Decision on Tsunami Evacuation Route in Tourism Area: A Case Study of Had Patong, Phuket

Decision on Tsunami Evacuation Route in Tourism Area: A Case Study of Had Patong, Phuket Thirayoot LIMANOND Assistant Professor School of Transportation Engineering Asian Institute of Technology P.O. Box 4, Klong Luang, Pathumthani 12120, Thailand Fax: +66-2-524-5681 E-mail: tlimanond@ait.ac.th Siradol SIRIDHARA Lecturer School of Transportation Engineering Suranaree University of Technology 111 University Avenue, T. Suranaree Nakhon Ratchasima, 30000 Thailand Fax: +66-0-4422-4608 E-mail: siradol74@yahoo.com Chutima CHERMKHUNTHOD Research Assistant School of Transportation Engineering Suranaree University of Technology 111 University Avenue, T. Suranaree Nakhon Ratchasima, 30000 Thailand Fax: +66-0-4422-4608 E-mail: jdeu_1415@hotmail.com Hyunmyung KIM Assistant Professor Department of Transportation Engineering Myong Ji University San 38-2 Nam-Dong, Cheoin-gu, Youngin-si Gyeonggi-do, 449-728, Korea Fax: 82-31-336-2885, E-mail: khclsy@gmail.com Chalat TIPAKORNKIAT Ph.D. Candidate School of Engineering and Technology Asian Institute of Technology P.O. Box 4, Klong Luang, Pathumthani 12120, Thailand Fax: +66-2-516-2126 E-mail: Chalat.Tipakornkiat@ait.ac.th Savalee UTTRA Research Assistant School of Transportation Engineering Suranaree University of Technology 111 University Avenue, T. Suranaree Nakhon Ratchasima, 30000 Thailand Fax: +66-0-4422-4608 E-mail: jangyoyo@hotlmail.com Abstract: In the wake of the devastating 2004 tsunami, a number of researches shifted their focus to disaster prevention, management and alleviation strategies. Most studies developed their plans through reviewing and modeling the characteristics of the great tidal wave. This study from Phuket Island, nonetheless, investigated the other side of the subject from the evacuees perspective. Route choice decision and its governing factors were determined. Their relationship to socioeconomic characteristics of local residents and foreign tourists were analyzed and quantified using a three-way contingency table technique. The study found an astonishing fact of an insignificant role of evacuation signage in guiding the evacuation route, as compared to individual instinct and the crowd reaction. The conclusion of the study would be the first step to help understand the crowd behavior and to help establish an effective evacuation measures under this specific emergency situation. Key Words: tsunami, evacuation plan, route choice, signage 16

1. INTRODUCTION In the morning of 26 th December 2004, a catastrophic tsunami struck the shores of 19 countries including Thailand. The disaster claimed dreadfully high death tolls and completely destroyed buildings and roadway infrastructure along its way. The most affected areas along the west coast of Thailand included Patong beach in Phuket, Kao Lak in Phan-nga, Phi Phi Island in Krabi (Ghobarah et al., 2006) all of which were among the most famous and beautiful beaches that had attracted hundred thousands of tourists from all over the world. Such locations were devastated with different degrees of severity depending on the geographic characteristic such as land forms, slope, elevation and available natural barriers (Srinivas and Nakagawa, 2008). In such famous tourist destinations, like Phuket, the unexpected disaster could cause casualties to as many foreigners as local residents. According to Srivichai et al. (2007), 279 casualties and 608 missing were reported in Phuket province alone. Out of these fatalities, 151 were Thais, 111 were foreigners, while 17 unidentified. Those who are missing comprise 245 Thais and 363 foreigners. Tsunami is one of the most devastating natural coastal disaster (Liu et al., 2009). A few key characteristics of tsunami separated the evacuation strategy from those of other natural disasters including hurricanes, wildfires or floods. First, it is nearly impossible to predict the occurrence of tsunami well in advance. The evacuees would receive a very short notice, if not at all, just before the tsunami strikes, leaving them no time to make sound judgment. Tsunami wave speed is approximately 500-800 km/hour (Ghobarah et al., 2006), thus it gives much less time for evacuation (Zhang et al., 2009). The warning system in Padang, Indonesia would only give a 20 40 minutes window before the tsunami reaches the shore (Lammel et al, 2010). Liu et al. (2009) used their numerical model to predict that tsunami, caused by future earthquake in South China Sea could arrive the coasts of China, Vietnam, Malaysia and Taiwan within 1-2 hours. On the other hand, the other natural disasters, such as hurricanes, flood and, to some extent, wildfire, with the state-of-the-art meteorological prediction technology, would provide considerable advance notice before the disaster really begins (Barret et al., 2000). Cancun residents were given as much as 16 hour in advance to evacuate during the Hurrican Gilbert in 1998 (Aguirre, 1991). Second, the destructive force of tsunami is so massive that can destroy anything along its way. People are left no choice but to quickly leave the affect area for their own safety. In the event of hurricane or flood, residents have a choice of staying or leaving their premise. The evacuation compliance rate for 1988 Hurricane Gilbert in Cancun and 1999 Hurricane Floyd in North Carolina was only 25 percent (Aguirre, 1991) and 65 percent (Dow and Cutter, 2002), respectively. Third, Tsunami often struck the tourist spots along the coast that draw many tourists. Tourists have long been known as susceptible to natural disaster (Drabek, 1996) due to the unfamiliarity of locations, weathers, cultures, and local hazards. In natural disaster events with a sufficient advance warning, tourists were typically evacuated first before local residents. In case of tsunami, on the contrary, both tourists and local residents must be evacuated simultaneously in an extremely limited time window. The impacts of the tsunami went well beyond the affected areas, causing crucial changes in many fields including the academia. Up until late 2004, most of the emergency evacuation research had focused on the recurring natural disasters that were regular for specific areas (Aguirre, 1991; Barret et al., 2000; Dow and Cutter, 2002; Lindell et al., 2005). After the 2004 tsunami, more attentions then shifted towards emergency evacuation research. Zhang et al. (2009) developed a linear wave model to predict the tsunami arrival time with satisfied accuracy. Tsai et al. (2009) developed an integrated application system of fuzzy logic and 17

GIS to model tourism disaster evacuation. Liu et al. (2009) used a numerical model to predict the tsunami impact in the South China Sea to surrounding countries. Lammel et al. (2010) developed a multi-agent transportation framework to simulate pedestrian evacuation for a large area. Although there were on-going efforts to study or analyze the evacuation for tsunami, little has been appreciated how people would actually react to the tsunami, and how they would choose their evacuation route. This study hypothesized that people with different characteristics would make a decision on the evacuation route differently. Some may choose their own route, some may follow the evacuation route signage, while some may run following the crowd. Each of the decisions has its own character, and would require specific policies or measures in order to achieve the highest evacuation efficiency. The adopted policies and measures must be well suited to the existing transportation system and the evacuation alarming system in the area. The understanding of the evacuation behavior nature of local residents and tourists would allow the authority to develop and manage an effective evacuation system that is suitable for the area. Given a short notice time and the destructive force of tsunami, the authority has to provide appropriate measures, infrastructure, aids to facilitate crowd reaction properly and quickly in order to save lives and injuries, especially in tourist favorite spots. A few consensus efforts prepared tsunami management system in the tsunami-prone areas, such as, Srivichai et al., 2007; Thanawood et al., 2006. Such system usually included the raising public awareness of residents on tsunami, the identification of the safe evacuation route, the preparation of evacuation drills, and the installation of the advance warning system. The objective of this study was to assess the decision of local residents and foreigners in a tourist spot on the evacuation routes in case of tsunami. More specifically, the characteristics of evacuees will be analyzed to investigate how it will affect the decision for the evacuation route. The results were expected provide an insight evacuation behavior; how people with different degree of the familiarity of the area would react and choose their route under the chaotic situations. The results were the first major stepping stone to evaluate the adequacy of the current adopted evacuation plans. The evaluation would consider whether the evacuation routes were able to accommodate the evacuees sufficiently, and could identify some pitfalls/loopholes under the current plans. In this study, we utilized a stated-preference survey to collect evacuation behavior more than 300 local residents and tourists in Patong beach, and used statistical contingency tables to investigate the major characteristics that would affect the evacuation decision. 2. METHODOLOGY The project study area was the town of Patong on Phuket Island, Thailand. Patong was located west coast and was one of the most famous tourist places in Thailand. The town covered 16.4 square kilometers with a 3.5 kilometers long beach. The total population of Patong was 18,423 persons in April 2008. There was no report on the exact number of visitors of Patong city. However, as many as 4.7 million tourists were reported to visit Phuket in 2007, with 3.1 million foreigners and 1.5 million Thais. Since Patong Beach was one of the most famous tourist spots on Phuket Island, thus, presumably the majority of the tourists visiting Phuket Island would spend some times in Patong area as well. It is estimated that there are approximately 8-9,000 tourists a day in Had Patong beach during the peak season, and the number is dropped to 2-3,000 tourists a day during the off-peak season. 18

In the morning of 26 December 2004, this are was completely washed out by the gigantic tidal wave. Tsunami rambled through the beautiful Patong beach and destroyed infrastructures and buildings of the city. After a year of rehabilitation, Patong area rapidly recovered to normal. The city Most importantly the town was well-prepared for the tsunami phenomenon by developing a tsunami warning system on the beach and local news medias. A several tsunami evacuation drills were conducted. Local residents and tourists were educated and trained how to react to such disastrous event. The researchers conducted an interview of more than 300 local Thais and foreigners randomly selected from the Patong beach area both at the beachfront as well as on the city road network. The questionnaire queried personal data of the respondents (including gender, age, physical condition, education, Thai language skill and the length of stay in Patong), their familiarity with the Patong area, and their regular mode of travel while in Patong. In addition, the surveyors inquired the respondent stated-preference on the evacuation routes upon the upcoming Tsunami case, which included 5 alternatives: 1) confidently evacuate on a familiar route, 2) follow the evacuation route guide sign, 3) follow the crowd, 4) seek assistance from the officers, and 5) others. The final sample includes 306 respondents, consisting of 125 Thais and 182 foreigners. The distribution of characteristics of the sample, categorized by local Thais and foreigners are shown in Table 1. Thai respondents consisted of slightly more male than female. Over 90 percent of Thai respondents were in the age group of 19 59 years old with very few teenagers and persons older than 60 years old. They were predominantly high-school graduates, followed by Bachelor degree holders and junior high-school graduates. Most of them rated their Thai language skills as fluent and fair. Almost 80 percent of them stayed in Patong area for more than 1 year, while approximately 15 percent stayed in Patong less than 1 month. More than 80% of Thai respondents usually use private vehicle to travel in the Patong city area. The group of foreigner respondents comprises of 61.5 percent male and 38.5 percent female. Similarly to Thai samples, they were predominantly in the age group of 19-59 years old but with almost 10 percent of persons older than 60 years old. Foreigners tend to receive a higher education than Thai counterparts as 36.3 percent finished their Bachelor degree, and 16.2 percent completed a post-graduate degree. Only 5 percent of foreigners claimed themselves as moderate to fluent Thai languages, while the majority of them barely had or did not have any Thai language skills. More than 90 percent of them visited Patong as tourists, while almost 80 percent stayed in Patong less than 1 month. Majority of the foreigner respondents were from Europe, and the rest were equally split among Asia, the United States, and others continent. Their modes of transport in Patong included a rental vehicle (47.8 percent), bus (13.2 percent), private vehicle (11 percent) and motorcycle taxi (5.5 percent). This study utilized the contingency tables to investigate the decisions on Tsunami evacuation route. The 2-way contingency table is a tool generally used to analyze two categorical variables of interest. It lists the number of events that fell into two categories as a matrix: one lists by rows, and the other lists by columns. The 2-way table allowed us to understand a general pattern of the occurrences under the two categorical variables simultaneously, for example, evacuation route plan vs gender. The table usually comes with an associated Pearson goodness-of-fit statistics ( 2 ) testing the null hypothesis whether the two variables (listed by rows and by columns) are independent (Wickens, 1989). In this study, we used the two-way tables to analyze the mode choice with the other key factors, such as, activity classes, gender, and activity duration (at the destination). 19

Table 1 General characteristic of the respondents. Thai Foreigners Total no. of sample 125 182 Characteristics Number % Number % Sex Female 58 46.4 70 38.5 Male 67 53.6 112 61.5 Age group <19 yrs. 7 5.6 7 3.8 19-29 yrs. 54 43.2 66 36.3 30-59 yrs. 62 49.6 92 50.5 >60 yrs. 2 1.6 17 9.3 Education Primary Junior Hi school 33 26.6 6 3.4 Hi school/associate 53 42.7 79 44.1 Bachelor 37 29.8 65 36.3 Master Ph.D. 1 0.8 29 16.2 Thai Language Skill None 0 0.0 146 80.2 Barely 9 7.2 27 14.8 Moderate 10 8.0 4 2.2 Fair 30 24.0 3 1.6 Fluent 76 60.8 2 1.1 No. of days in Patong in the past 5 years < 5 days 13 10.9 72 39.6 5 days 1 month 4 3.4 70 38.5 1 month 6 month 3 2.5 23 12.6 6 month 1 year 6 5.0 11 6.0 > 1 year 93 78.2 6 3.3 Trip purpose to Patong Beach Tourism/recreation 9 7.2 163 91.1 Business 0 0.0 12 6.7 Residence 114 92.8 4 2.2 Countries of residence Thailand 125 100.0 0 0.0 Asia 0 0.0 23 12.6 Europe 0 0.0 115 63.2 United State 0 0.0 22 12.1 Others 0 0.0 22 12.1 Usual Mode of Travel Private car/motorcycle 102 82.3 20 11.0 Rental car/motorcycle 8 6.5 87 47.8 Bus/Songtaew 4 3.2 24 13.2 Motorcycle taxi 3 2.4 10 5.5 Others 7 5.6 41 22.5 20

A more complex approach is a 3-way contingency table, which is a powerful tool to analyze the relationship of the three variables simultaneously. It listed one variable by rows and another one variable by columns, and then classified the third variable by layers. The 3-way table can be used to test the independency among the variables in various combinations as follow: Three-way association:- every pair of variables was associated with the remaining one. Homogeneous association:- the conditional relationship between any pair of variables given the third one was constant at each level of the third one. Conditional independence:- one variable was associated to the other two variables. However, those two variables were not directly associated. They may be related only through their association with the first variable. One-factor independence:- One variable was completed independent of the other two variables. Complete independence: All of the three variables were completed independent of one another. In this study, we used the 3-way tables to analyze the relationships among evacuation route decision, residents/tourists and the other influencing factors, such as, gender, age group, education, Thai language skill, the length of stays in Patong area in the past 5 years, regular travel modes. 3. FINDINGS Table 2 illustrates the number of events of various evacuation route decisions by nationality, categorized as Thais and Foreigners. Two numbers were recorded for each nationality/evacuation route decision category. The left number represented the total number of sampled nationalities who chose respective evacuation route decision. The right number indicated the proportion of samples with the same nationality by various evacuation route decisions. It was simply the number of events in that cell, divided by the corresponding row total. The bottom of the table also showed the resulting Pearson goodness-of-fit statistics ( 2 ), which tested the independency of the two factors (by rows and by columns of the table). Table 2 Two-way Table Analysis for Evacuation Route Decision and Physical Condition Evacuation Route Decision Nationality Choose Seek Follow the Follow the familiar assistance guide sign crowd route from officers Others Total N % N % N % N % N % Thais 86 69.3 29 23.4 8 6.5 0 0.0 1 0.8 124 Foreigners 34 18.7 55 30.2 69 37.9 23 12.6 1 0.6 182 Total 120 39.2 84 27.4 77 25.2 23 7.5 2 0.7 306 Goodness of fit statistics 2 Null hypothesis Degree of Pearson χ freedom (DF) Sig. (p>χ 2 ) Complete independence 94.300 4 0.000 21

In general, the preferred evacuation strategies among Thai were to choose their own route (69.3 percent), then following the evacuation route guide signs (23.4 percent), and following the crowd (6.5 percent). Given that most of the Thai respondents were local residents who stayed in Patong beach for more than 1 year, they were presumably familiar with the road network and, logically, could find a way to run away from tsunami. Foreigners, due to unfamiliarity with the area, tended to rely on other sources. Approximately 37.9 percent of them said they would follow the crowd, 30.2 percent would follow the evacuation route guide sign, and 12.6 percent would seek assistance from officers. However, approximately 18.7 percent of them felt confident enough to choose their own familiar route. This group most possibly comprised the tourists who stayed in Patong area for a long time, or recent tourists who were very cautious and already explored the tsunami evacuation route in the area. The Pearson χ 2 showed that evacuation route decision is dependent of nationality. Table 3 showed descriptive statistics concerning evacuation route decision by nationality and gender. Both Thai and Foreigner males relied more on themselves than female counterparts in choosing their own route. Almost 78.8 percent of Thai male respondents chose their own familiar evacuation route, while only 58.6 percent of Thai female respondents would do so. Similarly, 21.4 percent of foreigner males decided their own evacuation routes, compared to 14.3 percent of their female counterpart. Most female foreigners preferred following the evacuation route guide sign (41.4 percent), and following the crowd (38.6 percent). Similarly, most Thai females also preferred following the evacuation route guide sign (32.8 percent), but much fewer relied on the crowd (8.6 percent). Table 3 Three-way Table Analysis of Evacuation Route Decision Classified by Nationality and Gender Evacuation Route Decision Gender Choose Seek Follow the Follow the familiar assistance guide sign crowd route from officers Others Total N % N % N % N % N % Thais Male 52 78.8 10 15.2 3 4.5 0 0.0 1 1.5 66 Female 34 58.6 19 32.8 5 8.6 0 0.0 0 0.0 58 Total 86 69.4 29 23.4 8 6.4 0 0.0 1 0.8 124 Foreigners Male 24 21.4 26 23.2 42 37.5 19 17.0 1 0.9 112 Female 10 14.3 29 41.4 27 38.6 4 5.7 0 0.0 70 Total 34 18.7 55 30.2 69 37.9 23 12.6 1 0.6 182 Grand Total 120 39.2 84 27.4 77 25.2 23 7.5 2 0.7 306 Goodness of fit statistics Model Independence Factor Pearson χ 2 df. Sig. (p>χ 2 ) Complete independence NA 115.963 13 0.000 Thais/Foreigners 97.018 9 0.000 Independent of Gender 20.407 9 0.016 one factor Route decision 115.841 12 0.000 Thais/Foreigners-Gender 3.956 5 0.556 Conditional independent Gender-Route decision 18.434 8 0.018 Thai/Foreigners-Route decision 95.983 8 0.000 Homogeneous association NA 0.328 4 0.988 22

Table 4 shows the descriptive statistics on evacuation route decision by nationality and age group. Individuals in the age group of 30-59 years old would choose the evacuation route than those in the age group of 19-29 years old. As shown, 74.2 percent of Thais in their 30-59 years of age would rather decide their evacuation route, while only 62.3 percent of Thais in theirs 19-29 years old would do so. Similarly, 22.8 percent of those foreigners with 30-59 years of age would confidently choose their own route, whereas only 15.1 percent of those in 19-29 years of age would do so, and many relied on other sources instead. For Thai respondents, a higher percent of younger-age group (32.1 percent) relied on the evacuation route guide sign than those matured group (16.1 percent). Foreigners in a younger-age group tended to follow the evacuation route guide sign (36.4 percent) slightly more than those matured group (29.4 percent), while the shares of those following the crowd are equally with 36.4 percent. Foreigners of 60 years of age or older would most likely follow the crowd (41 percent) and seek assistance for the officers (29.4 percent). Age may have reflected deteriorated physical ability to perceive or receive information and to confidently react upon it. Table 4 Three-way Table Analysis of Evacuation Route Decision Classified by Nationality and Age Group Evacuation Route Decision Age Group Seek Choose Follow the Follow the assistance familiar guide sign crowd from route officers Others Total N % N % N % N % N % Thais < 19 years old 5 71.4 2 28.6 0 0.0 0 0.0 0 0.0 7 19-29 years 33 62.3 17 32.1 3 5.6 0 0.0 0 0.0 53 30-59 years 46 74.2 10 16.1 5 8.1 0 0.0 1 1.6 62 > 60 years old 2 100.0 0 0.0 0 0.0 0 0.0 0 0.0 2 Total 86 69.4 29 23.4 8 6.4 0 0.0 1 0.8 124 Foreigners < 19 years old 1 14.3 2 28.6 3 42.8 1 14.3 0 0.0 7 19-29 years 10 15.1 24 36.4 24 36.4 8 12.1 0 0.0 66 30-59 years 21 22.8 27 29.4 35 38.0 9 9.8 0 0.0 92 > 60 years old 2 11.8 2 11.8 7 41.1 5 29.4 1 5.9 17 Total 34 18.7 55 30.2 69 37.9 23 12.6 1 0.6 182 Grand Total 12 0 39.2 84 27.4 77 25.2 23 7.5 2 0.7 306 Goodness of fit statistics Model Independence Factor Pearson χ 2 df. Sig. (p>χ 2 ) Complete independence NA 420.695 76 0.000 Thais/Foreigners 109.160 44 0.000 Independent of Age group 359.200 72 0.000 one factor Route decision 384.960 68 0.000 Conditional independent Thais/Foreigners-Age group 18.245 40 0.999 Age group-route decision 213.603 64 0.000 Thai/Foreigners-Route decision 97.865 36 0.000 Homogeneous association NA 6.947 32 1.000 * For the shaded area, the number of cases (row sum) is so small as to be, statistically, of doubtful value. 23

Table 5 shows descriptive statistics on evacuation route decision by nationality and education level. As they received a higher education, chances of a Thai choosing their own route decreased and chances of following a sign increased. Up to 88 percent of those who finished a primary school or a junior hi-school chose their own familiar route to evacuate, while only 64.1 percent of those who finished a high school or an associate degree, and 58 percent of those who received a bachelor degree would choose an evacuation route themselves. Increasing shares of the latter 2 groups relied on following the evacuation guide sign (28.3 and 30.6 percent). This did not necessarily imply that those with a bachelor degree tend to rely more on the evacuation route guide sign. There could be correlation between level of education and familiarity with the area. Those with higher education were predominantly short-time visitors who might not know the road network well. Nonetheless, choosing their own routes was still the most preferred alternative for overall Thai population. Most foreigners who finished a high school or an associate degree would follow the crowd (48.1 percent), while those who receive a bachelor degree or a higher education (Master/Ph.D.) mainly relied on the signage (34 and 38 percent, respectively), then on the crowd (28 and 31 percent, respectively). The share of those seeking assistance from officers also increased in the higher education groups. Table 5. Three-way Table Analysis of Evacuation Route Decision Classified by Nationality and Education Level Evacuation Route Decision Seek Choose Education Follow the Follow the assistance familiar Others Total level guide sign crowd from route officers N % N % N % N % N % Thais Primary/Junior High school 29 87.9 3 9.1 1 3.0 0 0.0 0 0.0 33 High school/ Associate 34 64.1 15 28.3 3 5.7 0 0.0 1 1.9 53 Bachelor 21 58.3 11 30.6 4 11.1 0 0.0 0 0.0 36 Master/Ph.D. 1 100.0 0 0.0 0 0.0 0 0.0 0 0.0 1 Total 85 69.1 29 23.6 8 6.5 0 0.0 1 0.8 123 Foreigners Primary/Junior High school 3 50.0 0 0.0 2 33.3 1 16.7 0 0.0 6 High school/ Associate 12 15.2 21 26.6 38 48.1 8 10.1 0 0.0 79 Bachelor 14 21.5 22 33.9 18 27.7 10 15.4 1 1.5 65 Master/Ph.D. 5 17.2 11 38.0 9 31.0 4 13.8 0 0.0 29 Total 34 19.0 54 30.2 67 37.4 23 12.8 1 0.6 179 Grand Total 119 39.4 83 27.5 75 24.8 23 7.6 2 0.7 302 Goodness of fit statistics Model Independence Factor Pearson χ 2 df. Sig. (p>χ 2 ) Complete independence NA 222.900 58 0.000 Thais/Foreigners 126.149 34 0.000 Independent of one factor Education level 91.512 54 0.001 Route decision 135.099 52 0.001 24

Thais/Foreigners-Education Conditional independent level 39.511 30 0.115 Education level-route decision 37.599 48 0.860 Thai/Foreigners-Route decision 87.379 28 0.000 Homogeneous association NA 9.480 24 0.996 * The shade area represents cases with small sample sizes that may not generate meaningful analysis. Table 6 shows descriptive statistics on evacuation route decision by nationality and local language skill. Most Thai respondents were fair and fluent in their native language. Thus, the study did not produce a meaningful analysis on the effect of language skill to their route choice. There were no obvious distinctive evacuation route patterns between those with fair and fluent Thailand language skill. Meanwhile the group of foreigners showed noticeable difference between language skill groups. Foreigner with no local language skill would follow the crowd (39.7 percent) rather than follow the evacuation route guide sign (29.5 percent). The share of those seeking assistance from the officers among this foreigner group was also as high as the choosing the own evacuation route (approximately 15 percent). On the other hand, those foreigners who knew a little bit of Thai preferred to follow the evacuation route guide sign (44.5 percent) and follow the crowd (29.6 percent). Approximately 22.2 percent of this foreigner group would choose their own evacuation routes. Table 6 Three-way Table Analysis of Evacuation Route Decision Classified by Nationality and Local Language Skill Evacuation Route Decision Choose Seek Local Follow the Follow the familiar assistance language skill guide sign crowd route from officers Others Total N % N % N % N % N % Thais None 0 0.0 0 0.0 0 0.0 0 0.0 0 0.0 0 Barely 5 55.6 4 44.4 0 0.0 0 0.0 0 0.0 9 Moderate 5 50.0 3 30.0 2 20.0 0 0.0 0 0.0 10 Fair 22 73.3 7 23.3 1 3.3 0 0.0 0 0.0 30 Fluent 54 72.0 15 20.0 5 6.7 0 0.0 1 1.3 75 Total 86 69.4 29 23.4 8 6.4 0 0.0 1 0.8 124 Foreigners None 22 15.1 43 29.5 58 39.7 22 15.0 1 0.7 146 Barely 6 22.2 12 44.5 8 29.6 1 3.7 0 0.0 27 Moderate 3 75.0 0 0.0 1 25.0 0 0.0 0 0.0 4 Fair 2 66.7 0 0.0 1 33.3 0 0.0 0 0.0 3 Fluent 1 50.0 0 0.0 1 50.0 0 0.0 0 0.0 2 Total 34 18.7 55 30.2 69 37.9 23 12.6 1 0.6 182 Grand Total 120 39.2 84 27.4 77 25.2 23 7.5 2 0.7 306 Goodness of fit statistics Model Independence Factor Pearson χ 2 df. Sig. (p>χ 2 ) Complete independence NA 467.817 40 0.000 Thais/Foreigners 254.479 24 0.000 Independent of Local language skill 271.901 36 0.000 one factor Route decision 118.228 36 0.000 25

Thais/Foreigners-Local language skill 211.718 20 0.000 Conditional independent Local language skill-route decision 29.683 32 0.584 Thai/Foreigners-Route decision 170.020 20 0.652 Homogeneous association NA 6.715 16 0.978 * The shade area represents cases with small sample sizes that may not generate meaningful analysis. Table 7 shows descriptive statistics on evacuation route decision by nationality and the length of stay in Patong Beach in the past 5 years. As one might expect, Thai who spent more than 1 year in Patong Beach would rather choose their own evacuation route (75.3 percent), most likely because of their familiarity with the road network in the area. The remaining would follow the evacuation route sign (20.4 percent) and follow the crowd (3.2 percent). Thai who spent less than 5 days in Patong area would primarily relied on the other sources, such as following the guide sign (46.2 percent) and following the crowd (30.8 percent), while only 23.1 percent of them chose their own route. Foreigners who spent less than 5 days and those who spent between 5 days and 1 month seem to prefer following the crowd, slightly over following the signs. Approximately 14 percent of these foreigners would choose their own evacuation routes, and another 10-17 percent would seek assistance from the officers. Those foreigners who spent more than 6 month in Patong Beach seemed to get familiarity with the area. A high share of those groups would choose their own route (45.5 percent). An interesting finding was that all of the 6 foreigners (100 percent) who stayed in Patong Beach more than 1 year would choose their own evacuation routes. Table 7 Three-way Table Analysis of Evacuation Route Decision Classified by Nationality and Length of Stay in Patong Area Evacuation Route Decision Total Length of stay Choose Seek Follow the Follow the in the past 5 familiar assistance guide sign crowd years route from officers Others N % N % N % N % N % Thais < 5 days 3 23.1 6 46.2 4 30.7 0 0 0 0 13 5 day 1 mo. 4 100 0 0 0 0 0 0 0 0 4 1 6 mo. 3 100 0 0 0 0 0 0 0 0 3 6 mo. 1 year 3 50.0 3 50.0 0 0 0 0 0 0 6 > 1 year 70 75.3 19 20.4 3 3.2 0 0 1 1.1 93 Total 83 69.7 28 23.6 7 5.9 0 0 1 0.8 119 Foreigners < 5 days 10 13.9 23 31.9 27 37.5 12 16.7 0 0 72 5 day 1 mo. 10 14.3 23 32.9 29 41.4 7 10.0 1 1.4 70 1 6 mo. 3 13.1 7 30.4 9 39.1 4 17.4 0 0 23 6 mo. 1 year 5 45.5 2 18.2 4 36.3 0 0 0 0 11 > 1 year 6 100 0 0 0 0 0 0 0 0 6 Total 34 18.7 55 30.2 69 37.9 23 12.6 1 0.6 182 Grand Total 117 38.9 83 27.6 76 25.2 23 7.6 2 0.7 301 Goodness of fit statistics Model Independence Factor Pearson χ 2 df. Sig. (p>χ 2 ) Complete independence NA 515.258 94 0.000 26

Independent of one factor Conditional independent Thais/Foreigners 207.706 54 0.000 Length of stay 302.564 90 0.000 Route decision 206.753 84 0.000 Thais/Foreigners-Length of stay 148.739 50 0.000 Length of stay-route decision 120.583 80 0.002 Thai/Foreigners-Route decision 40.342 44 0.629 Homogeneous association NA 20.311 40 0.996 * The shade area represents cases with small sample sizes that may not generate meaningful analysis. Table 8 shows descriptive statistics on evacuation route decision by nationality and the usual travel modes in the Patong area. The largest share of foreigners who drove tended to follow the evacuation route guide signs (35.0 percent), followed by choosing their own familiar route (30.0 percent) and relying on the crowd (20.0 percent). Those foreigners who use a rental vehicle, public transit, and paratransit were likely to following the crowd (40.0-45.8 percent), followed by relying on the evacuation route guide signs (25.0-30.0 percent). Those who use public transit and paratransit are more likely to seek assistance from the officers (16.7-20.0 percent), over those who drive a rental vehicle (9.2 percent). Nevertheless, the result indicated that foreigners who drove either their own vehicle or a rental vehicle had a higher chance of choosing their own evacuation route. This was likely because driving in the city gave them familiarity with the area and confidence to choose their own evacuation route in case of tsunami. Table 8 Three-way Table Analysis of Evacuation Route Decision Classified by Nationality and Travel Mode Evacuation Route Decision Usual mode of Choose Seek Follow the Follow the travel in the familiar assistance Others Total guide sign crowd area route from officers N % N % N % N % N % Thais Private car/motorcycle 72 70.6 25 24.5 4 3.9 0 0 1 1.0 102 Rental car/motorcycle 4 50.0 2 25.0 2 25.0 0 0 0 0 8 Bus/Songtaew 3 75.0 0 0 1 25.0 0 0 0 0 4 Motorcycle taxi 3 100 0 0 0 0 0 0 0 0 3 Others 4 57.1 2 28.6 1 14.3 0 0 1 0 7 Total 86 69.4 29 23.3 8 6.5 0 0 1 0.8 124 Foreigners Private car/motorcycle 6 30.0 7 35.0 4 20.0 3 15.0 0 0 20 Rental car/motorcycle 17 19.5 23 26.5 39 44.8 8 9.2 0 0 87 Bus/Songtaew 3 12.5 6 25.0 11 45.8 4 16.7 0 0 24 Motorcycle taxi 1 10.0 3 30.0 4 40.0 2 20.0 0 0 10 Others 7 17.1 16 39.1 11 26.8 6 14.6 1 2.4 41 Total 34 18.7 55 30.2 69 37.9 23 12.6 1 0.6 182 27

Grand Total 120 39.2 84 27.5 77 25.2 23 7.5 2 0.6 306 Goodness of fit statistics Model Independence Factor Pearson χ 2 df. Sig. (p>χ 2 ) Complete independence NA 375.078 58 0.000 Independent of one factor Conditional independent Thais/Foreigners 195.859 34 0.000 Mode of travel 206.620 54 0.000 Route decision 136.854 52 0.000 Thais/Foreigners-Travel mode 125.396 30 0.000 Travel mode-route decision 45.686 48 0.568 Thai/Foreigners-Route decision 59.716 28 0.000 Homogeneous association NA 11.211 24 0.987 * The shade area represents cases with small sample sizes that may not generate meaningful analysis. 4. CONCLUSION This study investigated the tsunami evacuation route decision among sample groups of Thai and foreigners in Patong area. The majority of Thais were local residents or stayed in the Patong Beach area for more than a year, while most foreigner respondents were tourists and spend less than a month in the area. The evacuation route decisions among these two groups were rather distinctive. Almost 70 percent of Thai samples preferred to choose their own evacuation routes, and another 23 percent would follow the evacuation route guide signs. The foreigner respondents mostly relied on the evacuating crowd (38 percent) and, to a lesser extent, on the evacuation route guide sign. Only 19 percent confidently decided their own evacuation route, while 12.6 percent would seek assistance from the officers. Other socioeconomic characteristics were found to influence the tsunami evacuation route decision among Thai and foreigners. Three-way contingency table analysis concluded the following trends: - Both Thai and Foreigner males were more confident in choosing their own route than the female counterparts. - Similarly, 30-59 year-old individuals old were more confident in choosing the evacuation route than those in the age group of 19-29 years old. Foreigners older than 60 years old, preferred to follow the crowd and to seek assistance for the officers. - Most foreigners who finished a high school or an associate degree were more likely to follow the crowd, while those with a bachelor degree or a higher education (Master/Ph.D.) mainly relied on the guide sign, then on the crowd. - Foreigner with no local language skill would decide to follow the crowd rather than following the evacuation route guide sign. On the other hand, those foreigners who knew some local language preferred to follow the evacuation route guide sign, then to follow the crowd. - Thais who stayed less than 5 days in Patong area would primarily rely on the guide sign and the crowd. Foreigners who spent less than 5 days and those who spent between 5 days and 1 month preferred to follow the crowd slightly over follow the guide signs. - Foreigners who drove either their own vehicle or a rental vehicle had a higher share of choosing their own evacuation route. Based on the study results, we can develop measures to reduce disaster risks involved in tsunami evacuation as follows. 28

Since a large portion of the local Thai residents (69.3 percent) tend to choose their own evacuation route, it is important to educate them a correct suitable evacuation path on the network. When the tsunami alarms go on, they can confidently identify the correct path quickly. This can be implemented through various pro-active activities, such as, regular dissemination of easy-to-read evacuation route maps, regular community trainings and public awareness, and frequent evacuation drills. These measures will not only benefits to the 69.3 percent of the local residents, but would lead another 6.5 percent of local residents as well as 37.9 percent of foreigners who tend to follow the crowd (mostly female, people with an age more than 60 years old, no local language skill) to a safe place along the suitable evacuation route. These proactive measures are perhaps most effective to the majority of the people at risk, and will reduce disaster risk from awkward evacuation route decision. Approximately 23.4 percent of local Thai people and 30.2 percent of foreigners tend to follow the evacuation sign. It is important that the local authorities to install a comprehensive evacuation sign system on the network, and maintain its functions, so that both Thais and foreigners can easily detect the sign in the time of tsunami. The signage system should be more extensive around the beachfront or near the beach, because both Thais and foreigners in that particularly area have to evacuate abruptly. The effective signage system can be achieved through the applications of signage or pavement markings. Approximately 12.6 percent of foreigners tend to seek assistance for the officer. The local authority might designate competent local residents who live or work in the area to become informal officers who assist or guide those foreigners in need during the evacuation time. They can be selected based on volunteering basis from various locations in the area, and trained to help facilitate the evacuation with the consideration of own safety during the disaster time. This will ensure that the foreigner tourists (especially those whose age over 60 years old) who are least familiar to the area, will also get assistance. REFERENCES Aquirre, B.E. (1991) Evacuation in Cancun during hurricane Gilbert, International Journal of Mass Emergencies and Disasters, Vol. 9, No. 1, 31-45. Barrett, B., Ran, B., and Pillai, R. (2000) Developing a dynamic traffic management modeling framework for hurricane evacuation, Transportation Research Record, 1733, 115-121. Dow, K. and Cutter, S. L. (2002) Emerging Hurricane Evacuation Issues: Hurricane Floyd and South Carolina, Natural Hazard Review, February 2002, 12-18. Drabek, T. E. (1996) Disaster Evacuation Behavior: Tourists and Other Transients. Monograph No. 58. Institute of Behavioral Science, University of Colorado. Ghobarah, A., Saatcioglu, M. and Nistor I. (2006) The impact of the 26 December 2004 earthquake and tsunami on structures and infrastructure, Engineering Structures, Vol. 28, 312-326. Lammel, G., Grether, D., and Nagel, K. (2010) The representation and implementation of time-dependent inundation in large-scale microscopic evacuation simulations, Transportation Research Part C, Vol. 18, 84-98. Lindell, M. K., Lu, J. C., and Prater, C. S. (2005) Household decision making and evacuation in response to hurricane Lili, Natural Hazards Review, November 2005, 171-179. Liu, P. L. F., Wang, X., and Salisbury A. (2009) Tsunami hazard and early warning system in South China Sea, Journal of Asian Earth Sciences, Vol. 36, 2-12. 29

Srinivas, H. and Nakagawa, Y. (2008) Environmental implications for disaster preparedness: Lessons Learnt from the Indian Ocean Tsunami, Journal of Environmental Management, Vol. 89, 4-13. Srivichai, M., Supharatid, S., and Imamura, F. (2007) Recovery process in Thailand after the 2004 Indian Ocean Tsunami, Journal of Natural Disaster Science, Vol. 29, No. 1, 3-12. Thanawood, C., Yongchalermchai, C., and Densrisereekul, O. (2006) Effects of the December 2004 tsunami and disaster management in Southern Thailand, Science of Tsunami Hazards, Vol. 24, No. 3, 206-217. Tsai, C. H., Wang, W. C., Chen, C. W., Chen, H. L., and Lin M. L. (2009) The construction of decision model for tourism disaster evacuation based on GIS and fuzzy theory, World Academy of Science, Enginering and Technology, Vol. 54, 81-85. Zhang, D. H., Yip, T. L., and Ng C. O. (2009) Predicting tsunami arrivals: Estimates and policy implications, Marine Policy, Vol. 33, 643-650. 30