A novel agent-based model for tsunami evacuation simulation and risk assessment

Tsunami evacuation is an effective way to save lives from the near-field tsunami. Realistic evacuation simulation can provide valuable information for accurate evacuation risk assessment and effective evacuation planning. Agent-based modeling is ideal for tsunami evacuation simulation due to its capability of capturing the emergent phenomena and modeling the individual-level interactions among agents and the agents’ interactions with the environment. However, existing models usually neglect or simplify some important factors and/or mechanisms in tsunami evacuation. For example, uncertainties in seismic damages to the transportation network are not probabilistically considered (e.g., by simply removing the damaged links (roads/bridges) from the network). Typically a relatively small population (i.e., evacuees) is considered (due to computational challenges) while neglecting population mobility. These simplifications may lead to inaccurate estimation of evacuation risk. Usually, only single traffic mode (e.g., on foot or by car) is considered, while pedestrian speed adjustment and multi-modal evacuation (e.g., on foot and by car) are not considered concurrently. Also, pedestrian–vehicle interaction is usually neglected in the multi-modal evacuation. To address the above limitations, this study proposes a novel and more realistic agent-based tsunami evacuation model for tsunami evacuation simulation and risk assessment. Uncertainties in seismic damages to all links in the transportation network as well as uncertainties in other evacuation parameters are explicitly modeled and considered. A novel and more realistic multi-modal evacuation model is proposed that explicitly considers the pedestrian–vehicle interaction, walking speed variability, and speed adjustment for both the pedestrian and car according to traffic density. In addition, several different population sizes are used to model population mobility and its impact on tsunami evacuation risk. The proposed model is applied within a simulation-based framework to assess the tsunami evacuation risk assessment for Seaside, Oregon.

     

Pre-landfall evacuee perception of the meteorological hazards associated with Hurricane Gustav

In this study, evacuees from the path of Hurricane Gustav were surveyed to determine which meteorological hazards most influenced their decision to leave. Surveys were conducted along two major evacuation routes on August 30 and 31, 2008, to collect time-sensitive data on individual evacuation decisions related to the meteorological hazards from Hurricane Gustav. The regions of New Orleans, Houma, and Lafayette were represented most frequently, as determined by zip code data collected from the surveys. Responses were evaluated first by meteorological hazard for the entire dataset and then by three-digit zip code region. Overall, storm surge was the most important meteorological variable, followed by the size of the storm, wind, rain, and tornadoes. When separated into three-digit zip code regions, analyses revealed evacuees from in and around New Orleans were driven to evacuate as a result of the perceived threat from storm surge and storm size; residents in the Houma, Louisiana region were motivated to leave due to the threat from storm surge; and Lafayette and the surrounding areas were most-concerned with the threats posed by hurricane-force winds. Given the forecast track and intensity, survey respondents understood the meteorological hazards from Gustav and were motivated to leave based on personal evaluations of risk associated with the storm.     

Advancing tsunami risk assessment by improving spatio-temporal population exposure and evacuation modeling

Tsunamis are among the most destructive and lethal of coastal hazards. These are time-specific events, and despite directly affecting a narrow strip of coastline, a single occurrence can have devastating effects and cause massive loss of life, especially in urbanized coastal areas. In this work, in order to consider the time dependence of population exposure to tsunami threat, the variation of spatio-temporal population distribution in the daily cycle is mapped and analyzed in the Lisbon Metropolitan Area. High-resolution daytime and nighttime population distribution maps are developed using ‘intelligent dasymetric mapping,’ that is, applying areal interpolation to combine best-available census data and statistics with land use and land cover data. Workplace information and mobility statistics are considered for mapping daytime distribution. In combination with a tsunami hazard map, information on infrastructure, land use and terrain slope, the modeled population distribution is used to assess people’s evacuation speed, applying a geospatial evacuation modeling approach to the city of Lisbon. The detailed dynamic population exposure assessment allows producing both daytime and nighttime evacuation time maps, which provide valuable input for evacuation planning and management. Results show that a significant amount of population is at risk, and its numbers increase dramatically from nighttime to daytime, especially in the zones of high tsunami flooding susceptibility. Also, full evacuation can be problematic in the daytime period, even if initiated immediately after a major tsunami-triggering earthquake. The presented approach greatly improves tsunami risk assessment and can benefit all phases of the disaster management process.     

Virtual tools for volcanic crisis management, and evacuation decision support: applications to El Chichón volcano (Chiapas, México)

Decision making regarding massive evacuation of a population threatened by a probable volcanic eruption is a major problem in crisis management. Such a decision is general on the number of people to be evacuated, available resources and infrastructure, quantity and quality of the escape routes and shelters, and the economic, social and political costs involved in the operation, coupled with the updated information provided by scientists about the forecast of future activity and probable eruption scenarios. Knowing time-lapse between the evacuation decision-making time and the time in which the evacuation is completed is another critical issue that must be carefully considered in densely populated areas. In such areas, it is really important to estimate in advance this time-lapse, as the forecast must be released with enough time to complete all the evacuation process before the destructive manifestations of the eruption begin. In this context, evacuation planning is a crucial component of emergency management. It is common for Emergency Plans to include pre-established strategies. However, an evacuation procedure should be flexible, depending on the above-mentioned timing, and on the decisions, evacuation schemes, environmental characteristics and other factors. In this work, several hazard models such as a lava flow model based on a Monte Carlo algorithm, a pyroclastic density current based on energy cone model, a semi-empirical inversion model to estimate the thickness of ash deposits, and all available information about the El Chión volcano have been used to obtain the area that should be evacuated in case of an eruption. Then, multiple evacuation strategies at El Chichón volcano have been designed, considering not only the characteristics of the eruption forecast, but also environmental factors (e.g., weather conditions) and social factors (e.g., tourism and farming seasons). The variable scale evacuation model has been used to estimate the evacuation time. In the paper, those virtual tools are briefly described as well as the information obtained from the drill of 2009. In addition to the optimization of evacuation under variable conditions and situations, one of the main objectives of this work is to provide a reliable estimation of the mitigation action time, for an Emergency Plan.     

Spatial and temporal variations in the incidence of dust events over Iran

Large near-field tsunamis pose a significant threat to the Canadian West Coast due to its proximity to the circum-Pacific belt where a significant tsunami-inducing earthquake event from the Cascadia subduction zone is expected. This study investigated the risks associated with such an event in terms of pedestrian evacuation needs and plans for the Town of Tofino, a small community located on the West Coast of Vancouver Island. The population-at-risk within the hazard zone and its ability to evacuate to safety is evaluated using anisotropic path-distance modelling. Mitigation measures, such as vertical evacuation buildings, are quantitatively evaluated. Site-specific inundation modelling was not performed as part of this study; tsunami hazard and safe zones were computed using a range of run-ups varying between 3 and 25 m. It was established that up to 80% of the population is within the maximum hazard zone considered. This evacuation modelling exercise indicates that a maximum of 13% of the population would have insufficient time to reach safety when using a mobility-impaired ambulatory speed. The use of three vertical evacuation buildings can reduce the risk of losing population in this category by 99%. Although some conservative assumptions were used (vertical datum at higher high water, reductions in safe zones by generalization process and mobility-impaired evacuation speeds), the evacuation potential is likely overestimated due to the coarseness of the topographic data used in the evacuation modelling and from an overestimated first wave arrival time. This is the first Canadian study which used anisotropic evacuation modelling to evaluate the vulnerability of a Canadian community to tsunami inundation.

     

Tsunami risk reduction for densely populated Southeast Asian cities: analysis of vehicular and pedestrian evacuation for the city of Padang, Indonesia, and assessment of interventions

A major tsunamigenic earthquake is expected in the near future along the coast of West Sumatra Province of Indonesia. In the city of Padang, the arrival time of the tsunami is expected to be ~30 min. Currently, there are approximately 400,000 people in the city living within the potential inundation zone. This study aimed to complement the existing research in appraising possible risk reduction interventions, specifically looking at enabling the timely evacuation of the area. This research, developed in consultation with national and local authorities, emergency planners and NGOs, analysed interventions for tsunami risk reduction in Padang through the development of a pedestrian and vehicular evacuation model and the appraisal of possible solutions to enhance the evacuation rates. Some of the conclusions from this research can be applied to other areas in Southeast Asia where the traffic patterns are similar to those in Padang and where the distance to safety is greater than 4–5 km. For the case of Padang, the results show that pedestrian evacuation is strongly preferable to vehicular evacuation due to the limited road capacity and the high population density. In the present situation, however, 70–80 % of the population in Padang could not evacuate within 30 min, even by foot. Common interventions such as widening roads and building bridges prove to be relatively ineffective in this case due to the large distance that has to be covered in a short time. These interventions would only have a decisive impact if a longer evacuation time was available (more than 60 min). In the case of Padang, the evacuation rate in the first 30 min is strongly dependent on the presence/absence of evacuation shelters, whose effectiveness is limited by the capacity of the structures. Building a few high-capacity and high-resilience structures such as evacuation hills is a more effective and robust evacuation strategy than constructing many small high-raised buildings. Even with evacuation structures, wider roads and bridges, about 20 % of the population would still be unable to reach safety by the time the tsunami arrives. This means about 70,000 people of Padang’s current population, which is rapidly increasing. The building of evacuation shelters may be a viable option for saving lives in the short term, but it is not a sustainable option in the medium to long term. It is therefore also necessary to set up and enforce regulations for land use planning that take into account the tsunami risk and prevent further urban development for the areas that may be affected by a tsunami.     

An assessment of educational tsunami evacuation map designs in Washington and Oregon

Educational tsunami evacuation map brochures in Washington and Oregon have been developed locally, resulting in significant differences between the types of tsunami hazard information they include. This paper identifies six tsunami hazard information types present in 38 brochures in Washington and Oregon: (1) tsunami hazard zone, (2) road network, (3) assembly areas, (4) evacuation guidance, (5) infrastructure, and (6) terrain. It compares and contrasts these information types in the maps and text of six of the brochures, including a proposed design standard in Oregon. Design differences of all 38 brochure maps are then organized using principles of cartographic abstraction, which describe mapmaker decisions about selection, generalization, and symbolization of information. We further use this framework to situate the information content of a new interactive Google Maps tool in Oregon. Our assessment identifies limitations of current tsunami hazard information that may be relevant to improving tsunami education. In theory, more advanced evacuation map tools can play an important role in reducing the limitations of tsunami hazard information relevant to the public. The new Google Maps tool addresses few of these limitations. Recognizing how map-making decisions define the underlying information content of evacuation maps can facilitate much needed future evaluations and developments in evacuation map design.     

Computer-based Model for Flood Evacuation Emergency Planning

A computerized simulation model for capturing human behavior during flood emergency evacuation is developed using a system dynamics approach. It simulates the acceptance of evacuation orders by the residents of the area under threat; number of families in the process of evacuation; and time required for all evacuees to reach safety. The model is conceptualized around the flooding conditions (physical and management) and the main set of social and mental factors that determine human behavior before and during the flood evacuation. The number of families under the flood threat, population in the process of evacuation, inundation of refuge routes, flood conditions (precipitation, river elevation, etc.), and different flood warnings and evacuation orders related variables are among the large set of variables included in the model. They are linked to the concern that leads to the danger recognition, which triggers evacuation decisions that determine the number of people being evacuated. The main purpose of the model is to assess the effectiveness of different flood emergency management procedures. Each procedure consists of the choice of flood warning method, warning consistency, timing of evacuation order, coherence of the community, upstream flooding conditions, and set of weights assigned to different warning distribution methods. Model use and effectiveness are tested through the evaluation of the effectiveness of different flood evacuation emergency options in the Red River Basin, Canada.     

A method for estimating casualties due to the tsunami inundation flow

This study develops a method for estimating the number of casualties that may occur while people evacuate from an inundation zone when a tsunami has inundated an area. The method is based on a simple model of hydrodynamic forces as they affect the human body. The method uses a Tsunami casualty index (TCI) computed at each grid point of a numerical tsunami model to determine locations and times within the tsunami inundation zone where evacuation during the tsunami inundation is not possible and therefore where casualties are likely to occur. The locations and times can be combined with information about population density to compute the potential number of casualties. This information is useful in developing tsunami evacuation routes that avoid such locations. To illustrate the method, it is applied to the Seattle waterfront in Washington State, USA, that is under the threat of possible tsunami disasters due to Seattle Fault earthquakes. Preliminary results suggest that the tsunami casualties may occur within the Seattle waterfront for 15 min, during the time interval from 3 to 18 min after a large Seattle Fault tsunami is generated when the background tide level is mean high water.     

A Human Damage Prediction Method for Tsunami Disasters Incorporating Evacuation Activities

This study presents a tsunami human damage prediction method employing numerical calculation and GIS (Geographical Information System) for Usa town, Tosa City, Shikoku Island, Japan. Sometime near the end of the first half of the twenty-first century, a huge earthquake is predicted to occur along the Nankai trough and costal areas facing the Pacific ocean of Shikoku Island. Much damage due to the resultant tsunamis will be caused, therefore, it is necessary to predict the extent of human damage for every town in high-risk areas.The number of tsunami victims was estimated by population in areas of maximum inundation. The number of deaths as a result of tsunami was estimated by a method which employed accumulated death toll of every area in terms of time and space, taking into account consideration of time necessary to begin to seek refuge after an earthquake, tsunami inundation depth on land, flow velocity and evacuation speed. As a result of this study a rapid decrease in death toll by early evacuation was shown quantitatively for the first time.Thus, with the method presented here, it is possible to estimate the extent of tsunami human damage on coastal regions, and may be useful as a tsunami human damage countermeasure.     

Method to determine the locations of tsunami vertical evacuation shelters

The 2004 Indian Ocean tsunami and the 2011 Great Tohoku Japan earthquake and tsunami focused a great deal of the world??s attention on the effect of tsunamis on buildings and infrastructure. When a tsunami impacts structures in a coastal community, the structures are often not strong enough to withstand the forces and may collapse. Therefore, to maximize the survival probability, people evacuate to higher ground or move outside the inundation zone. However, this is not always possible because of short warning times for near-field tsunamis. Thus, sheltering-in-place or ??sheltering-near-place?? using vertical evacuation should be considered as an alternative approach to lateral evacuation from a tsunami inundation zone. This paper presents the method and results of a study to develop and demonstrate a methodology that applied genetic optimization to determine optimal tsunami shelter locations with the goal of reducing evacuation time, thereby maximizing the probability of survival for the population in a coastal community. The City of Cannon Beach, Oregon, USA, was used as an illustrative example. Several cases were investigated ranging from a single shelter to multiple shelters with locations of high elevation already in place near the city. The method can provide decision-support for the determination of locations for tsunami vertical evacuation shelters. The optimum location of the shelter(s), which was found to vary depending on the number of shelters considered, can reduce the evacuation time significantly, thereby reducing the number of fatalities and increasing the safety of a community.     

Tsunami evacuation behavior of coastal residents in Kochi Prefecture during the 2014 Iyonada Earthquake

The 2014 Iyonada Earthquake, which occurred at 02:06 JST on 14 March, measured 6.2 on the Richter scale and originated in the Seto Inland Sea of Japan. To elucidate tsunami evacuation behavior, we examined two coastal communities in Kochi Prefecture, Okitsu and Mangyo, where residents evacuated to high ground in anticipation of a tsunami. In the event of a Nankai megathrust earthquake and tsunami, it is expected that a huge tsunami will be generated and these communities will be severely damaged. Before the Iyonada Earthquake, we had previously collected data about tsunami preparedness and evacuation plans from the residents of these communities, and after the earthquake, we conducted in-depth interviews and questionnaire surveys with the residents regarding the actual evacuation behaviors that they took. This enabled us to compare evacuation plans with evacuation behaviors. Results indicate that many residents responded quickly to the earthquake, either by immediately evacuating to emergency shelters on high ground or by preparing themselves for evacuation. Additionally, the earthquake revealed great differences between the prior evacuation plans and the actual situation of residents’ evacuation, such as specific triggers that significantly led residents to evacuate and the use of vehicles in evacuation.     

Evacuees’ reentry concerns and experiences in the aftermath of Hurricane Ike

Managing evacuees’ reentry into their communities after an evacuation can be a major challenge for emergency managers, especially in instances when evacuees return before the official all-clear message. Despite the frequency of post-evacuation reentry into evacuated areas, there have been few studies of this process and the issues returnees expect and experience during the return phase. A survey of evacuees after Hurricane Ike indicates that household compliance with reentry plans was low, with only a minority of returnees (38 %) complying with official reentry plans. An examination of reentry concerns shows that minority ethnicity, lower education, and lower income were associated with higher levels of reentry concerns and, to a lesser extent, with problems experienced after returning. Results also indicate that none of the demographic variables correlated significantly with compliance with official reentry plans and only higher income predicted later entry. However, concerns about reentry traffic predicted earlier reentry and concern about physical risk was related to reentry plan compliance. This study provides insight into the concerns that motivate households’ reentry decisions and can inform the creation of return strategies that account for people’s concerns about their hurricane-impacted communities.     

An agent-based evacuation model for the 2011 Brisbane City-scale riverine flood

In this study, an agent-based model is proposed in order to provide new insights into the policy analysis and strategy assessment of city-scale evacuation management. The proposed model is suitable for assessment of the influence of different departure times and communications among peer evacuees on the number of residents at risk who arrive at official shelters. A case study is applied to build a simulation model for the coastal city of Brisbane in Australia. The Brisbane River catchment experiences regular flooding almost every year; the second severest flood since the twentieth century occurred in 2011. During that event, over 15,000 properties were inundated and around 3600 households evacuated in metropolitan Brisbane alone. Making use of high-performance computing clusters, the evacuation simulation was coupled with results from a validated hydrodynamic model to test a variety of escaping scenarios based on the 2011 flood situation. This case study demonstrates the proposed model’s capacity to represent the dynamic evacuation process and also shows that the model is able to help develop flood emergency plans and evaluate response measures through exploring key elements in a range of scenarios.     

Approaches for tsunami risk assessment and application to the city of Cádiz, Spain

Tsunamis can represent a significant risk to the population and cause huge economic damage in many costal regions. In order to be able to identify risk hot spots and implement targeted risk reduction measures, decision makers need to have a clear picture of the risk situation in their countries or regions. This work reviews existing approaches for tsunami risk assessment and recommends a five-step process for assessing tsunami risk. As a case study, a qualitative risk assessment for a worst-case tsunami scenario was carried out to understand the tsunami risk to the population in Cádiz. Moreover, a sensitivity analysis of the tsunami hazard input parameters was performed as a strong influence of the variability of the input parameters on the resultant tsunami hazard and risk zonation maps was observed. The study shows that regardless of the assumptions made a non-negligible tsunami risk to Cádiz exists.     

Community variations in population exposure to near-field tsunami hazards as a function of pedestrian travel time to safety

Efforts to characterize population exposure to near-field tsunami threats typically focus on quantifying the number and type of people in tsunami-hazard zones. To develop and prioritize effective risk-reduction strategies, emergency managers also need information on the potential for successful evacuations and how this evacuation potential varies among communities. To improve efforts to properly characterize and differentiate near-field tsunami threats among multiple communities, we assess community variations in population exposure to tsunamis as a function of pedestrian travel time to safety. We focus our efforts on the multiple coastal communities in Grays Harbor and Pacific Counties (State of Washington, USA), where a substantial resident and visitor population is threatened by near-field tsunamis related to a potential Cascadia subduction zone earthquake. Anisotropic, path distance modeling is conducted to estimate travel times to safety, and results are merged with various population data, including residents, employees, public venues, and dependent-care facilities. Results suggest that there is substantial variability among communities in the number of people that may have insufficient time to evacuate. Successful evacuations may be possible in some communities assuming slow walking speeds, are plausible in others if travel speeds are increased, and are unlikely in another set of communities given the large distances and short time horizon. Emergency managers can use these results to prioritize the location and determine the most appropriate type of tsunami risk-reduction strategies, such as education and training in areas where evacuations are plausible and vertical-evacuation structures in areas where they are not.     

A method for tsunami risk assessment: a case study for Kamakura,Japan

This paper presents a methodology for tsunami risk assessment, which was applied to a case study in Kamakura, Japan. This methodology was developed in order to evaluate the effectiveness of a risk-reducing system against such hazards, also aiming to demonstrate that a risk assessment is possible for these episodic events. The tsunami risk assessment follows these general steps: (1) determination of the probability of flooding, (2) calculation of flood scenarios, (3) assessment of the consequences and (4) integration into a risk number or graph. The probability of flooding was approximated based on the data provided by local institutes, and the flood scenarios were modeled in 1D using the Simulating WAves till SHore model. Results showed that a tsunami in Kamakura can result in thousands of casualties. Interventions such as improvements in evacuation systems, which would directly reduce the number of casualties, would have a large influence in risk reduction. Although this method has its limits and constraints, it illustrates the value it can add to existing tsunami risk management in Japan.     

Large Scale Rockfall Reach Susceptibility Maps in La Cabrera Sierra (Madrid) performed with GIS and Dynamic Analysis at 1:5,000

For the testing of the effect on the tsunami prevention facilities, a simplified methodfor tsunami risk assessment was suggested without wave run-up analysis. This methodis proposed using calculated offshore tsunami waveform and field reconnaissance suchas the seawall height, time necessary for residents' evacuation and tsunami warninginsurance. Then, two normalized values are evaluated; one is the ratio of calculatedmaximum tsunami height to seawall height, the other is the ratio of time betweentsunami over-topping and evacuation completion to total time required for evacuation.These two values are used to qualitatively estimate the safety of residents and the effectof tsunami prevention facilities, eliminating the necessity to compute complicatedtsunami run-up onshore.     

Changes in population evacuation potential for tsunami hazards in Seward, Alaska, since the 1964 Good Friday earthquake

Pedestrian evacuation modeling for tsunami hazards typically focuses on current land-cover conditions and population distributions. To examine how post-disaster redevelopment may influence the evacuation potential of at-risk populations to future threats, we modeled pedestrian travel times to safety in Seward, Alaska, based on conditions before the 1964 Good Friday earthquake and tsunami disaster and on modern conditions. Anisotropic, path distance modeling is conducted to estimate travel times to safety during the 1964 event and in modern Seward, and results are merged with various population data, including the location and number of residents, employees, public venues, and dependent care facilities. Results suggest that modeled travel time estimates conform well to the fatality patterns of the 1964 event and that evacuation travel times have increased in modern Seward due to the relocation and expansion of port and harbor facilities after the disaster. The majority of individuals threatened by tsunamis today in Seward are employee, customer, and tourist populations, rather than residents in their homes. Modern evacuation travel times to safety for the majority of the region are less than wave arrival times for future tectonic tsunamis but greater than arrival times for landslide-related tsunamis. Evacuation travel times will likely be higher in the winter time, when the presence of snow may constrain evacuations to roads.     

Tsunami vulnerability assessment in urban areas using numerical model and GIS

Natural disasters can neither be predicted nor prevented. Urban areas with a high population density coupled with the construction of man-made structures are subjected to greater levels of risk to life and property in the event of natural hazards. One of the major and densely populated urban areas in the east coast of India is the city of Chennai (Madras), which was severely affected by the 2004 Tsunami, and mitigation efforts were severely dampened due to the non-availability of data on the vulnerability on the Chennai coast to tsunami hazard. Chennai is prone to coastal hazards and hence has hazard maps on its earth-quake prone areas, cyclone prone areas and flood prone areas but no information on areas vulnerable to tsunamis. Hence, mapping has to be done of the areas where the tsunami of December 2004 had directly hit and flooded the coastal areas in Chennai in order to develop tsunami vulnerability map for coastal Chennai. The objective of this study is to develop a GIS-based tsunami vulnerability map for Chennai by using a numerical model of tsunami propagation together with documented observations and field measurements of the evidence left behind by the tsunami in December 2004. World-renowned and the second-longest tourist beach in the world “Marina” present in this region witnessed maximum death toll due to its flat topography, resulting in an inundation of about 300 m landward with high flow velocity of the order of 2 m/s.