Volcanic Geohazard Analysis - Case Study from the Ujung Lemah Abang Nuclear Power Plant Area, Central West Java

The paper discusses the proposed development plan of a nuclear power plant in the Ujung Lemah Abang (ULA) region, located in the northern part of Mount Muria in Central Java, Indonesia. The study evaluates the potential volcanic disasters at the proposed site of the Ujung Lemah Abang nuclear power plant (NPP-ULA) near Mount Muria using the Bayesian method, which is based on a probabilistic model of the spatial distribution of volcanic events in the past and the geological processes model. Mount Mulia is a polygenetic volcano that undergoes periodic eruptions alternating between lateral and central eruption-dominated phases. Semivariogram analysis shows a range of 150,000 years. This corresponds to the average duration of volcanic activity and the time between eruptions. Mount Muria volcanic complex can be classified as an extinct volcano with limited future eruption potential based on its last activity 320,000 years ago. The probability of eruption occurrences in Muria within the next 100 years is 1.21856 x 10−5. With this probability of eruptions, it appears that the Muria Mtn is not going to be erupted anytime soon. However, when a new magma system emerges, resulting in epicenter concentrations beneath Mount Muria, micro-earthquake monitoring is required to detect magmatic activities beneath the Muria Volcanic Complex. To estimate the likelihood of phreatomagmatic eruptions, geothermometer monitoring in deep wells is required as an active monitoring of hydrothermal activity in the Muria Volcanic Complex.


Introduction
The proposed nuclear power plant site is north of the Muria Peninsula on the Muria Volcanic Complex (Figure 1), which includes Mount Muria in the center, Mount Genuk to the north, and Mount Patiayam to the south.To mitigate volcanic disasters, mitigation engineering solutions are required, particularly for energy generation facilities located in volcanic areas.According to IAEA recommendations, when determining the location of power plants, volcanic activity must be considered [1].The study assesses the potential volcanic disasters caused by Mount Muria's central eruptions, which can result in pyroclastic material fall, ballistic projectile, lava flow, pyroclastic flow, and volcanic material lahar.Maar formation has the potential for pyroclastic material fall and ballistic fragment disasters, accompanied by the material throwing process, lava flow, and release of toxic gases into the air.The simulation of the potential disaster caused by the maar eruption is carried out in the northwest part of Mount Muria's slope, where the farthest combination of maar occurs.Mount Muria has a lava dome in the crater and at least four lava pyroclastic domes on its slope.The parasitic lava dome on the Mount Muria slope is the source of the potential disaster of dome extrusion and its association.Except for lava flow, there is no geological or historical evidence that the dome extrusion event produces pyroclastic material.Regionally, the area surrounding the Muria Volcanic Complex has been affected by recent earthquake activity caused by some of the East Java Basin's fault systems, specifically the E-W Sakala -Sepanjang Faults and the NE -SW faults system, which are related to the older basement structural grain.However, there are significantly fewer earthquake activities in the Muria area.
Depending on the type of disaster assessment, evaluating volcanic and seismic disasters that affect nuclear facilities differs in several ways [2,3].The most significant distinction is that the primary focus is on disasters at the specific location of the nuclear facility, in this case, the Tapak PLTN location, rather than disasters in the larger area surrounding the volcano or fault.On the left end of Figure 2, the level of risk acceptance for the nuclear facility is shown, which means that volcanic or earthquake eruptions could result in a loss of control over nuclear power generation or nuclear material leakage.
Figure 1.Location of study area for the power plant.Earthquake map on top left (inzet) and the interpreted faults around Muria Volcanic Complex [4] and the earthquake data were acquired from USGS.
The estimation of disasters and risks based on a small number of volcanic events is challenging either because they occur infrequently or are unidentifiable [6].Because geological records are incomplete, a small number of volcanic and seismic events preserved in geological records can determine risk analysis.One approach to improving disaster assessment based on multiple models is to modify probabilistic analysis by incorporating additional data via Bayesian inference [6,7,8].
During the first stage of the investigation, all data is gathered in order to identify potentially hazardous volcanic phenomena at the proposed site.Volcanic disasters are evaluated using deterministic and probabilistic methods to determine the acceptability of proposed area.
Is there current volcanic activity?
Is there Holocene volcanic activity?

Evaluation of Volcanic Disasters
Deterministic conservative evaluations for any geological phenomenon need to be conducted in order to determine the SDV (distance value).The evaluation includes the following assumptions: (1) Every eruption occurs at every volcanic center or in the vicinity of a nuclear power plant.
(2) The maximum potential for eruptions for that volcano is postulated by the size and duration of eruptions.There are geological phenomena that can cause disasters at nuclear power plant sites, such as pyroclastic fall, where wind sources blow directly towards the site at maximum possible speeds and at heights with high eruption column density during the eruption duration.The sources of avalanches, lahars, debris flows, and pyroclastic flows are assumed to be the most observations of the form and activity of comparable volcanoes that can be compared.As a result of this assessment, an area surrounding the volcano has a specific SDV-related geological phenomenon.The presence of volcanic deposits around the location is direct evidence that the location is within the SDV for that phenomenon, which needs to be investigated further.This evaluation is part of determining whether technical solutions to protect the site can be implemented.
Further evaluation decisions are made based on whether the phenomenon is associated with active volcanoes during the Holocene, which necessitates deterministic evaluation, or with pre-Holocene or historical events, which necessitate statistical/stochastic analysis to determine the recurrence rate or probability that the event will occur [9].Statistical/stochastic evaluations of source activity or similar volcanoes can be performed to determine patterns and trends.

Petrological Analyses
The HK volcanic rock series or Muria Muda volcanic rocks formed at lower degrees of partial mantle melting than the K volcanic rock series or Muria Tua volcanic rocks [10] Since the Holocene, strikeslip faults have dominated the structural pattern in the Java Sea surrounding the Muria Volcanic Complex (10,000 years).This indicates that the primary tectonic force at the time was a compression regime.
Plotting SiO2 (Figure 3) volcanic rock versus historical age using available data reveals that the most recent eruption products resulted from the differentiation of previous magma sets.If no new magma is added, the Muria Volcanic Complex's volcanic activity will be caused by magma that cools from the previous magma set.When heat from the cooling magma contacts groundwater, it usually causes phreatic eruptions.Monitoring volcanic earthquakes beneath the Muria Volcanic Complex can detect new magma input.However, there is no indication of a concentration of epicenters beneath the Muria Volcanic Complex [10].Mount Muria is thus a non-capable volcano for magmatic eruptions soon.

Spatial and Temporal Probabilistic Analysis
To investigate the spatial and temporal correlation of volcanic eruptions, stochastic models such as the Poisson process and the Kernel are used.The development of stochastic models is used to determine the probability of new volcanic eruptions using two spatial variables and one temporal variable that can depict volcanic eruptions, namely the location of the eruption center, the area affected by eruption products, and the frequency of eruption occurrences.
To identify the characteristics of volcanic hazard threats, descriptive methods and deterministic models are used.In this case, information on the frequency of areas affected by volcanic hazards is required.Hazards are a random function [11] corresponds to a "moving average" of randomly generated location functions.
The number of eruptions within a given time frame is the temporal component.According to observations, eruption events are frequently clustered in time, indicating a potential temporal correlation.The stochastic process must take into account the timing of volcanic activity.In geological space and time, the Poisson process model can identify and quantify volcanic phenomena (spatial and temporal analysis).
The first step in understanding the nature of volcanoes is determining eruption data and constructing eruption chronologies or time series of past eruptions.A table with a specific time interval covering the eruption period is one simple way to illustrate eruption data.The chronology of Muria volcano eruptions based on rock age dating is shown in Figure 3 that also shows data of the number of eruptions with a 25,000-year time interval based on confirmed ages [10].A cumulative plot depicting the number of eruptions that occur during or before time "t" is used to conduct time series analysis of Muria volcano complex eruptions.This type of plot clearly shows the fluctuating pattern and eruption rate that occurred over a specific time period.Correlating the plotting points and measuring the correlation slope yields the average number of eruptions in each time unit over a given period.The cumulative plot of the number of eruptions in the Muria volcano complex (Figure 3) shows no regime differences and indicates a stationary pattern or a non-time-dependent process.

Additional information comes from the time intervals between eruptions. The distribution of time intervals between eruptions is defined as a time function (t):
 N (t) = the proportion of repose intervals longer than t.
(1) where N is the number of intervals [12] The number of repose periods N(T) with a duration greater than T (25,000 years) for Muria volcano is presented in Figure 4 which also shows the survivor function with a very good proportion for short and long repose periods.This diagram depicts an exponential distribution as an indication of the Poisson process that is identical to a stationary process.The Poisson distribution provides a suitable model for understanding the nature and hypotheses of random data.The Poisson process is used to explain stochastic phenomena that are highly variable and sporadically occur over time, generally under random conditions.The number of events, X, per unit time has a probability of: where λ is the average annual occurrence of volcanic eruptions, and x = 0, 1, 2,....The estimated average occurrence, λ, is the average number of eruptions during the observation period, defined as: λ = Total number of eruptions during the observation period / observation Period Assessing the potential occurrence or recurrence of volcanic eruptions in a sequence or interval of time is a fundamental problem in statistical calculations.As in natural processes, there are two possible events that can occur, either it occurs or does not occur.Similarly, in volcanic processes, there is a possibility of eruption or non-eruption.In the world of statistics, this type of process is known as a Bernoulli sequence, which is based on three main assumptions, which are: (1) there are only two possible events, occurrence or non-occurrence (eruption in this case) and ( 2) the probability of the event is constant; and (3) statistical experiments are independent.
The probability of success p (and the probability of non-occurrence is 1-p), the number of successes in a number of n Bernoulli trials is depicted as a Bernoulli distribution.The probability of the occurrence, x, is expressed as: where nCx is the number of combinations of n and x at a time.
Ignoring the time dependency, the initial estimate of the average probability of eruption occurrence in the Muria Volcano Complex during activity is m = 0.675 per 25,000 years during the observation period, or 2.7 x 10-5 per year.Considering that the youngest eruption of Muria occurred approximately 320,000 years ago, the probability of an eruption in Muria in the next 100 years, equivalent to the following 13 periods (320,000 + 100/25,000 years), is 1.21856 x 10-5.Meanwhile, the probability of one eruption event occurring within the same time interval is B(x = 0) + B(x = 1) = 1.80528 x 10-5.
In reality, volcanic eruptions can occur during a certain period of time.In this case, the occurrence of eruptions can be depicted as a Poisson process.Volcanic eruption events are often grouped by time, and they do not occur randomly but tend to be correlated with time.Detection and quantification of such events can be done using variogram analysis.Although this statistical analysis is often applied in spatial variables, variogram analysis can also be applied to the study of stochastic processes in calculating time properties within large class intervals.With the stationary hypothesis, the variogram γ(τ) is defined as: where: γ(τ) : the value of variogram with a time interval (t) to others zi : the number of eruptions at time interval i z i+h : the number of eruptions to other eruptions at the time interval Σ(zi -zi+h)2 : the sum of the square of differences between the number of eruptions of all pairs of data within a certain time interval n : the number of data pairs separated by a certain time interval During a period of 1 million years, eruptions at Muria Complex are grouped into time intervals of 25,000 years (Figure 5).The figure shows the "measurement scale" as the grouping of eruption events into time intervals.This grouping is done randomly to obtain maximum resolution and stability of statistical parameter estimates.The temporal correlation of data variables is analyzed by the semi-variogram.This correlation shows a "range" with a value of 150,000 years, which indicates the stabilization of the firstlevel variogram of a number of eruptions in different time intervals (Figure 5).The range corresponds to a memory effect that can be interpreted as the average duration of volcanic activities and repose periods.Volcano geological data can be used as the foundation for quantitative hazard forecasting.Volcanic hazard forecasting frequently requires evaluating extremely rare events with severe consequences.The acceptable probability of hazards for facilities with high consequences, such as nuclear facilities, ranges from 10-4 to 10-8 events per year [5].
The first step in developing this probability model is to specify the events that will be examined [8,13].In the case of the Muria Complex, the events are limited to the formation of new eruption points associated with the Patiayam, Genuk, and Muria volcanic systems.This study's hazard forecasting model excludes hazards associated with eruptions such as pyroclastic flows, lava, and lahars.The volcanic hazard forecast in this study focuses on the possibility of new eruption points forming in the Muria Peninsula, which could have an impact on the Ujung Lemah Abang nuclear power plant.
The volcanic data used in this model come from monogenetic and polygenetic volcanic models, and include craters, lava domes, and maars.Each eruption source is treated equally and is considered independent of the others, regardless of the type of eruption.The possibility of magma injection without a new eruption is ignored.An alternative model for determining events can be used by weighting factors such as age, eruption type, distribution, and activity episodes.
The next stage in this model is to develop the mathematical probability model.The probability of an event occurring in a specific location is expressed as: Where A is the study area, the Muria Volcanic Complex that may impact the Ujung Lemah Abang nuclear power plant.The time t used in this formula is assumed to be the 100 years of the nuclear power plant's operation.The next modeling stage is to determine the spatial density, λs (the number of events that occur in each km), and the temporal recurrence rate, λt (the events per year).For the Muria Volcanic Complex, the estimated temporal recurrence rate is 2.7 x 10-5 events per year according to the Poisson distribution.This means that one volcanic eruption event is likely to occur every 37,037 years.
Although the distribution of volcanic eruptions in the Muria Complex exhibits an exponential distribution as a Poisson process or stationary process, the spatial distribution is assumed as a separate discrete event in space and time.Therefore, the Gaussian kernel function is chosen for modeling, expressed as: The model calculates the probability density function based on the distance (di) between a point (x,y) and the nearest volcano, using a smoothing parameter (h) and the number of volcanoes (N) in the area and number of events (λs) .The spatial density is then integrated across the entire spatial map to obtain a bivariate probability density function.The probability calculations depend on the value of h, with smaller h values resulting in higher probabilities for nearby volcanoes and larger h values producing more uniform probability distributions.The Gaussian kernel factor is equivalent to the standard deviation of the bivariate Gaussian distribution, and the choice of kernel depends on geological and statistical assumptions.
Various smoothing parameter values ranging from 12 km to 35 km were used in this study (Figure 6).The kernel model was chosen because, unlike the homogeneous approach, it allows for the comparison of geological information and does not require the determination of volcanic activity zones.The resulting spatial density map depicts the likelihood of new volcanic eruptions in the Muria Peninsula, ranging from 1.42 x 10-4 to 1.50 x 10-5.It is important to note that the probability values are extremely sensitive to the smoothing parameter chosen, implying that there is no definitive method for determining the optimal h value.

Conclusions
Volcano geological data is used to create a quantitative hazard forecasting model.The focus is on evaluating rare events with severe consequences, particularly for high-consequence facilities like nuclear power plants.The acceptable probability range for such hazards is between 10-4 and 10-8 events per year.In the specific case of the Muria Complex, the model examines the formation of new eruption points associated with the Patiayam, Genuk, and Muria volcanic systems.Hazards like pyroclastic flows, lava, and lahars are not considered.The goal is to forecast the possibility of new eruption points in the Muria Peninsula that could affect the Ujung Lemah Abang nuclear power plant.
The model incorporates volcanic data from various sources and treats each eruption source equally and independently.The possibility of magma injection without a new eruption is disregarded.Alternative models could consider factors such as age, eruption type, distribution, and activity episodes.
The Muria Volcanic Complex's stratovolcano system and caldera are non-capable volcanoes, according to vulcanological data.The magma system beneath the Muria Volcanic Complex has cooled, leaving only magma-derived volcanic gases, altered rocks, and anomalous temperature gradients in some areas.Future magmatic eruptions are impossible unless new magma influxes into the Muria Volcanic Complex's magma chamber.Seismic studies using micro-earthquake monitoring show no concentration of epicenters beneath Mount Muria, implying that there is currently no new magma influx into the Muria Volcanic Complex's magma chamber.
While volcanic eruptions in the Muria Complex exhibit an exponential distribution, the spatial distribution is treated as a discrete event.The model utilizes a Gaussian kernel function, considering the distance between a point and the nearest volcano, along with a smoothing parameter (h) and the number of volcanoes and events in the area.The choice of kernel depends on geological and statistical assumptions.The Gaussian kernel model is preferred because it allows for the comparison of geological information and does not require the determination of volcanic activity zones.It is important to note that the probability values are highly sensitive to the chosen smoothing parameter, implying that there is no definitive method for determining the optimal value.The Poisson distribution shows an initial estimation of the average probability of eruption occurrences in the Muria Volcanic Complex during activity, which is m = 0.675 per 25,000 years during the observation period, or 2.7 x 10 -5 per year.Considering the youngest Muria eruption occurred approximately 320,000 years ago, the probability of eruption occurrences in Muria within the next 100 years is 1.21856 x 10 -5 .With this probability of eruptions, it appear that the Muria Mtn is not going to be erupted anytime soon.However, when a new magma system emerges, resulting in epicenter concentrations beneath Mount Muria, micro-earthquake monitoring is required to detect magmatic activities beneath the Muria Volcanic Complex.To estimate the likelihood of phreatomagmatic eruptions, geothermometer monitoring in deep wells is required as an active monitoring of hydrothermal activity in the Muria Volcanic Complex.

Figure 3 .
Figure 3. Plots of different geochemical composition of Gunung Muria (top) and its cumulative number of eruptions.

Figure 4 .
Figure 4. Distribution of observed and calculated repose period with a duration window of 25000 years in Muria Volcanic Complex.

8 Figure 5 .
Figure 5. Memory effect determination with variogram analyses showing the different number of intervals on eruption period.

Figure 6 .
Figure 6.Spatial density map shows the probability of new volcanic eruptions in the Muria Peninsula, with probabilities ranging from 1.42 x 10 -4 to 1.50 x 10 -5 .