Introduction

Bengkulu City, the capital of Bengkulu Province, which frequently undergoes seismic hazard, is one of the developing areas in Indonesia. Mase [1, 2] mentioned that the City is very vulnerable to undergo earthquake impact and liquefaction. Under these complex circumstances, Farid and Mase [3] suggested that the local government needs to revise spatial plan every 5 years and consider the mitigation of seismic hazards for city development. This needs to be addressed to reduce the impact of natural disasters, which may become worse in the future [4].

The downstream of Muara Bangkahulu River, known as Sungai Serut District, has grown significantly to be one of the prospective areas in Bengkulu City (Fig. 1). Puteri et al. [5] predicted a high distribution pattern for the population of Bengkulu City in 2032, especially in Sungai Serut District. Furthermore, Puteri et al. [5] mentioned that the local government needs to take action in response to this issue by strengthening the regulation of land use. Moreover, Farid and Mase [3] recommended the development of spatial plan based on seismic hazard mitigation. Before developing a policy to cover the issue, a study of the environmental condition on several specific areas needs to be first prioritised. For instance, the geophysical characteristic of subsoil’s condition and site classification need to be considered. This will help the local government to completely understand how to define the vulnerability level of the hazard.

Fig. 1
figure 1

Study area and seismotectonic setting of Bengkulu Province

Full size image

Several studies related to the investigation of local sites in some areas have been presented. Kanli et al. [6], Long and Donohue [7], Sitharam and Anbazhagan [8], and Chakraborrty et al. [9] had implemented an active method using multichannel analysis of surface wave (MASW) to observe several areas, such as Dinar (Turkey), New Delhi (India), Oslo (Norwegia), and Jaipur (India). Furthermore, the ambient noise of microtremor, which is another well-known method, was implemented in various research, such as Mase et al. [10], Koçkar and Akgün [11], El-Hady et al. [12] to investigate Mao Lao (Thailand), Ankara (Turkey), and Marsa Alam site (Egypt). Generally, these studies concluded that both MASW and the ambient noise of microtremor are reliable methods for the investigation of local sites. Consequently, this study presents local sites investigation on the Downstream Area of Muara Bangkahulu River, Bengkulu City, Indonesia. Several geophysical measurements, including MASW and the ambient noise of microtremor, were used to determine shear wave velocity (Vs) profile in the study area. Furthermore, time-averaged shear wave velocity for the first 30 m depth (Vs30) was presented along with site classification based on National Earthquake Hazards Reduction Program (NEHRP) [13]. Generally, the results are able to describe the condition of local site on the downstream area of Muara Bangkahulu River. Furthermore, it serves as a reference for local government in updating spatial plan, which is based on seismic hazard mitigation in Bengkulu City.

Background of Seismic Hazards in Bengkulu City

The layout of the study area as presented in Fig. 1 shows that several earthquake sources surrounding Bengkulu City, such as Sumatra Subduction, Mentawai Fault, and Sumatra Fault. Within 20 years, at least two strong earthquakes had occurred in Bengkulu City, i.e. the Mw 7.9, which occurred on 4 June 2000 and the Mw 8.6 of 12 September 2007. They are later known as the Bengkulu-Enggano and the Bengkulu-Mentawai Earthquake, respectively. Furthermore, Farid and Mase [3] and Hausler and Anderson [14] reported that liquefaction was found during these strong earthquakes. Farid and Mase [3] also mentioned that several areas along the downstream of Muara Bangkahulu River experienced seismic impacts, such as ground failures and liquefactions during the Bengkulu-Enggano Earthquake.

Theory and Method

Study Area

The geological map of Bengkulu City is presented in Fig. 2, and its condition is composed of several formations, such as bintunan (QTb), alluvium (Qa), reef limestone (Ql), swamp deposit (Qs), alluvium terrace (Qat), and andesit (Tpan) [15]. The study area is indicated by yellow-dashed rectangular shape and it is generally composed of three dominant geological formations, namely Qat, Qa, and QTb. They are dominated by boulder, gravel, sand, silt, mud, and clay, which are mostly categorised as uncompacted materials and are sensitively scraped away by river stream [16]. Furthermore, Tsukamoto et al. [17] mentioned that saturated loose sands along river bank on the downstream area are relatively sensitive to undergo liquefaction. In line with the characteristic of dominant materials along the downstream of Muara Bangkahulu River, it is roughly predicted that this area may become liquefied during earthquakes.

Fig. 2
figure 2

Geologic map of Bengkulu City (modified from National Agency of Natural Disasters or BPBD [15])

Full size image

The layout of the investigation points is presented in Fig. 3. In this study, MASW and microtremor measurements are performed to investigate geophysical characteristics. In addition, the geotechnical exploration was performed using cone penetration test (CPT). Generally, the study focused on Sungai Serut District, which is the subwatershed of Muara Bangkahulu Hilir. Since the area is also relatively close to the downtown, many settlements in Bengkulu City are found along the subwatershed of Muara Bangkahulu Hilir [18].

Fig. 3
figure 3

Study area and site investigation locations

Full size image

Geophysical Measurements

The Passive Method

The passive method using microtremor is widely known as one of the cheapest for site investigation [10]. Its implementation has been presented by several researchers, such as Lachet et al. [19], Bard [20], and El-Hady et al. [12]. It was found that the main observation result is the reflection of site condition, which is interpreted into spectral ratio of horizontal motion. Kanai and Tanaka [21] introduced the technique of estimating the spectral ratio based on the observation of microtremor. This technique was later popularised by Nakamura [22], especially for geotechnical and geophysical investigations. Furthermore, Atakan [23] suggested that the spectral ratio generated by the microtremor measurement is in line with the earthquake record. Also, Lachet and Bard [24] mentioned that this method is reliable in predicting predominant frequency. The performance of spectral ratio technique has been confirmed in various research. Lachet et al. [19] and Kockar and Akgun [11] mentioned that the spectral ratios from the measurement of microtremor were more stable than the raw noise spectra. However, several effects such as human noise, the sensitivity of equipment, and environmental conditions have an effect on the result of measurement [10]. Furthermore, Bonnefoy-Clauded et al. [25] and Raptakis et al. [26] suggested that spectral ratio technique is still widely used, especially to inspect geophysical parameters.

The Active Method

Park et al. [27] introduced the active method, which is known as MASW. It was developed based on spectral analysis of surface wave (SASW) method, which is originally proposed by Nazarian et al. [28]. Furthermore, Park et al. [27] explained that MASW is useful to identify and isolate noise in accordance with the trace-by-trace coherency for arrival time and amplitude. Moreover, the difference of travel length is used to estimate the thickness of soil layer. Eikmeier et al. [29] mentioned that there are three main stages in MASW method, which include acquisition, processing, and inversion. The acquisition process is first performed to produce energy. A sledge or drop hammer is used to generate propagated seismic wave, which is then recorded by 24 geophones. The next step is processing, where the noise from field measurement is transferred into dispersion curve, which presents phase velocities versus frequency. Subsequently, the inversion technique is performed to define the best velocity model. Then, the method of stiffness matrix for layered system proposed by Kausel and Roesset [30] is employed to determine Vs profile. This process is completed once the theoretical dispersion curve is consistent with the measured. Afterwards, the matched dispersion curve is then transferred into Vs profile.

Site Classification

National Earthquake Hazard Reduction Provisions (NEHRP) [12] suggested that Vs30 is useful as the indicator to determine site classification. This system is used to determine the site-dependent seismic coefficients for earthquake–resistant design [12]. Its formulation is expressed below:

$$ V_{\text{s30}} = \frac{30(m)}{{\sum\nolimits_{i = 1}^{n} {\frac{{d_{i} }}{{V_{\text{si}} }}} }} $$
(1)

where di is the thickness of each soil layer, Vsi is shear wave velocity and n is the total number of soil layers considered up to the first 30 m depth.

Vs30 is generally used for specific purposes including site characterisation and response analysis. The implementation of this system is also related to ground motion prediction (GMP). Also, the updated GMP equations known as Next Generation Attenuation (NGA) models [31,32,33,34,35] included Vs30. In line with the benefit of Vs30, the prediction of Vs is very important for many implementations in geotechnical earthquake engineering.

Results and Discussion

Measurement Results

The results of geophysical measurements were further analysed to generate Vs profile. In this study, some representative results at A4, A5, A20, and A22, are presented in Fig. 4. A4 is located at the southern side of Muara Bangkahu River (the estuary area) and A5 at the western site. Also, A20 and A22 are located at the central side. Generally, the geological condition of these sites is relatively similar. As presented in Fig. 2, the study area is composed of Qa and Qat formations, where granular and sediment materials are dominantly found. Also, silty sand (SM) is found at shallow depth. Furthermore, the thickness is observed to vary from 1 to 14 m, with Vs of about 160 to 240 m/s. Other sand layers exist under SM, such as clayey (SC) and dense (SW), which had Vs ranging from 320 to 380 m/s. Sandy gravel (GS) is also found on certain sites, such as A5 (at depth of 24 to 30 m) with Vs of about 380 m/s/. The clay layers are generally found at A22, especially at a depth ranging from 0 to 8 m and 26 to 30 m. Generally, A4, A20, and A22 have Vs ranging from 240 to 300 m/s, therefore, these sites are categorised as Class D. Meanwhile, A5 is indicated as Class C with Vs30 of about 384 m/s.

Fig. 4
figure 4

Examples of measurement results a Site A4, b Site A5, c Site A20, and d Site A22

Full size image

Vs30 Distribution

Figure 5 presents Vs30 map for the study area, which is divided into five ranges namely Vs30 of 180–240 m/s, Vs30 of 240 to 300 m/s, Vs30 of 300 to 360 m/s, Vs30 of 360 to 420 m/s, and Vs30 of 420 to 480 m/s. Generally, these ranges of Vs30 are derived from NEHRP [12]. However, to observe a more detailed range, the Vs30 from NEHRP is broken down every 60 m/s. This simplified procedure was adopted by several studies, such as Silva et al. [36], Thompson and Wald [37], and Cannon and Dutta [38]. There are three dominant ranges in the study area. The first, which is Vs30 of 240–300 m/s, is generally found on the middle to the western part. The second and the third, which are Vs30 of 360 to 420 m/s and Vs30 of 420 to 480 m/s, respectively are found on the eastern part of the downstream area of Muara Bangkahulu River. Based on the geological map in Fig. 2, it is observed that the dominant areas are composed of alluvium terrace and deposit. These formations are made up of boulders, sands, silts, clays, and gravels, which are either loose or stiff materials with low-to-high soil resistance. Some areas which have Vs30 of 420 to 480 m/s are composed of bintunan formation, which consists of very dense soils and soft rock materials with high density as indicated by Vs. such as conglomerate, breccia, and clay stone. There are also two small parts on middle and western parts that have Vs30 range of about 300–360 m/s and 180–240 m/s, respectively. The area with Vs30 of 320 to 360 m/s is composed of alluvium terrace and deposit, whereas Vs30 of 180 to 240 m/is composed of alluvium formation. In general, the distribution of Vs30 is relatively consistent with geological formation in the study area, in which a large value indicates a large soil resistance and vice versa.

Fig. 5
figure 5

Map of Vs30 distribution in Sungai Serut District

Full size image

Site Classification Zonation

Figure 6 presents site classification zone for the study area, which is generated based on NEHRP [12]. The criteria had been widely used to characterise the condition of local sites for seismic ground response [39]. Furthermore, Vs30 was implemented to predict ground motion [40]. In general, the study area consists of two main site classes, namely C and D. Class D represents areas categorised as stiff soils and C as very dense soils or soft rocks. In line with Fig. 6, Site Class C area is concentrated in middle to eastern part, whereas Site Class D is in the middle to western part. The interpretation also reveals that high-terrain area (eastern part) tends to have stiffer soil than the low (western part).

Fig. 6
figure 6

Site Classification Map of Sungai Serut District

Full size image

Furthermore, Wills et al. [41] mentioned that D is generally composed of holocene alluvial deposits, whereas C consists of cretaceous sedimentary rocks and coarse-grained materials. In line with the geological condition of Bengkulu City, the statement of Wills et al. [41] seems to be consistent with the finding, i.e. Site Class D is generally composed of Qa and Qat. These formations are made up of several uncompacted materials, such as loose sands, silts, and soft clays. Along the downstream area of Muara Bangkahulu River, levee units are mostly dominated by fine grain sediments which have low Vs30. Therefore, soil resistances are relatively low. Conversely, Site Class C is generally composed of QTb and Qa formations which have a higher soil resistance. In addition, Kockar et al. [42] mentioned that C and D are composed of fluvial deposits, such as alluvial. Therefore, C is also found on alluvium deposits (Qa). Thitimakorn and Chanoo [43] mentioned that a site with lower Vs30 experiences more ground shaking than those with higher Vs30. This indicates that the study area undergoes ground shaking impact during earthquakes. It is also concluded that the sites located at low terrain of downstream area experiences more seismic damage. Overall, the results imply that the low terrain area is relatively more vulnerable to seismic hazard. Furthermore, environmental settings and soil characteristics contribute to the control of the possible impact of seismic hazards in the study area. However, as elaborated by Putrie et al. [5], this area is still becoming the most preferred residential location in Bengkulu City.

Potential Seismic Impact

The response framework of the nonlinear seismic ground proposed by Elgamal et al. [44] is implemented to investigate soil behaviour at representative sites (A4, A5, A20, and A22). According to Misliniyati et al. [45], this method is also known as nonlinear effective stress model, which is originally derived from multi-yield surface plasticity that emphasises permanent shear strain. The benefit of this model lies on the evaluation of stiffness on each incremental step. Furthermore, the nonlinearity is simulated by plasticity increment, which is able to calculate the estimation of permanent deformation and generate soil behaviour. Several researchers, including Tonuk and Ansal [46], Pender et al. [47], and Vivek and Mohanty [48], mentioned that the method is reliable in estimating soil behaviour during earthquakes. Also, Likitlersuang et al. [49] stated that the framework of nonlinear seismic ground response is implemented to investigate soil response during remote earthquakes. Mase et al. [50] revealed that the soil deformations and behaviours from nonlinear effective stress model are generally consistent with field evidences from the 2011 Tarlay Earthquake in Northern Thailand. In this study, the ground motion of Bengkulu-Mentawai Earthquake in 2007, which is noted as the most significant in the city from Mase [1], was used as input motion. Therefore, it is realistic to consider the most significant earthquake for the ground response analysis in the study area.

Figures 7, 8, 9, and 10 present the results of ground response analysis using a nonlinear effective stress model. Furthermore, the response profile during seismic wave propagation on each site is presented. Also, three observed parameters were presented including maximum acceleration, pore pressure, and relative displacement. Generally, several sites, including A4 (Fig. 7), A20 (Fig. 9), and A22 (Fig. 10) show that excess pore exceeded effective confining pressure, especially for silty sand (SM) layer at shallow depth. However, when this happened on a soil mass, it will result in liquefaction [50]. Mase et al. [51] also mentioned that the mixture of sandy and fine soils (SM and SC) also undergo liquefaction. Therefore, it was predicted that these sites may undergo liquefaction during the Bengkulu-Mentawai Earthquake in 2007. Furthermore, several sites, such as A5 (Fig. 8) and A20 may also undergo amplification during seismic wave propagation. For others (A4 and A22), motion tends to decrease at the ground surface. This implies that there is deamplification phenomenon. Also, maximum acceleration is generally observed to vary from 0.1 g to 0.3 g at ground surface. According to Kramer [52], a minimum acceleration of 0.1 g is required to trigger liquefaction. Perhaps, this is why liquefaction occurred at several observed sites during the Bengkulu-Mentawai Earthquake in 2007. Also, the maximum relative displacement at the ground surface is observed to vary from 2 cm to 6 cm.

Fig. 7
figure 7

Ground response profiles at A4 a excess pore pressure b maximum acceleration, and c relative displacement

Full size image
Fig. 8
figure 8

Ground response profiles at A5 a excess pore pressure b maximum acceleration, and c relative displacement

Full size image
Fig. 9
figure 9

Ground response profiles at A20 a excess pore pressure b maximum acceleration, and c relative displacement

Full size image
Fig. 10
figure 10

Ground response profiles at A22 a excess pore pressure b maximum acceleration, and c relative displacement

Full size image

To observe soil behaviour on each observed site, some certain points indicated by red circles in Figs. 7, 8, 9, and 10 were selected based on the depth at which the excess pore water has exceeded the effective confining pressure. For A4, the representative point is at depth of 7.06 m, which is the mid-point of SM layer. For A5, the point is at depth of 1 m, A20 is at 0.54 m (mid-point of SM layer), and A22 at 10.02 m (mid-point of SM layer). Furthermore, soil behaviours, such as time history of acceleration, time history of excess pore pressure ratio (ru), hysteresis loop (shear stress-shear strain), and effective stress path on representative points as presented in Figs. 11, 12, 13, and 14. It was also observed that A5 (Fig. 12) and A20 (Fig. 13) tend to undergo amplification. On the other hand, A4 (Fig. 11) and A22 (Fig. 14) tend to undergo deamplification. In addition, liquefaction was confirmed at various sites including A4 and A20 during seismic wave propagation. This was proven by the excess pore pressure ratio (ru) that exceeded liquefaction threshold (ru ≈ 1). Based on soil profiles presented in Figs. 7, 8, 9, and 10, SM layers on both sites are indicated to undergo liquefaction. For others (A5 and A22), there is no liquefaction indication (ru≤ 1). Figures 11, 12, 13, and 14 also reveal that hysteresis loop (shear stress-shear strain) varies on each observed point. For example, site A4 and A20 show the irregular hysteresis loop. The curve tends to be more flattered, which indicates a significant reduction in shear modulus due to excess pore water pressure. In addition, the effective confining pressure also decreases during seismic wave propagation. This indicates a loss of shear strength due to excess pore pressure. For A5 and A20, the hysteresis loops show that there is no flattered tendency during seismic wave propagation. Furthermore, effective stress paths show that effective confining pressure is not significantly reduced during seismic wave propagation. The time history of excess pore water pressure ratio also shows that there is no liquefaction indication on both A5 and A20 since the ratio did not exceed the threshold.

Fig. 11
figure 11

Soil behaviour at A4 a time history of acceleration, b time history of excess pore pressure, c shear stress vs shear strain, and d effective stress path

Full size image
Fig. 12
figure 12

Soil behaviour at A5 a time history of acceleration, b time history of excess pore pressure, c shear stress vs shear strain, and d effective stress path

Full size image
Fig. 13
figure 13

Soil behaviour at A20 a time history of acceleration, b time history of excess pore pressure, c shear stress vs shear strain, and d effective stress path

Full size image
Fig. 14
figure 14

Soil behaviour at A22 a time history of acceleration, b time history of excess pore pressure, c shear stress vs shear strain, and d effective stress path

Full size image

Figure 15 presents a comparison of the spectral accelerations on the ground surface to seismic design code of Bengkulu City [53]. It was observed that spectral accelerations on each represented site are relatively lower than the designs. Furthermore, when the structural buildings are built based on the seismic design code, the damage experienced during earthquakes is minimised. Several studies performed by Misliniyati et al. [13] and Hausler and Anderson [14] confirmed that there was moderate to high structural damage found along the downstream area of Muara Bangkahulu City in 2007. However, the concern about soil damage should be prioritised to minimise the impact. This is because the results show a potentially liquefiable layer at shallow depth. Therefore, it is important to perform a more in-depth study focusing on the liquefaction potential along the downstream area of Muara Bangkahulu River.

Fig. 15
figure 15

Spectral accelerations comparison

Full size image

Conclusion

This study presents a local site investigation and ground response analysis on the downstream area of Muara Bangkahulu River, Bengkulu City, Indonesia. Furthermore, geophysical measurements were carried out to interpret the geological condition and geophysical characteristics of the study area. Vs30 and site classification maps were also presented. In addition, the response analysis of seismic ground was conducted to observe soil behaviour and potential damage. In line with the above, the following concluding remarks were drawn:

  1. (a)

    The geophysical interpretation of the downstream area of Muara Bangkahulu River is successfully interpreted. The Vs30 distribution is able to describe the tendency of soil resistance in the study area. Generally, Vs30 on the downstream area is observed to vary from 180 to 420 m/s. It also indicates that the area is dominated by Site Classes C and D. Furthermore, areas with Class D are dominated by alluvium terrace and deposit, whereas C is dominated by bintunan formation. Both site classes also reflect that the study area is composed of stiff to very dense soils and soft rocks.

  2. (b)

    Sands and clays were identified as the main materials in the study area. The main geological formation of the area, which is alluvium terrace, is composed of these materials. Furthermore, dense are generally found on bintunan formation, which are composed of rocks and boulders. Also, the study area is more vulnerable to undergo some seismic impacts, such as ground amplification and liquefaction. The observation of soil behaviour confirmed that loose sandy soils at shallow depth are vulnerable to undergo liquefaction.

  3. (c)

    In terms of the design of earthquake resistance, the seismic code is still reliable to be used for structural design. Generally, the results describe the environmental settings in the study area, which is useful as a recommendation for local government in developing spatial plan based on seismic hazard mitigation in Bengkulu City.