International Journal of Disaster Risk Reduction 14 (2015) 403–410

Contents lists available at ScienceDirect

International Journal of Disaster Risk Reduction

journal homepage: www.elsevier.com/locate/ijdrr

Development of accurate tsunami estimated times of arrival for tsunami-prone cities in ,

Syamsidik a,n, Teuku Muhammad Rasyif b, Shigeru Kato c a Tsunami Computation and Visualization Laboratory of Tsunami and Disaster Mitigation Research Center (TDMRC) and Civil Engineering Department, Syiah Kuala University, Jl. Syeh Abd Rauff, No. 7, 23111 Indonesia b Tsunami and Disaster Mitigation Research Center (TDMRC), Syiah Kuala University and Civil Engineering Master's Program, Syiah Kuala University, Jl. Syeh Abd. Rauff, No. 7, Banda Aceh 231111, Indonesia c Department of Architecture and Civil Engineering, Toyohashi University of Technology, 1-1 Tempaku-Cho, Aichi Prefecture, 441-8580, Japan article info abstract

Article history: Ten years after the Indian Ocean Tsunami of 2004, the Aceh Province in Indonesia has made progress Received 11 April 2015 toward becoming a more tsunami resilient area. The government of Indonesia has undertaken a number Received in revised form of efforts to facilitate a better evacuation process, but a smooth evacuation process is still at large. A 15 September 2015 major issue is that the disaster management office for the districts are yet to have clear, accurate esti- Accepted 16 September 2015 mations of what would be the shortest time of the arrival of the tsunami waves.The worst scenario for Available online 24 September 2015 these cities' tsunami evacuation procedures has yet to be determined. This study seeks to discern worst Keywords: tsunami estimated times of arrival for the tsunami-prone cities in Aceh. A number of numerical simu- City Evacuation lations were performed using data from past events of tsunamis in this area and a future near-shore Tsunami tsunami events anticipated by most of the disaster managers around the Mentawai Island. The Cornell Numerical simulation Multi-grid COupled Tsunami (COMCOT) model was used as the tool for the simulations. Nine selected ETA COMCOT cities were determined as observation points, i.e., Sabang, Banda Aceh, Calang, Lageun, Teunom, Meu- laboh, Singkil, Sinabang, and Tapaktuan. The results of this research show that the shortest ETAs are Banda Aceh 35 min, Sabang 22 min, Lageun 25 min, Calang 29 min, Teunom 29 min, 35 min, Sinabang 20 min, Singkil 53 min, and Tapaktuan 24 min. & 2015 Elsevier Ltd. All rights reserved.

1. Introduction guideline for related stakeholders and emergency responders be- fore and during emergency period [31]. Visible tsunami evacuation The Indian Ocean Tsunami on December 26, 2004 left behind a signs are found in most of the cities. Although not on a regular stark landscape and initiated a drastic period of change for Aceh basis, the government and population of Aceh have performed Province and the Nias Island of Indonesia. Approximately 230,000 some tsunami evacuation drills. However, disaster management people died and nearly 1,000,000 people suffering from clear authorities still have not figure out how to be prepared for a worst images of the extensive damage caused by the tsunami were case scenario for emergency response to future tsunamis. In ad- displaced [29]. Ten years after the tsunami, Aceh has been re- dition, they are unsure of how such a scenario should be in- covered from the severe destruction and is now in much better corporated into the existing SOP. One complication is that in cities that were not affected by the 2004 tsunami, the perception and condition. A four-year rehabilitation and reconstruction process the preparedness efforts are different or somewhat lacking, leav- conducted in Aceh and the Nias Islands of Indonesia has brought ing them ill-prepared [26]. significant changes to the area, including infrastructure improve- Concerns became even greater after the two major earthquakes ments, new settlements, and new livelihoods [22–25]. At the (8.6 Mw and 8.2 Mw) that occurred on April 11, 2012. The earth- present, Aceh has built eight tsunami escape buildings, a number quakes' epicenters were located off the western part of the Sunda- of siren towers, and several well-designed tsunami escape roads. A Subduction Zone [30]. Although they did not produce a giant provincial standard operating procedure (SOP) for tsunami eva- tsunami, they did cause massive panic in major coastal cities in cuation has been formulated since 2011 [16]. The SOP is an official Aceh. The newly-constructed evacuation facilities were not able to accommodate the sudden spike in traffic. People had not been

n trained on how to select the best modes of evacuation. The greater Corresponding author. E-mail addresses: [email protected] (Syamsidik), number of cars on the roads overloaded them and soon created [email protected] (T.M. Rasyif), [email protected] (S. Kato). traffic jams at many places in the cities. In contrast, in the 2004 http://dx.doi.org/10.1016/j.ijdrr.2015.09.006 2212-4209/& 2015 Elsevier Ltd. All rights reserved. 404 Syamsidik et al. / International Journal of Disaster Risk Reduction 14 (2015) 403–410

Fig. 1. Area simulation and points for numerical simulations. Simulations were done in several runs of the process. The color of the rectangles represent areas in the same run of the numerical simulation process. The numbers refer to number of layers of the simulation in Table 1. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) tsunami, the existing escape buildings were almost totally de- not only for nearfield tsunamis, where the ETA is faster than serted. And to make matters worse, the siren towers did not work 30 min, but also for farfield tsunamis. A study conducted around properly [17]. In the 2012 event, people rushed to evacuate due to the southern coast of Myanmar indicated that this area was their experience during the 2004 tsunami. Another good sign was stricken by a tsunami 2.5 h after the first earthquake that triggered that none of them went to the coastal area to pick fish unlike what the December 26, 2004 tsunami [12]. Since the people did not feel happened in the 2004 tsunami. Some schools who have regular any shaking from the earthquake, the ETA information would have training and SOPs for tsunami and earthquake disasters managed helped the community and the government to anticipate the to evacuate their students calmly [27]. However, most of the tsunami [12]. However, those study results are yet to be included evacuation facilities are found in Banda Aceh, the capital city of to improve future evacuation procedures for the study area. Also, Aceh Province. Other cities that have been affected by previous the results have not yet been discussed with emergency managers tsunamis, such as Sabang, Lageun, Calang, Sinabang, and Meulaboh to see the practicality of incorporating the ETA information for were not equipped with such facilities. Although Meulaboh uses those cities. This research uses the Cornell Multi-Grid Coupled two-story shop roofs as shelters, the city still does not have ample Tsunami (COMCOT), a numerical model, to produce ETAs for the escape roads or siren towers. In Tapaktuan and Singkil, the impacts selected cities along the western coasts of Aceh. The results of the of the 2004 tsunami were minor, so the tsunami preparedness modeling were brought into discussions with emergency stake- efforts there have been minimal. Another troubling issue is that holders in the cities to incorporate the results into evacuation these cities have not put any shortest evacuation period into their procedures. evacuation SOPs. Therefore, in this study the goal was to develop and provide worst case tsunami Estimated Times of Arrival (ETAs) for these tsunami-prone cities in Aceh. 2. The study area An ETA is defined as a computed arrival time of the first tsu- nami wave after the occurrence of an earthquake-generating tsu- Except for Banda Aceh, the western coastal cities of Aceh are nami [8,14,13]. The travel time of the tsunami is correlated with well-known to be less populated in comparison to the eastern evacuation time to improve the number of survivors in tsunami- coasts of Aceh. The population of this area is about one third that impacted coastal areas [15]. City evacuation planning should of the northeastern part. Off the western coast of this area is the consider the shortest time that it will take an anticipated tsunami Sunda Fault where the Indo-Australia tectonic plate and Eurasian to get to their coastal area in order to assure that they can evacuate plate collide with each other. According to United State Geological their residents in time. The ETA and number of evacuees will de- Survey (USGS) data, between 1976 and 2012, there have been 64 termine where tsunami shelters should be placed, evacuation earthquakes in this area with magnitudes greater than 6.5 Mw. It routes, evacuation roads' dimensions, as well as evacuation traffic is anticipated that this tectonically active region will produce fu- management [28]. Numerous models were used to study ETA at ture major earthquakes [6]. Off the coast of the western region various locations. ETAs based on historical tsunami scenarios were there are a number of small islands: Simeulue Island, the Banyak found to be key factors in saving lives [15,21]. Zhang et al. [21] Islands, and the Pulo Aceh Islands. These islands are believed to be introduced a rapid method for estimating the arrival time of tsu- a natural barrier providing protection from the propagating tsu- nami waves using linear wave theory and data from tsunamis nami waves generated around the subduction zone. There are nine around the Pacific Ocean that occurred in the 1960s. Another study major coastal cities along this area: Banda Aceh, Sabang, Lageun, conducted around the Tohoku area to revise the ETA during the Calang, Teunom, Meulaboh, Singkil, Sinabang, and Tapaktuan, as March 11, 2012 earthquake and tsunami showed that the ETAs for shown in Fig. 1. The largest population lives in Banda Aceh, the the observed cities are related to their distance relative to the capital city of Aceh Province. During the 2004 Indian Ocean Tsu- source and the seabed forms, especially around the shallow sea nami, more than 70,000 people died. In 2014, the population of near the coasts [7]. Knowing the ETA for a certain area is important Banda Aceh was approximately 249,282 people. The second most Syamsidik et al. / International Journal of Disaster Risk Reduction 14 (2015) 403–410 405 populated city in this area is Meulaboh with about 55,000 re- 3.2. Numerical simulations sidents, according to the 2013 Population Census [1]. Except for Tapaktuan and Singkil, the other seven cities were severely da- A tsunami wave propagation model was simulated using the maged by the 2004 Indian Ocean Tsunami. Cornel Multi-grid Coupled Tsunami (COMCOT) model developed The angle of slope along a coast line has a strong influence on by Liu et al. in 1995 [18]. The model has been successfully applied the tsunami waves. Coastal slopes around Banda Aceh, Lageun, in various tsunami cases, such as the Indian Ocean case [18,20] and Calang, Teunom, and Meulaboh are classified as mildslope. the Tohoku Earthquake and Tsunami. The COMCOT is able to si- Meanwhile, the coastal slope in front of Tapaktuan, Singkil, Sa- mulate multi-size grids (nested grid method). The model applies a fi bang, and Sinabang is rather steep. The coastal profile also influ- leap-frog nite difference scheme that also uses two modes of the ences the deformation of tsunami waves around the nearshore Boussinesq-Shallow Water Equation (SWE), i.e. a linear SWE and a area. The milder the coastal slope, the more destructive the tsu- nonlinear SWE. When the tsunami waves propagate from offshore fi nami waves are to the area [20]. Despite this knowledge, only to near-shore areas, the nonlinear effects increase signi cantly. Banda Aceh, Calang, and Meulaboh have well-designed evacuation However, employing a nonlinear SWE for the whole simulation domain is not effective and is time consuming. The COMCOT offers buildings. Banda Aceh has six such designated evacuation build- the advantage of being able to cluster the application of SWEs in a ings. Meanwhile, Calang and Meulaboh only have one each. The simulated domain with different types of the equation. The linear others do not have any designated evacuation shelters yet. Due to SWE equations adopted by COMCOT are as follows. a lack of funding, no significant evacuation infrastructure im- provements have been made since 2009. Each evacuation building ∂η 1 ⎧ ∂P ∂ ⎫ ∂h + ⎨ + (Q cos ∅)=⎬ − can only facilitate about 500 people should a mass evacuation ∂tRcos ∅ ⎩ ∂ψ ∂∅ ⎭ ∂t ()1 occur [16]. Faced with this fact, the Disaster Management Agency of the West Aceh District decided to utilize two-story shop roofs as tsunami shelters. In contrast, the other cities still rely exclusively ∂P gh ∂η + − fQ = 0 on horizontal evacuation procedures. This decision makes it cri- ∂t R cos φ ∂ψ ()2 tical to know how much time is available for a horizontal eva- cuation in order to save lives. ∂Q gH ∂η + +=fP 0 ∂t R ∂ψ ()3 3. Methods

This research included numerical simulations and Focus Group f = 2sinΩ (φ) ()4 Discussions. The numerical simulations were conducted to obtain where η is the water surface elevation; PQ, denote the volume the shortest ETAs for the selected cities based on historical tsu- fluxes in X (West–East) direction and Y (South–North) direction, nami data and an anticipated future tsunami around the Mentawai respectively; (φψ, ) denote the latitude and longitude of the earth; Islands. The shortest ETAs were later brought into discussions with R is the radius of the earth; g is the gravitational acceleration and fi District Disaster Management Of cers from the selected cities to h is the water depth. And the term ∂∂h/ t reflects the effect of encourage conformance with the existing tsunami evacuation SOP. transient seafloor motion. The f represents the Coriolis force Specific attention was given to showing the practicality of the SOP coefficient due to the rotation of the earth and Ω is the rotation and the ETAs when the dual major earthquakes occurred on April rate of the earth. 11, 2012. Meanwhile, the nonlinear SWEs employed by COMCOT are formulated as follows:

2 3.1. Fault model ∂P gh ∂ ⎧ P ⎫ 1 ∂ ⎧ PQ ⎫ gH ∂η + ⎨ ⎬ + ⎨ ⎬ + − fQ+ Fx ∂t R cos ϕψ∂ ⎩ HR⎭ ∂∅ ⎩ H ⎭ R cos ∅ ∂ψ To mimic the deformation of the seabed due to under-sea 5 earthquakes, this study applied an analytical function developed = 0 () by Masinha and Smylie combined with Okada theories [5,9]. The model uses finite rectangular sources where focal mechanism 2 ∂Q 1 ∂ ⎧ PQ ⎫ 1 ∂ ⎧ Q ⎫ gH ∂η parameters and rupture parameters are inserted for this study. The + ⎨ ⎬ + ⎨ ⎬ + ++=fP Fy 0 focal mechanism parameters were obtained from the USGS data- ∂tRcos ϕψ∂ ⎩ HR⎭ ∂∅ ⎩ H ⎭ R ∂ϕ ()6 base of previous earthquakes and tsunamis and the parameters for future tsunamis were obtained from parameters published by gn2 other researchers [11,4,2]. More details about the parameters can F PP22 Q1/2 x=+7/3 be seen in the simulation scenario section. The empirical re- H () ()7 lationship between the magnitude of an earthquake and the size of the rupture model has been applied [19]. The empirical equa- gn2 tion produces these dimensions of the rupture area; dislocation, F QP22 Q1/2 y=+7/3 () 8 fault length, and width of the area. H () The COMCOT fault models are classified as either instantaneous or transient seafloor deformation models. This study adopted the Hh 9 instantaneous fault model that calculates the seafloor deformation =+η () fi using elastic nite fault plane theory. The instantaneous fault where H is the total water depth; Fx and Fy represent the bottom model assumes the rupture area to be a rectangular fault plane friction in the ψ and φ directions, respectively; and n is the and also assumes that the water volume displaced by the earth- Manning's roughness coefficient. quake is equal to the shape of the deformed seabed. In addition, the dual-modes of the SWEs in COMCOT enable us 406 Syamsidik et al. / International Journal of Disaster Risk Reduction 14 (2015) 403–410

Table 1 Model parameters for the nested grids employed in the numerical simulation.

LayerNo. Extent of Grid GridSize Number ofGrid CoordinateSystem SWEType

Longitude Latitude

1 From 79.2 to 107.6 From 13.6 to 18.2 1 min 1704 Spherical Linear À 1851 m 1908 Spherical 21 From 91.734 to 98.126 From 1.864 to 8.125 0.2 min 1917 Spherical Linear 370.2 m 1284 Spherical 22 From 94.81 to 98.39 From 0.31 to 4.59 0.25 min 859 Spherical Linear 462.75 m 1027 Spherical 31 From 94.5 to 97.212 From 3.0 to 4.875 0.0667 min 2439 Spherical Non-linear 123.4 m 1686 Spherical 32 From 94.92 to 95.573 From 5.41 to 5.98 0.04 min 979 Spherical Non-linear 74.04 m 855 Spherical 33 From 96.821 to 97.99 From 1.669 to 3.79 0.0833 min 842 Spherical Non-linear 154.25 m 1527 Spherical 34 From 95.42 to 96.9 From 2.02 to 3.18 0.0833 min 1066 Spherical Non-linear 154.25 m 835 Spherical to develop faster and more accurate results, especially for near- 3.3. Simulation scenario shore wave diffraction effects where it takes place mostly around the coastal cities. In this study, we used three layers of simulation Numerical simulations were performed based on two groups of grids. The locations of the simulation domain and the simulation simulation scenarios, i.e., a Past Event Scenario and a Mentawai layers can be seen in Fig. 1. Details of the simulation layers and the Scenario. The Past Event Scenario was developed from information type of SWE applied to each layer can be seen in Table 1. Due to on eight past major earthquakes with magnitudes greater than 7.0 Mw. Not all of them did generate a destructive tsunami, but all the large simulation domain area adopted in the simulation, a of them created massive evacuations throughout the south- spherical coordinate system was used. Bathymetric data was ob- western cities of Aceh. The focal mechanism for each of the tained from two sources; GEBCO for offshore bathymetry data and earthquakes was obtained from the USGS database. Table 2 shows data from the Nautical Chart produced by the Indonesia Geospatial the Past Event Scenario applied in the numerical simulations. The Information Agency (BIG) was used for near-shore area close to the empirical equation as explained previously was not used for the study area. One detailed bathymetric measurement was done 2004 Indian Ocean tsunami, the 2005 Nias, and the Mentawai around the Meulaboh coast in September 2011 to update the data megathrust. The first two events used the multi-fault model de- on the coastal bed form. Topography data for the simulation used rived by Romano [11] and Banerjee et al. [32]. The Mentawai SRTM30 data with the resolution of the data at about 0.5 min. The Scenario was based on several studies conducted around the Su- Manning roughness coefficient was applied for the seawater area matra Subduction Zone. One of the most anticipated scenarios was in the domain, which was assumed to be equal to 0.02. As this a tsunami generated from around the Mentawai Megathrust [35]. research did not consider the tsunami run-up process on the Detail data for the fault parameter of the Mentawai Scenario was landside area, types of land use for the study area were not based on Kongko [4]. significant. The observation points in the simulations were located at a 3.4. Focus group discussions contour depth of 10 m in the near-shore zones of the study cities. A number of Disaster Managers from Aceh Jaya Districts, the The definition of the ETA was taken from the Indonesia Geophysic, West Aceh District, and the South Aceh District were gathered Climate, and Meteorological Agency (BMKG). The ETA is confirmed separately for Focus Group Discussions (FGDs). The FGDs were when the tsunami waves reach 0.5 m from the Mean Sea Level done between September 2014 and October 2014. The West Aceh near the coastal area [31]. Therefore, in these simulations, the District and the Aceh Jaya District had already formulated their fi same de nition was adopted. Validation of the numerical simu- tsunami evacuation SOPs. However, the South Aceh District has yet lations was made to assure agreement of the results with the to compose any document related to such emergency procedures. measured data. Details of the validation process are discussed in The ETAs produced from these numerical simulations were the other sections. brought into the FGDs as one of the key components of questions

Table 2 List of fault parameters for past scenarios.

No Date Coordinate Fault parameters Depth(km) Mw Length(km) Width(km) Dislocation(m)

Latitude (deg) Longitude (deg) Strike Dip Slip

1 26/12/2004 94.26 3.09 329 8 110 28.6 9 [11] [11] [11] 2 21/03/2005 97.07 1.67 333 8 118 25.8 8.6 [32] [32] [32] 3 12/09/2007 100.99 3.78 328 9 114 24.4 8.5 328 47 7.3 À 4 12/09/2007 100.13 2.46 317 19 102 43.1 7.9 145 42.6 4.12 À 5 06/04/2010 96.74 2.07 307 7 88 17.6 7.8 145.2 30 4.1 6 25/10/2010 99.32 3.71 316 8 96 12 7.8 145 30 4.1 À 7 10/01/2012 93.209 2.452 195 79 16 10 7.2 79 16 1.9 8 11/04/2012 93.1 2.244 201 75 6 40 8.6 580 40 13.5 Syamsidik et al. / International Journal of Disaster Risk Reduction 14 (2015) 403–410 407 for the FGD participants. Key questions delivered in the FGDs were wave propagations to the observed study area are relatively structured into three main parts. The first part of the questions reliable. was about the main city's facilities assigned for tsunami evacua- tion and future development program. The second part was about their perception on the ETAs produced by these simulations and 4. Estimated time of arrival (ETAs) how the ETA could be incorporated into the district's evacuation procedures. These questions were also about how much time Observation points of the ETAs were placed at several contour would be needed to evacuate an entire city population to desig- lines close to the Aceh cities. However, the simulation results show nated safe points. The last of the questions were about what would that a contour depth of 10 m gives the most stable results for a be the best way to communicate the ETA information and the ci- coastal area [10]. Based on this, the observation points used for ty's tsunami preparedness to their community members. determining the ETAs were deployed at a 10 m depth close enough to the coast lines of the cities. For the observation area, the contour 3.5. Validation of numerical simulations lines of 10 m are located between 50 m and 150 m from the coast line. The distance of the observation points to the nearest coast Validation of the numerical simulations was performed for five line of the city depends on the coastal slope. ETAs from all of the tsunamis that occurred on: simulation scenarios for the selected cities are shown in Table 3. The minimum and moderate ETAs are critical factors in the December 26, 2004 time necessary to complete evacuation procedures and for in-  March 21, 2005 tegration into city planning respectively. Aceh's cities were divided  September 12, 2007 into the two categories; Minimum and Moderate ETAs. Minimun fi  April 6, 2010 ETA is the shortest ETA. Meanwhile, the Moderate ETA is classi ed  April 11, 2012. as the second shortest ETA for the observed area if available.  During FGDs with disaster management agency officers at district Validation of the 2004 Indian Ocean tsunami was conducted level, the moderate ETA was asked. This is for alternative solutions using Jason 1 Satellite data [3]. The satellite measured the sea level to the district in providing tsunami evacuation in case the shortest fi around the Indian Ocean about two hours after the tsunami was ETA is dif cult to meet. Moderate ETAs were established for La- geun 62 min, Calang 72 min, Meulaboh 78 min, and Tapaktuan first generated. The other tsunamis were validated by data from 37 min. The minimum (shortest) ETAs obtained from the simula- Deep Ocean Assessment and Reporting of Tsunami's (DART) buoys tions were for: Sabang – 22 min, Banda Aceh – 35 min, Lageun – operated by NOAA of the United States. Fig. 2 shows a map of the 25 min, Teunom – 29 min, Calang – 29 min, Meulaboh – 35 min, buoys used for validations the 2005, 2007, 2010, and 2012 events. Singkil 53 min, Sinabang 20 min, and Tapaktuan – 24 min. The Under normal conditions, the buoys record the sea level every 2004 Indian Ocean tsunami produced the shortest ETAs for Sa- 15 min. During a major earthquake, the DART buoys measure sea bang, Banda Aceh, Lageun, Teunom, Calang, Sinabang, and Meu- level changes every 15 s. Fig. 3a illustrates validation for the 2004 laboh. But the minimum ETAs for Singkil and Tapaktuan were tsunami. produced by the 2005 tsunami, where the epicenter was located Fig. 3b–e represents the tsunami validations for 2005, 2007, near Nias Island. The minimum ETA for the six cities clarifies that 2010, and 2012 tsunamis, respectively. All of the validation graphs the 2004 tsunami was the worst scenario for most of the cities. confirm that the simulated results strongly correlate with either The other tsunamis generated off the western coast did not pro- the Jason 1 satellite data (2004 tsunami) or the DART buoy mea- duce the shortest ETAs, although they were close to these cities. sured data for the other tsunamis. They clarify that the tsunami Shoreline shapes around Meulaboh are rather concave from the landside, which results in a longer tsunami arrival time compared to where the shoreline is convex, as found in the case of Ta- paktuan. The ETA for Meulaboh is longer than for cities nearby, such as Teunom and Calang [10]. The results from the Mentawai anticipating scenario did pro- duced an ETA of more than two hours to the Aceh's coastal cities. This knowledge will enable the cities to be adequately prepared for tsunami evacuations should a Mentawai Megathrust occur. However, it is important to highlight that a tsunami generated in the Mentawai scenario will not be accompanied by receding water around the coast of the western cities of Aceh. This is in stark contrast to the past tsunami scenarios that have occurred around the western coast of Aceh. It should also be noted that the ETAs produced in this study are not aimed at providing real time data, but rather at providing initial information for authorities to un- derstand quickly what would be the escape window timeframe for their cities in the face of an oncoming tsunami.

5. Tsunami evacuation plans

The data obtained from the numerical simulations were pro- vided to the disaster management authorities at Calang, Lageun, Meulaboh, and Tapaktuan. The tsunami evacuation process for Banda Aceh was obtained from a report produced by TIM based on Fig. 2. Location of the DART buoys for numerical results validations. the April 11, 2012 earthquakes [17]. The report was also validated 408 Syamsidik et al. / International Journal of Disaster Risk Reduction 14 (2015) 403–410

Fig. 3. Tsunami height comparisons between the COMCOT model and; (a) Satellite Jason1 in 2004, (b) DART at Cocos Island in 2005, (c) DART at Phuket, September 12, 2007, (d) DART at Phuket, April 6, 2010, and (e) DART at Phuket, April 11, 2012.

Table 3 jams around Banda Aceh triggered by tsunami events have been The ETA values in the study cities. confirmed through a number of tsunami evacuation modeling [33,34]. This was also confirmed by Local Disaster Management No Name of the city Minimum (min) Moderate (min) Agency staffs who are in charge of the city's evacuation process. 1 Sabang 22 22 Although, there are six designated escape buildings around the 2 Banda Aceh 35 35 city, it was hard to attract the maximum number of evacuees that 3 Lageun 25 62 the facilities could shelter during the evacuation. 4 Calang 29 72 5 Teunom 29 72 In Calang and Lageun, the two cities have smaller populations 6 Meulaboh 35 78 and they have several hills around the cities that can be used for 7 Tapaktuan 24 37 evacuation. Although Calang only has one escape building, the 8 Singkil 53 53 surrounding hills can accommodate a large number of evacuees. 9 Sinabang 20 26 On April 11, 2012, the Local Disaster Management Agency revealed that a large number of people used the hills during the evacua- through an interview with the Local Disaster Management Agency tions. The escape building in this city was also used by a significant in Aceh. number of evacuees. On normal days, the escape building is also In Banda Aceh, it was estimated that one hour (60 min) would used as the office for the Local Disaster Management Agency. A be needed to evacuate the city's population to a safer area. The number of community-based disaster risk reduction activities tsunami safer areas in Banda Aceh are located about 4 km from the were performed at this building. This helped the Agency to con- coast line. During normal conditions, it takes about 30 min by nect the use of the escape building to the local community. During motorcycle to reach these areas. With a sudden transport volume an FGD, it was estimated that the city would need about 35 min to increase, the time to safety would increase due to traffic jams as evacuate the entire population of the city to safer points. observed after the two earthquakes on April 11, 2012. The traffic In Meulaboh, the evacuation time is estimated to be about Syamsidik et al. / International Journal of Disaster Risk Reduction 14 (2015) 403–410 409

50 min. This city does not have higher ground suitable for tsunami enact tsunami mitigation based spatial planning around their evacuation. During panic situations triggered by earthquakes in coastal areas. March 2005, there was severe traffic congestion during the eva- cuation. This situation inspired the Local Disaster Management Agency to improve the evacuation procedure of the city with measures such as, providing enough staff to direct the evacuation 6. Conclusions process, especially at critical junctions or roads. These attempts were proven effective at reducing the evacuation time and mini- This study explored the opportunity to develop tsunami Esti- mizing traffic accidents during the evacuation. However, the eva- mated Times of Arrival for these southwestern cities of Aceh; La- cuation time of Meulaboh was still longer than the simulated ETA geun, Calang, Meulaboh, and Tapaktuan. Except for Tapaktuan, all for this city. of them were severely damaged in the 2004 Indian Ocean Tsu- Conversely, Tapaktuan is one of the cities that does not have nami. The ETAs were produced using numerical simulations that any preparation for anticipated future tsunamis. In previous tsu- incorporated several past tsunami scenarios. The ETAs were di- namis, this city received only minor impacts, which has led to only vided into two categories, i.e., Minimum ETA and Moderate ETA. minimum preparation by local government. Based on an FGD The minimum ETAs were Lageun 25 min, Calang 29 min, Meula- conducted in Tapaktuan, it was found that the city would need boh 35 min, and Tapaktuan 24 min. The 2004 Indian Ocean tsu- about 60 min to evacuate its population to safer points. Although nami produced these minimum ETAs, except for Tapaktuan where this city is surrounded by hills, in this case, the higher ground the minimum ETA was generated from the 2005 tsunami. Mod- areas are not well suited to accommodate tsunami evacuees. There erate ETAs were confirmed for Lageun 62 min, Calang 72 min, are no roads, stairways, or tsunami evacuation signs. No tsunami evacuation drills have been performed in this city. In fact, in 2005, Meulaboh 78 min, and Tapaktuan 37 min. In terms of tsunami a tsunami struck this area and resulted in damage to a number of evacuation procedures, none of the authorities from these cities houses around its coastal areas. Comparing this information to the are ready to anticipate their minimum ETAs. Despite several efforts simulated ETA has revealed a significant gap in the tsunami eva- to revise the tsunami evacuation procedures for these cities, the cuation time for Tapaktuan. A comparison of the Time of Official emergency facilities are still insufficient to provide evacuees with Warning Released by the Indonesia Geophysical, Climate and safety in the face of their minimum ETAs. These cities need more Meteorological Agency (BMKG) [17] and the simulated ETAs for evacuation roads and shelters and more actual tsunami drills to each city and Estimated time for Tsunami Evacuation (ETE) of the perform effective immediate tsunami evacuation procedures. city's population can be seen in Fig. 4. This figure shows that all of the selected cities have a shorter ETA than the city's current eva- cuation time (ETE). Banda Aceh, Meulaboh, and Tapaktuan have a Acknowledgments serious challenge to reduce this gap. To do so, they could construct more escape buildings around the coastal areas, widen evacuation The authors are thankful to the Toyohashi University of Tech- roads, utilize the hills around the city (as does Tapaktuan), and nology who facilitated support during training and researcher conduct regular tsunami drills. Among the nine cities considered exchanges with the Civil Engineering Master's Program of Syiah in this study, Banda Aceh, Lageun, Calang, Teunom, Meulaboh, and Kuala University under the JASSO Program. This study was par- Singkil already have their official Tsunami Risk Maps. However, tially funded by the Directorate of Higher Education (DIKTI) of the Tapaktuan and Sinabang have yet to formulate official tsunami risk Ministry of Education of Indonesia through a research Grant from maps to further develop their evacuation procedures. STRANAS No. 1702 APPN in 2012. FGDs for this study were funded Based on the FGDs and the numerical simulations, this study by a Hibah Pasca Sarjana research Grant no. 2940 IHPS from the revealed a number of serious gaps in evacuation process at most of the cities in Aceh. In addition, this area is still classified to be Syiah Kuala University in 2014. Authors are also grateful to USAID tsunami prone area where tsunami wave heights higher than and National Academy of Sciences who fund this publication 0.5 m have a higher probability of occurrence compared to other process through PEER Cycle 3, Sponsor Grant Award Number: AID- regions in Indonesia [36]. The local disaster management agencies OAAA-A-11-00012 and Sub Grant Number PGA-2000004893 with acknowledged there are a number of challenges to fill the gaps. research project title: Tsunami Waves Impacts on Coastal Mor- Among them are to regulate emergency traffic management and to phological Changes Based on Sediment Transport Simulations.

Fig. 4. Comparison between the Official Warning Time released by BMKG, simulated ETAs, and estimated tsunami evacuation times for Banda Aceh, Lageun, Calang, Meulaboh, and Tapaktuan. P is the estimated population of the city. W1 the first tsunami early warning after 5 min of an earthquake, and W2 is the second warning after ¼ ¼ 10 min of an earthquake. 410 Syamsidik et al. / International Journal of Disaster Risk Reduction 14 (2015) 403–410

References [18] X. Wang, Liu PLF, An analysis of the 2004 earthquake fault plane mechanism and Indian Ocean Tsunami, J. Hydraulic Res. 44 (2) (2006) 147–154. [1] BPS Aceh (Aceh Provincial Statistical Bearau), Aceh in Numbers 2012 (in In- [19] D.L. Wells, K.J. Coppersmith, New empirical relationship among magnitude, donesia), Banda Aceh, Indonesia, 2012. rupture length, rupture width, rupture area, and surface displacement, Bull. [2] Z. Duputel, H. Kanamori, V.C. Tsai, L. Rivera, L. Meng, J.P. Ampuero, J.M. Stock, Seismol. Soc. Am. 84 (4) (1994) 974–1002. The 2012 Sumatra great earthquake sequence, Earth Planet. Sci. Lett. 351 [20] J.J. Wijetunge, X. Wang, Liu PLF, Indian Ocean Tsunami on December 26, 2004: (2012) 247–257. numerical modeling of inundation in three cities on the south coast of Sri [3] J. Gower, Jason-1 detects the December 26, 2004 tsunami, EOSTrans. AGU 86 Lanka, J. Earthquake Tsunami 2 (2) (2008) 133–155. (2005) 37–38. [21] D.H. Zhang, T.L. Yip, C.O. Ng, Predicting Tsunami arrivals: estimates and policy [4] W. Kongko, South Java Tsunami Model using Highly Resolved Data and implications, Mar. Policy 333 (2009) 643650. Probable Tsunamigenic Sources (Doctoral Thesis), Leibniz Universitat, Hann- [22] BRR NAD-NIAS, Infrastructure: Stimulating the Triggering Factor, (Aceh Nias over, Germany, 2011. Rehabilitation and Reconstruction Agency), Banda Aceh, Indonesia, 2009. [5] L. Masinha, D.E. Smylie, The displacement fields of incline faults, Bull. Seismol. [23] BRR NAD-NIAS, TSUNAMI From Disaster to the Emergence of Light, (Aceh Nias Soc. Am. 61 (5) (1971) 1433–1440. Rehabilitation and Reconstruction Agency), Banda Aceh, Indonesia, 2009. [6] R. McCaffrey, The tectonic framework of the Sumatran subduction zone, Annu. [24] BRR NAD-NIAS, HOUSING: Roofing the Pillars of Hope, (Aceh Nias Re- Rev. Earth Planet. Sci. 37 (2009) 345–366. habilitation and Reconstruction Agency), Banda Aceh, Indonesia, 2009. [7] A. Muhari, F. Imamura, A. Suppasri, E. Mas, Tsunami arrival time characteristics [25] BRR NAD-NIAS, CASE STUDY: the Scattered Beads, (Aceh Nias Rehabilitation of the 2011 East Japan Tsunami obtained from eyewitness accounts, evidence, and Reconstruction Agency), Banda Aceh, Indonesia, 2009. and numerical simulation, J. Natural Disaster Sci. 34 (1) (2012) 91–104. [26] Indra, Z.A. Muchlisin, A. Sabti, N. Fadli, Syamsidik, Bustamam, Ibrahim, A Study [8] National Oceanic and Atmospheric Association (NOAA). Tsunami Terminology. on Disaster Risk Reduction and Community Preparedness in West Aceh Dis- The National Tsunami Hazard Mitigation Program, History 1995–2005. trict towards Tsunami Disaster. A Report, Tsunami and Disaster Mitigation 〈Nthmp-history.pmel.noaa.gov/terms.html〉, 2014 (accessed 02.07.14). Research Center, Banda Aceh, Indonesia, 2012 (in Indonesian). [9] Y. Okada, Surface deformation due to shear and tensile faults in a half-space, [27] R.S. Oktari, K. Shiwaku, K. Munadi, Syamsidik, R. Shaw, A conceptual model of Bull. Seismol. Soc. Am. 75 (4) (1985) 1135–1154. [10] T.M. Rasyif, Database development of estimated time of arrival for tsunami a school-community collaborative network in enhancing coastal community disaster mitigation in southwestern cities of Aceh, Indonesia (Master's Thesis), resilience in Banda Aceh, Indonesia, Int. J. Disaster Risk Reduct. 12 (2015) Civil Engineering Master's Program of Syiah Kuala University, Banda Aceh, 300–310. Indonesia, 2014. [28] BSN (Indonesia National Standardization Agency). Indonesia National Stan- [11] F. Romano, The Rupture Process of Recent Tsunamigenic Earthquake by Geo- dard on Tsunami Evacuation Routes (in Indonesian). Jakarta, Indonesia, 2012. physical Data Inversion (Doctoral Thesis), Universita Degli, Bologna, Italy, [29] USGS (United States Geological Survey). Magnitude 9.1-Off the West Coast of 2009. the Northern Sumatra. 〈http://earthquake.usgs.gov/earthquakes/eqinthenews/ [12] K. Satake, T.T. Aung, Y. Sawai, Y. Okamura, K.S. Win, W. Swe, C. Swe, T.L. Swe, S. 2004/us2004slav/#summary〉, 2005 (accessed 01.04.15). T. Tun, M.M. Soe, T.Z. Oo, S.H. Zaw, Tsunami heights and damage along the [30] USGS (United States Geological Survey). Magnitude 9.6 Earthquake strikes Myanmar coast from the December 2004 Sumatra–Andaman earthquake, Northern Sumatra, Indonesia. 〈http://www.usgs.gov/blogs/features/usgs_top_ Earth Planets Space 58 (2006) 243–252. story/magnitude-8-6-earthquake-strikes-northern-sumatra-indonesia/〉,2012 [13] T. Steinmetz, U. Raape, S. Tebmann, C. Strobl, M. Friedemann, T. Kukofka, (accessed 01.04.15). T. Riedlinger, E. Mikusch, S. Dech, Tsunami early warning and decision sup- [31] BMKG (Indonesia Geophysical, Climate and Meteorological Agency). Tsunami port, Nat. Hazards Earth Syst. Sci. 10 (2010) 1839–1850. Early Warning Service Guideline (in Indonesian), Jakarta, Indonesia, 2012. [14] G. Strunz, J. Post, K. Zosseder, S. Wegscheider, M. Muck, T. Riedlinger, H. Mehl, [32] P. Banerjee, F.F. Pollitz, B. Nagarajan, R. Burgmann, Coseis- mic slip distribu- S. Dech, J. Birkman, N. Gebert, H. Harjono, H.Z. Anwar, Sumaryono, R. tions of the 26 December 2004 Sumatra–Andaman and 28 March 2005 Nias M. KhomaruRMddin, A. Muhari, Tsunami risk assessment in Indonesia, Nat. earthquake from GPS static offsets, Bull. Seism. Soc. Am. 97 (1A) (2007) Hazards Earth Syst. Sci. 11 (2011) 67–82. S86–S102. [15] Syamsidik and Istiyanto DC, Tsunami mitigation measures for tsunami prone [33] Y. Goto, M. Affan, Agussabti, Y. Nurdin, D.K. Yuliana, Ardiansyah, Tsunami small islands: lessons learned from the 2010 tsunami around the Mentawai evacuation simulation for disaster education and city planning, J. Disaster Res. Islands of Indonesia, J. Earthquake and Tsunami 7 (1) (2013) 1350002. 7 (1) (2012) 92–101. [16] TDMRC (Tsunami and Disaster Mitigation Research Center). Standard Opera- [34] Y. Goto, Y. Ogawa, T. Komura, Tsunami disaster reduction using town watching tional Procedure (SOP) for the Use of Evacuating Building/Escape Building, and moving tsunami evacuation animation trial in Banda Aceh, J. Earthquake Banda Aceh Indonesia, 2011. Tsunami 04 (02) (2010) 115–126. [17] Rapid Assessment Team (TIM), Evaluation of Tsunami Early Warning System [35] K. Megawati, T.-C. Pan, Regional seismic hazard posed by the Mentawai seg- during Earthquakes and Tsunamis in Aceh on April 11, 2012, in: E. Yulianto I. ment of the Sumatran megathrust, Bull. Seismological Soc. Am. 99 (2A) (2009) Rafliana(Eds.), published by GITEWS, Jakarta, Indonesia. 〈http://www.gitews. 566–584. org/tsunami-kit/en/E4/further_resources/case_study/Evaluasi%20Sistem% [36] N. Horspool, I. Pranantyo, J. Griffin, H. Latief, D.H. Natawidjaja, W. Kongko, 20Peringatan%20Dini%20Tsunami%20pada%20Kejadian%20Gempabumi%20% A. Cipta, B. Bustamam, S.D. Anugrah, H.K. Thio, A probabilistic tsunami hazard 20Tsunami%20Aceh%2011%20April%202012.pdf〉, 2012 (accessed 02.03.15). assessment for Indonesia, Nat. Hazards Earth Syst. Sci. 14 (2014) 3105–3122.