Particle Swarm Optimization-based Inversion of HVSR Measurement for Estimating Sediment Thickness in Paleovolcanoes around Bakauheni

doi:10.21203/rs.3.rs-3568044/v1

Download PDF

Research Article

Particle Swarm Optimization-based Inversion of HVSR Measurement for Estimating Sediment Thickness in Paleovolcanoes around Bakauheni

https://doi.org/10.21203/rs.3.rs-3568044/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

In the geotechnical field, determining the thickness of the sediment layer is very important. The thickness of the sediment layer can provide invaluable information in the planning and design of building structures, infrastructure, and other construction projects. Bakauheni is an area that has many calderas and ancient volcanic deposits from the Pliocene - Holocene era. It is fascinating to study how thick the sediment layers are in the area. We used 64 Horizontal to Vertical Spectral Ratio (HVSR) measurement points to determine the thickness of the sediment layer and how it correlates with the presence of an ancient caldera in the Bakauheni area. Next, to obtain a 1D shear wave velocity model (Vs), an inversion process was carried out using the Particle Swarm Optimization (PSO) algorithm. The inversion results show that the average Vs value in the study area is around ~ 600 m/s. This value is relatively high for a sedimentary layer, which indicates that the sedimentary layer has started to become dense and compact or consists of a layer of weathered bedrock. Based on the results of calculating the average HVSR curve, a natural frequency (f_o) value of 15.12 Hz was obtained. Assuming an average Vs of the sediment layer of 600 m/s, the sediment thickness is estimated at 9.92 meters. This result shows an excellent correlation with the median value of sediment thickness calculated at 64 measurement points, which is 10.55 meters. The minimum and maximum sediment thickness in the study area was 4.39 and 103.57 meters, with an average sediment thickness reaching 18.22 meters. From these results, we conclude that the thickness of the sediment layer in the Bakauheni area ranges from 10–18 meters. The thickest sediment layers (> 30 meters) are associated with the presence of calderas and low topography. It shows that a relatively thick layer of sediment covered the ancient caldera. Based on the results obtained, the HVSR method provides quite good results in determining sediment thickness to identify the presence of a caldera. However, it should be noted that Vs values may vary throughout the study area, depending on the nature and composition of the sedimentary rocks present. Therefore, further interpretation and research are needed to understand more deeply the nature and characteristics of the sediment layers in the study area.

sediment

HVSR

shear wave velocity

caldera

Bakauheni is located in South Lampung Regency, Lampung Province, the southern tip of Sumatra Island. This area continues to experience development, both infrastructure for land and sea transportation crossing areas as well as tourism and harbour city areas. Geologically, this area also has a complex structure, consisting of several local faults and several volcanic rock formations (QTl, Qhv(rb) and Tpv), which covered the entire region from the Pliocene - Holocene era (Mangga et al., 1993; Prasetio et al., 2013; Zaenudin et al., 2022). Moreover, there are several calderas of ancient volcanoes where the relationship between volcanic activity in this area is unclear. However, based on the results of a volcano-stratigraphic and remote sensing analysis, it can be concluded that the Lampung tuff formation (QTl) originated from a large eruption forming the pre-Rajabasa Caldera (Bronto et al., 2012). Based on this complex geological structure, it is necessary to carry out geophysical measurements to find out what the subsurface model is in the Bakauheni area which is located in the paleovolcanoes area.

The main purpose of the present study was to develop subsurface models and the thickness of sediment layers and to determine the presence of calderas in the paleovolcano area. In engineering geology, knowledge about the sediment layer's thickness is a crucial parameter. It is important for construction engineering and plays an essential role in the stability of building foundations. However, near-surface sediments may be complex because they are partly influenced and modified by human activities such as construction and development (Tian et al., 2019). In this paper, we used measurement based on ambient noise records to obtain natural frequency (${f}_{o}$) and shear wave velocity (Vs) structure. This method was first introduced Nogoshi and Igarashi (1970; 1971) and then developed by Nakamura (1989) and is usually known as the Horizontal to Vertical Spectral Ratio (H/V or HVSR).

In recent years, passive seismic methods (HVSR) have gained popularity as an alternative to active seismic sources in weathering zone investigations or sediment thickness and bedrock investigation (Ghione et al., 2023; Maghami et al., 2021; Moon et al., 2019; Tian et al., 2019; Zaenudin et al., 2022). Passive seismic methods gather data from naturally occurring seismic sources, such as earthquakes or ambient noise. Passive seismic methods offer several advantages over active seismic methods. Firstly, they do not require the use of controlled seismic sources, eliminating the need for extensive equipment and associated logistical challenges (Arai & Tokimatsu, 2005; Mucciarelli & Gallipoli, 2001). These advantages make passive seismic methods more cost-effective and suitable for remote or inaccessible study sites. Secondly, passive seismic methods can be conducted in conservation areas without needing high-impact seismic sources. It is crucial in environmentally sensitive areas where the use of active seismic sources may be restricted to minimize disturbance to wildlife and ecosystems (Ghione et al., 2023).

The application of the HVSR method, which originates from ambient noise, can be used to obtain 1D models (i.e., layered) shear wave velocity (Vs) profile (Raptakis & Makra, 2010). It has proven to be a valuable tool in geotechnical engineering for obtaining reliable shear wave velocity profiles, which are essential for seismic hazard analysis and other geotechnical applications (Fäh et al., 2003). To obtain a subsurface Vs profile based on HVSR data, it is necessary to carry out the HVSR curve inversion process. In this paper, we discuss the inversion of the HVSR curve using the particle swarm optimization (PSO) to get Vs profile. The PSO algorithm was chosen because of its advantages, firstly PSO is a global approach which shows that PSO has the ability to find optimal solutions globally. In the context of inversion, this means that PSO can search for solutions that approximate the best overall solution in a complex parameter space. The second reason is that PSO has a relatively fast convergence speed compared to other optimization methods. It means PSO can reach a good solution in fewer iterations. Lastly, PSO is relatively easy to implement and use compared to other optimization methods. The PSO algorithm can be easily modified and adapted to the specific needs of the inversion problem being faced (Farduwin et al., 2021; Farduwin & Yudistira, 2021; Zaenudin et al., 2022). Although PSO has many advantages, it is important to remember that its results depend on proper parameter selection and accurate modelling of the inversion problem. Apart from that, PSO can also be trapped in a local optimum if it is not appropriately managed. Therefore, careful parameter selection and effective parameter adjustment strategies are essential to achieve good results with the PSO inversion method.

The research area is located in Bakauheni, the tip of Sumatra Island, which is close to the Sunda Strait (Fig. 1). Geologically, Bakauheni is located on the Tanjungkarang geological sheet map (Mangga et al., 1993). Bakauheni is a hilly area consisting of deposits resulting from volcanic activity between the Pliocene and Holocene. The oldest rocks (Tpv) in the research area are andesite volcanic rocks, which are Pliocene (Tertiary) in age. On top of it, Lampung Tufa rocks (QTl) of Late Pliocene to Pleistocene age were deposited. Then, even younger, there is a formation of young volcanic deposits (Qhv) originating from Mount Rajabasa (rb), consisting of lava (andesite-basalt), breccia, and tuff deposited from the Late Pleistocene to the Holocene (Prasetio et al., 2013). The youngest rocks are scattered along the coast to the east of the research location and are included in the alluvial (Qa) formation.

The Qa formation in the Bakauheni area comprises cobble, pebble, sand, clay, and peats. The formation Qhvrb or young volcanic deposits are composed of andesite-basalt lava, breccia, and tuff, which are spread locally in the Mount Rajabasa area, south of Kalianda City. The Lampung formation (QTl) is composed of pumiceous tuff, rhyolite tuff, welded tuff tuffit, tuffaceous claystone, and tuffaceous sandstone, which is spread very widely from Bakauheni to the north of Kalianda to the north of the city of Bandar Lampung. Pumiceous tuff is yellowish-grey-white, medium-coarse-grained, poorly sorted, mainly consisting of pumice and rock fragments. Tuff is brownish white, composed of rhyolite, relatively hard, almost circular, partly pumice, somewhat soft, and has a cross-bedding structure, generally composed of dacite. Andesite (Tpv) is dark-grey in colour, consisting of andesite lava as a product of tertiary volcanism, spread to the east to southeast of Rajabasa Volcano (Bronto et al., 2012; Darmawan et al., 2023).

Based on its structure, Sumatra Island has an exciting structure. This structure results from the subduction process of the Indian Ocean Plate boundary beneath the Eurasian Continental Plate. This subduction occurred since the Cenozoic Period convergently forming a subduction zone along the collision boundary (McCaffrey, 2009) and caused a change in the direction of the position of Sumatera Island, which was initially trending west-east to northwest-southeast (Hamilton, 1979). The island of Sumatra has one significant fault, the Semangko Fault, a dextral strike-slip fault that moves obliquely to the northwest (Sieh & Natawidjaja, 2000). This fault extends from Aceh (the northern part of Sumatra) to Semangko Bay (the southern part of Sumatra) as fault segments. During the Neogene, this fault was reactivated with powerful right-hand horizontal movements, which resulted in the orogenic strip being folded and broken. As a result, in this research area, there are also active faults shown by red dotted lines and calderas shown by black dotted lines seen in Fig. 1. The structures that develop in the Bakauheni area are tensile joints with a southwest-southeast direction following the direction of the Semangko Fault.

The Peterjajar Fault is an oblique fault with sinistral movement and an upward vertical component that can be found in the Bakauheni area. The existence of this fault is indicated by the presence of sinistral offsets on several ridges in the Bakauheni area. This fault also extends relatively north-south in the research area. It is considered to have the potential to be a source of earthquakes in this area in the future. Prasetio et al. (2013) state that the Peterjajar Fault is a counterbalance if the Semangko Fault experiences movement.

The area of the South Lampung district consists of volcanic cones, hills, and the Rajabasa plains. The Rajabasa volcanic cone is still active and characterized by solfatara fields, including fumaroles, hot springs, and other geothermal features. The morphology of the Bakauheni area is generally in the form of undulating solid hills in the northern part. Intense undulating hills are composed of hard rock, namely igneous rock. The morphology of volcanic mountains is interpreted as lithology composed of volcanic material (a mixture of breccia and tuff lava), while the plain areas to the east it is alluvial plains and soft sediments (Bronto et al., 2012).

The most exciting part of this area is the morphological presence of ancient calderas, interpreted as the former Bakauheni ancient volcanic complex (Bronto, 2013). The rocks around the Bakauheni caldera complex are dominated by andesite lava rock and welded tuff with fracture and fold structures (Bronto et al., 2012). The widespread occurrence of welded tuff in the Lampung formation is believed to be a pyroclastic product resulting from the destruction of the ancient volcanic caldera of the Bakauheni group (Bronto, 2013; Mangga et al., 1993). However, the pattern of the caldera walls does not appear consistent with the geological rock formations in the area, as reflected in Fig. 1. Therefore, an approach using the HVSR method was taken to look at sediment distribution to identify the accumulation zone for volcanic sediment resulting from caldera volcanism activity in the area. This approach is expected to describe the relationship between the geomorphology of the paleovolcano caldera area and subsurface conditions, primarily based on sediment thickness.

In this study, we acquired HVSR measurement using a 3-component seismometer at 64 measurement points. The recording duration was 30–40 minutes, with a sampling frequency of 100 Hz. The distance between measurement points is around 1 km. However, several points are more than 1 km because the microtremor measurements were forbidden on those locations. It is due to the activities of residents or industrial factories, so we chose a safer location to reduce noise in the data.

3.1 Extraction of HVSR curve

To obtain the HVSR curve, we used the Geopsy software (SESAME, 2004) concerning the standard HVSR processing procedure to obtain a reliable HVSR curve. We use a Butterworth filter of order 2 with a cut-off frequency of about 0.3 Hz to reduce the effect of wind around the measurement location and the effect of drift caused by the tool (A. Mihaylov et al., 2019; D. Mihaylov et al., 2016). We set the minimum and maximum frequencies at 0.5–50 Hz, a window length of 30 s with 25% overlap. Then, we use an anti-trigger algorithm to remove very low amplitude or transient signals (STA = 1s; LTA = 30s) with STA/LTA between 1.0–2.5. We apply the Konno and Ohmachi (1998) smoothing filter with a coefficient value of 20. The purpose of smoothing the HVSR curve is to remove the effect of spike and modulation values on the HVSR curve.

3.2 Inversion process

At the inversion of the HVSR curve, we use Octave software and the particle swarm optimization (PSO) algorithm to estimate the Vs and thickness (h) value of sub-surface layers. This algorithm was used by (Zaenudin et al., 2022) to inverse the HVSR curve and shows better results than the genetic algorithm (GA). Apart from that, PSO is relatively stable and fast toward convergence (Ding et al., 2019; Fernández Martínez et al., 2010; Pallero et al., 2017). PSO is a population-based optimization algorithm inspired by the social behaviour of bird flocks or swarms. It was initially proposed by Kennedy and Eberhart in 1995. In PSO, a group of particles, also known as individuals, represents potential solutions to an optimization problem. Each particle moves through the search space, representing the possible solutions, and tries to find the best solution (also known as the global best) by imitating the social behaviour of birds searching for food.

The movement of each particle is influenced by its own best solution (personal best) and the best solution found by any particle in the swarm (global best). These two pieces of information guide the particles towards better solutions. The personal best represents the best solution a particle has found. In contrast, the global best represents the best solution any particle in the swarm finds. The movement of particles in PSO is determined by two main components: the cognitive component and the social component. The cognitive component represents the particle's tendency to move towards its personal best. In contrast, the social component represents its tendency to move towards the global best found by any particle in the swarm. PSO explores the search space and converges towards the optimal solution by iteratively updating the position and velocity of particles based on these components. The algorithm continues until a stopping criterion is met, such as reaching a maximum number of iterations or achieving a desired level of solution quality (Elhussein, 2021).

First, we have to set the tuning parameter values to get the best results using PSO. We use the inertia moment ($\omega$) = 0.8. This value calculates how the particle's velocity will be changed in each iteration. The higher value (closer to 1) means the particle will have a more consistent velocity and maintain the same direction of movement from the previous iteration. So, the particles will explore a wide search area. Then, we set the value of local acceleration (${a}_{l}$) and global acceleration (${ a}_{g}$) equals to 1.8 and 2. The number of populations generated is 300, and the maximum iteration is up to 100 times. The number of iterations is small because PSO is fast towards convergence.

The number of layers used during the inversion process is five layers. This number refers to research by (Farduwin et al., 2021), which was carried out in Kalianda (near the research area) using Electrical Resistivity Tomography (ERT), which obtained the results of 3 layers to a depth of 30 meters. So, for depths greater than 100 meters, we try to increase the layer to 5 layers. Additionally,Zaenudin et al. (2022) carried out an inversion of the HVSR curve in Kalianda using a total of 5 layers until 120 meters of depth. For the model search space, we set the Vs model in layer 1 at 100–800 m/s, layers 2–2 at 300–1500 m/s, in layers 4–5 at 500–2500 m/s. This search space refers to EC8 soil classification (European Commitee for Standardization, 2004), where soil deposits or soil layers with soft to hard have Vs30 < 180 m/s. We think that the Vs30 still has the same range as that Vs value, so we set the minimum search space below 180 m/s. For the maximum Vs value, we limit it up to 800 m/s. This value refers to the dense sand, gravel, or clay deposits that do not exceed 800 m/s. For deeper layers, we set a higher value because we assume that the velocity value will also increase if the depth increases. So, we set it up to 2500 m/s for the last layer. We use this value because the geological conditions of the research area are the product of volcanic activity, especially andesite lava and compact tuff, which have a velocity value higher than the sediment or soil layer.

The search space model of thickness (h) we set for h₁ and h₂ at 1–20 meters, while for h₃ – h_5, we set at 5–50 meters. The thickness of layers 1 and 2 is made shallower to obtain a thin layer (sediment) near the surface. Furthermore, layers 3–5 are made to a maximum of 50 meters because the volcanic layer is assumed to have a thick layer. The forward function used in this inversion stage is based on the work of (Herak, 2008), which assumes that the one-dimensional model of the Horizontal to Vertical Spectral Ratio (HVSR) curve is solely influenced by the amplification spectra resulting from the vertical propagation of body waves. While some researchers employ the surface wave approach, incorporating factors such as Rayleigh wave ellipticity, more complex modelling involving Rayleigh and Love waves, and the complete wavefield approach, both approaches yield similar but not identical curves. The discrepancy lies primarily in the amplitude value, specifically the H/V ratio (Albarello et al., 2023).

In HVSR inversion modelling, several physical parameters will be identified. However, in this research, the main focus is to determine the parameters Vs and h (layer thickness). Other parameters, such as qp (P-wave attenuation coefficient) and qs (S-wave attenuation coefficient), are considered fixed with values of 30 and 10. In addition, the values of Vp (P-wave velocity) and (ρ) (density) are estimated using Brocher Eq. (2008). It can be defined as:

$${V}_{P}=\text{0,9409}+\text{2,094}{7V}_{s}- {\text{0,8206}V}_{s}^{2}+{\text{0,2683}V}_{s}^{3}+{\text{0,0251}V}_{s}^{4}$$

$$\rho ={\text{1,6612}V}_{p}-{\text{0,4721}V}_{p}^{2}+{\text{0,0671}V}_{p}^{3}-{\text{0,0043}V}_{p}^{4}+{\text{0,000106}V}_{p}^{5}$$

where Vp in km/s and $\rho$ in g/cm³. Eq. 1 is valid for a range of Vs values from 0 to 4.5 km/s. This equation can be used to determine the value of Vp from Vs in HVSR (Albarello et al., 2023).

4.1 Vs to depth profile from PSO inversion result

The HVSR curve can be used to reconstruct a 1D shear wave velocity (Vs) model at the station's location by assuming a 1D structure. The curve provides information about the amplification of seismic waves at different frequencies, which the subsurface structure influences. By analyzing the HVSR curve and comparing it to theoretical models or empirical relationships, it is possible to estimate the shear wave velocity profile at the station's location. This condition allows for the reconstruction of a simplified 1D model of the subsurface, providing insights into the geological layers and their properties.

The results of the 1D inversion modelling that carried out can be seen in Figs. 3 and 4. Figure 3 shows the HVSR observation curve and calculations resulting from the inversion process. The two curves show an excellent correlation. It is characterized by a relatively good fit between the observation curve and calculations. From several HVSR curve samples, it was found that the dominant ${f}_{o}$ is at a frequency of 10–20 Hz. The relatively high ${f}_{o}$ value indicates that the sediment layer in the study area is not very thick (not very deep). The HVSR curve also shows a low H/V (amplification) value (the majority is below 3). It means the research area is relatively safe when seismic activity occurs because there is no significant amplification of seismic waves. Areas that are vulnerable when an earthquake occurs have a high amplification value. This is because the seismic vulnerability index is proportional to the square of the amplification.

Figure 4 displays the subsurface layer model obtained from the HVSR curve inversion. Overall, the inversion results provide a maximum depth profile of up to 120 meters, with an average depth of 100 meters. Within the top 5 meters of the model, the estimated Vs values range from 200 to 300 m/s, suggesting the presence of a soil layer at this depth. Then, at depth > 5 meters, it shows a significant increase in Vs value. This velocity increase reaches 800–900 m/s. Then, the third layer to the last layer shows an increase in velocity ranging from 1000 m/s to 2000 m/s. Specifically, at point M76 in the fifth layer, the velocity in that particular layer can exceed 2000 m/s. Overall, it is assumed that the bedrock layer can be found in the third layer, starting at a depth of 20 meters.

The black lines in Fig. 4 represent the best model obtained during the inversion process. From this figure, it is clear that the PSO algorithm is quickly converging. This is marked by black lines that continue to coincide so that in inversion using PSO, significant iterations are unnecessary. To assess the uncertainty in the model, standard deviation values were calculated for both the thickness and Vs velocity values. These standard deviations were obtained by considering all the models generated during the inversion process. The results obtained show a relatively small standard deviation value. This shows that the model is collected in a point area relatively close to the best model globally.

Overall, the inversion results using the PSO algorithm have provided excellent results. The fast convergence of the PSO algorithm is shown by the overlapping black lines in Fig. 4, indicating that the algorithm quickly finds the optimal solution. It shows there is no need for many iterations when using the PSO inversion method. In addition, the relatively small standard deviation values for thickness and velocity Vs indicate that the models generated in the inversion process cluster closely around the best global model. This shows a high level of confidence in the results obtained. Therefore, it can be said that the PSO inversion results provide outstanding and reliable results.

To understand how local sites are in the Bakauheni area, we modeled the average HVSR curve over the entire measurement area. We modelled the mean Vs profile to the depth. These results can be seen in Fig. 5(a-b). However, it should be noted that this mean model only provides general information and does not consider local variations in the Bakauheni area.

Figure 5(a) shows the mean model HVSR curve over the entire measurement area in the Bakauheni area. In this model, the mean value of the HVSR curve is measured at each frequency and then plotted as a function of frequency. This model creation refers to Nelson & McBride (2019) and Trichandi et al. (2023), who created an average spectrum model of the HVSR curve. The subsurface layers, sediment or weathered layers, and bedrock can be separated from this curve. The boundary between layers is located at the top of the HVSR curve obtained, so it can be said that the location of the ${f}_{o}$ value marks the layer boundary between the sediment or bedrock. In this mean HVSR curve, the ${f}_{o}$ value is obtained at 15.12 Hz. The bedrock layer is characterized by low frequency towards ${f}_{o}$, while the sedimentary layer (weathering zone) is at ${f}_{o}$ to high frequency. By knowing the ${f}_{o}$ value, the depth of the sediment layer (including the weathered layer) can be determined using the following equation:

$${f}_{o}=\frac{{V}_{s}}{4d}$$

where ${f}_{o}$ = natural frequency associated with sediment layer/bedrock, Vs = the mean Vs (shear wave velocity) of sediment, and d = sediment layer thickness.

In Eq. (3), the thickness value of the sediment layer can be known if the mean Vs value in the sediment layer is known. It can be estimated based on the mean Vs profile against depth in Fig. 5(b) to determine the mean Vs value. The figure shows the model of the mean Vs profile against depth in the Bakauheni area. In this model, the mean value of velocity Vs is measured at each depth and then plotted as a function of depth.

Based on the mean Vs curve obtained, it can be seen that the mean Vs curve shows exciting changes in seismic wave velocity in the Bakauheni area. At a depth of up to 5 meters, the mean Vs value obtained is around 600 m/s, which is assumed to be the mean Vs value in the sediment layer. This shows that the material below the surface consists of sedimentary layers with relatively low seismic wave velocities at this depth. Thus, if it is assumed that the mean Vs in the sediment layer is ~ 600 m/s, then based on Eq. (3), the thickness of the sediment layer is 9.92 meters. It shows that the Bakauheni area has a thin sediment layer.

However, at depths > 5 meters, there is a significant increase in the mean Vs value, reaching around ~ 900 m/s at a depth of 20 meters. This indicates significant geological changes beneath the sediment layer, and perhaps a denser or harder rock layer exists in this area. Furthermore, the mean Vs increases slowly at a depth of 20 meters to 70 meters. This suggests a gradual change in the geological properties of the area, perhaps a transition from sedimentary layers to denser rock. The mean Vs appears constant at depths of more than 70 meters to 120 meters. This shows that the subsurface material tends to be homogeneous in this area with relatively stable seismic wave velocities.

Figure 6(a) shows a natural frequency map in the research area, showing the contrast of high and low anomalies. Overall, the study area is dominated by high frequencies, indicating that this area has relatively shallow sediment thickness. The sediment thickness map in Fig. 5b confirms this, which shows low anomalies (blue < 20 meters). High f_o anomaly > 15 Hz is distributed in the SW-middle-NE and southern parts. While low f_o anomalies < 15 Hz dominate in the central, eastern, and NW parts. In general, the f_o anomaly map does not clearly show the existence of a caldera in the research area. However, in areas B-2, B-4, B-5, and B-6, there is a low f_o anomaly in the caldera, indicating that this area has a thick layer of sediment.

Table 1

Statistic of the sediment thickness
min	max	mean	median
4.39	103.57	18.22	10.55

Based on Fig. 6(b), it is obvious that the B-2, B-4, B-5, and B-6 caldera areas have a thickness of > 30 meters (yellow to red). Based on statistics on the thickness of the sediment layer in the research area (Table 1), overall, Bakauheni has an average thickness of 18.22 meters with a median value of 10.55 meters. The median value correlates very well with the average thickness value calculated from the average HVSR curve. On the other hand, the average value calculated at each point shows a more considerable value. The data has a high thickness value (data has a wide range), making the average value more extensive than the median value.

The minimum thickness is 4.39, and the maximum thickness reaches 103.57 meters. This very thick area is located in the eastern part, which is an area with low topography and near the beach (alluvium formation). This result also matches the morphological form of Bakauheni (Fig. 2), which shows that the area with a thick layer of sediment is located at low topography. This shows that areas B-2, B-4, B-5 and B-6 have a lot of sedimentary material.

The research study in Bakauheni has geological formations dominated by volcanic deposits, making it very interesting to carry out measurements using passive seismic methods, especially HVSR. This method is straightforward to measure, does not damage the environment, requires low costs, and can model the subsurface well. Apart from that, applying the HVSR method can be used as a preliminary survey method in subsurface imaging or geophysical exploration. Our main objective in carrying out HVSR measurements at Bakauheni is to determine the sediment layer's thickness and identify the presence of a caldera from an ancient volcano. To achieve these results, we ensure that the HVSR curve measurement results comply with the standards determined by SESAME for a reliable curve and clear peak.

After obtaining the HVSR curve, we perform inversion using the PSO algorithm to reconstruct the 1D Vs model. The results of this inversion will be used to determine the average Vs velocity value in the sediment layer. The average Vs value of ~ 600 m/s obtained from the inversion results is relatively high for the sediment layer. In this case, the Vs value of 600 m/s indicates that the sediment layer in the study area has relatively high density and stiffness. This can indicate the presence of dense sedimentary rock, well-compacted sedimentary layers, or possibly a layer of weathered bedrock. However, it should be noted that Vs values may vary across the study area, depending on the nature and composition of the sedimentary rocks present. Therefore, further interpretation and research are needed to understand the nature and characteristics of the sediment layers in the study area.

In the next stage, we calculated the average HVSR curve throughout the measurement area to obtain an average f_o value to determine the average thickness value of the sediment layer. With the average f_o value obtained at 15.12 Hz, the average sediment layer thickness value in the Bakauheni area is 9.92 meters. This result is quite valid, with the median value of sediment thickness calculated at each measurement point. The median value obtained is 10.55 meters. This value shows a good agreement between the thickness of the sediment layer from the calculation of f_o on the average HVSR curve and the median value of the thickness of the sediment layer calculated at each measurement point. However, based on the average value, the calculation results at each measurement point show a value of 18.22 meters. From these results, we conclude that the average thickness of the sediment layer in the Bakauheni area is around ~ 10–18 meters. Apart from that, the location of the thick sediment layer is very consistent with the low topography and is in a caldera environment. This condition shows that the caldera of an ancient volcano has become a depositional environment for sedimentary material, and HVSR measurements provide excellent results in identifying the caldera through the thickness of the sediment layer.

Albarello, D., Herak, M., Lunedei, E., Paolucci, E., & Tanzini, A. (2023). Simulating H/V spectral ratios (HVSR) of ambient vibrations: a comparison among numerical models. Geophysical Journal International, 234(2). https://doi.org/10.1093/gji/ggad109
Arai, H., & Tokimatsu, K. (2005). S-wave velocity profiling by joint inversion of microtremor dispersion curve and horizontal-to-vertical (H/V) spectrum. Bulletin of the Seismological Society of America, 95(5). https://doi.org/10.1785/0120040243
Brocher, T. M. (2008). Compressional and shear-wave velocity versus depth relations for common rock types in northern California. Bulletin of the Seismological Society of America, 98(2). https://doi.org/10.1785/0120060403
Bronto, S. (2013). Geologi Gunung Api Purba (Second Edition). Badan Geologi - Kementerian Energi dan Sumber Daya Mineral.
Bronto, S., Asmoro, P., Hartono, G., & Sulistiyono, S. (2012). Evolution of Rajabasa Volcano in Kalianda Area and Its Vicinity, South Lampung Regency. Indonesian Journal on Geoscience, 7(1), 11–25. https://doi.org/10.17014/ijog.v7i1.132
Darmawan, I. G. B., Rustadi, & Karyanto. (2023). Hydrothermal fluid flow modeling using numerical simulation to identify potential reservoir zones in the Mount Rajabasa geothermal field. Kuwait Journal of Science. https://doi.org/10.1016/j.kjs.2023.02.025
Ding, Y., Zhang, W., Yu, L., & Lu, K. (2019). The accuracy and efficiency of GA and PSO optimization schemes on estimating reaction kinetic parameters of biomass pyrolysis. Energy, 176. https://doi.org/10.1016/j.energy.2019.04.030
Elhussein, M. (2021). New Inversion Approach for Interpreting Gravity Data Caused by Dipping Faults. Earth and Space Science, 8(2). https://doi.org/10.1029/2020EA001075
European Commitee for Standardization. (2004). Eurocode 8: Design of structures for earthquake resistance - Part 1 : General rules, seismic actions and rules for buildings. In European Committee for Standardization (Vol. 1, Issue English). The European Union Per Regulation 305/2011, Directive 98/34/EC, Directive 2004/18/EC.
Fäh, D., Kind, F., & Giardini, D. (2003). Inversion of local S-wave velocity structures from average H/V ratios, and their use for the estimation of site-effects. Journal of Seismology, 7(4). https://doi.org/10.1023/B:JOSE.0000005712.86058.42
Farduwin, A., Lumbatoruan, P. G., Karyanto, & Triyanto, D. (2021). Identification of zeolite using electrical resistivity tomography in Campang Tiga, South Lampung Regency. IOP Conference Series: Earth and Environmental Science, 882(1). https://doi.org/10.1088/1755-1315/882/1/012046
Farduwin, A., & Yudistira, T. (2021). Shear velocity inversion from ambient seismic noise using RR-PSO: A case study of Nusa Tenggara Island. Journal of Physics: Conference Series, 1949(1). https://doi.org/10.1088/1742-6596/1949/1/012022
Fernández Martínez, J. L., García Gonzalo, E., Fernández Álvarez, J. P., Kuzma, H. A., & Menéndez Pérez, C. O. (2010). PSO: A powerful algorithm to solve geophysical inverse problems Application to a 1D-DC resistivity case. Journal of Applied Geophysics, 71(1). https://doi.org/10.1016/j.jappgeo.2010.02.001
Ghione, F., Köhler, A., Dichiarante, A. M., Aarnes, I., & Oye, V. (2023). Vs30 and depth to bedrock estimates from integrating HVSR measurements and geology-slope approach in the Oslo area, Norway. Frontiers in Earth Science, 11. https://doi.org/10.3389/feart.2023.1242679
Hamilton, W. (1979). Tectonics of the Indonesian Region. In Geological Society of Malaysia, Bulletin (Vol. 6).
Herak, M. (2008). ModelHVSR-A Matlab® tool to model horizontal-to-vertical spectral ratio of ambient noise. Computers and Geosciences, 34(11). https://doi.org/10.1016/j.cageo.2007.07.009
Kennedy, J., & Eberhart, R. (1995). Particle swarm optimization. Proceedings of ICNN’95 - International Conference on Neural Networks, 1942–1948. https://doi.org/10.1109/ICNN.1995.488968
Konno, K., & Ohmachi, T. (1998). Ground-motion characteristics estimated from spectral ratio between horizontal and vertical components of microtremor. Bulletin of the Seismological Society of America, 88(1). https://doi.org/10.1785/bssa0880010228
Maghami, S., Sohrabi-Bidar, A., Bignardi, S., Zarean, A., & Kamalian, M. (2021). Extracting the shear wave velocity structure of deep alluviums of “Qom” Basin (Iran) employing HVSR inversion of microtremor recordings. Journal of Applied Geophysics, 185. https://doi.org/10.1016/j.jappgeo.2020.104246
Mangga, S. A., Amirudin, Suwarti, T., Gafoer, S., & Sidarto. (1993). Geological Map of Tanjungkarang, Sumatra. Bandung: Geological Research and Development Centre.
McCaffrey, R. (2009). The tectonic framework of the sumatran subduction zone. In Annual Review of Earth and Planetary Sciences (Vol. 37). https://doi.org/10.1146/annurev.earth.031208.100212
Mihaylov, A., El Naggar, H., Mihaylov, D., & Dineva, S. (2019). Approximate analytical HVSR curve using multiple band-pass filters and potential applications. Soil Dynamics and Earthquake Engineering, 127. https://doi.org/10.1016/j.soildyn.2019.105840
Mihaylov, D., El Naggar, M. H., & Dineva, S. (2016). Separation of high- and low-level ambient noise for HVSR: Application in city conditions for Greater Toronto area. Bulletin of the Seismological Society of America, 106(5). https://doi.org/10.1785/0120150389
Moon, S. W., Subramaniam, P., Zhang, Y., Vinoth, G., & Ku, T. (2019). Bedrock depth evaluation using microtremor measurement: empirical guidelines at weathered granite formation in Singapore. Journal of Applied Geophysics, 171. https://doi.org/10.1016/j.jappgeo.2019.103866
Mucciarelli, M., & Gallipoli, M. R. (2001). A critical review of 10 years of microtremor HVSR technique. In Bollettino di Geofisica Teorica ed Applicata (Vol. 42, Issues 3–4).
Nakamura, Y. (1989). Method for dynamic characteristics estimation of subsurface using microtremor on the ground surface. Quarterly Report of RTRI (Railway Technical Research Institute) (Japan), 30(1).
Nelson, S., & McBride, J. (2019). Application of HVSR to estimating thickness of laterite weathering profiles in basalt. Earth Surface Processes and Landforms, 44(7). https://doi.org/10.1002/esp.4580
NOGOSHI, M., & IGARASHI, T. (1970). On the Propagation Characteristics of Microtremor. Zisin (Journal of the Seismological Society of Japan. 2nd Ser.), 23(4). https://doi.org/10.4294/zisin1948.23.4_264
NOGOSHI, M., & IGARASHI, T. (1971). On the Amplitude Characteristics of Microtremor (Part 2). Zisin (Journal of the Seismological Society of Japan. 2nd Ser.), 24(1). https://doi.org/10.4294/zisin1948.24.1_26
Pallero, J. L. G., Fernández-Martínez, J. L., Bonvalot, S., & Fudym, O. (2017). 3D gravity inversion and uncertainty assessment of basement relief via Particle Swarm Optimization. Journal of Applied Geophysics, 139. https://doi.org/10.1016/j.jappgeo.2017.02.004
Prasetio, A. D. , Muslim, D. , Marjiyono, M. , & Soehaimi, A. (2013). Peterjajar active fault study at Bakauheni area, South Lampung. Journal of Environment and Geological Hazards, 4(1), 15–31. https://doi.org/http://dx.doi.org/10.34126/jlbg.v4i1.48
Raptakis, D., & Makra, K. (2010). Shear wave velocity structure in western Thessaloniki (Greece) using mainly alternative SPAC method. Soil Dynamics and Earthquake Engineering, 30(4). https://doi.org/10.1016/j.soildyn.2009.10.006
SESAME. (2004). Guidelines for The Implementation of The H/V Spectral Ratio Technique on Ambient Vibrations-Measurements, Processing and Interpretations, SESAME European Research Project. SESAME: Site EffectS Assessment Using Ambient Excitations, March.
Sieh, K., & Natawidjaja, D. (2000). Neotectonics of the Sumatran fault, Indonesia. Journal of Geophysical Research: Solid Earth, 105(B12). https://doi.org/10.1029/2000jb900120
Tian, B., Du, Y., You, Z., & Zhang, R. (2019). Measuring the sediment thickness in urban areas using revised H/V spectral ratio method. Engineering Geology, 260. https://doi.org/10.1016/j.enggeo.2019.105223
Trichandi, R., Bauer, K., Ryberg, T., Wawerzinek, B., Vargas, J. A., von Blanckenburg, F., & Krawczyk, C. M. (2023). Shear-wave velocity imaging of weathered granite in La Campana (Chile) from Bayesian inversion of micro-tremor H/V spectral ratios. Journal of Applied Geophysics. https://doi.org/10.1016/j.jappgeo.2023.105191
Zaenudin, A., Darmawan, I. G. B., Farduwin, A., & Wibowo, R. C. (2022). Shear wave velocity estimation based on the particle swarm optimization method of HVSR curve inversion in Bakauheni district, Indonesia. Turkish Journal of Earth Sciences, 31(5). https://doi.org/10.55730/1300-0985.1815

No competing interests reported.

Download PDF

Editor assigned by journal
07 Nov, 2023
Submission checks completed at journal
07 Nov, 2023
First submitted to journal
06 Nov, 2023

You are reading this latest preprint version

Particle Swarm Optimization-based Inversion of HVSR Measurement for Estimating Sediment Thickness in Paleovolcanoes around Bakauheni

Status:

Version 1

Abstract

Figures

1. Introduction

2. Geologic Setting

3. HVSR acquisition and processing

3.1 Extraction of HVSR curve

3.2 Inversion process

4. Result and Discussions

4.1 Vs to depth profile from PSO inversion result

5. Conclusions

References

Additional Declarations

Status:

Version 1