Weather and Society Watch
Improving Risk Characterization for NWS Decision Support for Emergency Managers
By Jessica L. Losego*, Kenneth J. Galluppi**, Burrell E. Montz***, and Catherine F. Smith****
In anticipation of extreme weather events, there are two communities that come together to plan and prepare: the National Weather Service (NWS), which informs about hazards, and the emergency management community, which works to minimize risk to life and property due to hazards. These groups try to form a common understanding about the impending dangers associated with an event regarding the anticipated magnitude of the event, where and when it will occur, and who may be impacted. However, there is a gulf between what NWS communicates and what emergency managers (EMs) need with respect to the nature of the risk and vulnerable areas. EMs must characterize and manage the risk to understand what lives and infrastructure will be impacted and to communicate that information to others. NWS often tries to characterize the hazard with the assumption that the relationship to risk is implied. This leaves the EMs to interpret for themselves the nature of the hazard and to assess the associated risks.
The connection between these two communities can be described using the National Research Council's (NRC) risk paradigm concept (Fig 1, 1983). The left side of the diagram connects knowledge about a hazard, characterized in terms of significance, and relates the hazard to what is actually vulnerable or exposed to the hazard. This result is termed risk characterization. The critical link between risk characterization and management of that risk—the job of EMs—is risk communication, which includes issues such as uncertainty, confidence, perception, and modes of communication. Effective risk communication enables EMs to understand all the pertinent physical conditions needing to be managed and to pass that information along to others.
The NWS currently focuses on hazard identification and characterization of weather and does so rather effectively. However, to reduce the impact on life and property, EMs need to understand the hazard's significance sufficiently well to make the appropriate connections as to who is exposed or vulnerable in order to make appropriate decisions to reduce the risk. As other socio-economic factors weigh into risk management, it is critical to have a complete and understandable risk characterization communicated to EMs so that proper and prioritized actions can be taken. The process of understanding hazards, exposure, and risk can only be achieved by a joint effort between the NWS and EMs that leaves little of the process open to interpretation and assumption by personnel not optimally prepared to make those connections.
To assist the NWS in understanding how it might better facilitate EMs' decision making processes within the NRC risk paradigm, the University of North Carolina and East Carolina University have joined with the NWS Office of Science and Technology (OS&T) and Office of Climate, Water, and Weather Services (OCWWS) in a cooperative pilot project. The goal is to understand, from an EM perspective, what critical decisions EMs operate towards, what knowledge about weather hazards and vulnerability underpins these decisions, and how to best package and deliver this information to meet their needs.
To develop this appreciation, we are looking at the entire EM community, which includes the 15 Federal Emergency Management Association (FEMA) Emergency Support Functions (ESF) listed in Table 1, of which only one is the “emergency manager”. Each ESF has its own responsiblities, decisions, and timelines during an event, creating dynamic and ad hoc networks of EMs and comunications among ESFs (Fig 2). The county EM director's main function is to coordinate information and resources that enable the ESFs to do their jobs in support of accomplishing common objectives in managing risk or crisis. This information coordination includes hazard information obtained from the NWS and numerous other sources. This project examines the EM processes in order to understand 1) how information provided by the NWS is translated into actionable knowledge and decisions and 2) how the weather information process can be made more direct and influential when connected more strongly to risk characterization.
We examine both EM and NWS current practices to understand how information flows and is used in EM processes and to establish a baseline to allow us to highlight gaps between knowledge needs of EMs and information provided by the NWS. The next step is to see when and how hazard information can be better linked to risk and, thus, better communicated. Improving the process includes providing information more pertinent to EMs' needs, getting it more directly into the hands of decision makers, improving timeliness, and solving other information and knowledge bottlenecks.
A variety of techniques, including focus groups, interviews, observation of practices, and surveys, are used in an iterative cycle of gathering and hypothesizing to collect data about what is actually occuring. To date we have found that, in general, EMs struggle with finding and understanding relevant weather information and, more importantly, with properly connecting it to their decisions and operations centered on reducing risk and crisis impacts. NWS products and services describe hazards and, at times, connect them to hazard assessments to provide meaning, but stop well short of connecting to vulnerability and risk characterization. This leaves EMs to interpret its context.
To better understand the needs of the EM community and learn about NWS current practices, we examine real situations or “use cases” in an iterative manner, building an understanding one step at a time, starting with small groups. The first use case was on winter weather and the second on tropical weather.
Winter Weather Use Case
We undertook the first case during the winter of 2010-11 and will resume this winter. After conducting mutiple focus groups in North Carolina, with various ESFs to establish critical decisions, collaborators, and timelines, three critical sub-groups whose decisions can have a major societal impact emerged for winter weather: power companies, the state Department of Transportation, and schools. We narrowed our focus to schools, since we believe they are the most underserved by weather information.
We first established a baseline of current schools' processes and their use of weather information and then worked to identify gaps in information. We did this by conducting multiple interviews of nine counties in central and western North Carolina via phone or in person before any winter storms occurred, as well as after storms. We also conducted a statewide survey of school administrators who make the decision to close schools. Findings include:
Critical decisions to close schools are often based on the safety of 16-year old drivers
Forecasts for road conditions, the most critical parameter, are not available for most roads
School staff have little to no training in meteorology yet interpret weather and advise on critical decisions
Location-specific onset time of wintry conditions is crucial for school closing decisions but is hard to find and interpret, and the information that is provided does not work well with the timelines of schools
Most school officials are unfamiliar with useful NWS products (e.g. Hourly Weather Graph)
Relevant information needs to be in one place for easy access on the web
School staff often drive the roads from 3:00-5:00 a.m. to determine conditions but have no way to easily gather and share observations or forecasts.
Our next step was to take what we had heard from the school reps and develop prototypes to translate what we learned into products that might more directly meet their needs. Fig 3 shows two of these prototypes.
After testing these prototypes with our initial group of schools, we visited several more schools to verify our findings and present our prototype ideas. Many of our findings were confirmed, and new ideas emerged for us to test. We will continue the use case this winter by working with schools to allow them to use some prototypes during events.
Tropical Weather Use Case
Our second use case is examing EM processes during tropical weather. We followed steps similar to the winter use case and first established a baseline of current processes, decisions, and timelines for all ESFs. We gathered this information through four focus groups conducted in coastal counties in North and South Carolina. We also interviewed 35 ESF representatives in five North Carolina coastal counties via phone. None of these were county EM directors.
Hurricane Irene made landfall in North Carolina as we were beginning our next phase of information collection, which was talking with county EM directors. This was advantageous because it refreshed in the EMs' minds their processes and use of weather information, making it easier to identify gaps in information and knowledge between NWS and EMs.
Some preliminary findings include:
EMs' top operational concern is the onset time of tropical storm force winds; products to easily get this information do not exist.
Surge information is needed at 72 hours to be operationally useful. EMs can deal with uncertainty, so having a best guess at 72 hours is better than not having any information. This type of information needs to be easy to access,understandable, and zoomable to a local level.
Maximum Envelope of Water (MEOW) and Maximum of MEOW (MOM) surge products are not widely used in North Carolina because they are hard to find and understand.
Briefing packages released by many NWS offices to county EM directors are a critical method of communication. These packages are then passed on to ESFs so that everyone in the county is getting the same information.
EMs who do not receive briefing packages to pass on to others struggle to find the time to gather the relevant information themselves, as it is not easy to find.
EMs go to several different weather sites to get what they need to gain confidence in their understanding. They currently cannot find everything they need in any one place.
Based on these findings, we developed, along with local NWS offices, several prototypes with direct input from EMs. Prototypes included an onset time of tropical storm force winds map, a graphic showing the extent of the potential impact of hazards beyond the cone of uncertainty, and a one-page point and click text summary of “where” and “when” information that is most relevant to EMs.
During this use case, we also focused on learning about the current practices of local NWS offices during events. Because we often hear about EMs struggling to find relevant weather information, we inventoried briefing pages on local NWS offices' sites nationwide. We then conducted a nationwide survey of Warning Coordination Meteorologists (WCMs) asking respondents: 1) if they had a special briefing web page for EMs (and to list it if they had one), 2) if they had a separate briefing package, 3) their opinions on what content is most important for these briefings, and 4) what they believe is the best mode to communicate this information. Of the 98 respondents, more than half reported that they have a special webpage for the EM community, and approximately 93% provide briefing materials beyond this special web page to EMs. PowerPoint and videos are the most used forms for the briefing materials. These results coincide with what we have heard from the EM community.
In addition to this national look at NWS practices, we have also actively engaged our local coastal offices in our focus groups, graphics brainstorming sessions, prototyping, and shared the results of our interviews.
The emergency management community consists of 15 different support functions that manage risks, not hazards, and have varying needs for weather information. NWS currently provides mostly hazard information that, while helpful, often is not meeting the knowledge needs of EMs, forcing them to interpret for themselves weather information that in itself can be hard to find and understand. Many EMs are unaware of existing NWS products (e.g., Hourly Weather Graph), and often times the most crucial information an EM needs is not presented in a direct way, if at all.
Through this project we are working to understand how weather risk information can be communicated from NWS to EMs. Using various social science methods, such as focus groups, interviews, and surveys, we are studying the complex, dynamic, and ad hoc nature of the EM community. An iterative rapid prototyping approach allows us to test various ideas to determine if they meet EM needs. From these methods, we identify the gaps and needs for weather knowledge that NWS can provide, and begin moving towards communication and characterization of risk that will support improved decision making.
* Jessica L. Losego (firstname.lastname@example.org) is a research scientist at the University of North Carolina at Chapel Hill Institute for the Environment.
** Kenneth J. Galluppi (email@example.com) is a senior research scientist at the University of North Carolina at Chapel Hill Institute for the Environment.
*** Burrell E. Montz (firstname.lastname@example.org) is a professor and Chair of the Department of Geography at East Carolina University.
**** Catherine Smith (email@example.com) is a retired professor of English at East Carolina University.
Committee on the Institutional Means for Assessment of Risks to Public Health, National Research Council, 1983. Risk Assessment in the Federal Government: Managing the Process. Available at http://www.nap.edu/catalog.php?record_id=366.
Federal Emergency Management Agency, 2008. National Response Framework. Available at http://www.fema.gov/emergency/nrf/#.
Table 1: Emergency Support Functions (ESFs) established by FEMA
Figure 2: Networking diagram showing communication two days before a winter storm. Red circles indicate multiple NWS injection points. The county EM director is in the center and is connected to many ESFs.
Figure 3: (Left) EMs prefer relevant information to be placed in one easy-to-find and use location. Currently they are not aware that many products exist and can have difficulty finding products they are familiar with. (Right) At 9:00 p.m. the night before a storm, schools officials would like to receive a forecast of road conditions for 4:00 a.m. the next day. Combining this information with what officials see when they “ride the roads” provides critical information that schools need to make their decisions.