Abstract

Payangan Beach in Ambulu sub-district is prone to tsunamis because it is facing the Indian Ocean where a megathrust lies, making the identification of suitable Temporary Evacuation Sites (TES) essential for disaster mitigation. This study integrates Geographic Information Systems (GIS) with the Analytical Hierarchy Process (AHP) and the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) that aims to evaluate nine TES candidates based on eight criteria. The AHP–TOPSIS results show that evacuation time, evacuation route width, and tsunami risk are the most dominant variables. The initial AHP–TOPSIS ranking identified Mount Watangan, Sumberejo 03 Elementary School, and Sumberejo Village Field as the most suitable TES. However, Focus Group Discussion (FGD) findings revealed that these mathematically optimal locations were not feasible due to steep access, proximity to the river, and other micro-topographic constraints not represented in the model. Community validation, therefore, led to the replacement of three TES with safer and more accessible alternatives. This study demonstrates that while AHP–TOPSIS provides a rapid analytical framework for prioritizing evacuation sites, final decisions must integrate local knowledge to ensure realistic evacuation planning. Therefore, this study supports the installation of the first tsunami evacuation route maps in Ambulu District.

1 Introduction

Tsunamis are one of the deadliest coastal hazards and pose a serious threat to areas located near active subduction zones (Armono et al. 2021; Maramai and Tinti 1997; Tsuji et al. 1995). The southern coast of Java Island is known to have a high level of tsunami susceptibility due to intense tectonic activity. Jember Regency, particularly the Ambulu Sub-district and the Payangan Beach area, is located within this high-risk zone and has a history of high-wave events and tidal flooding. Payangan Beach is a coastal tourism area with a high intensity of human activities. The high-wave event in 2020 demonstrated the clear vulnerability of this area to tsunami threats, while the high level of activity and the presence of coastal settlements have the potential to increase the risk of casualties if evacuation processes are not adequately planned (Gloria et al. 2023).

The success of tsunami evacuation largely depends on the availability of safe locations that can be reached quickly and the clarity of evacuation routes (Muhammad et al. 2021; Marfai et al. 2021). However, horizontal evacuation is not always feasible due to the limited tsunami arrival time and the topographic conditions of coastal areas. Therefore, Temporary Evacuation Sites (TES) become an important component of the tsunami evacuation system. In the early stages, the determination of TES was generally based on local knowledge and simple considerations. Along with the development of disaster mitigation, TES determination is now carried out using more measurable parameters, such as building or land elevation, distance from the coastline, accessibility and evacuation routes, safety aspects, and evacuation travel time. In this context, Geographic Information System (GIS) plays an important role as the main tool for spatially analysing and integrating these parameters (Bonilauri et al. 2021). Local communities have an important role in the process of determining TES, particularly through the provision of local knowledge, evacuation experience, and assessment of the accessibility and feasibility of TES locations under tsunami emergency conditions (Villagra et al. 2021; Benazir et al. 2023).

Several studies related to the determination of TES in the Jember region have been conducted (Sari et al. 2020). However, these studies generally still use relatively simple scoring methods and have not been complemented by validation from the perspective of user acceptance and understanding, namely the local community. In fact, there are many combinations of parameters used for selecting TES such as building heights, building placements (León et al. 2023; Hafizh et al. 2023), distance from coastline, emergency infrastructure, and evacuation routes (Villagra et al. 2021; Ganjehi and Khatiri 2021; Laksono et al. 2022), topography and maximum inundation height (Hafizh et al. 2023), safety (Arif et al. 2024), and Estimated Times of Arrival (ETA) (Benazir and Oktari 2024; Ferreira et al. 2025). Therefore, TES determination constitutes a multi-criteria problem that includes both technical and social aspects. In multi-criteria decision-making (MCDM) problems such as TES selection, criteria weighting is a very crucial stage. The Analytical Hierarchy Process (AHP) is one of the most used methods for determining criteria weights because it can accommodate expert judgment in a structured and consistent manner (Depari et al. 2023; Ganjehi and Khatiri 2021). However, AHP has limitations in objectively ranking alternatives. To overcome these limitations, AHP is often combined with other MCDM methods such as Multi-Attribute Incident Response and Consequence Analysis-MAIRCA (Ulandari et al. 2024), ELimination Et Choix Traduisant la REalité-ELECTRE (Sigar et al. 2018), and Technique for Order Preference by Similarity to Ideal Solution-TOPSIS (Uslu et al. 2016; Du et al. 2020). TOPSIS is widely used because of its ability to identify alternatives that are closest to the ideal condition and farthest from the worst condition (Liu 2022). The AHP–TOPSIS combination is considered an effective approach because it integrates expert-based weighting with alternative ranking that is more objective (Basar 2024).

Based on these issues, this study aims to identify TES and tsunami evacuation routes in the Payangan Beach area using an AHP–TOPSIS approach supported by community-based validation. The validation is conducted through Focus Group Discussions (FGD) with the local community to enhance the reliability of the results and to adjust TES according to user preferences and understanding within the tsunami early warning system.

2 Methods

The overall research framework combines tsunami hazard analysis, multi-criteria decision-making, and participatory validation to determine suitable TES and evacuation routes in Payangan Beach, Jember as shown in Figure 1.

Figure 1 Research flowchart.

2.1 Data

This study uses both spatial and non-spatial data obtained from expert judgment and existing datasets, as shown in Table 1. In the AHP approach, a purposive sampling strategy was used to select experts directly involved in tsunami mitigation, spatial planning, and community-level preparedness. Because AHP prioritizes expertise over sample size, the 25 respondents included represent all relevant stakeholder groups required to generate reliable and consistent pairwise judgments. The respondents consisted of 5 from the East Java Regional Disaster Management Agency, 5 from the Jember Regional Disaster Management Agency, 5 from the Ministry of Public Works and Housing, and 10 members of the Disaster-Resilient Village (Destana) program. These institutions were selected due to their direct involvement in disaster preparedness, spatial planning, and community-level evacuation management within the study area.

Table 1 Data sources.

No.	Data	Data Source
1	Tsunami height at Watu Ulo Beach	Tsunami propagation modeling result by Sofiana et al. (2022)
2	Digital Elevation Model data	EARTHDATA (2024)
3	Land use data	EARTHDATA (2024)
4	Topography map of Jember Regency	EARTHDATA (2024)
5	Weight of criteria	Field questionnaire

The collected datasets were subsequently applied in three stages of analysis:

1. Tsunami hazard assessment using DEM, land-use data, and tsunami simulation results.
2. TES evaluation using expert judgment through AHP–TOPSIS, which was then validated through an FGD involving local stakeholders.
3. Evacuation route mapping using GIS based on the road network, elevation model, and community perception.

2.2 Tsunami hazard assessment

The tsunami hazard assessment was generated using the Berryman (2005) method to spatially model the inundation using DEM, land use maps, coastline data, and the tsunami height in the coastal area. The tsunami height as the input for inundation modeling was taken from Sofiana et al. (2022), who modeled the Mw 9,1 megathrust tsunami propagation using 250-m-grid size Delft3D Flow and reported a tsunami height of 12.57 m in Watu Ulo Beach, Ambulu District with an estimated arrival time of 29 minutes. The tsunami propagation value was used to spatially model the inundation based on the height-loss spatial analysis by Berryman (2005). Inundation areas are classified into three zones (low, medium, and high) based on the inundation depth that may occur in the area.

2.3 TES evaluation using AHP–TOPSIS

The TES evaluation was conducted using AHP to determine the relative importance of the criteria, followed by TOPSIS to rank the TES alternatives. Eight evaluation criteria were adopted based on the 2013 guideline for TES planning issued by the Indonesia National Agency for Disaster Management (BNPB 2013), as shown in Table 2. Field observations supported by Google Maps, Google Earth, and the Altimeter Ler application were used to obtain data on TES distance from the shoreline, elevation, evacuation time, and available facilities. The maximum allowable TES distance for a weak walker traveling at 3.22 km/h within 30 minutes is 1.61 km. Distance measurements were taken at nine evacuation points: Hidayatullah Moqsue, Al-Hidayah Mosque, Al-Ishlah Mosque, Al-Huda Mosque, Assasul Huda Mosque, Sumberejo Village Field, Sumberejo 03 Elementary School, Mount Teluk Love, and Mount Watangan. The storage capacity at each evacuation point was calculated using Equation 1:

(1)

Table 2 TES evaluation criteria.

Code	Criterion	Code	Criterion
P1	TES distance (km)	P5	Evacuation road width (m)
P2	Elevation (m)	P6	Numbers of toilet and clean water
P3	Capacity (people)	P7	Numbers of emergency equipment warehouse
P4	Evacuation time (minute)	P8	Tsunami disaster risk

Evacuation routes, toilet availability, and clean water facilities were also identified through field surveys. Tsunami risk levels were classified using the 2021 tsunami risk map from the Indonesia National Agency for Disaster Management (BNPB 2021), resulting in categories ranging from very low to very high for the nine points. All observation results were categorized based on the evaluation variables, and the classifications were then used to determine the performance values for each variable across the alternatives.

To standardize the different measurement units and convert raw observations into comparable performance scores, each variable was classified using the Equal Interval Method. In this approach, the range of observed values for each criterion was divided into five equal classes by subtracting the minimum value from the maximum value and dividing the result by the number of classes. The interval width is calculated using Equation 2:

(2)

Where:

Xmax	=	maximum observed value,
Xmin	=	minimum observed value, and
n	=	number of classes applied in this study (n = 5).

The decision-making structure was organized into a two-level hierarchy, consisting of Level 1 (Goal), being the identification of the most suitable TES, and Level 2 (Criteria), comprising eight evaluation parameters (P1–P8). The TES candidate locations used as alternatives in the AHP–TOPSIS evaluation were subsequently identified through preliminary spatial assessment and field observations. A pairwise comparison matrix A = [a_ij] was constructed using the aggregated expert responses based on Saaty’s 1–9 scale (Saaty 1980; Donegan and Dodd 1991). The normalized principal eigenvector provided the criterion weights w_j, which matrix consistency was assessed using Equations 3 and 4:

(3)

(4)

Where:

Consistency Index (CI)	=	measure of the degree of inconsistency in the pairwise comparison judgements,
Consistency Ratio (CR)	=	ratio of the consistency index to the random index, used to assess whether the level of consistency is acceptable,
	=	maximum eigenvalue,
n	=	8, the number of criteria, and
Random Index (RI)	=	1.41 was adopted based on Saaty for n = 8 (Saaty 1980).

The TOPSIS analysis was then applied. First, an evaluation matrix consists of m alternatives, and n criteria is created. The matrix X = [x_ij] was normalized using Equation 5:

(5)

The next step was to create a weighted normalized decision matrix. To find the value of the weighted normalized decision matrix, each element in the normalized decision matrix is multiplied by the weight of each criterion that has been determined by the AHP method using Equation 6:

(6)

After that, the positive ideal solution matrix and negative ideal solution matrix of the ideal solution are determined. The positive ideal solution is the maximum value of each criterion, and the negative ideal solution is the minimum (Equations 7 and 8):

(7)

(8)

Finally, the preference value for each alternative was derived using Equation 9:

(9)

Where:

xij	=	original value of alternative i under criterion j,
rij	=	normalized value,
wj	=	AHP-derived weight of criterion j,
yij	=	weighted normalized value,
and	=	positive and negative ideal values for criterion j, respectively,
and	=	separation of alternative i from the positive and negative ideal solutions, and
Vi	=	final preference score of alternative i. Higher values of Vi indicate more suitable TES alternatives.

2.4 Evacuation route mapping

The evacuation routes were generated using GIS-based spatial modeling. Initially, the road and pedestrian pathway datasets were converted into a topologically connected network to enable routing analysis. Areas located within the tsunami inundation extent identified in the hazard assessment were excluded from the network to ensure that the evacuation paths avoided high-risk zones. A least-cost path analysis was then applied to determine the most efficient evacuation routes from residential clusters to the prioritized TES locations derived from the AHP-TOPSIS evaluation. To improve evacuation safety, route suitability was refined by incorporating terrain slope information extracted from the DEM, allowing for the identification of feasible walking paths under disaster conditions. The GIS-generated evacuation routes were subsequently validated through a focus group discussion involving the Secretary of Sumberejo Village, representatives of the Destana program, the Village Consultative Body, the Jember Regional Disaster Management Agency and Neighbourhood and Community Unit Leaders as shown in Figure 2. During this process, adjustments were made to accommodate locally preferred access routes, informal pathways, and practical considerations not captured in the spatial model, thus enhancing the contextual reliability of the final evacuation plan.

Figure 2 FGD activities in Sumberejo Village.

3 Results

3.1 Analysis of tsunami hazard maps

The tsunami hazard map used in this study identifies the distribution of potential inundation zones along the Payangan coastal area, as shown in Figure 3. Based on inundation analysis using Berryman (2005), tsunamis inundated two villages in Ambulu District, namely Sabrang Village in the west, and Sumberejo Village in the eastern area, with a total inundation area of 829 ha. The inundated areas are 444.1 ha in Sabrang Village, and 384.9 ha in Sumberejo Village. The inundation areas lay in a coastal area, and in an area close to the river. The high-hazard zone appears with red shading along a large portion of the coastline with the total area of 240 ha, while the medium-hazard zone (yellow to orange shading) extends inland behind it with the total area of 283.8 ha. The low-hazard zone (green shading) occupies the outermost inland areas with a total area of 305.2 ha. These spatial patterns reflect the gradation of tsunami height produced by the run-up scenario applied in this study. The hazard map serves as a spatial reference for identifying potential TES and for assessing the relative exposure of different parts of the study area. The classification provides the basis for determining which locations lie within high-hazard zones, and which areas may be more suitable for evacuation purposes.

Figure 3 Tsunami hazard map.

3.2 TES determination variable index

The observations indicate that the nine TES candidates exhibit substantial variation in their physical and functional characteristics, providing a clear picture of the strengths and limitations of each location, as shown in Table 3. In terms of P1, the nearest site lies within 0.5 km of the coast, whereas the farthest site is located more than 6 km away. Nevertheless, all sites remain within ≤1.5 km of residential areas, indicating that they are generally reachable during emergency evacuations. Notable differences also appear in P2 (elevation) and P3 (capacity). Although all sites are situated above the historical run-up height, their accommodation capacity varies widely, ranging from locations that can host only a few dozen evacuees to others capable of accommodating nearly 300 individuals. Accessibility differences are reflected in P4 (evacuation time), which ranges from only several minutes to substantially longer durations for sites located farther inland or on steep terrain. In addition, based on field observations, it is known that the smallest evacuation road width (P5) is in Mount Teluk Love, and the widest evacuation road is in Hidayatullah Mosque. The P8 (tsunami risk category) of each TES was taken from calculation results of hazard and vulnerability, which were calculated by the Indonesia National Agency for Disaster Management (BNPB 2021), with five risk categories: very low, low, medium, high, and very high. The values range from 0 to 16 and are classified using an equal interval approach. The minimum value is 0 and maximum value is 16, divided into five categories:

1. Very low (0 to 3),
2. Low (more than 3 to 6),
3. Medium (more than 6 to 9),
4. High (more than 9 to 12), and
5. Very high (more than 12 to 16).  

Table 3 Locations of proposed TES in Payangan Beach.

No	Alternative	P1a	P1b	P2	P3	P4a	P4b	P5	P6	P7	P8	Risk	Figure
1	Hidayatullah Mosque	6.70	0.57	45.5	42	90	10.6	0.84	2	1	6	Low
2	Al-Hidayah Mosque	6.20	1.10	42	23	84	20.5	0.51	2	1	6	Low
3	Al-Ishlah Mosque	5.10	1.11	43.1	67	70	20.7	0.84	4	0	6	Low
4	Al-Huda Mosque	4.60	0.90	42	42	63	16.8	0.84	1	1	6	Low
5	Assasul Huda Mosque	4.20	1.10	40.2	33	58	20.5	0.84	2	1	7	Medium
6	Sumberejo 03 Elementary School	4.10	1.25	39.5	297	56	23.3	0.67	4	1	6	Low
7	Sumberejo Village Field	4.50	0.97	42.3	187	61	18.1	0.84	3	1	6	Low
8	Mount Teluk Love	0.45	1.48	94.4	97	6	27.6	0.34	0	0	15	Very High
9	Mount Watangan	3.90	1.14	114.4	281	54	21.2	0.17	0	0	5	Low

NOTE: (a) from the shoreline, and (b) from the residential area.

Substantial variation is also evident in P6 (toilets and clean water availability), which ranges from 0 to 4 units. Two TES have no toilet or clean water facilities at all, while others provide between 1 and 4 units, indicating that basic facility readiness is uneven across the alternatives. Similarly, P7 (emergency equipment warehouse) also varies, with only a limited number of sites having dedicated storage for emergency supplies. P8 (tsunami disaster risk classification) further distinguishes the alternatives with most sites falling into the low category, while one is classified as medium and another as high. Each evacuation point observation result was classified for each variable into 5 classes, based on the Equal Interval Method. The results of the assessment for each variable classification were then used to determine the performance value for each variable on each alternative, which will be used in the TOPSIS calculation as shown in Table 4.

Table 4 Performance rating of each alternative. 

Code	Alternative	P1	P2	P3	P4	P5	P6	P7	P8
A1	Hidayatullah Mosque	5	1	1	1	5	3	5	2
A2	Al-Hidayah Mosque	5	1	1	1	3	3	5	2
A3	Al-Ishlah Mosque	4	1	1	2	5	5	1	2
A4	Al-Huda Mosque	4	1	1	2	5	2	5	2
A5	Assasul Huda Mosque	3	1	1	2	5	3	5	3
A6	Sumberejo 03 Elementary School	3	1	5	3	4	5	5	2
A7	Sumberejo Village Field	4	1	4	2	5	4	5	2
A8	Mount Teluk Love	1	4	2	5	2	1	1	5
A9	Mount Watangan	3	5	5	3	1	1	1	2

3.3 AHP-TOPSIS result for TES alternatives

The pairwise comparison matrix used to assess the relative importance of the eight variables for determining the TES is shown in Table 5. This matrix serves as the basis for calculating the eigenvalues for each variable (P1 to P8) and determining the relative priority among the variables.

Table 5 Pairwise comparison matrix.   

Variable	P1	P2	P3	P4	P5	P6	P7	P8
P1	1.00	0.84	1.13	0.81	0.45	1.16	1.27	1.26
P2	1.18	1.00	1.16	0.79	0.72	1.65	1.47	0.87
P3	0.88	0.86	1.00	0.66	0.60	1.90	1.43	0.86
P4	1.54	1.27	1.52	1.00	1.14	2.74	2.38	1.21
P5	2.22	1.38	1.65	0.88	1.00	1.81	2.31	0.79
P6	0.94	0.60	0.53	0.36	0.55	1.00	0.85	0.50
P7	0.78	0.68	0.62	0.36	0.38	1.02	1.00	0.60
P8	0.97	1.15	1.16	0.82	1.27	1.99	1.89	1.00
Total	9.51	7.79	8.76	5.69	6.12	13.27	12.64	7.10
Eigen Value	0.115	0.126	0.114	0.178	0.168	0.076	0.076	0.147
Weight (%)	11.54	12.57	11.38	17.81	16.85	7.56	7.62	14.68

The eigenvalue calculation produced the initial weights, with a range of 0.0076 to 0.178. Accordingly, the maximum λ obtained was 8.127, which resulted in a Consistency Index (CI) of 0.0182. When compared to the Random Index value of 1.41 for eight variables, the Consistency Ratio (CR) value was 0.0129, which is far below the 0.1 threshold, indicating that the respondents’ judgments are consistent, and the weights can be accepted. The result shows P4, P5, and P8 have the highest weights, respectively 14.68% to 17.81%. Meanwhile, the two variables with the lowest weights are P7 at 7.62% and P6 at 7.56%. The total weight of all variables sums to 100 percent, indicating that the entire priority structure can be used as the basis for weighting in the TOPSIS stage.

Six variables, namely P1, P2, P3, P5, P6, and P7, function as benefit variables, so alternatives with higher values on these variables receive a positive contribution to their suitability. In contrast, P4 and P8 are cost variables that reduce suitability when their values increase. This pattern is important because it determines the form of the ideal solution that then becomes the basis for comparing all alternatives. The TOPSIS normalization results are shown in Table 6; thus, the weighted decision matrix was obtained from the multiplication of the TOPSIS normalization values with the AHP weights. Based on the calculation results in the weighted matrix, it is known that the variable that has the most significant influences determining TES are P4, P5, and P8, while the variable that has the least influence is P7, as shown in Figure 4.

Table 6 TOPSIS normalization result.

Code	P1	P2	P3	P4	P5	P6	P7	P8
A1	0.45	0.14	0.12	0.13	0.40	0.30	0.40	0.25
A2	0.45	0.14	0.12	0.13	0.24	0.30	0.40	0.25
A3	0.36	0.14	0.12	0.26	0.40	0.50	0.08	0.25
A4	0.36	0.14	0.12	0.26	0.40	0.20	0.40	0.25
A5	0.27	0.14	0.12	0.26	0.40	0.30	0.40	0.38
A6	0.27	0.14	0.58	0.38	0.32	0.50	0.40	0.25
A7	0.36	0.14	0.46	0.26	0.40	0.40	0.40	0.25
A8	0.09	0.58	0.23	0.64	0.16	0.10	0.08	0.64
A9	0.27	0.72	0.58	0.38	0.08	0.10	0.08	0.25

Figure 4 AHP-TOPSIS weight of TES’s variables.

The relative closeness value of each alternative to the ideal condition is presented in the form of the preference value V_i, as shown in Table 7. The preference values are categorized into three levels of suitability based on the equal interval approach between 0 and 1, as follows: low (0.00–0.33), medium (0.34–0.66), and high (0.67–1.00). The preference values show that all TES alternatives are very closely clustered between 0.4882 to 0.4978, which falls within the medium category. Therefore, no single TES achieves the ideal level of suitability for all criteria.

Table 7 Preference value.

Code	Alternative	Preference (Vi)	Rank	Category
A7	Sumberejo Village Field	0.4978	1	Medium
A9	Mount Watangan	0.4975	2	Medium
A1	Hidayatullah Mosque	0.4971	3	Medium
A6	Sumberejo 03 Elementary School	0.4965	4	Medium
A2	Al-Hidayah Mosque	0.4962	5	Medium
A3	Al-Ishlah Mosque	0.4951	6	Medium
A4	Al-Huda Mosque	0.4950	7	Medium
A5	Assasul Huda Mosque	0.4941	8	Medium
A8	Mount Teluk Love	0.4882	9	Medium

However, the ranking results from the TOPSIS method show differences among all TES alternatives. Sumberejo Village Field (A7) occupies the highest rank with a value of 0.4978, followed by Mount Watangan (A9) with 0.4975, and Hidayatullah Mosque (A1) with 0.4971. On the other hand, Mount Teluk Love (A8) has the lowest preference value of 0.4882, which indicates that this location has the largest deviation from the ideal solution, particularly in the variables of distance, evacuation time, and availability of facilities. All alternatives fall into the medium suitability group with no substantial differences in performance. These results confirm that trade-offs exist among the eight criteria, so that all TES locations are suitable to function as a TES based on the ranking calculated by the TOPSIS method.

The TES locations based on the AHP-TOPSIS results and evacuation routes map in Payangan Beach, Ambulu District, Jember Regency are shown in Figure 5.

Figure 5 AHP-TOPSIS TES and evacuation routes map.

3.4 Community participation for evacuation map

The participation validation resulted in several important adjustments to the initial list of temporary evacuation sites. Local stakeholders provided input based on actual field conditions, particularly regarding the safety of evacuation routes, the presence of new infrastructure, and proximity to hazard sources. Several changes to TES locations were agreed upon during the validation process, as shown in Table 8.

Table 8 TES Adjustments after community validation.

TES Location	Before FGD	After FGD
Hidayatullah Mosque	✓	✓
Al-Hidayah Mosque	✓	✓
Al-Islah Mosque	✓	✓
Assasul Huda Mosque	✓	✓
Al-Huda Mosque	✓	✓
Sumberejo 03 Elementary School	✓	-
Sumberejo Village Field	✓	✓
Love Bay Mountain	✓	-
Mount Watangan	✓	-
Sumberejo Village Office	-	✓
Mount Samboja	-	✓
Mount Gamping	-	✓

First, Mount Teluk Love is no longer recommended as a TES. The route to this location is now affected by the construction of a jetty, and there is direct water flow toward the area, making it potentially obstructive for evacuation and increasing risk during a tsunami. As a replacement, the community proposed Mount Samboja (on the northern side of Teluk Love) because it is closer to settlements and has safer access. Second, Sumberejo 03 Elementary School was removed from the TES list because it is located near the large Mayang River, which poses a risk of overflow or backwash during a tsunami. The river flow direction is also considered likely to accelerate the concentration of water mass toward the school area, making it unsafe as an evacuation point.

Third, Mount Watangan was also not retained as a TES, even though it technically has sufficient elevation. The community considered the access to this location too steep and unsuitable for rapid evacuation, especially for vulnerable groups. This site was replaced with Mount Gamping, which is considered to have safer topographic characteristics and better accessibility. Additionally, the validation participants proposed the Sumberejo Village Office as an alternative evacuation point. This location was chosen because it is easily accessible to the community, has sufficient open space, and is considered strategic as an emergency coordination centre. These changes emphasize that the selection of TES is not determined solely by technical criteria, but must also consider social conditions, actual accessibility, and local knowledge, as shown in Figure 6. Participatory validation ensures that the resulting TES list is more realistic and can be effectively implemented in an emergency.

Figure 6 AHP-TOPSIS–community participatory TES and evacuation routes map.

3.5 Discussion

Based on the AHP–TOPSIS results, the variables that emerged as the most dominant were evacuation time (P4), evacuation route width (P5), and tsunami risk (P8). This is related to the mathematical characteristics of TOPSIS, which is sensitive to variations among criteria. Previous studies show that criteria with high standard deviation or coefficient of variation contribute more to the overall Euclidean distance to the ideal solution (Chen 2019). In this study, Table 6 shows that P4 has the widest normalization range (≈0.13–0.64), followed by P5 (≈0.10–0.50) and P8 (≈0.25–0.64). This high level of variability makes these three variables the most determinant factors in differentiating the quality of alternatives. Conversely, facility-related variables such as P6 (≈0.08–0.40) and P7 (which is relatively uniform across TES) show low variation, resulting in a small contribution to the selection of the ideal solution. These findings indicate that TOPSIS reveals real performance variation in the field (Vavrek 2019), making it suitable to be combined with AHP, which captures subjective perceptions of variable importance.

This result is clearly seen in the difference between the assessment of Mount Watangan (A9) and Mount Teluk Love (A8). Although both are in a high-elevation zone, Teluk Love obtained the worst normalization values for evacuation time (≈0.64) and tsunami risk (≈0.64). The combination of long travel time and high exposure risk places Teluk Love far from the ideal solution. In contrast, Mount Watangan performs better in travel time and tsunami risk, which are mathematically the most determining variables, even though its normalized value for evacuation route width is smaller than that of Mount Teluk Love. This analysis confirms that high elevation does not automatically make a location suitable as a TES, instead, accessibility, disaster risk level, and route quality have much greater influence in the objective AHP–TOPSIS assessment.

However, the FGD results show that mathematical performance does not always align with field conditions. Although A9 ranks at the top according to TOPSIS due to its good elevation, capacity, and travel time, the community rejected this location after considering real risks not captured in the model, such as extreme slope access and proximity to the Mayang River, which is prone to overflow and backflow during a tsunami. These micro-scale risks are consistent with the findings of Sabani et al. (2021), which note that technical mapping often fails to capture micro-topographic changes, access barriers, and new development along evacuation routes. Through community participation, dynamic field information such as damaged roads, new inundation areas, and changes in water flow direction can be updated and integrated so that evacuation plans become more accurate and contextual (Wulan Mei and Rachmawati 2016). In addition, community validation also increases the legitimacy and social acceptance of the evacuation plan because the community can directly assess its technical suitability with real field conditions (Sofyan et al. 2025). It is therefore unsurprising that the community instead promoted Mount Gamping, Mount Samboja, and the Sumberejo Village Office as locations with safer access, better reachability, and more realistic use during emergencies.

4 Conclusion

Based on the results, this study shows that the variation in inundation levels along the Payangan coast requires Temporary Evacuation Sites that are not only elevated but also accessible, safe, and supported by adequate facilities. The AHP TOPSIS analysis identified evacuation time, road width, and tsunami risk as the most influential criteria in differentiating the suitability of the nine TES candidates, producing initial rankings that favored Mount Watangan, Sumberejo 03 Elementary School, and Sumberejo Village Field. However, the community validation demonstrated that these mathematically optimal sites are not always feasible due to steep access, river proximity, and changing site conditions that pose real evacuation constraints. Through this participatory process, alternative locations such as Mount Gamping, Mount Samboja, and the Sumberejo Village Office were identified as safer and more realistic options for emergency use. These findings confirm that rapid analytical methods like AHP-TOPSIS are effective for prioritizing evacuation sites, yet final decisions must incorporate local knowledge to ensure that the selected TES truly support community preparedness and practical tsunami evacuation planning.

Acknowledgments

This study is funded by the Internal Research Grant of the University of Jember for 2024, under contract number: 2909/UN25.3.1/LT/2024.

References

1. Arif, Z., M. Ella, and Y. Halida. 2024. “Analysis of School Infrastructure Potential as Temporary Evacuation Centers During Emergency Disaster Conditions in Palu and Surrounding Areas.” E3S Web of Conferences 476, 01040. https://doi.org/10.1051/e3sconf/202447601040
2. Armono, H.D., A.R. Putra, and Wahyudi. 2021. “Analysis of Tsunami Wave Height, Run-up, and Inundation in the Coastal of Blitar and Malang Regency.” IOP Conference Series: Earth and Environmental Science 936, 012013. https://doi.org/10.1088/1755-1315/936/1/012013
3. Basar, A.A. 2024. “Participatory Urban Planning for Social Sustainability: A Combination of the Analytic Hierarchy Process, with Strengths, Weaknesses, Opportunities, Threats Analysis, and the Technique for Order Preference by Similarity to Ideal Solution (A’WOT-TOPSIS).” Sustainability 16 (24): 10862. https://doi.org/10.3390/su162410862
4. Benazir and R.S. Oktari. 2024. “Assessing tsunami risk along the Aceh coast, Indonesia: a quantitative analysis of fault rupture potential and early warning system efficacy for predicting arrival time and flood extent.” Natural Hazards 120, 4875–4900. https://doi.org/10.1007/s11069-024-06401-x
5. Benazir, Syamsidik, Y. Idris, and N.P. Putra. 2023. “Connecting community’s perspectives on tsunami risk to anticipated future tsunamis: a reflection from a progress of tsunami preparedness from a coastal community in Aceh-Indonesia after 19 years of the 2004 Indian Ocean Tsunami.” Geoenvironmental Disasters 10, 21. https://doi.org/10.1186/s40677-023-00252-7
6. Berryman, K. 2005. Review of Tsunami Hazard and Risk in New Zealand. Ministry of Civil Defence and Emergency Management, Institute of Geological & Nuclear Sciences Limited, Lower Hutt, New Zealand.
7. BNPB (Indonesia National Agency for Disaster Management). 2013. Pedoman Teknik 2: Perencanaan Tempat Evakuasi Sementara (TES) Tsunami (Technical Guidelines 2: Planning for Tsunami Temporary Evacuation Sites). BNPB, Jakarta, Indonesia. In Indonesian. Available online: https://perpustakaan.bnpb.go.id/inlislite/opac/detail-opac?id=1902
8. BNPB (Indonesia National Agency for Disaster Management). 2021. Peta Risiko Tsunami Indonesia (Tsunami Risk Map of Indonesia). BNPB, Jakarta, Indonesia. In Indonesian. Available online: https://inarisk.bnpb.go.id/pdf/INDONESIA/risiko_tsunami_INDONESIA.pdf
9. Bonilauri, E.M., A.J.L. Harris, J. Morin, M. Ripepe, D. Mangione, G. Lacanna, et al. 2021. “Tsunami evacuation times and routes to safe zones: a GIS-based approach to tsunami evacuation planning on the island of Stromboli, Italy.” Journal of Applied Volcanology 10, 4. https://doi.org/10.1186/s13617-021-00104-9
10. Chen, P. 2019. “A Novel Coordinated TOPSIS Based on Coefficient of Variation.” Mathematics 7 (7): 614. https://doi.org/10.3390/math7070614
11. Depari, C.P.S., J. Manalu, and J. Siallagan. 2023. “Strategi Mitigasi Bencana Tsunami Kawasan Pariwisata Teluk Youtefa Kota Jayapura.” Lentera: Multidisciplinary Studies 2 (1): 140–147. In Indonesian. https://doi.org/10.57096/lentera.v2i1.69
12. Donegan, H.A. and F.J. Dodd. 1991. “A Note on Saaty’s Random Indexes.” In Mathematical and Computer Modelling 15 (10): 135–137. https://doi.org/10.1016/0895-7177(91)90098-R
13. Du, Y., W. Chen, K. Cui, and K. Zhang. 2020. “Study on Damage Assessment of Earthen Sites of the Ming Great Wall in Qinghai Province Based on Fuzzy-AHP and AHP-TOPSIS.” International Journal of Architectural Heritage 14 (6): 903–916. https://doi.org/10.1080/15583058.2019.1576241
14. EARTHDATA. 2024. Digital Elevation Model Data. NASA, USA. Available online: https://www.earthdata.nasa.gov/topics/land-surface/digital-elevation-terrain-model-dem
15. Ferreira, M.A., C.S. Oliveira, and R. Francisco. 2025. “Tsunami risk mitigation: the role of evacuation routes, preparedness and urban planning.” Natural Hazards 121, 6719–6751. https://doi.org/10.1007/s11069-024-07061-7
16. Ganjehi, S. and K.N. Khatiri. 2021. “Determination of emergency roads to emergency accommodation using loss analysis results.” Geoenvironmental Disasters 8, 15. https://doi.org/10.1186/s40677-021-00190-2
17. Gloria, L.E., R.U.A. Wiyono, and G. Halik. 2023. “Simulation of High Wave Inundation at Payangan Beach Using Delft3D for Coastal Mitigation.” UKaRsT 7 (1): 60–74. https://doi.org/10.30737/ukarst.v7i1.3233
18. Hafizh, F.R., B.N. Marsekal, C.A. Da Silva Nusantara, and W. Windupranata. 2023. “Mapping of Tsunami Evacuation Routes and Shelters in Batukaras Village, Pangandaran Regency, Indonesia.” IOP Conference Series: Earth and Environmental Science 1245, 012044. https://doi.org/10.1088/1755-1315/1245/1/012044
19. Laksono, F.A.T., A. Widagdo, M.R. Aditama, M.R. Fauzan, and J. Kovács. 2022. “Tsunami Hazard Zone and Multiple Scenarios of Tsunami Evacuation Route at Jetis Beach, Cilacap Regency, Indonesia.” Sustainability 14 (5): 2726. https://doi.org/10.3390/su14052726
20. León, J., A. Gubler, P. Catalán, M. Correa, J. Castañeda, G. Beninati, and A. Ogueda. 2023. “Assessing potential tsunami vertical-evacuation practices: A study of four cases in Chile using virtual reality and GIS.” International Journal of Disaster Risk Reduction 98, 104098. https://doi.org/10.1016/j.ijdrr.2023.104098
21. Liu, K. 2022. “GIS-based MCDM framework combined with coupled multi-hazard assessment for site selection of post-earthquake emergency medical service facilities in Wenchuan, China.” International Journal of Disaster Risk Reduction 73, 102873. https://doi.org/10.1016/j.ijdrr.2022.102873
22. Maramai, A. and S. Tinti. 1997. “The 3 June 1994 Java Tsunami: A Post-Event Survey of the Coastal Effects.” Natural Hazards 15, 31–49. https://doi.org/10.1023/A:1007957224367
23. Marfai, M.A., N. Khakim, H. Fatchurohman, and A.D. Salma. 2021. “Planning tsunami vertical evacuation routes using high-resolution UAV digital elevation model: case study in Drini Coastal Area, Java, Indonesia.” Arabian Journal of Geosciences 14, 2028. https://doi.org/10.1007/s12517-021-08357-9
24. Muhammad, A., R. De Risi, F. De Luca, N. Mori, T. Yasuda, and K. Goda. 2021. “Are current tsunami evacuation approaches safe enough?” Stochastic Environmental Research and Risk Assessment 35, 759–779. https://doi.org/10.1007/s00477-021-02000-5
25. Saaty, T.L. 1980. The Analytic Hierarchy Process: Planning, Priority Setting, Resource Allocation. McGraw-Hill International Book Co., New York, London.
26. Sabani, W., Juhadi, and E. Trihatmoko. 2021. “Participatory Mapping of Tsunami Evacuation Routes (Case Study of Karangbenda Village Cilacap Regency).” IOP Conference Series: Earth and Environmental Science 884, 012033. https://doi.org/10.1088/1755-1315/884/1/012033
27. Sari, I.C., I.N.S. Wijaya, and F. Usman. 2020. “Penentuan Titik Evakuasi Dan Arahan Jalur Evakuasidesa-Desa Di Sepanjangpesisir Kabupaten Jember.” Planning for Urban Region and Environment Journal (PURE) 9 (3): 121–132. In Indonesian. Available online: https://purejournal.ub.ac.id/index.php/pure/article/view/134
28. Sigar, T.R., V. Gunawan, and J.E. Suseno. 2018. “Implementation of the ELECTRE Method for Determining the Location of Evacuation of Web-based Tsunami Disaster.” International Journal of Computer Applications 180 (52): 25–31. https://doi.org/10.5120/ijca2018917362
29. Sofiana, R., R.U.A. Wiyono, and I. Nurtjahjaningtyas. 2022. “Tsunami Mitigation Strategy at Watu Ulo Beach Based on Numerical Modeling Using Delft3D-Flow.” UKaRsT 6 (2): 158. https://doi.org/10.30737/ukarst.v6i2.2959
30. Sofyan, H., R.S. Oktari, J. Fisaini, G. Rahmatina, and A. Hidayati. 2025. “Analysis of Community Preparedness and Tsunami Evacuation Routes: A Case Study in Banda Aceh.” IOP Conference Series: Earth and Environmental Science 1510, 012013. https://doi.org/10.1088/1755-1315/1510/1/012013
31. Tsuji, Y., F. Imamura, H. Matsutomi, C.E. Synolakis, P.T. Nanang, Jumadi, S. Harada, et al. 1995. “Field Survey of the East Java Earthquake and Tsunami of June 3, 1994.” Pure and Applied Geography 144, 839–854. https://doi.org/10.1007/BF00874397
32. Ulandari, N.W.A., N.M. Astiti, I.P. Ramayasa, and I.P.W. Putra. 2024. “Sistem Pendukung Keputusan Evakuasi Prioritas Menggunakan Metode Ahp Dan Mairca Di Bpbd Provinsi Bali.” Naratif: Jurnal Nasional Riset Aolikasi dan Teknik Informatika 6, 2. In Indonesian. https://doi.org/10.53580/naratif.v6i2.290
33. Uslu, A., A. Aktepe, and S. Ersoz. 2016. ”Afet Sonrası Hizmet Verecek Ekiplerin Konuşlanma Yerlerinin Belirlenmesi.” 4th International Symposium on Innovative Technologies in Engineering and Science, 3–5 November 2016 (ISITES 2016 Alanya/Antalya – Turkey). Available online: https://www.isites.info/PastConferences/ISITES2016/ISITES2016/papers/A14-ISITES2016ID32.pdf
34. Vavrek, R. 2019. “Evaluation of the Impact of Selected Weighting Methods on the Results of the TOPSIS Technique.” International Journal of Information Technology and Decision Making 18 (6): 1821–1843. https://doi.org/10.1142/S021962201950041X
35. Villagra, P., C. Quintana, S. Ariccio, and M. Bonaiuto. 2021. “Evacuation intention on the Southern Chilean coast: A psychological and spatial study approach.” Habitat International 117, 102443. https://doi.org/10.1016/j.habitatint.2021.102443
36. Wulan Mei, E.T. and R. Rachmawati. 2016. “Participatory Planning for Evacuation Routes and Meeting Points in Ngargomulyo Village as a Disaster Risk Reduction Effort of Mount Merapi.” Journal of Regional and City Planning 27 (1): 34–48. https://doi.org/10.5614/jrcp.2016.27.1.4