A variety of emergency management organizations and other groups that communicate with the public during disasters have been using social media to disseminate information and observe the public response to events. In this workshop session, four speakers shared firsthand experiences with the use of social media: Brian Humphrey, Los Angeles Fire Department; Brad Panovich, WCNC-TV in Charlotte, North Carolina; Paul Earle, United States Geological Survey National Earthquake Information Center; and Keri Lubell, Centers for Disease Control and Prevention.
Brian Humphrey described the wide array of social media tools the Los Angeles Fire Department (LAFD) uses to both disseminate and monitor information before, during, and after fire emergencies, including two Twitter accounts: @LAFD,1 used solely for alerts, and @LAFDtalk,2 used for conversations with the public. The latter account is used to help build trust between the public and the fire department. For example, when followers share personal news, such as birthdays or anniversaries, via Twitter, @LAFDtalk may acknowledge these occasions to engage with fol-
lowers socially. Status updates often include names or initials to reinforce that there are real people behind the account.
The value of the LAFD’s engagement of the public through social media was highlighted during an incident in which an explosion was reported at the Los Angeles International Airport. The LAFD could see who had recently indicated via Foursquare3 that they were at the terminal where the explosion was reported, and asked these individuals (via Twitter) to contact its public affairs office by telephone. By asking these individuals what they had observed, the public affairs office could provide first responders with information about the event even before they were on the scene; in this case, the public affairs office was able to tell the responders that the explosion was in fact the result of a lithium battery overheating, a relatively minor incident.
Nevertheless, some potentially useful tools for using social media have proven too expensive to acquire, and some uses of certain tools are precluded by the terms of their end user license agreements, commented Humphrey, but he noted that the LAFD has identified a set of free or low-cost tools that are useful for disseminating and monitoring information. Many of the LAFD’s social media accounts are fed by email using the services Ping.FM4 and HelloTXT.com,5 which makes it possible to quickly provide or request information through the approximately 80 social media accounts managed by the LAFD. These accounts, including @LAFD and @LAFDtalk, can also be used individually to interact with the public.
Different social media tools are useful during different stages of a disaster. For example, according to Humphrey, Blog Talk Radio6 is a tool that is particularly helpful during the recovery stage. Blog Talk Radio can be used to create Internet radio stations, which can take calls from users. Other related tools are also useful. Standards such as RSS (Really Simple Syndication) and various XML (Extensible Markup Language) schema make it easier to represent, distribute, and analyze information.
One task for which better tools would be helpful, Humphrey observed, is in extraction of information from photographs distributed over social media. Currently, the LAFD relies on manual searches for potentially useful pictures based on the social media tags they are associated with. Another possible source of information is the metadata included in image files, such as the time or location. It would also be helpful to have tools
3 Foursquare is a location-based, social media application for mobile devices. Users “check in” to locations found near their current location, which is detected using the GPS hardware in the mobile device. See https://foursquare.com.
that could automatically identify pictures that might enhance the LAFD’s situational awareness about events, such as pictures that show fire or smoke.
Humphrey concluded his remarks by identifying several important lessons from the LAFD’s experience with social media:
• A partnership with information technology (IT) staff in the emergency management organization is important in part so that IT staff understand how emergency professionals are using the computers and network.
• Appropriate management of and collaboration with traditional and new media are critical components of each phase of a given disaster. There is often a gap between response (usually an acute situation) and recovery (a more ongoing process). Social media can smooth the gaps in the cycle by providing two-way communication between the affected population and both first responders and emergency personnel.
• Messages disseminated using social media need to be clear, concise, and, most important, actionable.
• The more one is willing to empower people and engage them, the more information the public is willing to provide.
• Understanding how people communicate with social media is important. People often do not simply state “help,” “fire,” or “explosion” but instead use such exclaimers as “OMG!” (oh my god!) or other slang. When one has three or more people within a 20-mile radius saying “OMG,” this can be a signal to look more closely at what might be happening in the area.
• Different phrasing suggests quite different meaning, as the following examples regarding a fictional shooting illustrate:
—I heard there was a shooting at 5th and Elm St.
—I heard there is a shooting at 5th and Elm St.
—I heard about a shooting at 5th and Elm St.
—I heard shots fired at 5th and Elm St.
—I saw a shooting at 5th and Elm St.
—I just saw a person get shot at 5th and Elm St.
Brad Panovich began by describing the multi-tiered warning process he uses to inform the public when potentially severe weather is forecast. The first step is a blog post several days in advance to increase public awareness. As severe weather approaches, Panovich begins issuing alerts or warnings via Twitter and Facebook. Many of the alerts and warnings
originate from iNWS, a service of the National Weather Service, which sends information via text and email to emergency managers, local county officials, and the media.7
Panovich, using his desktop computer, laptop, or other mobile device, needs to be able to quickly post alerts and warnings from iNWS to his social media accounts. In Panovich’s workflow, these messages are sent to an email account that has a rule set to automatically forward iNWS messages to TwitterMail, which in turn posts the alerts to Twitter. Panovich noted that a more complicated setup is needed to automatically post messages to Facebook because only the subject of an email is included in a status update when a message is emailed to Facebook. Instead, tools like Ping.FM and Tumblr8 can be used to automatically post information to Facebook. A text alert from iNWS can be pasted into Ping.FM, which is configured to send the message to any of the six Facebook pages that Panovich manages. Tumblr blogs allow for posts via email, and then these posts can be sent directly to an RSS feed, a Web-based tool that can display various updated Internet sites in a standard format, or automatically posted to Facebook or Twitter.
WCNC uses YouTube and uStream to transmit video over the Internet. One use is to supplement the broadcast forecasts with additional background on the science behind the weather, which is appreciated by audiences and helps strengthen Panovich’s credibility with them. Streaming also allows the public to view forecasts and other information on computers and mobile devices9 and can be used as a supplemental source of information to point people to an alert displayed over normal television programming, allowing viewers who seek additional information to quickly find it using a cell phone or other mobile device. In general, Panovich believes, the public finds this approach less disruptive than interrupting programming to provide information to supplement an alert, especially when many of those receiving the broadcast may not be in the affected geographical area.
Panovich’s viewers and readers access his information using both traditional and social media channels. Understanding an audience’s needs is important, Panovich observed. He uses a tool called SocialBro to analyze the activities of his followers. For example, information on when users are most active helps inform decisions about when to create new content. The
9 To provide mobile device streaming, it is important to use the correct codex for mobile devices.
tool also can be used to understand other aspects of users, such as what words are used to describe significant weather events.
Panovich identified several important lessons from his experience using social media as a broadcast weather forecaster:
• It is important to build a social network before an emergency occurs.
• Although automatically transmitted alerts are useful, the best results come through a dialog with the public.
• Because people use social media around the clock, a forecaster must be quite active to fully engage the public.
• Social media and other new tools for disseminating information are important, but not everyone uses social media, so broadcasting and other traditional media also play an important role in informing the public about weather events.
The National Earthquake Information Center (NEIC) monitors earthquakes worldwide. Additionally, the center provides backup for the network of regional seismic centers. The NEIC’s customers include relief organizations, government agencies, and research and financial institutes, as well as the media and general public. The main products include real-time monitoring (which is used mostly by researchers), estimates of the magnitude of ground shaking that are based on seismic monitor readings and damage reports, and rapid estimates of fatalities and economic damage on a green-yellow-orange-red scale. The rapid estimates are emailed to approximately 200,000 subscribers and are posted on the NEIC Web site. The NEIC also uses Twitter to disseminate alerts and detect earthquakes.
Paul Earle described an event detector developed by the NEIC that examines tweets10 to detect when an earthquake has occurred. It watches for a sharp increase in the rate at which keywords associated with earthquakes are used. This tool detects between 1 and 4 earthquakes a day worldwide (seismic instruments detect approximately 50 a day). Although not as accurate as seismographs in determining either magnitude or location, it extends coverage in areas with little instrumentation and also provides a backup should instruments fail or be off-line. Where detectors are sparse, earthquake detection based on instrument readings can take up to 5 minutes, whereas detection using Twitter can take less than 2.
10 Status updates posted on Twitter are often referred to as “tweets.”
Earle described several limitations to Twitter-based detection. First, fewer than 5 percent of all tweets used are accurately geocoded. Second, the keyword detection algorithms have approximately a 10 percent false rate. Detection thresholds can be positively adjusted, but there will always be tradeoffs between false positives and false negatives.
To distribute earthquake alerts using Twitter, the NEIC manages a verified11 Twitter account, @USGSted (US. Geological Survey Tweet Earthquake Dispatch). Although the Twitter feed currently reaches fewer people than does NEIC’s email distribution list, the alerts are distributed almost instantly (whereas it can take up to 20 minutes to send emails to the approximately 200,000 email subscribers), and the Twitter feed is not as hard to manage as an email list.
The Twitter alerts contain the region name and geocoded information if available, the data, the time, and the rate at which Twitter messages used to detect the earthquake were sent. The Twitter messages do not include the quake magnitude because this information can change rapidly as measurements are verified, and retweeted messages may continue to circulate long after their content is no longer current. Instead, the messages contain a link to a Web page that provides current information on the earthquake. An example of a Twitter alert and of the Web page it links to is shown in Figure 2.1.
The mission of the Centers for Disease Control and Prevention (CDC) includes the detection of emerging health threats, informing the public about these threats, and informing public officials about potential responses and the associated risks and benefits. Sometimes this involves circumstances where knowledge about a potential threat is incomplete and where there is considerable uncertainty about its implications.
The CDC uses social media both to help detect emerging threats and to disseminate information to the public about how to respond, explained Keri Lubell. Social media can provide clues not only about emerging events but also about how people are responding to those events and to the information that the CDC and others are providing. As a result, there
11 Verified Twitter accounts are marked with a blue verified badge. Verified users have authenticated identities, and verified accounts are generally not available to the general public but instead to those who are considered high-quality sources. A list of frequently asked questions on verified accounts is available from Twitter at https://support.twitter.com/articles/119135#.
is interest in automated approaches that could take greater advantage of all the potential information available from social media.
In terms of the use of social media to disseminate information, Lubell said that the CDC had been an early adopter but that the agency’s use of social media and associated policies are still evolving. The CDC’s Facebook page for emergencies (https://www.facebook.com/cdcemergency) has approximately 14,000 “likes,”12 and the main CDC Facebook page (https://www.facebook.com/CDC) has about 210,000 likes.
The Twitter handle @CDCemergency was established in January 2009 and was first used during the 2009 Salmonella typhimurium outbreak that was associated with peanut butter. That account has approximately 1,350,000 followers, and the general CDC account, @CDCgov, has approximately 97,000 followers. (Up until NASA launched the last shuttle, the CDC had the highest number of followers in a single government Twitter account, Lubell noted.) If the agency can determine good ways to mobilize them, 1.3 million followers could prove a useful resource.
A recent use of social media in conjunction with the CDC’s worldwide polio eradication effort illustrated both the public’s interest in and potential pitfalls of using social media, observed Lubell. As part of this push, the CDC had activated its emergency operations center for an 18- to 24-month period. As a part of the campaign’s communication strategy, the CDC along with its global partners used Twitter to engage the public and volunteers across the world. (Such prearranged public discussions are frequently used by the CDC to engage the public in important health issues.) During the discussion, several items were retweeted by @CDCemergency, with the goal of reaching the wider audience connected to @CDCemergency. However, within a few minutes, followers began complaining that because these messages about polio were not about an emergency, they were not an appropriate use of that feed (Figure 2.2).
In response, the @CDCemergency team provided a brief response and stopped using that channel. Lubell observed that the incident demonstrated that people who followed @CDCemergency were engaged and that it was important to adapt in the face of public response. It also pointed to the need for organizations like the CDC to develop strategies for how best to use the various social media channels at their disposal.
12 A Facebook “like” is similar to “friending” or “following.” By liking a Facebook page, a user may receive updates on the liked organization on his or her news feed. The information frequently changes. See https://www.facebook.com/cdcemergency or https://www.facebook.com/CDC for an up-to-date count of “likes.”
FIGURE 2.2 Examples of the public’s response to the use of @CDCemergency to send messages about the Centers for Disease Control and Prevention’s global polio eradication campaign.
Observations on current uses of social media offered by workshop panelists and participants in the discussion that followed the panel session included the following:
• Emergency response agencies, notably in the areas of fire and public health response, have been successful in building considerable trust with the public. There are potential risks involved in using social media, and government agencies tend to be very risk-averse. However, social media can significantly enhance public response capabilities and can also be used to enhance public trust.
• Social media offer a partnership through which officials can provide the best information they have available at the moment and their followers can help disseminate the information.
• The public response to events, including both individual messages and trends such as increased social media traffic in a location or about a particular topic, can help responders understand an event as it unfolds.
• Local officials, who already have credibility advantages because they are perceived as being “in the same boat” as the public they serve, have had success in fostering public trust in advance of crisis events and in using social media to communicate effectively with the public during such events.
• As communication becomes increasingly mobile and Internet media services are substituted for broadcasting services, social media applications can provide an increasingly important way for emergency managers to reach the public.
• Although a number of existing tools have been successfully adapted, there is also a need for tools for both information dissemination and monitoring that are better matched to the needs of emergency managers.
• The use of social media for disaster response requires significant advance planning. This includes experimenting with various workflows and technologies to assist in the rapid dissemination of information.