Journal of Japan Society of Civil Engineers, Ser. F3 (Civil Engineering Informatics), Vol. 72, No. 2, I_1-I_13, 2016. ANALYSIS ON CAR COMMUTERS’ BEHAVIOR DURING A MASSIVE DOWNPOUR BASED ON PROBE DATA AND QUESTIONNAIRE SURVEY Mohammad Hannan Mahmud KHAN1, Motohiro FUJITA2 and Wisinee WISETJINDAWAT3 1Student member of JSCE, Graduate Student, Dept. of Civil Eng., Nagoya Institute of Technology (Gokiso, Showa, Nagoya 466-8555, Japan) E-mail: [email protected] 2 Member of JSCE, Professor, Dept. of Civil Eng., Nagoya Institute of Technology (Gokiso, Showa, Nagoya 466-8555, Japan) E-mail: [email protected] 3 Member of JSCE, Assistant Prof., Dept. of Civil Eng., Nagoya Institute of Technology (Gokiso, Showa, Nagoya 466-8555, Japan) E-mail: [email protected] A massive downpour due to Typhoon Roke attacked the Tokai region on 20th September, 2011. Several roads in the northeastern part of Nagoya city and the adjacent areas were closed to traffic, resulting in a serious commuter chaos. In this research, we attempted to explore the effects of departure hours, early or late departure, the significance of acquiring proper traffic information as well as the impacts of road closures on the level of difficulty of home returning trips. Regression models were developed using both questionnaire survey and taxi probe data. Questionnaire survey can gather drivers’ information; however, it is difficult to gather the actual changes in travel condition. On the other hand, probe data can demonstrate a real time change in travel condition at every couple of minutes. Therefore, this study presents a combined usage of both data for a clearer explanation on the travel condition and the behavior of drivers during the typhoon. The findings revealed that the difficulty to gather reliable and comprehensive information on the traffic regulations enforced temporarily during the typhoon had led drivers facing the closure sections and the consequent very long travel delay. In fact, these delays can be avoided if the drivers had chosen the right detour route from the beginning. Based on the findings, this paper also suggests some proper counter- measures to reduce the severity of commuter chaos for the future disaster. Key Words: downpour, transportation system failure, car user, traffic information, commuter chaos 1. INTRODUCTION which was conducted soon after Typhoon Roke as well as the probe data of taxis operating on the same The damages due to typhoons had caused much day were used to analyze the impacts of the disaster economic losses to Japan, although the severity of an on the returning home commuters. occurrence is much less than that of a large scale Generally, it is difficult for a questionnaire survey earthquake. This is because this type of disaster oc- to gather such detailed information as the changes in curred rather frequently. As an example, the massive travel condition, congestion level, and road closures downpour due to Typhoon Roke which attacked the at every minute along the travel route. However, Tokai region from the afternoon until late evening of questionnaire survey can gather other drivers’ in- 20th September 2011 caused a loss of more than 1 formation such as the differences on departure time, trillion Yen1). During the event, several transporta- travel route, and travel time between on a regular day tion services were closed. The inundations and the and on the day of typhoon. Also, the questionnaire consequent road closures at several locations re- survey can obtain the attribution of drivers, their sulted in a very serious commuter chaos during home habits, as well as their opinions regarding counter- returning trips. measures to the disaster. On the other hand, although In this study, the data from a questionnaire survey taxi probe data cannot provide such information; 1 I_1 Journal of Japan Society of Civil Engineers, Ser. F3 (Civil Engineering Informatics), Vol. 72, No. 2, I_1-I_13, 2016. however, the data can demonstrate the real time model was developed to explain the choice that the changes in travel condition at every couple of sec- stranded commuters chose for returning home on the onds or minutes. Also, the data can be visualized on a day of earthquake. Later, this model was used to GIS map. estimate the number of populations who walked Utilizing both information can complete each home on the day. Similarly, Fujiu et al.5) used a other drawbacks. However, there is only a few web-based survey to clarify the walking routes that studies considering this approach. Therefore, this the stranded commuters used in the aftermath of the paper aims at providing an analysis using both same earthquake. They also investigated fire danger questionnaire and probe data in order to obtain a and available support facilities at different walking better representation of the road situation and driv- routes. Ito et al.6) used a similar survey method and ers’ circumstances, and improve the quality of the investigated further the availability of facilities and analysis on the difficulty of returning home. We services for the people who walked home in the af- focus on the problem on stranded commuters during termath of the same earthquake. These studies pro- the 2011 Typhoon Roke in Tokai region. Based on vided an insight into the problems faced by the both data sources, several regression models repre- stranded commuters and their travel conditions in the senting the difficulty to return home are developed. aftermath of the 2011 Great East Japan Earthquake. Furthermore, the effects of the acquisition of the The impacts of the same disaster may be different information and countermeasures related to road in other regions considering the different geographic closures on the returning home of stranded com- and transport facilities. Some works investigated the muters during the heavy downpour are investigated. impacts of the earthquake to Tokai region. Koyama and Okada7) conducted a questionnaire survey to companies in Nagoya regarding their workers’ home 2. LITERATURE REVIEW returning behaviors by assuming public transport services were shut down after a large scale earth- In recent years, there are an increasing number of quake. The questionnaire asked the workers’ be- researches on commute changes due to disasters. haviors both when receiving an earthquake warning Depending on the characteristics of disaster, the and after the actual occurrence of the earthquake. changes in commute patterns are rather different. The data was used to estimate the number of stranded Some disasters such as earthquakes and hurricanes commuters. They concluded that the current shelters can cause heavy damages to infrastructures. Often and evacuation facilities are inadequate for the up- the damages are so severe and required a long period coming large scale earthquake in the region. of time until the infrastructures can be operable Different disaster causes different consequences. again, and hence this situation results in a long pe- The damages by a downpour disaster also have riod disruption on commuters’ travel pattern. In the unique characteristics. The damages can be tempo- US, several works related to commute pattern rary until the heavy rain stops and the floods are changes after a hurricane or earthquake can be found. drained. Some works regarding the commuter chaos To name a few, Guiliano and Golob2) studied the during a massive downpour can be found. Mitamura impacts on commuting behaviors after two major et al.8) examined the problems public transport users transport corridors were heavily damaged by the experienced when returning home during the Tokai Northridge Earthquake. They found that automobile torrential rain. They also developed a binary choice users preferred to shift route, travel schedules, and model whether or not to return home on the date, destination rather than to shift to public transport based on a Stated Preference data assuming the tor- mode or share riding. rential rain happened again. One of the interesting Commuter chaos occurred often even after a small findings was that the worse experience the commuter scale incident. Especially after a major disaster, the has, the less likely they will return home. Regarding adverse impacts due to the chaos can be enormous. the consequences of the torrential rain to automobile After the 2011 Great East Japan Earthquake, in To- users, Sakamoto et al.9) analyzed the impacts of kyo alone, the disaster had caused more than 3 mil- traffic information on drivers’ decision whether to lion commuters having trouble to return home3). return home or to stay at workplace based on a data Hence, this problem drew much attention from many collected from a questionnaire survey. They found researchers to investigate further the problems re- that if the travel time is 4-5 times of the usual travel garding the commuter chaos after the earthquake. time, 80% of the drivers’ prefer to stay at office. For example, Hiroi et al.4) used a web-based survey Most researches in this topic were conducted to collect the information on home returning trips of based on a questionnaire survey either telephone, people in Tokyo during the earthquake. A choice internet, or paper based. With today technology, it is 2 I_2 Journal of Japan Society of Civil Engineers, Ser. F3 (Civil Engineering Informatics), Vol. 72, No. 2, I_1-I_13, 2016. possible to analyze the impacts of disaster on travel behavior using GPS data from car navigation system. Chuno The precision of the GPS data is excellent, proven by the work by Liang et al.10) which successfully used Tono Gifu the GPS data from taxis to model the arterial speed. R19 However, only very few studies used GPS data for Ichinomiya Tajimi studying traffic condition during a disaster. Khan et Kasugai al.11) was the first to demonstrate the potential of TOMEI Central using a taxi probe data to study the traffic condition Moriyama Expway Nagoya during home returning in a disaster situation.