Following a series of natural disasters, including Hurricane Katrina, that revealed shortcomings in the nation’s ability to effectively alert populations at risk, Congress passed the Warning, Alert, and Response Network (WARN) Act in 2006. The law prompted the first significant changes to national alerting systems since the mid-1990s when the Emergency Alert System (EAS) replaced the Emergency Broadcast System used for radio and television alerting. The resulting Integrated Public Alert and Warning System (IPAWS) combined existing systems such as the Emergency Alert System and the National Oceanic and Atmospheric Administration Weather Radio All Hazards system and would come to include the Wireless Emergency Alerts (WEA) system,1 which delivers short alert messages to cell phone subscribers. (See Box S.1.)
Today, new technologies such as smartphones and social media platforms offer new ways to communicate with the public, and the information ecosystem is much broader, including additional official channels, such as government social media accounts, opt-in short message service (SMS)-based alerting systems, and reverse 911 systems; less official channels, such as mainstream media outlets and weather applications on connected devices; and unofficial channels, such as first person reports via social media. Traditional media have also taken advantage of these new tools, including their own mobile applications, to extend their reach beyond broadcast radio, television, and cable; many even develop their
1 This text was modified after prepublication.
own mobile applications to deliver information. Furthermore, private companies have begun to take advantage of the large amounts of data about users they possess to detect events and provide alerts and warnings and other hazard-related information to their users; for example, Google provides alerts along with search results, and Facebook Safety Check detects emergencies and provides its users with an opportunity to register their status. Applications like Waze provide automated alerting regarding traffic situations to oncoming drivers and may also be used by government agencies to provide evacuation information. As a result, there are numerous opportunities to better deliver, target, and tailor emergency alerts.
More than 60 years of research on the public response to alerts and warnings has yielded many insights about how people respond to information that they are at risk and the circumstances under which they are most likely to take appropriate protective action. Some, but not all, of these results have been used to inform the design and operation of alert and warning systems, and new insights continue to emerge. In particular, a recent body of research, including work funded by the U.S. Department of Homeland Security (DHS; summarized in Box S.2 and described in more detail in Appendix B), has examined the implications of current and emerging technologies for the public response to alerts and warnings with a focus in part on how the public would respond to messages that could be delivered by future versions of WEA. Some of these results have already been used to enhance WEA, as evidenced by references to the research in the Federal Communication Commission’s (FCC’s) 2016 Report and Order on WEA,2 including the key insight that the 90-character message length afforded by the previous WEA system was not sufficient to accommodate the quality and quantity of information necessary for yielding a quick public response.3 Through this and other research, a good deal has been learned about how people use other tools, such as social media, during hazards and disasters.4
This report reviews the results of past research, considers new possibilities for realizing more effective alert and warning5 systems, explores how a more effective national alert and warning system might be created and some of the gaps in our present knowledge, and sets forth a research agenda to advance the nation’s alert and warning capabilities.
2 Federal Communications Commission, 2016, Report and Order and Further Notice of Rulemaking, FCC 16-127, September 29, https://apps.fcc.gov/edocs_public/attachmatch/FCC-16-127A1.pdf.
3 J. Sorensen and D. Mileti, 1987, Decision making uncertainties in emergency warning system organizations, International Journal of Mass Emergencies and Disasters 5(1):33-61.
4 L. Palen, K.M. Anderson, G. Mark, J. Martin, D. Sicker, M. Palmer, D. Grunwalk, 2010, A Vision for Technology-Mediated Support for Public Participation and Assistance in Mass Emergencies and Disasters, Association of Computing Machinery and British Computing Society’s 2010 Conference on Visions of Computer Science, Proceedings of the 2010 ACM-BCS Visions of Computer Science Conference, Article No. 8; and B.R. Lindsay, 2011, Social Media and Disasters: Current Uses, Future Options, and Policy Considerations, No. R41987, Congressional Research Service.
5 An alert notifies the recipient that something significant has happened or may happen, and a warning, which typically follows an alert, provides more detailed information describing the event and indicates what protective action should be taken by the recipient. The distinction between alerts and warnings is not always clear-cut because a warning can also serve as an alert, and an alert may include some information about protective measures. Technology has further eroded the distinction. However, this distinction may be important as some tools are better designed to provide what has traditionally been called an alert, or vice versa.
AN INTEGRATED ALERT AND WARNING ECOSYSTEM FOR THE FUTURE
The development and deployment of IPAWS and WEA established a valuable new tool for public alerting, and has been credited with saving lives.6 It leverages the ubiquity of cell phones (92 percent of American
6 National Weather Service, “Wireless Emergency Alerts: Real Stories,” release date May 28, 2014, https://www.weather.gov/news/130313-wea-stories.
adults own a cell phone, and 90 percent of those owners carry their phone with them frequently).7 However, there are many opportunities to go beyond WEA to make use of next-generation broadcast and multicast technologies, the emerging Internet of Things (IoT), and the ability of mobile devices to decide which messages to present based on user needs
7 L. Rainie and K. Zickuhr, “Americans’ Views on Mobile Etiquette,” release date August 26, 2015, http://www.pewinternet.org/2015/08/26/americans-views-on-mobile-etiquette/.
or contextual information the device has about the user or environmental and other contextual information. Moreover, the availability of new tools and technologies will likely generate new expectations among the public. Alerts and warnings that reach people through tools and communication devices they are using and present information in a way they are accustomed to will be the most effective. For an increasingly connected population using communication media in diverse ways, any methodology that relies solely on the current (cell) broadcast technology will no longer be sufficient to serve as the primary alert and warning system.
FINDING: Alert and warning systems exist within a larger communication and technical ecosystem, and government-designed and maintained systems must fit within this larger ecosystem.
FINDING: A more cohesive and all-encompassing alert and warning system is needed that will integrate public and private communications mechanisms and sources of information, and continue to provide the necessary information for the purpose of preserving the health and safety of people, while being technologically agnostic—such that new technologies for alerts and warnings can be adopted quickly.
FINDING: The nation’s alerting capabilities, such as WEA and IPAWS, will need to evolve and progress as the capabilities of smartphones and other mobile broadband devices improve and newer technologies become available. This evolution will need to be informed by both technical research and social and behavioral science research.
EVOLUTION OF AN INTEGRATED ALERT AND WARNING ECOSYSTEM
The committee envisions an alert and warning system that continually takes advantage of new technologies and reflects new knowledge that emerges from events and research. In the near term, this may mean increasing adoption of WEA and other existing alert and warning systems, incorporation of current knowledge about public response to craft more effective alert messages, and research focusing on verifying technology implementation and may also involve adapting existing technologies—such as new technologies for delivering and geotargeting messages—for use in alert and warning systems. Long term, this will involve gaining a better understanding of existing technologies, exploring new technologies, and continued sociotechnical research to inform the design and operation of future alerting capabilities. These near- and long-term visions
for an alerting system are fleshed out in this section and underpin the research agenda described in the next section.
Near Term: Adopt Existing Technologies for Alerts and Warnings
As of August 8, 2016, just under a third of U.S. counties have registered to use the Integrated Public Alert and Warning System8 gateway, the system that allows message originators to send WEA messages. As of the same date, state or local governments had originated only 387 wireless emergency alerts since WEA came online; by comparison the National Weather Service sent approximately 2 million alerts.9 An increased use of WEA by local emergency officials could not only mean reaching additional populations, but also increased use (for events other than weather) would improve familiarity with the systems, which could improve public response times.
Pending new FCC rules for WEA would expand the message length to 360 characters and allow the use of Web links (URLs) in messages.10 Although DHS research studied a range of message lengths, none of the studies looked specifically at 360-character messages, the capability introduced by the FCC rulemaking, or the use of URLs, which would point users to supplemental information. As a result, although the new rules provide new opportunities for emergency managers who have struggled to provide useful information in 90 characters, research is needed to determine what information to include and how to best display additional information in a WEA message itself and on any media it links to.
WEA was developed prior to the wide use of smartphones and newer cellular network technologies. New technologies could address the shortcomings of WEA, including a host of accessibility, security, functionality, and other concerns. These advances include the following:
- Modernize delivery technologies. The immediate opportunity to modernize is to switch from second- or third-generation Short Message Service-based Cell Broadcast to fourth-generation long-term evolution
8 IPAWS was created under Executive Order 13407 to integrate various alerting systems—Emergency Alert System, National Warning System, Wireless Emergency Alerts, and NOAA Weather Radio All Hazards—into one modern network. IPAWS takes advantage of the Common Alerting Protocol (CAP), an XML-based data format for exchanging alerts and warnings.
9 M. Lucero, FEMA IPAWS Division, “IPAWS Evolution,” presentation to the committee on August 9, 2016.
10 See Federal Communications Commission, 2016, Report and Order and Further Notice of Rulemaking, FCC 16-127, September 29, https://apps.fcc.gov/edocs_public/attachmatch/FCC-16-127A1.pdf.
- (LTE) broadcast, which provides faster delivery and longer messages, to deliver alerts.11
- Diversify communications technologies in handsets to help distribute alert messages when cellular network congestion or failure occurs. Short-range communications technology, such as Bluetooth and WiFi, could be used to forward messages locally, while FM radio provides an alternate and longer-range communication technology.
- Support the use of location information stored in handsets to improve the precision of geotargeting. Targeting based on relevance could be enhanced further by leveraging the ability of smartphones to determine not only where a phone is but also where it has been and thus where it is likely to be in the future.
- Incorporate more adaptability so that alert and warning capabilities can be upgraded more easily as understanding of public response and technology capabilities change. For example, the software on smartphones that supports receipt and presentation of WEA alerts could be moved from the operating system (which on some phones may not be frequently updated) to an app, which is more readily upgraded through the usual software update mechanisms.
- Provide mechanisms for performance monitoring and user feedback to facilitate studies related to perceived relevance (by seeking user feedback and/or inferring action taken), coverage (how many users did and did not receive a message), and message delivery latency.
Long Term: Build an Integrated Alert and Warning Ecosystem
In the longer term, IPAWS could be augmented so that it draws on a wide variety of data sources, enhances public understanding of emergencies and public response, and uses a wider range of potential technologies and devices for delivering messages. Envisioning such an advanced system requires exploring questions around technical feasibility and implementation and an understanding of how these tools will affect public response. However, past technical, social, and behavioral research already informs us of some of the properties that an ecosystem should have. These include the following:
- Using technologies that are privacy preserving. For example, location and other contextual information can be stored locally on a smartphone, and applications can use this information to decide when and how to display messages.
11 LTE Broadcast (or multicast) provides faster delivery and supports a larger content size.
- Assuring end-to-end service availability and the validity and integrity of messages.
- Giving users as much control as possible over what kinds of messages they receive, rather than limiting them to simply on or off.
- Including metadata in alerting systems that can be used in combination with user preference to determine when and how to present alerts.
- Integrating messages across a wider array of available communication channels. For example, IPAWS messages could be made widely available as a data stream for private industry to use freely in weather applications, navigation systems, social media streams, and the like.
- Making alerting systems device agnostic and able to support more than one modality of information presentation. For example, both text and voice alerts can be provided on mobile devices.
- Reflecting a better understanding of the information needs of emergency managers to quickly analyze data generated via social media.
- Using IoT devices and other embedded sensors to detect, analyze, and categorize potential events, send alerts, and potentially automate certain protective actions.
- Incorporating available communications technologies, such as mesh networking and FM broadcast signals,12 to increase the ability to deliver information in the event that primary communication networks fail.
- Adapting message content and format to the context and needs of the end user—for example, considering location of the device, known home location of the device owner, language of the device owner, disability status, and other context (as selected or entered by the user).
These desirable system properties and goals have the potential to inform research investments and to inform future system requirements.
A RESEARCH AGENDA
To realize the above-envisioned alert and warning system, additional research questions will need to be answered. Given that alerts and warnings are inherently interdisciplinary—both a social science phenomenon (their goal is to change public behavior) and a technical phenomenon (technology is required for their dissemination), this research agenda includes a wide range of sociotechnical questions and highlights the need for social and behavioral scientists and technologists to interact frequently with each
12 Many smartphones have FM radio receiver hardware built into them. There is potential for these to be used to provide information if a cellular network is not functioning; however, enabling this function requires the consideration of a number of technical and business issues.
other. The areas of research are described below briefly and explained in greater detail in Chapter 3.
As outlined in Chapter 1, much has been learned about the public response to alerts and warnings from years of research. However, many long-standing questions remain, and new technologies have introduced new questions. Key open topics include the following:
- Message characteristics. How message length and inclusion of protective guidance as hyperlinks affect public response, how to best express lead-time to a hazard, and how to best manage opt-in and opt-out preferences.
- Accessibility. How to most effectively provide messages in languages and dialects other than English, how to adapt to differing physical abilities, and how to account in emergency planning for disparities in access to technologies.
- Geotargeting. How to use the improved geotargeting capabilities afforded by WEA and the Common Alerting Protocol (CAP) best to communicate location, determine locations of interest (e.g., an individual’s location might not be at risk but their residence is), make use of improving indoor location capabilities, and determine and communicate protective action based on location.
- Community engagement. New tools and technologies support communications among members of a community; for example, NextDoor allows people to quickly identify neighbors and communicate with those people who live nearby. NextDoor is already being used by public safety organizations to educate the public;13 however, little is known about how such tools have been or could be used during disasters and after emergencies.
Expressing Time until Hazard Impact
Different hazards have different lead times. When too much lead-time is provided, people are less likely to follow protective guidance. Understanding how to best express lead time in WEA messages, and other alerting tools, is an important area for future research as well as the ideal lead times by hazard type.
13 M. Helft, “A Facebook for Crime Fighters,” Fortune.com, July 1, 2014, http://fortune.com/2014/07/01/nextdoor-local-neighborhood-social-network-police/.
Opt-In versus Opt-Out
Current WEA guidelines allow an individual to opt out of most emergency alerts. Past research suggests that alerts and warnings should be sent through as many channels as possible, but new research is needed to explore whether receiving the same message on numerous channels might prompt people to opt out of messages from WEA, such as third-party applications, or local text alerting systems.
Message Length and Protective Guidance in Enhanced Media Links
Although much is known about what information a message should contain, less is known about how to best communicate this information given constraints on message length and content. Research is needed, for example, to understand the public response to messages that fit into the new 360-character length, and further research is needed to determine the optimal minimum length, including personalized risk visualizations and/or URL links that can elicit the appropriate protective action from an alerted population. At this point, it is unclear what information is best included in a WEA message and what information is best included in linked content. Furthermore, concerns remain as network congestion could be caused by people accessing an included link within seconds of receiving a WEA that includes a URL.14
Language and Dialect
What technical challenges exist in transmitting messages in multiple languages or relying on the receiving device to translate messages? Are there practical limits to the number of different languages that can be supported? Additionally, key language elements, such as descriptions of protective actions, may be challenging to translate accurately to various languages and dialects. Research is needed to understand whether templates can be created so that messages can be automatically translated with sufficient fidelity.
14 Federal Communications Commission, “Improving Wireless Emergency Alerts and Community-Initiated Alerting,” release date November 19, 2015, https://apps.fcc.gov/edocs_public/attachmatch/FCC-15-154A1.pdf.
Adapting to Differing Abilities
A variety of technologies, such as vibration cadences and Braille interfaces, have been developed to allow mobile phones to be used by a wide range of physically and cognitively challenged message recipients. Research is needed to understand how best to customize the content and means of delivery to physically and cognitively challenged individuals. What other technologies exist to support information dissemination to differently abled individuals? How can protective action instructions be tailored or customized to support diverse populations—including those of differing ages and abilities—and their caregivers?
Although a large and growing portion of the population uses smartphones, there are still others who cannot afford or choose not to use them. Considering the diversity in communication habits and availability of technology, alert and warning systems will need to consider various technologies to reach individuals facing hazards.
What graphics most effectively indicate that an individual is in an at-risk location? How can visualizations be used to best illustrate the location of the message receiver relative to the area of impact? What is the best way to communicate to someone who is unfamiliar with the area they are in?
Determining Locations of Interest
Individuals want to be alerted not only when they are at risk, but also, for example, when their children may be at risk or their home may be at risk. How can locations of interest be determined and updated automatically rather than manually specified and updated by the end-user?
Location-Based Protective Action
The best protective action—for example, shelter in place versus evacuate—for an individual may vary across the affected area. Furthermore, individuals could be assigned diverse routes to enhance evacuation traffic flows. What are the technical challenges to providing such precision?
What are the limitations to implementing these for disaster response? How might we encourage use of these tools?
Knowledge of a person’s location within a building could be used to determine the best evacuation route or if the individual should instead shelter in place. Limited indoor location capabilities are already being deployed in some areas, chiefly for marketing purposes; determining building floor is a bit harder. What location techniques are emerging, and how could they best be used?
Hazard and Alerting Education
Very limited research has been performed looking at what makes hazard and alerting public education effective. Research to date has found current public education campaigns are generally ineffective because they are not specific enough and do not contain content that motivates behavior change.15 More research is needed to determine how to motivate behavior change as well as what other factors contribute to successful public disaster education campaigns.
Post-Alert Feedback and Monitoring
Technology is needed that solicits feedback from message recipients to help understand better who has received alert messages, how the public is responding to the messages, and what additional information might be needed. WEA’s cell broadcast technology may not reach all phones in the target area, and today’s systems are one-way, meaning that alert originators have no way to know who has actually received an alert. An acknowledgment mechanism would be a useful element as part of any feedback and monitoring system. Some information about the public response can also be obtained using tools that extract information from social media. More direct feedback mechanisms could be built into alerting applications on mobile devices, and these tools will need to be more readily available. Perhaps more importantly, research is needed to under-
15 B.J. Adame, and C.H. Miller, 2015, Vested interest, disaster preparedness, and strategic campaign message design, Health Communication 30(3):271-281; J.D. Fraustino, and L. Ma, 2015, CDC’s use of social media and humor in a risk campaign – Preparedness 101: Zombie apocalypse, Journal of Applied Communication Research 43(2):222-241; and M.M. Turner, and J.C. Underhill, 2012, Motivating emergency preparedness behaviors: The differential effects of guild appeals and actually anticipating guilty feelings, Communication Quarterly 60(4):545-559.
stand what information would be more helpful to emergency managers. Tools, including those that employ machine learning and other artificial intelligence techniques, are also needed to quickly understand and process feedback to ensure emergency managers are not overwhelmed with information.
A future alerts and warnings ecosystem that includes consistent, well-understood, and insightful measurements could inform (and improve) response to future hazards. Such a data-driven experimental framework would be of great interest to multiple stakeholders, including emergency managers and researchers. By building measurement into the alerts and warning system itself, researchers could gain supporting evidence for findings made in lab studies (e.g., which message length is appropriate? should we include a map or not?). Feedback during the life cycle of a hazard could also be integrated into future responses within the same incident. For example, low response rates to an initial message could lead to more aggressive message content in a follow-on message.
Alerting System Trustworthiness
Existing and new technologies can increase the likelihood that messages reach populations at risk, enable richer interactions between emergency managers and those at risk, and automate certain time-sensitive functions. At the same time, current and future systems will be at risk of malicious attacks and attention will be needed to protect individual privacy.
Today, WEA is designed to use only cellular communications. While there are still several technical research questions around cell broadcast technologies, such as the use of next-generation networks, cell phones can also receive data through a variety of other wireless communications technologies that could be adapted for message dissemination. Additionally, during hazards, some cellular networks may not function properly, so other technologies are needed to deliver messages (e.g., peer-to-peer, FM radio). Battery life management on end-user devices is also a rich area of research.
Role of Connected Devices
As the IoT grows, more devices in homes and throughout the environment will be available not only as an alerting channel but also to detect emergencies and potential risk. It may be helpful to ensure that many
connected devices that have a sensory output can be triggered to provide an alert or warning. To make the most effective use of these opportunities, several questions around aggregating data, potential use of automation, which devices are best suited for alerting, and the potential role of IoT devices in milling. Machine learning and other artificial intelligence techniques will play a role in the ability to automate the sending of alerts in short-fuse events, such as earthquakes and active shooter situations, and also to provide responders with better information during and post-events.
Trust, Security, and Privacy
A system that instructs large populations to take a particular action may represent a significant target for attacks on service availability, compromises of the integrity of valid messages, and spoofed messages.16 As emergency managers begin harnessing information—including personal and geographically relevant information—from social media, security and privacy concerns will increase. How can we take advantage of these tools while still protecting end-user privacy?
Furthermore, in a system that makes use of user-generated (public-generated) content, misinformation becomes an increasing concern as well. Quickly detecting and correcting poor information will be valuable system capabilities.
CHALLENGES TO BUILDING BETTER ALERTING SYSTEMS
Beyond the specific research topics listed above, the committee noted several challenges to building better alert and warning systems.
Slow Adoption of New Systems
Reasons for lack of adoption include system costs for jurisdictions and message originators’ education, but even those with access to the IPAWS gateway can be hesitant to use the system. Moreover, in smaller jurisdictions, sending alerts may be a part-time job, and a person may only be active in the emergency response community during events; in the largest jurisdictions, public alerting may be the responsibility of a large team of individuals who are trained emergency management professionals immersed in disaster response full time.
16 C. Woody, Software Engineering Institute of Carnegie Mellon University, “SEI Wireless Emergency Alerts (WEA) Research 2013 through 2016,” presentation to the committee on September 1, 2016.
Limitations on Weather Forecasts and Other Information about Natural Hazards
Agencies that distribute weather-related messages at the state, local, regional, or federal level must ultimately rely on forecasts and weather information from the National Weather Service and National Oceanic and Atmospheric Administration and information provided by the U.S. Geological Survey. These agencies in turn rely on the infrastructure that collects, models, and distributes information about weather, earthquakes, air quality, and other environmental conditions. Information provided by these agencies also supports an array of private-sector alerting services. Effective alerting depends on modeling and data collection and analysis capabilities being maintained and advanced.
The array of information and communications technologies used by the public are continually evolving. At the same time, old and new technologies coexist for long periods of time. To reach the majority of individuals, alerting systems must not only evolve but continue to make use of legacy technologies. Furthermore, both the technology of emergency alerts and citizens’ capacity to comprehend the alerts and use messaging functions also continue to evolve. The interaction between the developing technologies and citizens’ capacity to use these technologies effectively on a community scale is itself an issue for future research.
Difficulty of Interdisciplinary Research and Converting Research to Practice
Public response to alerts is a highly interdisciplinary activity, and it is also an activity closely coupled to the practice of emergency management, which takes place primarily at the state and local level in the United States. Yet technologists, social science researchers, and emergency managers have had few opportunities for ongoing interactions to consider how to apply current knowledge or fill gaps in our understanding.
Incentives to Participate
An alert and warning ecosystem incorporates numerous official sources of information as well as numerous other information providers, such as social media companies, navigation companies, local media, and hardware makers. For example, WEA relies on cellular service providers to implement the necessary capabilities in their infrastructure and for cell phone manufacturers to include the necessary software in smartphones
(although participation is voluntary, all major carriers currently participate). Incorporating these various pieces, and ensuring that information about how the system is working is shared, will be an increasing challenge. How do we encourage openness among stakeholders and encourage participation by those who operate other valuable computer and communications capabilities?
Our nation’s ability to respond effectively to natural hazards and humanmade disasters depends on our ability to deploy improved alerting systems that take advantage of new technologies, informed by a better understanding of the way in which the public uses and responds to these systems. Doing so will depend on addressing the challenges and research areas listed above.