite symmetrical with a peak at 15, 0.12. All values approximated.

FFT for Melli: (i) R, (ii) T

Fig. 7
Two graphs of Fourier amplitude versus frequency in hertz. The frequency is mainly between 0 and 20. In R, the graph is skewed to the right with a peak at 10, 0.2. In T, the graph is also skewed to the right with a peak at 10, 0.09. All values approximated.

FFT for Singtham: (i) R, (ii) T

5 Continuous Wavelet Transforms of Strong Ground Motions

Figures 8, 9, 10, 11, 12, and 13 (i, ii) show the CWTs for the six stations. In general, the CWT figures show the presence of high-frequency waves. In all the six towns, the high-amplitude waves contained high-frequency content of around 10 Hz, representing a low period of 0.1 s. However, in some towns, the high-amplitudes waves were observed at multiple frequency ranges. For example, the transverse record from Gezing town shows high-amplitude waves at multiple low periods, ranging from 0.04 to 0.2 s. Similarly, in the radial record from Singtam town, the high magnitude seismic waves are distributed over a low period range of 0.05–0.2 s. The damage due to this horizontal ground motion in Singtam town is also clearly corroborated with the observation from the reconnaissance surveys (rows 16, 17, Table 1). The CWTs of ground motions from Mangan town show high-amplitude waves having a period of 0.1–1.2 s. This is also clearly corroborated with the observed damage to 1 and 2 storey structures in Mangan town, as recorded in the reconnaissance survey (row 14–15, Table 1). Similarly, Gangtok town experienced multiple high-amplitude waves at various multiple frequencies causing damage to 1–4 storey structures (row 5–13, Table 1). In Chungthang town, the high-frequency–high-amplitude content was observed for a very short duration. Further, the response spectra as observed (Fig. 14) exhibit a peaky behaviour, corroborating the above observations. Overall, these observations indicate the high hazard to low-rise structures, prevalent in the region, which typically have low natural periods.

Fig. 8
R and T figures of C W T for Chungthang. The image of period versus time, both in seconds, has a color scale on the side that indicates magnitude.

CWT for Chungthang: (i) R, (ii) T

Fig. 9
R and T figures of C W T for Gangtok. The image of period versus time, both in seconds, has a color scale on the side that indicates magnitude.

CWT for Gangtok: (i) R, (ii) T

Fig. 10
R and T figures of C W T for Gezing. The image of period versus time, both in seconds, has a color scale on the side that indicates magnitude.

CWT for Gezing: (i) R, (ii) T

Fig. 11
R and T figures of C W T for Mangan. The image of period versus time, both in seconds, has a color scale on the side that indicates magnitude.

CWT for Mangan: (i) R, (ii) T

Fig. 12
R and T figures of C W T for Melli. The image of period versus time, both in seconds, has a color scale on the side that indicates magnitude.

CWT for Melli: (i) R, (ii) T

Fig. 13
R and T figures of C W T for Singtam. The image of period versus time, both in seconds, has a color scale on the side that indicates magnitude.

CWT for Singtam: (i) R, (ii) T

Fig. 14
A graph of response acceleration, in g, versus period, in seconds. The radial and transverse graphs of all six stations are highly skewed to the right and have a peak at 0.02 seconds

Response spectra for six stations (R and T components)

6 Summary and Conclusions

A massive Mw 6.9 earthquake struck the Himalayan state of Sikkim in India on 18th September 2011 causing severe damage to the built environment in several towns. In this paper, accelerograms recorded at six stations, which represent information in the time-domain, are analysed to obtain key characteristics, such as PGA, PGV, and AI of each ground motion. The response spectra of the horizontal ground motions from the six stations exhibit high peaks at very low periods and almost reduce to zero at a period of 1 sec. Fast Fourier Transforms (FFTs) of recorded ground motions from the six towns indicate that the frequency content mostly lay in the range of 0–20 Hz. The analysis of Continuous Wavelet Transforms (CWTs) correlates well with the observed damage to low-rise structures recorded in various reconnaissance surveys. It is observed that CWTs convey improved information about the destructive nature of earthquake waves as compared to the time domain or frequency domain representations.