1 Introduction

Chiang Rai Province is a well-known trade and tourist zone in Northern Thailand which experienced an economic boom in the 1980s. Unfortunately, earthquake threats have become a pressing issue in this area. A recent significant earthquake that took place in Northern Thailand on March 24, 2011 was the 6.8 Mw Tarlay Earthquake positioned near the Thailand-Myanmar border. This earthquake resulted in very serious damage, especially in Mae Sai District, Chiang Rai Province, Northern Thailand (Fig. 1), 33 km from the earthquake’s epicentre. The hypocentre was located at 20.705 N, 99.949E with the focal depth of 10 km below ground surface (Phodee et al. 2015). Mae Sai District is also the most northern district in Thailand and a gateway between Thailand and Myanmar. The Tarlay earthquake affected general activities in several neighbouring countries, including Myanmar and Laos PDR (People’s Democratic Republic). In addition, Likitlersuang et al. (2020) reported that the Tarlay Earthquake shaking was also felt by people staying in Bangkok, the capital city of Thailand, which is located about 1000 km from the epicenter.

Fig. 1
figure 1

Modified from Google Earth (2016)

Area of study, epicentre of Tarlay Earthquake, and surrounding active faults.

Full size image

The Thai Meteorological Department (2011) and Soralump and Feungaugsorn (2013) reported that a maximum peak ground acceleration (PGAmax) of about 0.207 g was recorded during the earthquake at the closest seismic station (Mae Sai Seismic Station or MSAA) to the earthquake epicentre. Moreover, this earthquake triggered a soil damage phenomenon called liquefaction. Soralump and Feungaugsorn (2013) noted that this was the first case of observable evidence of earthquake-induced liquefaction found in Thailand since 1943. Through site response analysis during the Tarlay Earthquake, Mase et al. (2018a) mentioned that the Thailand–Myanmar border region is very vulnerable to liquefaction. Mase et al. (2018b) also reported that the Tarlay Earthquake triggered liquefaction at the border of Thailand–Myanmar because of the existence of loose sand layers shallow depth. Meanwhile, structural damage during the Tarlay Earthquake was reported by Ruangrassamee et al. (2012), with the damage dominantly observed on low to medium stories buildings in the vicinity. Similar studies in other areas, such as Flores et al. (1987) and Fischer and Fischer (1998) also reported that the damage could be worse due to the interaction between site effect and propagated wave during the Mexico Earthquake and Northridge Earthquake. This phenomenon is also known as resonance effect. In line with the fact finding found during the Tarlay Earthquake, it is worth to know the detail how the characteristics of sites and ground motion is. This research goes on to suggest that the damage found during the earthquake might have been caused by the resonance effect between the site and propagated waves, a phenomenon well-known for its contribution to structural and soil damage in several other areas previously hit by earthquakes.

Several intensive studies related to earthquake characteristic and liquefaction identification are also performed by several local researchers. The studies by Ruangrassamee et al. (2012), Mase et al. (2017, 2018a, b) and had presented the general impacts of the Tarlay Earthquake, focusing on reports of structural damage and liquefaction potential estimated by empirical approaches (Seed and Idriss 1971; Youd and Idriss 2001; Idriss and Boulanger 2006). Mase et al. (2020a) performed a verification of liquefaction potential during the Tarlay Earthquake at the border of Thailand–Myanmar and found that several sites could be potentially experiencing liquefaction at shallow depth. Tanapalungkorn et al. (2020) conducted a statistical analysis to observe the characteristic of earthquake ground motion in Northern Thailand, including the Tarlay Earthquake and suggested NGA-West2 are the most appropriate attenuation models to predict ground motion in Northern Thailand. However, what these previous studies lack are detailed specific description of the local site conditions, and ground motion parameters alongside verification of resonance effects during the Tarlay Earthquake.

This study presents the ground motion parameters and local site observations during the Tarlay Earthquake based on microtremor measurements taken at the closest seismic station to the rupture (Fig. 1). Interpretations of ground motions, such as peak ground motion amplitude, duration, and energy content are presented. The earthquake’s predominant frequency (f0) and amplification ratio (horizontal to vertical spectral ratio or H/V ratio) are also presented. The study then compares the spectral ratio between measured H/V with recorded ground motion for resonance prediction. A comparison of this study’s results to the guideline design spectral acceleration stipulated in the Seismic Design Code of Thailand is presented at the conclusion of this study. In general, the results from this study would give a better understanding of resonance effects during strong earthquakes in Northern Thailand, and provide recommendations to prepare against earthquake impacts in the region.

2 Geological Condition Study Area

The geological map of the area under study is presented in Fig. 2. There are generally three dominant geological formations in the Mae Sai Basin, Northern Thailand. The first one is a fluvial deposit (Qff), which is composed of sediments and coarse materials such as sands, gravels, silts, and clay materials. MSAA seismic station (M-1), and Borehole-2 (BH-2) where core sampling was taken to characterise the Mae Sai Basin, are both located on this fluvial deposit formation. The other two geological formations are a chert deposit (CPk), made up of sandstone, limestone, and shale, and basic igneous rock (PTrv). These formations are generally found in Thailand’s mountainous areas which were formed during the Permian to the Carboniferous ages (Thai Department of Mineral Resources or TDMR 2017).

Fig. 2
figure 2

The geologic map and observed points

Full size image

Geological condition of Northern Thailand had been presented by previous studies, such as Thay et al. (2013), Mase et al. (2018a, b, 2020a, b). Generally, sand layers followed by thin clay layer are found in Northern Thailand. In line with these previous studies, Fig. 3 presents the geological condition of the Mae Sai Basin. Site investigations at BH-2 have determined that the Mae Sai Basin is dominated by sandy soils. Ground water is found at a depth of 1.2 m. At shallow depth, loose sandy soils classified as SP-SM with average The corrected value of standard penetration test or N1(60) about 8 blows/ft and fines content (FC) of 18% are found up to 9 m depth, then followed by silty sands (SM) with average N1(60) about 10 blows/ft, FC = 18% from 9 to 16.5 m. Mixed-gravelly-sandy soils (SM-GM and GP) with average N1(60) ~ 15 blows/ft, FC = 19% lie at 16.5–19 m and Clayey sand (SC) layers with average N1(60) ~ 15 blows/ft, FC = 20% lie at 19–20.5 m. SM and SC layers are again found at 20.5–23.5 m and 23.5–30 m of depth, respectively. The “time averaged shear wave velocity for the first 30 m of depth” (Vs30) at BH-2 was about 200 m/s; this makes the basin “Site Class D” based on the National Earthquake Hazard Reduction Program (NEHRP 1998). Of note are the saturated sand layers at shallow depth, which, as Mase et al. (2018b) reported, were liquefied during the Tarlay Earthquake near the Thailand–Myanmar border.

Fig. 3
figure 3

Typical subsoil condition of Mae Sai Basin

Full size image

3 Theory and Method

3.1 Ambient Noise Measurement Using Microtremor

Lachet and Bard (1994) previously stated that geophysical measurements obtained with microtremor equipment can provide insight for application in seismic ground response between sediment and bedrock. The observation of local site using microtremor was introducted by Kanai and Tanaka (1954). Both Lachet et al. (1996), Bard (2004), El-Hady et al. (2012), Mase et al. (2020b) have utilised microtremor measurements to characterise site conditions; the most popular method of characterisation is to compare horizontal to vertical spectral accelerations, also known as an H/V ratio. The empirical correlation to calculate H/V ratio from microtremor measurement is based on the work of Nakamura (1989). H/V is the horizontal to vertical spectral ratio derived from the fast Fourier’s Transform (FFT) of the ambient noise from microtremor measurements. This parameter is generated from the transfer function of ground motion, which is also capable of predicting the predominant period (T0) and amplification of peak H/V. It is also able to estimate sediment thickness probabilities and be used to describe the impedance contrast. Atakan (2009) has found that there is a consistency between H/V yielded from microtremor measurements and H/V recorded during an earthquake.

In this study, the result of microtremor measurements, or seismographs, are further analysed using the Fast Fourier Transform Method to obtain H/V. Microtremor and earthquake H/V values are then compared to each other, corresponding to frequency. A site’s H/V ratio is estimated based on comparing the shear wave (Vs) on the vertical and horizontal directions, as expressed by the equation:

$$ H /V = \sqrt {\frac{{H_{(EW)}^{2} + H_{(NS)}^{2} }}{{2V^{2} }}} $$
(1)

where H(EW) and H(NS) are the horizontal spectral values in the east–west (EW) and north–south (NS) directions, respectively, and V is the vertical spectral value.

In general, H/V is relatively more stable than raw noise spectra, as confirmed by the works of Lachet et al. (1996) and Koçkar and Akgün (2012). Several researchers have stated that H/V is normally interpreted with the predominant frequency, or f0. However, this interpretation is still debated and frequently reviewed, especially when related to determining the H/V. This is due to the fact that noise caused by human factors and environmental settings could influence the quality of measurements. Other factors, such as Poisson’s ratio (ν) and receiver restriction are also capable of affecting the reliability of H/V (Bonnefoy-Claudet et al. 2006). Due to these nature and human influences, H/V is then not overall clarified. Despite these complications and limitations of the reliability of a local site’s H/V values, Raptakis et al. (2005) states that there is a common perception where f0 values obtained from microtremor measurements are reliable.

Ambient noise was measured with a seismometer, model DATAMARK JU410, equipped with a 24-bit digital acquisition device, fabricated by Hakusan Co.Ltd. This microtremor equipment has 3 accelerometers component, i.e. north–south (NS), east west (EW), and up–down (UD), that were used to observe the strong and weak motions on the investigated site (2008). The microtremor equipment tends to have lower sensitivity than compared with the seismometer sensors (Bard 2004; Koçkar and Akgün 2012). The JU410 model is relatively more sensitive to detecting ambient noise and vibration that are used to determine spectral ratios than other kinds of accelerometer sensors. In addition, electrical noise during measurements were minute and could be neglected. Signal duration was set at 1000 s, and sampling frequency at 100 Hz, based on the suggestions of Koçkar and Akgün (2012). Digitizers were warmed-up for 5 min prior to measurements to avoid problems associated with the low frequency range and to ensure the quality of a digitized waveform. After that, measurements of ambient noise were performed for 30 min (Mase et al. 2020b). In this study, the expected minimum frequency from each measurement is about 1 Hz (Koçkar and Akgün 2012).

In line with the investigation to the effect of ground motion to the sites, the resonance effect between the ground and the earthquake is also aimed to measure in this study. Gosar (2010) stated that the resonance effect can be observed by comparing the predominant frequency of the investigated site’s and the earthquake’s horizontal-to-vertical spectral acceleration ratios, or H/Vs. Equation 2 expresses the relationship between site H/V and earthquake H/V that is used to predict resonance vulnerability:

$$ R = \frac{{\left| {f_{0} Site - f_{0} Earthquake} \right|}}{{f_{0} Earthquake}} \times 100\% $$
(2)

where R is the resonance vulnerability between site and earthquake motion, and f0Site and f0Earthquake are the site’s and earthquake’s predominant frequency, respectively.

3.2 The Importance of Ground Motion Parameters

Bommer and Martínez-Pereira (2000) have stated that ground motion parameters are important for interpreting ground motion characteristics. Misliniyati et al. (2019) also mentioned that the ground motion parameters could be used as one of methods to determine the appropriate soil models in one dimensional seismic ground response analysis. Mase et al. (2018a) assigned the main ground motion parameters as the indicator to determine the suitable attenuation. In line the use of ground motion parameters elaborated in previous studies, it is therefore important to observe ground motion parameters, especially during the strong earthquakes.

The importance of understanding ground motion characteristics lies in identifying potential damage during seismic wave propagation. Kramer (1996) listed three main parameters to describe ground motion characteristics: amplitude parameters of vertical array, frequency content, and shaking duration. Seismosoft (2018) explains that ground motion parameters have to be first identified to obtain the specific characteristics of ground motion. Seismosoft (2018) also proposed additional ground motion parameters which can help engineers to understand the characteristic of ground motion; these parameters are maximum acceleration (PGAmax), maximum velocity (PGVmax), maximum displacement (PGDmax), time values of PGAmax, PGVmax, PGDmax, the ratio Vmax/Amax, root-mean-square (RMS) values of acceleration, velocity, displacement, Arias intensity, characteristic intensity, specific energy density, intensity of spectrum responses, Housner intensity, effective duration, and energy flux. One of these parameters, the Arias intensity, is usually recorded as the level of acceleration containing up to 95% of detected Arias intensities, or A95. The significance of the Arias intensity was noted by Arias (1970) for its usefulness in predicting the significance of shaking duration from each ground motion. Several of these parameters are analysed in this study as they were recorded by the MSAA station during the Tarlay Earthquake.

4 Results and Discussion

4.1 Ground Motion Parameters of Tarlay Earthquake

The peak ground accelerations on each direction during the Tarlay Earthquake are interpreted in Fig. 4. There are three peak ground acceleration records representing the ground motion on the directions of EW (East–West), NS (North–South), and UD (Vertical). All records show that peak ground acceleration on a defined direction range from 0.114 to 0.207 g, with the largest value coming from the horizontal EW ground motion. According to Mase et al. (2018a), this ground motion falls into the typical range as recorded at rock sites. Other ground motion parameters recorded by the MSAA station which were analysed are presented in Table 1. Parameters such as maximum value, time of maximum value, energy, ground motion intensity, and root mean square (RMS) value are observed. Based on analysis, ground motion along the NS direction has the largest maximum peak ground velocity (PGVmax) at 101.299 cm/s. The maximum time of PGVmax was equal in all directions at 59.990 s. The largest maximum peak ground displacement (PGDmax) is 5.111 m in the NS-direction. The time of PGAmax during the Tarlay Earthquake ranges from 14.065 to 19.955 s. In regard to energy, the largest recorded specific energy density was in the NS-direction.

Fig. 4
figure 4

Ground motion records in: a EW direction b NS direction c UD direction (TMD 2011)

Full size image
Table 1 Analysed ground motion parameters of Tarlay Earthquake recorded at MSAA
Full size table

The Arias intensity in each direction during the Tarlay Earthquake is presented in Fig. 5, as are the recorded parameters of energy flux and significant duration. As seen in Fig. 5, the Arias intensity in each direction significantly increased at 10 s and continued to climb to a peak at 35 s. Energy flux increases more gradually for the first 52 s in each direction of ground motion, before significantly increasing up to 60 s. For maximum values of energy flux in each ground motion direction ranges from 1800 to 35,000 cm2/s. A significant duration is defined as the time duration between when 5% of Arias intensity (t5%) and 95% of Arias intensity (t95%) are reached. From Fig. 5, it can be seen that the significant duration of the Tarlay Earthquake ranges from 14 to 21 s. In general, the NS direction had the largest energy flux and significant duration.

Fig. 5
figure 5

Ground motion records in: a EW direction b NS direction c UD direction

Full size image

4.2 Resonance of Tarlay Earthquake and Spectral Acceleration Comparison

Figure 6 presents the comparison of H/V ratios between ground and earthquake motions at M-1. Generally, H/V from the Tarlay Earthquake and from microtremor measurements are consistent and similar with each other, and thus can be compared to one another. From Fig. 6, it can be seen that the predominant frequency (f0) of ground motion during the Tarlay Earthquake from MSAA was 1.82 Hz, and from M-1 was 2.1 Hz, which are relatively similar. The predominant period (T0) of the Tarlay Earthquake was 0.46 s, and of the M-1 site was 0.55 s. In other words, resonance effects are plausible within this frequency range. These results also imply that the M-1 site tends to suffer from more impactful shaking during earthquake events since the site vibrates along with wave propagation. Based on the estimation from Eq. 2, the resonance value between the local ground and the Tarlay Earthquake’s motions at M-1 is about 15%, which is categorised as a medium-high level of resonance vulnerability.

Fig. 6
figure 6

Comparison of predominant frequency between ground structure and Tarlay Earthquake ground motion

Full size image

The Thai Design Seismic–Seismic Design Code of Thailand (2009) outlines design spectral accelerations based on a 2% probability of exceedance or return period of 2475 years, and a 10% probability of exceedance or return period of 475 years. Figure 7 compares both design spectral accelerations outlined in Thai Design Seismic to the spectral accelerations recorded at MSAA during the Tarlay Earthquake in each direction (NS, EW, and Vertical). It can be seen in Fig. 7 that a maximum spectral acceleration of 0.72 g was recorded during the Tarlay Earthquake, which exceeds the design spectral acceleration based on 10% probability of exceedance. This maximum spectral acceleration was recorded at bedrock, since the MSAA is located on a rock site (Mase et al. 2018a). Because this maximum spectral acceleration was at recorded at the bedrock where the MSAA is located, it is expected that the spectral acceleration at ground surface may be even larger due to amplification effects, which would yield even worse damage. Figure 7 also shows that during the resonance period (0.4 to 0.6 s), spectral acceleration ranged from 0.28 to 0.33 g; because this spectral acceleration value range was recorded during the resonance period, it can be interpreted as the dominant spectral acceleration which buildings were subjected to during the Tarlay Earthquake. It is customary to simplify a building’s natural period as Tn = 0.1n, where n is the number stories in the building. Therefore, a period of 0.4 to 0.6 s can be described as a natural period for low to medium rise buildings 4 to 6 floors high. Generally, the spectral acceleration of the Tarlay Earthquake was slightly larger in the NS direction than other directions; this finding agrees with Ornthammarath (2013) and Mase et al. (2018b), which stated that the NS directional ground motion of the Tarlay Earthquake recorded at the MSAA most accurately reflected possible earthquake impact due to ground displacement.

Fig. 7
figure 7

Comparison of recorded ground motion spectral accelerations during the Tarlay Earthquake to Thai Design Seismic (2009) for earthquake resistance buildings

Full size image

4.3 Ground Motion Effect to Liquefaction at Investigated Site

Ground motion interpretations also explained that PGA on each direction has exceeded the threshold of peak ground acceleration to yield liquefaction, i.e. 0.1 g (Kramer 1996). It seems to be the reason why liquefaction could occur in the region, as reported by Mase et al. (2018b, 2020a). Resonance between ground motion and site could enhance ground motion, particularly once it is approaching ground surface (Paudyal et al. 2012). Yang et al. (2002) mentioned that vertical ground motion could be significantly amplified at ground surface, but horizontal acceleration could not, especially during strong shaking. However, once ground motions on both components take place together (mostly in real cases), the increase of plasticity can happen and it can lead to liquefaction (Tsaparli et al. 2016). It may be caused by the combination between amplification and reduction of effective confining pressure. Horizontal component significantly contribute in pore pressure build-up and vertical component contribute in resulting large acceleration through the layers due to the amplifcation effect.

In line with the observations, it can be observed that ground motions of Tarlay Eartqhuake are relatively large, i.e. 0.2 g. Several studies (Mase et al. 2018b, 2020a, b) considering horizontal component of Tarlay Earthquake ground motion to observe liquefaction potential had confirmed liquefaction in Northern Thailand. Based on the observation, both components show peak values of large horizontal ground motion at base (beyond 0.2 g). It is in line with Yang et al. (2002) who mentioned that horizontal component could significantly induce liquefaction at saturated sandy soils. In the study area, the similar effect of horizontal component could also contribute to trigger pore pressure build-up in saturated sandy soils. The resonance effect also plays important role in yielding liquefaction in the study area. The resonance effect due to vertical ground motion could lead to liquefaction earlier (Tsaparli et al. 2017). Mase et al. (2018b, 2020a, b) showed that liquefaction at the border of Thailand–Myanmar during the Tarlay Earthquake first occurred within first 12 s (the duration of shaking is 60 s) in which peak input motion at the period is about 0.1 g. It means that liquefaction could occur before maximum peak ground acceleration taking place (see Table 1). Therefore, it can be concluded that liquefaction in the region during the Tarlay Earthquake had been induced by the combination effect between both vertical and horizontal ground motions.

5 Conclusion

This study presents the analysis of ground motion parameters and observation of resonance effect during the strong earthquake in Northern Thailand. Several research, such as the dominant ground motion during the earthquake and the tendency of site comparison in identification of resonance are presented. It is identified that during the Tarlay Earthquake of March 2011, the NS component of ground motion recorded at the MSAA is the most impactful direction of ground motion. A large peak amplitude of acceleration, a longer significant duration, and the maximum shaking energy all verify that during the Tarlay Earthquake, the NS component of ground motion played a more significant role than the other directional ground motion components. Results also reveal that soil damage during the Tarlay Earthquake probably more seriously impacted low to medium high-rise buildings. The implementation of H/V ratio is successfully implemented to determine the possibility of resonance effect during the strong earthquake. Resonance effects are suspected to have more contribution to the damage of structural buildings during the Tarlay Earthquake along the Northern Border of Thailand with Myanmar. The results of this study have successfully interpreted the ground motion parameters and the effect of resonance between site and ground motion. Both horizontal and vertical components are predicted to contribute in trigering liquqfeaction in the region. The results of this study would suggest to investigate the study area by performing seismic response analysis due to the NS component of ground motion, specifically and investigate the role of vertical component of ground motion in triggering liquefaction in the study area. In addition, the results also would suggest the local engineers to consider earthquake impact in Northern Thailand. Finally, these results are expected to provide a better understanding of earthquake characteristics in Northern Thailand.