Managing Pedestrians During Evacuation of Metropolitan Areas > Table of Contents

Executive Summary

Transportation Agency Interest in Pedestrian Evacuation

The September 11, 2001, (or 9/11) attacks on the high-profile workplaces of the World Trade Center (WTC) in New York City and the Pentagon in the Washington, D.C. area, made real the impact of an unexpected, or “no-notice,” event in a metropolitan setting.  The news coverage of the events of 9/11 showed thousands of people leaving the area of the WTC on foot.  The evacuation from the borough of Manhattan included not only the typical traffic congestion expected in an evacuation in the United States, but thousands of pedestrians moving along with, or among, the vehicles.

When a large-scale, damaging event has occurred or the imminent threat of one has become known, transportation agencies working with public safety officials have traditionally had two principal objectives:

• Minimize the time it takes to get an adequate force of emergency responders to the scene where they can help victims, provide assessments, and control access.
• Maximize the proportion of the population moved away from the hazardous area without being subjected to other risks (e.g., traffic crashes, prolonged exposure to the danger). 

Evidence that large numbers of pedestrians may be part of an evacuation raised questions within the Federal Highway Administration (FHWA) about what actions are needed to manage pedestrian traffic during metropolitan evacuations and what FHWA can contribute in this area to ensure safe and effective movement of pedestrians while minimizing their impact on vehicular movement. 

Scope of the Research

The term “no-notice metropolitan evacuation” here refers to an emergency evacuation taken as a protective action that is implemented for a portion of a densely built-up downtown area in a large city in the United States.  The basic situation assumed for the discussion of the issues is the following: 

• The triggering event will occur without warning (e.g.: terrorist attack, explosion, or toxic release);
• The scope of the event will be such that an emergency evacuation activity will be critical for a portion of the metropolitan area;
• Emergency response actions will occur to mitigate the injury and damage from the destructive or threatening event even as the evacuation is taking place;
• The evacuation is expected to be temporary for those parts of the area that were not badly damaged or contaminated. 

The term “pedestrian evacuation” generally refers to masses of people who leave a suddenly dangerous area in order to reach a safer place and do so on foot.  For pedestrian evacuation to be of concern to transportation agencies, it entails the combination of masses of people on foot along with the corresponding congestion of the evacuation of others in private vehicles, always or at times moving along the same routes.  Examination of the 9/11 situations in New York City and Washington, D.C., also brings out the need to consider the situation of large numbers of people who are initially on foot but have chosen their route because it takes them to a point for accessing mass transit that can take them out of the city.  These two examples also suggest that when a downtown business district is the source of the evacuation streams, a large proportion of the evacuees will have as their intended destination their residence, even if it is far beyond the distance they need to travel to be in a safe location.

The reader should consider this report as an initial and cursory effort to identify what is already known about managing pedestrian traffic in U.S. metropolitan evacuations.  The research tapped a large variety of information sources for insights.  These included published studies representing a variety of disciplines (Chapter 2), telephone interviews with practitioners in cities representing examples of emergency evacuation or evacuation planning for the city (Chapter 3), and a discussion of variables and issues with a multi-disciplinary panel of experts convened for a day-long meeting on the topic (Chapter 4). In total, 23 practitioners and other subject matter experts provided information.

The inquiry began with a review of literature that tapped into the topic from several perspectives, including: transportation planning and operations, social and behavioral research on evacuation decision making and behavior, and crowd and pedestrian flow modeling.  The research materials included a series of case study reports done for the FHWA to examine transportation operations during catastrophes that included the 9/11 attacks, the Northeast power outage in 2003, and the 1994 Northridge Earthquake in southern California.  Several emergency response after-action reports also were found and reviewed, related to the 1993 World Trade Center bombing, the 1995 Oklahoma City federal building bombing, and the 1995 Tokyo subway sarin gas attack.

Findings from the Literature Review

Pedestrian evacuees apparently are a fairly rare phenomenon since car ownership became widespread in the United States.  The principal observation from the literature review is that studies of evacuations do not mention pedestrians and no research was found that was focused on the specific topic of pedestrians in U.S. evacuations prior to 9/11. Similarly, no systematic research was found from 9/11 that focused on the unusual phenomenon of large numbers of people who left the city on foot.  On the other hand, at least between 1950 and 2000, there have been no sudden and catastrophic events in the larger U.S. cities that would have provided a context for examining pedestrian behavior in no-notice emergency evacuation from a large city. 

Practitioner Interviews

The research also identified five cities where emergency evacuations or evacuation planning had taken place and conducted phone interviews with public safety, emergency management, and transportation practitioners in these cities:  Washington, DC metropolitan area; Cleveland, Ohio; San Francisco-Bay Area, California; New York City, New York; and Seattle, Washington. Some selected points of importance for practitioners are the following:

Observations from Washington, D.C., practitioner interviews:

• Selected points important for practitioners:  The evacuation on 9/11 was spontaneous rather than recommended, causing the traffic control measures to prevent gridlocked traffic to lag behind the initial stages of the evacuation.  The large numbers of people leaving office buildings at approximately the same time created even larger crowds on the sidewalks and around mass transit loading areas than does an end-of-day commute. The sudden deployment of security barriers around government facilities created unfamiliar detours for both drivers and pedestrians.  Since 9/11, traffic cameras have become more common along major routes, and can be of considerable value in the event of a future evacuation. The transportation agency also intends to continue to deploy “spotters” to watch the evacuation traffic at key points in order to be able to resolve traffic conflicts or delays before they worsen.  It is critical for transportation agencies to work closely with the mass media to ensure that no misinformation gets reported about the status of the mass transit system.  Mass transit, especially rail service, can be an important mode for use by evacuees because it is not affected by traffic congestion.

Observations from Cleveland practitioner interviews:

• Selected points important for practitioners:  The evacuation of just a few downtown buildings in the face of a general threat (as happened in Cleveland on 9/11) is likely to prompt evacuation of the entire downtown area so it is important for responders to be prepared to manage it.  When the development of a downtown evacuation plan was undertaken shortly after 9/11, the mayor involved the downtown building owners and managers groups as well as public safety and other government agencies.  This helps to spread the education efforts among many participants about the evacuation plan.  The normal contingency plans that the transit authority has for quickly creating “bus bridges” if some portion of the rail system becomes inoperable are easy to activate should they be needed to effect some part of a city evacuation.  When Cleveland was affected by the Northeast blackout of 2003, emergency managers recognized the exodus from downtown as an early commute rather than an evacuation and did not try to invoke the downtown evacuation plan because of its basically different underlying strategy for clearing people from downtown.

Observations from San Francisco Bay practitioner interviews:

• Selected points important for practitioners:  Experience with portions of highways being suddenly closed due to earthquake damage has made local authorities aware of the situation of drivers suddenly finding themselves in areas they are totally unfamiliar with and in need of staging areas, service, and information.  Media convergence on areas with dramatic damage can create a bigger problem for emergency responders than do pedestrian evacuees who are more likely to stay out of the way. Because earthquakes often cause damage to some parts of the transportation infrastructure, the transportation agency is likely to be focused on addressing these problems and not available for assistance with evacuation streams in other locations. The Oakland Hills fire demonstrated that people accustomed to driving out of their neighborhoods may not think to evacuate on foot along footpaths even when it is evident that driving out will take too long. One law enforcement officer interviewed stated his belief that pedestrians will know what to do and will not need a lot of direct instruction from officials.

Observations from New York City practitioners and other materials:

• Selected points important for practitioners:  The large number of people who left the downtown area on foot in the face of the attack on and collapse of the World Trade Towers on 9/11 may be a reflection of the fairly large percentage of people who use mass transit to get to work in the business district so are car-less when downtown. The importance of maintaining mass transit as much as possible, and devising means for providing information to people wanting to use mass transit for evacuation, is critical in this situation. The fact that so many of the people who evacuated on foot walked much further than was necessary to be out of any danger suggests that when people evacuate from their offices they are most likely to prefer to get home, even if it is several miles, so response organizations need to prepare to accommodate this behavior and make safer and easier than it would be otherwise. Many ad hoc arrangements were made to accommodate evacuees such as other boats engaging in ferrying people off Manhattan Island, and law enforcement requesting drivers to take as passengers evacuees who might otherwise try to walk across a bridge or through a tunnel to leave Manhattan. Much of what initially looked like pedestrian evacuation was likely to be more multi-modal in that many people walked to where they could get mass transit and completed part of their journey in that manner.

Observations of Variables Affecting Evacuation Management

Using the information from the literature review and the interviews, concepts and practices that would be useful in planning and managing metropolitan evacuations were delineated. Once the various contextual elements of a metropolitan evacuation are established, the planning effort can focus on the combination of response elements that are bested suited to the characteristics of the specific event and the area affected by it.  Basic variables that combine to affect the evacuation process, and which form the basis for planning and managing a no-notice urban evacuation, include the following:

• Characteristics and geographic scope of the threat to personal safety;
• Location and number of evacuees;
• Configuration and capacity of evacuation routes and mass transit systems.

This context becomes the basis for:

• Strategic intent of evacuation implementation (e.g., minimize vehicle traffic, maximize evacuees going north, etc.);
• Primary evacuation instructions as formulated by the authorities;
• Available options for communicating instructions to evacuees.

As any transportation or public safety official knows, real events and exercise scenarios seldom fit what would be the “best case” situation for implementing a large scale evacuation from a city center.  There obviously can be multiple combinations of variables related to the type of event, characteristic and current conditions of the area, transportation and routing alternatives, whether evacuees are leaving home or going home, and the degree of preparedness and cooperation among the agencies involved in facilitating the evacuation.

As another effort to gather ideas, a multi-disciplinary group of subject matter experts was convened to address issues related to pedestrian evacuation from various perspectives.  The panel participants included practitioners and subject matter experts in firefighting and incident command, evacuation planning, public safety and law enforcement, traffic operations, modeling of pedestrian movement, and social science research on evacuation and disaster response.  (It should be noted that invited transit agency representatives were unable to attend.)

A major recurring theme was that best practice for managing pedestrian evacuees lies with the application of best practice in overall emergency preparedness and response.  For most participants in the discussion the emergence of a large contingent of evacuees on foot was viewed as an element that may or may not occur in conjunction with an urban evacuation.  The most important factor for effective management of pedestrian evacuation as a variable related to context and situation is that all relevant agencies have, first, a shared understanding of the concept of operations for the jurisdiction’s agencies, and second, established working relationships with other agencies either during actual response activities or preparedness activities such as evacuation planning and emergency exercises.  At the same time, another important theme was that there may be conditions under which a largely pedestrian evacuation might be a preferred strategy for getting people moved a short distance to safety, with the associated issue of whether evacuees could be expected to accept instructions to abandon familiar transportation modes or walking routes.

The ability to do this would depend on a shared understanding of applying decision criteria in a timely manner.

Other themes and issues raised by the panel were:

• Preparation for special considerations related to pedestrian evacuation may be particularly important in cities with large numbers of tourists and high commuter use of mass transit, because many people will be unfamiliar with routes and landmark facilities named in evacuation messages.
• Interagency planning for how best to provide clear and consistent information to pedestrians that enables them to make quick decisions about evacuation and route selection can help to reduce the amount of responder attention needed to manage pedestrian evacuation.
• In cities where Intelligent Transportation System (ITS) assets are well developed, the transportation agency will have some of the best information and the broadest overview of the traffic and pedestrian flow, making it imperative that the transportation agency participates in emergency management functions and not simply in a support role.

Chapter 5 presents conceptual approaches for transportation and emergency planners to consider for managing and accommodating pedestrian evacuation.  These are a reflection of solutions implemented by New York City to keep evacuating pedestrians, vehicles, and emergency vehicles separated and of evacuation concepts reflected in two downtown evacuation plans identified during the project. These plans included public education about how people evacuating on foot would be accommodated.  Because of the location of the main business district of New York City being on the island borough of Manhattan, it is possible for emergency response planners to assume that evacuating vehicles and pedestrians will converge at the few transportation facilities for leaving the island, and so can focus attention on how to balance mobility and safety at those points.  The two downtown evacuation plans discussed by the panel have not as yet had to be implemented, but are designed around a basic strategic intent of getting much of the downtown population to immediately move on foot directly away from the danger area with the knowledge that there are several pre-determined locations where evacuation buses will be staged.  

In order to outline generic options for managing or accommodating pedestrian evacuees, it is necessary to hold constant the underlying scenario of the event.  The following evacuation planning concepts suggested here extend beyond the familiar one of evacuees all driving or riding in private vehicles. For the sake of a simple presentation, these are predicated on the existence of the basic, best case scenario of an impact that has not created an airborne hazard, in a downtown district that is mostly office buildings, daytime and good weather conditions, and operable mass transit options. These are referred to as conceptual approaches to indicate that they are presented as pure types, without an attempt to address actual variations in the context or the nature of the agent prompting the evacuation.

Three approaches assume there will be appreciable numbers of pedestrians among the population making a temporary evacuation from an area because of an unanticipated and major impact in the densely developed downtown area of a large city.  The reader is cautioned that this basically is just an assumption, there having been too few such events in large U.S. cities to know if the scenes from the 9/11 attacks in New York City and Washington, D.C., are likely for those cities only because of some particular characteristics or the particular 9/11 events, or if they are likely to occur in other large U.S. cities. 

The three general modes available to individuals whether they are commuting from home to work or going about a city for errands or social engagements are: driving a private vehicle, riding in a private vehicle or public transit, or walking.  People who are car-less in a large city normally have several options for making trips within the city or back and forth between home and work.  All of these modes are suitable to situations requiring evacuation of a portion of a city.  In densely developed urban areas evacuation in one’s private vehicle may not be the most effective for making a rapid evacuation. Under certain circumstances, greater use of other modes may be more suitable for achieving a balance of rapidity and safety for the largest number of people.

The three approaches share a common objective: to ensure the safety and mobility of pedestrians while minimizing the likelihood that they may contribute to evacuation traffic congestion.  Crowd and traffic management techniques can be used to separate vehicle and pedestrian streams, or, taking a different angle, the pedestrian stream can be reduced as the evacuation progresses.

 

The other three theoretically logical approaches evacuation planners might consider for reducing or eliminating risks created by evacuation streams including both vehicles and pedestrians are the following:

• Prohibit or minimize evacuation on foot, other than for moving on foot to transit options or private vehicles, which is the form of evacuation that seems to emerge naturally in the United States, given generally low density towns and cities, and high levels of car ownership.
• Prohibit or minimize the choice of evacuation by vehicle, as a way to eliminate evacuation delays caused by the extreme traffic congestion due to the sudden increase of vehicles on the streets within a very short time period, in particular in high density urban places. For this to happen, authorities would have to prohibit people from accessing private vehicles parked in the area. 
• Prohibit or minimize evacuation, which is the option of sheltering-in-place that is adequate as protection under some types of circumstances or, even if imperfect for complete safety, the best option when imminent danger creates a greater risk for those who are on the streets rather than in buildings at the moment.

The first of the three is the most likely to be feasible to implement because it is already a behavior that typically emerges in the United States when evacuation is recommended by authorities. Evacuating in one’s private vehicle allows the evacuees some discretion in deciding how far to go and permits the evacuee to keep at least this one valuable possession at hand.  The latter two call for authorities to recommend and expect that individuals will follow instruction that runs counter to the typical social behavior observed by disaster researchers.

Findings from many decades of disaster research lead social scientists to be skeptical of the likely effectiveness of emergency planning based on the assumption that the population will do what makes the most sense to planners and responders.  Behavioral research suggests that emergency planning instead needs to acknowledge what makes the most sense to people whose first concern will be for assuring the safety of other family members or friends even at the risk of delaying appropriate protective action for oneself.  However, the social scientists are also quick to observe that, in actuality, there is little empirical evidence at this time for knowing if what is known about evacuation behavior in smaller cities will apply in concentrated metropolitan areas.  The most commonly observed response and evacuation behavior has been in response to the traditionally structured environment of smaller cities.  Persons living in high density urban environments and more used to transportation alternatives can be expected to collectively adjust their behavior to that context. The unanticipated emergence of pedestrian evacuees in New York City and Washington, D.C., in the face of the 9/11 situation provides some evidence that there will be differences in future events in similar settings, because of the 9/11 experience. Not only was the option of evacuating on foot “publicized” through media accounts as an option during urban evacuation events, but the delays drivers experienced because of the initially gridlocked traffic may result in more urban evacuees being willing to try other options in the future.  Circumstances that differ in many high density downtown areas, compared to smaller cities, include better walking facilities, more mass transit options, and a greater likelihood that a large proportion of the evacuees are headed home, not leaving home. Evacuees headed home will have a route in mind, because of this preferred destination.

Research Needs

Up to this time, evacuations from large, highly concentrated U.S. cities similar to New York City, Chicago, or San Francisco are virtually lacking to provide sufficient information upon which to base expectations. Systematic study of large-scale pedestrian evacuation in U.S. cities still is needed to better understand how pedestrian evacuees react to various types of settings, situations, and instructions. At the highest level of generality, previous research indicates they will behave rationally; what is not yet systematically documented is the specific adaptive and problem-focused evacuation behavior from the point of view of urban occupants.

The research for this report was conducted prior to Hurricanes Katrina and Rita in September 2005.  The evacuation of the New Orleans Metropolitan Area in response to the several-day advance warning about the approaching hurricanes followed the typical pattern for hurricane evacuations.  Approximately 75% of the occupants evacuated, most driving or sharing private vehicles, during the warning period that preceded the hurricane impact.  Almost half of the 25% who did not evacuate thought it would be safe to stay.  The unanticipated flooding from the levy breaches necessitated rescue of many of those who had not evacuated.

Transportation managers and researchers need to be prepared to conduct studies following the next case of a major metropolitan evacuation from a no-notice event.  FHWA sponsored relatively detailed case studies of the 2001 and 2003 evacuations from New York City and the 2001 evacuation from Washington, D.C., but the objective of the case studies was to determine the disruptions to transportation activities and adjustments made by the transportation agencies. Consequently, the large number of people evacuating on foot was more a finding than a focus of the inquiries.

At the moment of this report, there are two major opportunities for gathering more systematic detail about urban evacuations involving pedestrian evacuees:  the last case of a large scale pedestrian evacuation and the next one.  The most recent cases would be that of people leaving Lower Manhattan after it became evident that two airliners hitting the World Trade Center towers was not an accident, and the government and other office workers spontaneously leaving downtown Washington, D.C., as news of the attacks reached them. Various types of research were found in relation to New York City, but none was identified that provided an overview of the decision points of the pedestrian evacuees or the interactions between pedestrians and the agency officials with roles related to facilitating the temporary evacuations.  Now that issues have been raised about pedestrian evacuation, preparations should be made for research on the next case as well.  General areas of recommended research about the management and accommodation of pedestrians from these two past cases and in the next case of no-notice urban evacuation are the following:

• Surveys with samples of people who work in office buildings in the downtown area to obtain more detailed information about their evacuation decision, the transportation mode (including walking) that they selected, what information they used to make these decisions, what other information they wish they had had, and if they intend to make different decisions in the event of another evacuation situation.  This information can be compared to findings from earlier research on evacuations from cities and towns of lesser population density to determine similarities and differences.
• Interviews with key staff from relevant agencies that can be used to triangulate the observations of officials (e.g., law enforcement, traffic control, or transit personnel) who were located in different parts of the city and operations centers when the evacuation began and as it progressed.  This could help to pinpoint if, how and where pedestrians, private vehicle traffic, and emergency vehicle traffic were intermixed and how situations of traffic and pedestrian congestion that emerged were resolved. Findings from such research would help identify the extent of such congestion, factors related to its existence, the ways in which safety and mobility were affected, what might be effective approaches for mitigating such congestion if it presents an actual problem, and what specific messages for evacuees were provided through what means.
• Compilation of photographic records (both still-images and video captured by private, mass media, and even surveillance cameras) identified with respect to specific location. These could provide important details about the routes and behaviors of the pedestrian evacuees and of interactions between pedestrians and officials as the basis for providing scenarios for consideration by the city’s evacuation planners.  Such otherwise quickly disappearing “data sets” might exist with the mass media, private citizens’ video recordings, traffic and security surveillance cameras at intersections and on buildings, ATMs, fortuitously-timed satellite photographs, and so forth.  This technique could provide a wide scale and graphic overview of a phenomenon that is difficult to make sense of when viewed from only one or two locations. This information could be used by evacuation planners to consider appropriate measures for facilitating and managing both traffic and pedestrians.

If transportation and emergency management agencies view as high priority the possession of better information on which to base planning assumptions about evacuations with large numbers of people on foot as well as in vehicles, it is important that the decision to collect this information be made in advance of the next event.  Timely post-disaster research calls for consideration of how to make resources immediately available for data collection.

Summary and Conclusion

In summary, the main intent of this research was to try to identify practices or ideas that transportation managers and public safety agencies were planning or maybe had actually used, to ensure that people evacuating on foot and people evacuating in vehicles did not hamper each other’s mobility in relation to a preferred destination.  Public safety and emergency response practitioners expressed concern that they had not given adequate thought to planning for large numbers of people evacuating on foot, now that there was some indication this may occur in large cities. The discussions with the practitioners also indicated they believe the usual techniques for traffic congestion management can be used.  However, they assert that emergency planners need to focus much more attention during preparedness planning on evacuation scenarios that provide planning assumptions for pedestrian evacuation. The public safety practitioners emphasize that the basis of effective emergency response of any type is the development of close and continuous working relationships across agencies.  The transportation practitioners add to this that the transportation agencies have traditionally responded based on decisions and directives of others.  The development of intelligent transportation systems (ITS) in many cities now puts the local departments of transportation in the position of being able to make major contributions of information during disaster response as well as to provide information to the public, and especially travelers.

In conclusion, based on the information collected for this report, it seems reasonable to say that the factors associated with the emergence of a large scale pedestrian evacuation from a section of a large U.S. city are not as yet very well understood.  Further focused research is needed to help clarify the planning assumptions for managing or accommodating pedestrians in a metropolitan evacuation.  Pedestrian evacuation can be seen as a phenomenon that responders should consider in order to be prepared for the potential that large numbers of people may decide to evacuate on foot, along with vehicle traffic.  On the other hand, evacuation planners may want to consider under what circumstances the deliberate implementation of a pedestrian evacuation might be a preferred and feasible strategy in highly urbanized areas for most quickly getting the largest number of people out of harm’s way.