Comparison of peak ground acceleration using deterministic and probabilistic approach in the coastal area of Loh Buaya, Rinca Island, Indonesia

An area located near an active volcano, subduction zone, and/or active fault has a risk of an earthquake. Earthquakes can cause damage to buildings with a large number of losses. One of the earthquake-related parameters required in planning an earthquake-resistant building is the Peak Ground Acceleration (PGA). The objective of this study is to compare the PGA value at the coastal area of Loh Buaya, Rinca Island, Indonesia, using Probabilistic Seismic Hazard Analysis (PSHA) and Deterministic Seismic Hazard Analysis (DSHA). The DSHA is calculated based on earthquake data series from the United States Geological Survey (USGS) in the period between 1922 to 2022. Also, it is calculated using 10 data of Ground Motion Prediction Equation (GMPE). Data from the USGS is constrained with magnitude > 5.0 MW and earthquake distance < 200 km from the study site. The PSHA is based on the Indonesian earthquake hazard map and the de-aggregation map from the National Earthquake Study Centre of Indonesia with a probability exceeding 2% in 50 years. Based on the parameters of PGA maximum, distance, and magnitude, the DSHA by Kanno method has the closest result to the PSHA. It is concluded that this area has a high earthquake risk with a PGA maximum of more than 0.5 g.


Introduction
Indonesia is one of the countries frequently affected by earthquakes.Earthquakes are disasters that can occur in areas close to volcanoes, subduction zones, and active faults.Shock motion is detected on the Earth's surface due to seismic waves.It caused damage to structures; subsidence and liquefaction, and slope failure; and tsunami [1].In the last two decades, there have been several major earthquakes in Indonesia, including the Banda Aceh earthquake with 9.2 M W in 2004 [2]; Nias with 8.7 M W in 2005 [3]; Pangandaran with 7.7 MW in 2006 [4]; Yogyakarta with 6.3 MW in 2009 [5]; Padang with 7.6 MW in 2009 [6]; Lombok with 7.0 MW in 2018 [7]; Palu with 7.5 MW in 2018 [8]; dan Flores with 7.4 MW in 2021 [9].One of the main earthquake parameters used to measure ground motion strength is Peak Ground Acceleration [10].1298 (2024) 012014 IOP Publishing doi:10.1088/1755-1315/1298/1/012014 2 Peak ground acceleration (PGA) is the maximum ground acceleration that occurred during earthquake shaking at a site.It is equal to the amplitude of the maximum acceleration measured on an accelerogram at a location during an earthquake [11].Determining PGA is an important requirement in planning earthquake-resistant buildings [12].Earthquakes with identic magnitude can offer different results of PGA.Even moderate-magnitude earthquakes can produce larger PGA than larger-magnitude earthquakes [13].
There is no Ground Motion Prediction Equation (GMPE) to calculate PGA in specific locations in Indonesia yet [14].Indonesian seismic hazard analysis was adapted GMPE from other locations with similar geology, tectonics, and earthquake mechanisms [15].Probabilistic Seismic Hazard Analysis (PSHA) method and de-aggregation map from the Indonesian Earthquake Study Center (Pusgen) have been able to provide information on the magnitude of the PGA value at a location, but the information on earthquake magnitude and distance provided is still in the form of a range [16].Knowing the specific magnitude and distance of an earthquake to produce a PGA value can improve accuracy in earthquake risk planning.
This research has two objectives.First, to compare the PGA, magnitude, and distance of earthquakes by Deterministic Seismic Hazard Analysis (DSHA) through existing GMPEs from various countries with PSHA from Pusgen.Second, to find out the factors that make the research location prone to earthquakes.The research is located in Loh Buaya, Rinca Island, Indonesia.The research was conducted at Loh Buaya, Rinca Island, Indonesia.It is expected that this research can provide an overview of which GMPE is closest to the PSHA results and the cause of the earthquake.

Peak Ground Acceleration (PGA)
PGA is one of the variables in estimating ground motion intensity by considering independent variables such as magnitude, source-to-site distance, site class, and fault type [17].While earthquake magnitude is used to estimate the losses caused by an earthquake, PGA is more widely used concerning the damage to buildings caused by the earthquake [14].The value of PGA will be directly proportional to the damage to the building.
There are several methods to determine the PGA value, which include the probabilistic and deterministic approach.The probabilistic approach to determine PGA is also called Probabilistic Seismic Hazard Analysis (PSHA) method whereas deterministic approach is also called Deterministic Seismic Hazard Analysis (DSHA) method.PSHA method in Indonesia can refer to the Indonesian standard in SNI 1726:2019 [12] and has been updated by Pusgen 2022 [16] to determine the dominant magnitude and epicenter distance.Meanwhile, the DSHA method uses Ground Motion Prediction Equation (GMPE) equations based on earthquake records at specific locations to predict PGA.From 1964-2010, 289 GMPEs from all around the world were formulated based on their characteristics of location, geology, fault type, fault mechanism, and location respectively [18].

Probabilistic Seismic Hazard Analysis (PSHA)
Probabilistic approach or PSHA method is a method of analyzing earthquake hazards by considering the uncertainty of location, magnitude, and time of occurrence using the probability theorem.This research uses the calculation of PGA from SNI 1726-2019 which is the determination of PGA.In planning earthquake-resistant buildings using a return scale of 2,500 years or probability exceeding 2% of 50 years [12].The PGA value is obtained from the response spectra (RSA) Cipta Karya website by selecting the site coordinates.The PGA value is the PGA at the bedrock, so it needs to be multiplied by the site class coefficient according to equation (1).Methods for determining the classification of site classes according to (Table 1).
According to [16], De-aggregation Map complements the probabilistic method of RSA Cipta Karya to provide information on representative magnitude (M) and epicenter distance (R) from the site to the reviewed earthquake source.In determining the magnitude and epicenter distance, a logic tree scheme with various GMPEs was used to provide quantification of uncertainties in the model and its input parameters (Table 2).

Deterministic Seismic Hazard Analysis (DSHA)
The calculation of PGA in deterministic approach or DSHA is carried out to calculate the level of ground acceleration that occurs due to earthquakes based on fault mechanism, hypocenter distance, earthquake depth, magnitude, and local soil conditions.GMPE is DSHA method that is formulated based on a collection of seismic data at a location and then approached with the characteristic equation in the form of geological, tectonic conditions and fault mechanism.That are also consideration in the application of GMPE to different locations.Because of limited ground motion records, there is no GMPE in specific sites available in Indonesia [14].The following GMPE is used to predict PGA in the Loh Buaya, Rinca Island: [19].The GMPE equation from Donovan used earthquake data from 214 earthquake records located in San Fernando, California, USA.The proposed equation is according to equation ( 2):

GMPE by Donovan
with y is PGA (gal);   1 is 1080;   2 is 0.5; R is distance; M is magnitude.[20].The GMPE equation from Donovan and Bornstein was used to characterize earthquake records in California with the review of soil types of rock and hard soil.The proposed equation is according to equation ( 3)-( 6):

GMPE by Bornstein and Donovan
with y is PGA (gal);  1 is 2,154,000;  2 is 2.10;  1 is 0.046;  2 is 0.445; R is distance; M is magnitude.Mcguire [21].This GMPE equation used characteristics of 11 earthquake records on rocky ground and 69 earthquake records on alluvial ground.The magnitude under review is M < 6.5.

GMPE by
The proposed equation is according to equation (7): with x is PGA (cm/s2);   1 is 3.40;   2 is 0.89;   3 is -1.17;   4 is -0.20;   is 0 for gravel or 1 for soil; M is magnitude; R is distance.[22].The GMPE equation from Campbell used earthquake data characteristic of the western United States located 50 km from the fault.The magnitudes of the earthquakes reviewed were between 5.0-7.7 M. The earthquakes were shallow earthquakes without considering the influence of subduction.The proposed equation is according to the equation ( 8):

GMPE by Campbell
with M is magnitude and R is distance.[23].The GMPE equation from Youngs used the characteristics of 389 earthquake records from Alaska, Chile, Peru, Japan, Mexico, and Solomon.The proposed equation is according to (9):
= 51.5 0.516 (19) with N is SPT blows.[29].This GMPE equation used earthquake data characteristic of hypocenter depth 3.94 < H < 161 km.Earthquake records are from Taiwan and the addition of subduction earthquakes from Mexico, the western United States, and New Zealand.The proposed equation is according to equation (20).

Geology and seismic review
Rinca Island is located between several active faults: Flores back arc thrust in the north, Sape strikeslip in the west, and the Bondowatu fault in the south.Among the three active faults, the Flores back arc thrust fault is the longest fault.The Flores back arc thrust fault consists of 6 segments stretching from the north of Bali Island to Wetar Island in Maluku Province (Table 3).Flores back arc thrust is a reverse fault, while Bondowatu fault is a normal fault.Flores back arc thrust has the highest slip rate compared to Bondowatu fault and Sape strike-slip, in the range of 7.0-18.3mm/yr.This results in the Flores back arc being more dominant as an earthquake trigger.Several large earthquakes have occurred due to the Flores back arc thrust activity, such as the Lombok earthquake in 2018 with a magnitude of 6.9 MW and with hypocenter depth < 40 km [30], the Flores earthquake on December 12, 1992, with 7.7 MW [31] and the Flores earthquake on December 14, 2021, with 7.4 MW [9].Besides being caused by active faults, Rinca Island is located near the subduction zone between the Australian plate and the Eurasian plate.The subduction zone extends from the western part of Sumatra Island to the southern part of the Nusa Tenggara islands.The NTB and Bali segments are the closest segments to Rinca Island (Table 4).The potential for earthquakes due to plate subduction in both segments is up to 9.0 MW with a historical earthquake magnitude of 7.0 MW recorded in 1977 [32].Loh Buaya Rinca Island is included in an area that has the potential to experience strong earthquakes with an intensity of more than VIII Modified Mercalli Intensity (MMI) or PGA of 0.34-0.65g[33] This intensity can cause ground cracks, landslides, and ground shifts.Based on earthquake data from the USGS, it is found that 330 earthquakes have occurred in the range of 1922-2022 with magnitude limits > 5.0 MW and epicenter distance < 200 km (Figure 1).Earthquakes that have occurred are dominated by shallow earthquakes with a total of 191 events (Table 5).Generally, Shallow earthquakes generate stronger PGA than intermediate and deep earthquakes because the energy is released closer to the surface.

Geotechnical investigation
Geotechnical investigation is required to calculate PGA by PSHA or DSHA.In the PSHA method, it is required to determine the site class classification.While in DSHA as in GMPE Kanno, Geotechnical investigation is required to calculate the correction factor.Standard Penetration Test (SPT) data was used as the basis for finding the VS30.SPT was conducted on soil dominated by sandy soil.SPT results up to 28m obtained average SPT blows are 7.34 and VS30 conversion value using equation ( 18) are 144.06m/s (Figure 2).Based on the following results, the site class classification is SE.This type of Ground conditions can result in PGA amplification so that the impact of shocks due to the earthquake will be more pronounced.In SPT data, the depth of data has a limitation of only 28m depth.To increase the accuracy of the correlation from SPT to VS30, it would be better to collect data up to 30m in depth.

Comparison of PSHA and DSHA
A comparison of maximum PGA calculations using PSHA Pusgen and DSHA with 10 GMPE is provided in (Table 6).This comparison aims to determine which GMPE is the closest to the results of PSHA Pusgen.PSHA uses earthquake sources from a combination of shallow crustal, Benioff, and megathrust.The PSHA method found that the PGAm was 0.59 g, magnitude 64-66 km, and earthquake distance 70-80 km.From 10 GMPEs, various results were obtained.The largest PGA was obtained using GMPE from Donovan and Bornstein of 0.85 g and the smallest from Youngs of 0.06 g.By comparing with PSHA, GMPE Kanno is the closest based on PGA, magnitude, and distance criteria.
The difference in PGA is 0.01 g, the difference in distance is 5.19 km and the difference in magnitude is 0.3 Mw.The factor that caused GMPE Kanno close to Pusgen's PSHA results is the use of 91,371 earthquake records from 4,967 earthquakes in Japan, California, and Turkey.Employing many earthquake records will make the GMPE equation better.In addition, there is a correction factor based on the soil type at the location under review based on VS30 data like Kanno.This correction factor causes the PGA amplification based on soil type to be considered.In another study, the GMPE from Kanno had the best reliability using the Euclidean Distance Ranking (EDR) method by comparing four GMPEs with the PGA recorded from the accelerograph at the Mataram Geophysical Station [35].Based on active faults and subduction zones, the Loh Buaya, Rinca Island, and Mataram City earthquakes were triggered by the Flores back arc fault and the Sumba subduction zone of the Bali and West Nusa Tenggara segments.

Conclusion
The earthquake that occurred around Loh Buaya, Rinca Island was influenced by faulting and subduction activities.The Flores back arc is the dominant fault that causes earthquakes because it has the largest slip rate among other faults.In addition, the subduction zone in the south of Rinca Island has the potential to produce large earthquakes.Soil conditions included in site class SE cause the PGA would be amplified so that the impact of the earthquake will be increased.Using probabilistic approach or PSHA method, the PGA value from Pusgen resulted in 0.59 g.By deterministic approach or DSHA method, GMPE Kanno is the closest to PSHA Pusgen results compared to the other 10 GMPE.The result based on the variables of earthquake distance, magnitude, and PGA.PGA values from both methods show PGA values of more than 0.5 or equivalent to VIII MMI.By considering the PGA value and soil conditions, the earthquake that occurred has the potential to cause another disaster, such as liquefaction.Further research needs to be conducted regarding the potential for liquefaction disasters.

Figure 2 .
Figure 2. SPT test results with sandy soil conditions with groundwater level (a) SPT blows (b) Shear wave velocity (Vs)

Table 2 .
[16]equation used in the logic tree in the De-aggregation map[16]

Table 6 .
Comparison of earthquake variable results between PSHA and DSHA