1. Introduction and background
Ice storms are a serious hazard that cause millions of dollars in damage each year, and catastrophic events are becoming more common (Changnon 2003). One way to reduce the costs associated with ice storms is to improve warning procedures, which aids community leaders, first responders, and individual citizens in preparing. Through an analysis of data collected from a questionnaire and from observation of recent ice storm warnings, this paper will offer insights into National Weather Service (NWS) ice storm warning procedures and suggestions for improvement.
An ice storm produces freezing liquid precipitation (ice) in the form of freezing rain that creates a glaze of ice on objects such as trees, roads, and utility lines (Glickman 2000). Generally, a storm with sleet is less catastrophic than one with freezing rain because sleet provides some traction for motorists (Kocin and Uccellini 2004, p. 151) and does not coat terrestrial objects in the same manner as freezing rain. Nonetheless, both sleet and freezing rain are hazardous. Furthermore, an ice storm’s disruption may be extended if cold air follows the storm, which limits melting and forces people without electricity to use generators (which can cause carbon monoxide poisoning) or to travel to shelters.
a. Climatology of freezing rain and ice storm catastrophes
While there is some disagreement about the exact frequency of freezing rain [cf. maps in Bennett (1959) with those in Changnon (2003)], there is general agreement that most of New England, New York, and Pennsylvania annually experience ≥4 days of freezing precipitation, on average. Areas that typically experience 3 days of freezing rain annually include the upper Midwest, Appalachians, and Columbia and Snake River valleys (see also Cortinas et al. 2004).
In an analysis of property insurance loss data for the period 1949–2000, Changnon (2003) found that the overwhelming majority of “catastrophic” storms affected states from Texas to Maine, as shown in Fig. 1. States with 19 or more catastrophic storms during that time include North Carolina, Virginia, Maryland, Pennsylvania, New York, New Jersey, and all of New England.
Two regions show a discontinuity between the risk of freezing rain and the occurrence of ice storm catastrophe: the northern Midwest and the Deep South. While most of Minnesota, for example, averages 4 or 5 days of freezing rain per year (Bennett 1959; Changnon 2003; Cortinas et al. 2004), only two ice storm catastrophes occurred there between 1949 and 2000, which is fewer than in Florida (see Fig. 1 again). Louisiana, Mississippi, Alabama, and Georgia, which typically experience 1 day of freezing rain per year, registered an average of 15 disasters for the same 52-yr period. Rauber et al. (2001) suggested two reasons for this: first, storms in the South have a higher moisture content than those in the upper Midwest, and, second, their durations are often longer. Thus, ice storms in the South, while less common than those elsewhere, are more likely to cause ice accumulations sufficient for severe damage. Sociological and other environmental factors may also contribute, such as variations in the amount of equipment available for mitigation, population density, and forest cover.
b. Role of the National Weather Service
The National Weather Service is the primary source for disaster warning information in the United States. Its employees issue tens of thousands of warnings in any given year; a small fraction of these are for ice storms. According to NWS regulations, an ice storm warning should be issued when more than 0.64 cm (0.25 in.) of freezing rain is expected (NWS 2005a). The NWS is divided into six regions (Fig. 2), and several of these have issued supplemental directives regarding ice storms. For example, in the northern portion of the Eastern Region, warnings should not be issued unless at least 1.27 cm (0.50 in.) of freezing rain is forecast (NWS 2007). Some regions and individual forecast offices also incorporate nonmeteorological criteria, such as time of day, in deciding whether to issue warnings (see NWS 2007 for some examples).
To make accurate forecasts and decide whether or not to issue warnings, forecasters must understand the hazards that affect their area and the risks of each (Alexander 1993, p. 405; Lopez et al. 1993; Smith 2004, p. 69). Their understanding is influenced by prior experience with risks (Halpern-Felsher et al. 2001; Peacock et al. 2005), but expert and lay perceptions of risk often vary (Slovic 1987; Peacock et al. 2005; Siegrist and Gutscher 2006). More specifically, if an NWS employee has little experience with a hazard, warning for the hazard will be more challenging. Thus, individual employees may overwarn or underwarn, affecting the ability of the larger society to prepare and respond.
While much research has shown the importance of accurately understanding risk, the process used to make the decision to warn is less well understood (Mileti and Sorensen 1990, 2–8). But, assuming that a local forecaster makes the decision to warn, it is important that the warning not only be accurate, clear, and confident, but also that it offer guidance about what to do (Mileti and Sorensen 1990, 2–9). Alexander (1993, p. 401) has argued that a warning should include information about how long a potential disaster will last and the expected impacts. It follows, then, that a study of NWS warnings should examine both the decision-making process behind the warnings and the warning statements themselves.
Timely dissemination of warning information has reduced the impact of hazards such as severe convective storms and snow (Corfidi 1999; Call 2005). Parker and Neal (1990) note that, in general, accurate warnings are almost always issued, but warning dissemination is the more problematic aspect. In other words, an accurate warning that is not heard or acted upon is of limited value. While NWS warnings are available through National Oceanic and Atmospheric Administration (NOAA) Weather Radio, previous research has shown dismally low awareness of this system (Redmond 1995; Hammer and Schmidlin 2002), with many more people getting warnings through mass media and directly from friends and family (Hammer and Schmidlin 2002). Thus, some offices have become active in communicating with constituents (see Morris et al. 2002; Troutman et al. 2001), and the NWS itself has pushed for improved communication with emergency management officials (NWS 1998). Nonetheless, prior research has not established how widespread such types of communications are. It is also unclear what effect the Internet and other modern technologies have had on warning dissemination, although they probably have helped.
c. Research questions
In an attempt to address some of the above issues, the author examined how employees of the NWS perceive the ice storm hazard and how they warn both the general public and specific audiences, such as emergency managers. To focus the research, a series of specific questions were devised as listed here:
How seriously do NWS warning coordination meteorologists (WCMs) perceive the hazardousness of ice storms on both absolute and relative scales?
What products do they, or other members of their respective offices, issue to warn people in advance of a storm?
How much contact do they have with others in need of information regarding ice storms, such as emergency managers?
Finally, what educational products and programs do they provide to prepare people well in advance of a storm situation?
2. Methodology
To answer the research questions, documents about NWS warning guidelines and procedures were obtained from the NOAA Central Library and various NWS Web sites, and a survey was mailed to 54 NWS WCMs in April 2006. WCMs oversee warning programs at individual forecast offices, engage in community outreach, and coordinate public education campaigns and training sessions with members of the hazards community. WCMs only issue a small fraction of the warnings from an individual office; this is typically done by the on-duty forecaster, who may or may not be the WCM at any given time. Nonetheless, because of their leadership positions, detailed knowledge of warning procedures, and above-average length of service, WCMs were surveyed. Additionally, it was believed that WCMs, as a group, would be more likely to respond given their regular engagement with the hazards community and others outside of their office.
The WCMs surveyed forecast for the 29 states most likely to experience a catastrophic ice storm; Fig. 3 shows the exact area served by these WCMs. All states within this area experienced greater than 10 catastrophic ice storms between 1949 and 2000 (refer to Fig. 1). This survey area includes portions of Texas and all parts of Oklahoma, Missouri, Arkansas, Louisiana, and all states east of the Mississippi River except Michigan, Wisconsin, and Florida. Because the focus of this study was on an assessment of ice storm impacts and not on the frequency of ice, the study area was delimited by ice storm catastrophes rather than the climatology of freezing rain; this eliminated states where freezing rain is common but catastrophes are relatively rare (e.g., Minnesota). Nonetheless, the author believes that many of the conclusions are equally applicable to these states.
The survey group included all 23 WCMs in the NWS’s Eastern Region, 20 of 32 WCMs in the Southern Region, and 11 of 38 in the Central Region. Southern and Central Region WCMs not surveyed either had offices that primarily served areas outside the study region or were located in portions of states that rarely experience an ice storm (such as southern Florida). A key objective of such a large sample was the ability to explore local and regional differences in the NWS response to ice storms.
To maximize the utility of the survey tool while minimizing the burden on respondents, survey design resources such as the work of Dillman (2000) and books by Linda Bourque and Eve Fielder (1995) and Robert Peterson (2000) were used to design the survey. For example, most survey questions were closed-ended since these are less ambiguous and easier to answer (Peterson 2000, p. 54); for examples of specific questions, see the appendix. Other admonitions common to the survey design literature, such as “be brief,” “be specific,” and “be objective” were also followed (Peterson 2000, p. 74). Many of Dillman’s (2000) specific suggestions for survey construction, such as minimizing the number of “jumps” and having respondents check boxes rather than circling responses were also implemented. Finally, the surveys were pretested by several WCMs not in the pool of participants and revised accordingly. The Institutional Review Board of Syracuse University (the author’s home institution at the time) also approved the survey. This research involved human “subjects,” and all universities require IRB approval to ensure that federal regulations protecting such subjects are followed.
At the suggestion of Dillman (2000, p. 155), each participant was contacted four times. This contact occurred in March and April 2006. The initial contact was a prenotice e-mail, and several potential participants responded either positively or with suggestions of alternate, but qualified, people to survey. A week after the prenotice e-mail, all potential participants were mailed a paper survey. After another week had elapsed, “thank you” reminder postcards were mailed; again, several participants responded to let the survey administrator know that they needed additional time or another survey copy. Finally, after 1 month’s time, a final e-mail reminder (with an attached electronic copy) was transmitted to any remaining nonrespondents. Only two WCMs responded to this last request for assistance.
The author is unaware of any systemic problems with either the survey or its implementation. Most respondents replied within 3 weeks, and the last response arrived 6 weeks after the first contact. Two participants chose to complete the survey electronically instead of with pencil and paper, but this did not seem to affect their responses. Any respondent who indicated an interest in the results was sent a summary of the responses in late 2006.
Survey data were summarized and analyzed using Microsoft Excel and MINITAB. The results were also displayed graphically on maps. However, to protect respondent confidentiality, WCMs responses were combined within 12 subregions as shown in Fig. 4. Each subregion consists of at least three offices, and in all subregions at least half of the WCMs responded. These subregions are clearly too small for detailed analysis; rather, they were created to help visualize the data spatially.
3. Results
Thirty-nine of 54 potential participants (72%) responded, which is much better than typical mail survey response rates, which are on the order of 30%–50% (Goyder 1985; Baruch 1999). As shown in Table 1, regional response rates were 65% for the East, 85% for the South, and 64% for the Central. Basic demographic information about the respondents is also displayed in Table 1. The average participant has been a WCM at his or her current office for 7.7 yr, but some respondents had just begun serving as a WCM. All survey respondents have substantial work experience with the NWS; the average length of service is more than 21 yr, and the minimum length is 11 yr. Interregional variations in length of service are small and had no apparent effect on the results.
a. WCM opinions about hazards
Survey respondents were asked to rate the seriousness of nine hazards from “not serious” to “very serious”; “not applicable” was also an option. (To view the actual text of this and other survey questions, please refer to the appendix.) The seriousness terms corresponded to numerical values, where 0 represented not applicable and 5 indicated very serious. If a WCM did not rate a hazard, a value of 0 (not applicable) was assumed, although this was uncommon.
Table 2 shows the mean rating value for the nine hazards. Flooding was universally considered a high threat. Not only was flooding considered the most serious hazard in the Eastern Region (and second most serious in the other regions), but the ratings for flooding had the smallest standard deviation and range for all regions. WCMs in the Southern and Central Regions rated tornadoes as more serious than flooding, but WCMs in the Eastern Region viewed these as less serious than flooding.
The winter hazards of ice storms and snowstorms were considered serious in all three regions, but less so than tornadoes and flooding. Not surprisingly, Southern Region WCMs rated ice storms and snowstorms as less of a threat relative to other hazards due to their less frequent occurrence.
The assessments of the seriousness of other hazards, such as hurricane winds, storm surge, chemical spill, terrorism, and wildfire, showed more inter- and intraregional variation. Hurricane-related hazards had such large deviations that the means are irrelevant; WCMs near coastal regions tended to rate hurricane hazards with values of 5, while their inland counterparts chose values at the opposite end of the scale such as 0 or 1.
Figure 5a shows the mean absolute ratings for ice storms by subregion. Because freezing rain is less common in the south, WCMs in the southern United States tended to rate ice storms as less serious than their northern counterparts. Nonetheless, the highest seriousness ratings for ice occurred in the Oklahoma–Arkansas subregion. This is probably because an exceptional pair of ice storms adversely affected the region in December 2000.
Figure 5b shows the rank of ice storms relative to other hazards for each subregion. Additionally, the highest ranking hazard in each subregion is also displayed. Like Fig. 5a, WCMs in the north and west tend to consider ice storms among the most serious hazards they face, while those in the south focus on more frequent threats, such as tornadoes and hurricanes. A lack of recent ice catastrophes may explain why WCMs in the mid-Atlantic ranked ice storms low in comparison to other hazards. As a group, WCMs in the Ohio Valley subregion considered ice storms to be the most significant hazard they faced (equal to tornadoes) but there is no clear explanation for why; it may simply be a sampling artifact.
Meteorologists were also asked about the recurrence interval of ice storms. Most WCMs in the Eastern Region reported a recurrence interval of every 0.5–2 years. The large range in this number reflects the latitudinal extent of the Eastern Region of the NWS, which stretches from Maine to South Carolina. WCMs in the Southern Region generally reported an ice storm recurrence interval of anywhere from 1 to 5 yr. In the Central Region, the respondents reported experiencing, on average, one or two ice storms per year. These values are similar to those reported by Bennett (1959) and Changnon (2003), indicating that study participants are well aware of the frequency of ice storm occurrence in their individual forecast areas.
b. Ice storm products issued by NWS offices
Section 3 of the survey requested information about ice warning products issued by the NWS and the criteria used in the decision-making process. A scenario involving a hypothetical ice storm was presented. Participants then responded to a series of questions regarding the products they would issue, when they would issue those products, and what criteria they would use in deciding to issue such products.
If an ice storm was “expected” in 72 h, most WCMs surveyed would issue an advisory or special weather statement (SPS) immediately, except for several Southern WCMs who preferred to wait for another run or two of the computer models. In general, a 30% confidence level was expected before an SPS would be issued, although 10% of respondents would also consider nonmeteorological criteria, such as the day of the week or potential threat to utility operations in making their decision. The majority of WCMs would also discuss the potential for ice storm in a hazardous weather outlook (HWO); see Fig. 6 for an example. A hazardous weather outlook, issued daily, provides specific details about the potential for hazardous weather in the next day or two and then describes, in general terms, the potential for severe weather during the remainder of the next week. Several WCMs, all in the South, would also consult with other offices before issuing an SPS or HWO that discussed the potential for an ice storm. One wrote that it is important for adjacent forecast offices to issue a “unified statement of issuance.” In this case, whether an HWO or SPS would be issued largely depended on the proactive behavior of adjacent offices.1
As the time before the hypothetical storm decreased, WCMs would issue a winter storm watch, similar to that shown in Fig. 7. A watch product, unlike an HWO, includes a headline and more detailed, but jargon-free, weather discussion. Central Region participants were more likely to issue an early watch than their counterparts elsewhere (see Fig. 8). This probably reflects less uncertainty in predicting ice storms for that region, while in the Eastern Region, minor wobbles in the track of a nor’easter can make forecasting the location of icing difficult (Kocin and Uccellini 2004, p. 177). Another idea, suggested by a reviewer of this article, is that this difference could reflect different internal policies at the NWS regional level. Southern WCMs would wait slightly longer to issue a watch. This likely has to do with the rarity of freezing rain there.
Objectively, most WCMs would not issue a watch until they were at least 50% confident that 0.64 cm (0.25 in.) or more of ice would occur. In comparison to the criteria used to decide whether or not to issue an HWO or SPS, a greater number of WCMs in the South would also incorporate nonmeteorological factors, such as expected impacts of the storm, in their decision-making process. By contrast, WCMs in other regions were not more (or less) inclined to consider nonmeteorological factors.
As the storm drew closer, WCMs would issue warnings or advisories depending on how much ice was expected; see Fig. 9 for a warning. Like watches, warnings and advisories include headlines and a nontechnical description of the weather. However, they also have specific details, such as expected accumulation amounts and call-to-action statements, such as a request to avoid unnecessary travel. Additionally, warnings and advisories cover a smaller geographical area than a watch.
Generally, WCMs would issue a warning or advisory about 24 h in advance, as shown in Fig. 10. Just under 40% of WCMs (15/39) would issue a warning 12–24 h prior to the storm, while about 46% (18/39) would issue a warning 24–36 h in advance. The entire group of WCMs was slightly more inclined to issue a warning earlier than an advisory, but this did not hold true for individual regions, which all showed slightly different distributions (cf. Figs. 10a and 10b).
The criteria used to choose between a warning or an advisory were very clear. In almost all cases, anticipation of at least 0.64 cm (0.25 in.) of ice was needed for a warning, while a lesser amount corresponded to an advisory. A few WCMs would use a higher threshold of 1.27 cm (0.5 in.) in deciding whether to issue an advisory or a warning, while two WCMs (one each in the East and South) would use a lower threshold of either 0.13 cm (0.05 in.) or 0.32 cm (0.125 in.), respectively. These findings are in line with NWS regulations for ice storm warnings, which suggest the same threshold, subject to “local definition” (NWS 2005a); the Eastern Region, as of the time of this survey, had adopted a more stringent definition of 1.27 cm (0.5 in.) for New England, New York, and northeastern Pennsylvania (NWS 2007). Although these criteria seem exact, measuring and verifying ice accumulations is quite challenging (Ryerson and Ramsay 2007). This survey did not specifically address whether the current thresholds are appropriate or how the use of them may affect the warning process. Future researchers may wish to investigate these questions.
Besides the quantitative criteria, several WCMs, mainly in the East, noted that impact was a factor that they would also consider in this part of the decision-making process—in contrast to the Southern WCMs, who were more likely to consider impact in earlier stages of the warning process. Nonetheless, this difference is relatively minor, as WCMs in all three regions noted that forecaster discretion was also a criterion used to make the decision between warning and advisory. Both the Eastern Region (NWS 2007) and Central Region (NWS 2005b) supplements to the NWS regulations for winter weather products (NWS 2005a) encourage this behavior.
In conclusion, assuming foreknowledge of an ice storm, WCMs would generally discuss the product first in hazardous weather outlooks, and then in special weather statements and winter storm watches, in line with the “tiered approach” outlined in NWS directives. Shortly before the storm began (and while it was occurring) they would issue winter storm warnings or advisories depending on how much ice they expected. Although the objective criteria for issuing watches, warnings, and advisories are fairly clear, some WCMs, in all three regions, commented that they considered nonmeteorological factors in deciding what products to issue.
c. Warning procedures
Section 4 of the survey asked how WCMs communicate information regarding forecasted ice storms to specific groups likely to take action, such as emergency managers. Questions fell within two categories: those relating to “call to action” statements and those related to “calling outsiders” (specific groups of customers).
Out of 39 survey respondents, just over half (20/39) reported the existence of a prepackaged call-to-action statement. The low percentage was surprising because prepackaged statements improve warning communication, insofar as forecasters can compose and revise the wording on a quiet day instead of during a busier severe weather situation. Prepackaged statements can also be produced for a variety of different ice storm scenarios. One WCM in the Central Region provided five sample paragraphs as examples of prepackaged statements from his office. Individual forecasters could then choose the text that best fit the forecasted scenario, and assuming that the forecast did not change, the message would not change from shift to shift.
There is no geographic clustering of offices with prepackaged statements. While all offices in the lower Mississippi subregion that responded to this survey had statements, only 25% of those in the adjacent Southeast did. No offices in the Carolina or mid-Atlantic subregions had prepackaged statements, but most in the Virginias did. Since the mapped pattern of such statements (not shown) cannot be attributed to climatology, recent ice storm disasters, or NWS regions, perhaps the existence of such statements is simply the result of individual forecaster initiative.2
Thirty-five WCMs work at offices that directly contact specific groups to warn of ice storms, and an additional participant noted that he or she is called instead. This finding was encouraging because good communication between government agencies is important, and it helps the NWS disseminate warnings. As shown in Table 3, nearly all WCMs contact local and state emergency managers in advance of an ice storm, and the vast majority speak with local media. Few WCMs directly contact political leaders, but a substantial number were unsure. Many offices conduct conference calls with emergency managers and other first responders, and the WCMs are not sure if politicians are listening in. Another reason for the uncertainty was because the amount of contact between emergency managers and politicians varies from place to place—something beyond the survey respondents’ knowledge.
Contacts between WCMs and utility companies varied geographically. About half of the Eastern WCMs contacted utility companies and more than half of the Southern WCMs did. No Central WCMs contacted utility companies, but only three Central WCMs answered this question directly (the rest left it blank), so it is unclear if this finding is a sampling artifact, a result of a different regional policy, or due to some other factor. In fact, more than a dozen WCMs did not respond to this question at all. Several WCMs added comments of interest. An Eastern WCM reported that there was no need to contact utilities because that is the job of private vendors and not the role of the NWS. Another WCM reported a practice of contacting small utilities but not large ones, perhaps because smaller companies would be less likely to have meteorologists on staff or contracts with private vendors.
Finally, nearly all WCMs reported that the decision to call outsiders rests with the lead forecaster on duty, although that person often seeks the opinion of other meteorologists at the local office [such as the WCM or the meteorologist in charge (MIC)]. Respondents in the Eastern and Central Regions often have some guidelines (or rules of thumb) that suggest a consideration of the forecasted amount of ice or the expected impact. In contrast, most offices in the Southern Region have no formal criteria that determine whether or not to call outsiders, and two Southern Region NWS Forecast Offices (NWSFOs) have a very low threshold (e.g., any threat of winter precipitation), probably due to the rarity of winter precipitation in their forecast areas. In sum, it seems that individual forecasters have substantial discretion when deciding whether or not to call outsiders.
d. Educational and outreach programs
This section of the survey was concerned with the NWS’s educational offerings for the general public and specific groups. The purpose was to see what types of educational products various NWS offices provide well in advance of ice storms.
Many state governments sponsor an annual winter weather awareness week (some may sponsor just a “day”), and all NWSFOs involved in this study participate—if their home state sponsors one. NWS employees participate by issuing public information statements discussing various winter hazards and ways for constituents to stay safe. Respondents devote either 21%–40% or 41%–60% of the discussion to ice storms. Assuming a 5-day week, this equates to 1–3 days.
Because state governments proclaim winter awareness events, the NWS lists participating governments on their Web site (Magnus 2006). Figure 11 shows which states held winter weather awareness events in 2004 and 2006; data for 2005 were not available. While it is not surprising that Florida and Arizona do not have winter awareness events, several northeastern states, such as Pennsylvania and Massachusetts, also lack these events. Winter awareness weeks are a good opportunity for members of the NWS, emergency managers, planners, and the general public to review procedures and prepare for winter events (Peterson and Perry 1999). Governors or emergency management directors in Connecticut, Massachusetts, Pennsylvania, Mississippi, and Georgia should consider instituting winter awareness events, especially since these states have all experienced catastrophic ice storms within the past half-century. Another interesting pattern is that states with winter awareness days (as opposed to weeks) are clustered in the Great Plains and along the west bank of the Mississippi River. This is likely because this area includes “tornado alley,” and governments and meteorologists prefer to focus on weather hazards that are more probable.
Additional questions in this section sought information about educational newsletters and workshops. Thirty-one WCMs have workshops that include a discussion of ice storms, while six WCMs (including four in the Southern Region) do not. Of the 33 WCMs who have newsletters, nearly half (17) mail out newsletters with information about ice storms. Interestingly, almost two-thirds of WCMs in the Eastern Region did not mail out newsletters, while in the other regions one-third of WCMs did not, although a few WCMs noted that they post items to their forecast office’s Web site in addition to or instead of mailing. (Since the survey did not specifically ask about online newsletters, it is possible that some WCMs who post online exclusively may have indicated that they have no newsletter.) The majority of WCMs (62.5%) send newsletters to 100–500 people, although one Eastern Region WCM (in an area that experiences lake-effect snow) reported sending newsletters to more than 1000 people! When asked about what specific groups received their newsletters (e.g., emergency managers, media), most WCMs did not respond. Since few know what groups are represented on their mailing lists or visiting their Web sites, a survey or other assessment method may help WCMs better understand their audience.
The last group of survey questions asked for greater detail about workshops that include a winter weather discussion. A typical workshop might feature presentations by NWS employees of the latest product innovations and safety information, as well as dialogue (often informal) between NWS employees and constituents, but there are no official specifications for workshops. Twenty-three WCMs host such workshops (including all seven Central Region respondents), nine do not (with about equal numbers in both the East and South), and the remaining seven WCMs skipped the question (presumably because they had no workshops). If an office has a workshop, the media are almost always invited; in fact, three offices have workshops exclusively for media. Other groups commonly invited to attend workshops include emergency managers, and, less commonly, SKYWARN observers. A few WCMs invite other government officials, utility companies, or any member of the general public, but these WCMs compose less than one-third of the total group. Nearly two-thirds of WCMs host one to four workshops annually, with equal portions of the remainder hosting fewer or more workshops. The median attendance at workshops varied greatly within each region, but WCMs in the Eastern Region were more likely to host a workshop for a large group (see Table 4).3
e. Miscellaneous comments
In the final section of the survey, participants could offer additional comments, feedback, or elaboration on previous answers. There was an unexpected geographic variation in response: almost no Eastern WCMs wrote anything, while about one-third of other WCMs provided feedback. Several WCMs in the South noted that the relative rarity of storms in their forecast areas causes them to approach ice storms differently from other weather events. For example, one Southern WCM commented that because any winter weather is rare, “we have taken a stance that public impact is a higher priority over accumulation amounts.” Another noted that “communication is critical because local government is not set up to deal with impacts.” A third WCM commented that ice storms are challenging to forecast since any mention of snow or ice shuts the area down, regardless of whether it occurs. These comments indicate that WCMs in the Southern Region are well aware of the limitations of local government and the general public, and they are independently working to reduce impacts arising from these limitations.
4. Analysis of actual ice storm warnings
In early 2007, soon after the survey results were analyzed, a catastrophic ice storm—perhaps the most severe since the aforementioned December 2000 storms—affected a broad area of the United States from Oklahoma to the eastern Great Lakes. One area hit especially hard was southern Missouri. This storm provided an opportunity to view actual NWS warning procedures and text. Since NWS employees followed the warning procedures as discussed in the previous section, this discussion will focus on the warning text itself.
Figure 12 shows portions of ice storm warnings issued in early 2007 by National Weather Service Forecast Offices in Tulsa, Oklahoma; Springfield, Missouri; and St. Louis, Missouri. These products are disseminated on NOAA All-Hazards Radio, the World Wide Web, and by media (but often edited). Only the portions that discuss anticipated impacts are included here. As shown in Fig. 12, NWS meteorologists appropriately focus on power outages and transportation disruptions—the most likely problems associated with an ice storm—but need to better direct their products toward similar audiences and determine the appropriate length and tone.
There are three major differences between the descriptions of expected impacts. Perhaps the most obvious difference is in the various lengths. The Tulsa forecaster summarizes everything in two short paragraphs, while the St. Louis forecaster uses more than 250 words to discuss expected impacts. Another major difference is in the intended audience. The Tulsa warning is primarily directed at government officials in Oklahoma and Arkansas, as it calls on them to “prepare to house many people” and notes that “supplemental heat sources will be required.” In contrast, the St. Louis warning is directed at individual members of the general public; it includes six imperative sentences and two other sentences with the subject “you.” The Springfield warning does not seem to be directed at any particular group. Finally, the intensity of the language varies as well. While the Springfield warning mentions use of “extreme caution” with “improvised heating devices” and suggests avoiding travel because it is a “dangerous situation,” the warning from St. Louis says “You can die!” in regard to burning charcoal in enclosed areas. In addition, the St. Louis warning includes numerous call-to-action statements, such as methods to travel safely, precautionary plans for power outages, and a suggestion to stock up on firewood, water, and candy bars. Perhaps the most unusual admonition comes from the St. Louis office in urging people to seek shelter if power is lost: “Do not be too proud to accept public assistance such as heated shelter areas.” It is unclear whether these call-to-action statements originated from the specific office, the specific forecaster, or more general software used by various NWS offices; nonetheless, these statements are much more numerous and forceful than those used by the other offices. To summarize, there are major differences in the three warnings with regard to length, intended audience, and tone.
A positive finding is that all the statements have information about power outages, which have the longest-lasting and most widespread impacts in ice storms (Call 2007, p. 45). The statements from the Missouri offices also warn against unnecessary travel, a sensible suggestion when roads are ice covered. However, the forecaster from Springfield devotes more text to discussing travel disruptions than to power outages, which is not appropriate given that power outages are a much more substantial disruption, often lasting for weeks as opposed to days (Call 2007, p. 46). On the other hand, the St. Louis and Tulsa statements include numerous call-to-action statements, which are greatly beneficial to end users. Thus, while all three statements focus on power outages, which is appropriate, the Springfield statement should have less information regarding travel disruption and should include at least a few call-to-action statements.
More generally, there are several steps that the NWS can take to improve the impact statements for ice storms. First, it should assess what the audience is for these statements and then tailor them accordingly. In other words, should the NWS focus on warning government officials (as Tulsa does), individuals (as St. Louis does), speak generally (as Springfield does), or attempt to focus on another group? Second, the NWS should determine how much warning information to include. The St. Louis impact statements were part of a winter storm warning with more than 500 words, which is probably an excessive length. Finally, the intensity of the warnings varied, with the St. Louis warning being the most dramatic. While this storm was, in fact, severe enough to warrant strong language, the NWS may wish to provide guidance for offices to avoid overplaying or underplaying the potential for disaster (see discussion in Troutman et al. 2001). Additionally, some coordination between adjacent offices would help with adjusting the intensity of the warning language to an appropriate level for the audience and forecasted situation.
5. Discussion and conclusions
Thirty-nine National Weather Service warning coordination meteorologists completed a survey asking for their opinions of the ice storm hazard and about the techniques they use to warn the government and the general public. Participants completed the survey either with pencil and paper or electronically in April 2006, and there were no known major problems with the survey’s design or implementation.
Because severe ice storms are less common than other meteorological hazards (especially in the southern United States), it was interesting to observe that WCMs in all regions accorded them a relatively high level of seriousness. On average, the participants considered ice storms the third most serious hazard they face. However, there were significant regional differences; WCMs in the Southern Region were more concerned about tornadoes and wildfires than WCMs elsewhere, which agrees with climatology. Even so, Southern WCMs, as a group, seem realistic in assessing the limitations of local government in dealing with both ice storms and winter weather in general. Evidence of this comes from the criteria they use in deciding what watch and warning products to issue and their additional comments provided at the end of the survey.
WCMs are also aware of the criteria they consider in issuing HWOs, SPSs, watches, and warnings relating to ice storms due to the fact that NWS Headquarters and NWS regions have established clear criteria separating the products (NWS 2005a,b, 2007). Even so, local forecasters still have a relatively high level of discretion in deciding what products to issue and whom to call, rather than being required to get approval from supervisors. This probably comes out of the high levels of autonomy each individual forecast office has. Although it was promising to see that a number of WCMs already consider social factors such as potential impact, timing, and the preparation of their region in deciding what products to issue and when, more WCMs should incorporate such factors into their decision-making process.
An encouraging finding was the high level of communication between NWS meteorologists and people outside their offices. Nearly 95% of participants communicate with outsiders in preparation for an ice storm, with many contacting local emergency managers. WCMs as a group also respect the division of labor between government and private industry in regard to contacting utility companies. Many do not contact utility companies and several noted an unwillingness to contact large companies. Nonetheless, it may be prudent for them to track contact with utility companies to avoid complaints from private forecasters.
Many WCMs are proactive in providing educational information for interested members of the general public through newsletters (either mailed or electronic) and through workshops. Those that are not providing such information should consider this part of their mission in helping to protect life and property. Those already providing information should consider expanding their mailing list or hosting more workshops; perhaps they could learn from their colleagues who already host multiple workshops or send information to more than 1000 customers. Workshops especially are a great way to build ties between NWS employees and important community figures involved in warning and protecting the public, such as broadcasters and emergency managers. These relationships may prove helpful during later severe weather events (see Reaugh and Mowry 2005 for an example).
The size and configuration of the study area undoubtedly affected the results insofar as WCMs in the northern part of the Central Region or the Western Region might have different ice storm opinions. But, since catastrophic ice storms are less of a threat in those areas, it seems unlikely that their perceptions of the seriousness or outreach programs would be significantly greater. Given that the WCMs in the Southern Region seem as adequately prepared as their counterparts elsewhere (despite less frequent storms), it does not seem likely that WCMs outside the study area would be much less prepared. Additionally, the regional differences in this study were generally small. The main exception was in the ratings given to hazards, which greatly varied from region to region, but generally agreed with the climatology of hazards.
Another limitation of this study was its focus on the communication of ice storm information. There was no assessment of whether products are issued with sufficient lead time to allow for preparation or whether people receiving the warnings actually take action. Future research may wish to examine these issues. These are important parts of the warning process, and a failure in timeliness or action diminishes the efficacy of the issued warnings.
To conclude, 39 WCMs responded to a survey in April 2006 regarding how they communicate information about ice storms. Most participants are well aware of the seriousness of the ice storm hazard and are taking appropriate actions in warning and preparing the public, but an assessment of the actual warning products is needed to determine the audience and appropriate length and tone of the warnings. Most WCMs are proactive in communicating the threat of ice storms with government officials, media, and others that can take actions to protect the general public, but more WCMs should consider nonmeteorological factors when issuing watches and warnings. Some WCMs should also create or expand their educational offerings regarding ice storms. These criticisms aside, it seems that when aiming to protect life and property in relation to ice storms, WCMs are accurately assessing the ice storm risk and issuing appropriate forecast products.
Acknowledgments
The author thanks the participants for taking the time to offer thoughtful and insightful responses. Additional thanks are due to the Roscoe Martin Fund of Syracuse University for financial support, and Mark Monmonier, Susan Millar, Adam Burnett, and several reviewers for helpful feedback on this paper.
REFERENCES
Alexander, D., 1993: Natural Disasters. Chapman and Hall, 632 pp.
Baruch, Y., 1999: Response rate in academic studies—A comparative analysis. Human Relations, 54 , 421–438.
Bennett, I., 1959: Glaze: Its meteorology and climatology, geographical distribution, and economic effects. Quartermaster Research and Engineering Center Tech. Rep. EP-105, 234 pp.
Bourque, L. B., and Fielder E. P. , 1995: How to Conduct Self-Administered and Mail Surveys. Sage, 264 pp.
Call, D. A., 2005: Rethinking snowstorms as snow events: A regional case study from upstate New York. Bull. Amer. Meteor. Soc., 86 , 1783–1793.
Call, D. A., 2007: An integrated hazard analysis of ice storm impacts, warnings, and emergency plans. Ph.D. dissertation, Dept. of Geography, Syracuse University, 200 pp.
Changnon, S. A., 2003: Characteristics of ice storms in the United States. J. Appl. Meteor., 42 , 630–639.
Corfidi, S. F., 1999: The birth and early years of the Storm Prediction Center. Wea. Forecasting, 14 , 507–525.
Cortinas J. V. Jr., , Bernstein B. C. , Robbins C. C. , and Strapp J. W. , 2004: An analysis of freezing rain, freezing drizzle, and ice pellets across the United States and Canada: 1976–90. Wea. Forecasting, 19 , 377–390.
Dillman, D. A., 2000: Mail and Internet Surveys: The Tailored Design Method. 2nd ed. John Wiley and Sons, 480 pp.
Glickman, T. S., 2000: Glossary of Meteorology. 2nd ed. Amer. Meteor. Soc., 855 pp.
Goyder, J., 1985: Face-to-face interviews and mailed questionnaires: The net difference in response rate. Public Opinon Quart., 49 , 234–252.
Halpern-Felsher, B. L., Millstein S. G. , Ellen J. M. , Adler N. E. , Tschann J. M. , and Biehl M. , 2001: The role of behavioral experience in judging risks. Health Psychol., 20 , 120–126.
Hammer, B., and Schmidlin T. W. , 2002: Response to warnings during the 3 May 1999 Oklahoma City tornado: Reasons and relative injury rates. Wea. Forecasting, 17 , 577–581.
Kocin, P. J., and Uccellini L. W. , 2004: Northeast Snowstorms. Meteor. Monogr., No. 54, Amer. Meteor. Soc., 818 pp.
Lopez, R. E., Holle R. L. , Heitkamp T. A. , Boyson M. , Cherington M. , and Langford K. , 1993: The underreporting of lightning injuries and deaths in Colorado. Bull. Amer. Meteor. Soc., 74 , 2171–2178.
Magnus, M., cited. 2006: Winter weather awareness weeks. NWS/Office of Climate, Water, and Weather Services. [Available online at http://www.weather.gov/os/winter/wintercal.shtml.].
Mileti, D., and Sorensen J. , 1990: Communication of emergency warnings: A social science perspective and state-of-the-art assessment. Oak Ridge National Laboratory Rep. ORNL/TM-6609, 166 pp.
Morris, D. A., Crawford K. C. , Kloesel K. A. , and Kitch G. , 2002: OK-FIRST: An example of successful collaboration between the meteorological and emergency response communities on 3 May 1999. Wea. Forecasting, 17 , 567–576.
NWS, 1998: The central Texas tornadoes of May 27, 1997. NOAA/NWS Service Assessment, 99 pp.
NWS, cited. 2004: Winter weather awareness weeks. [Available online at http://www.weather.gov/os/winter/wintercal.shtml.].
NWS, 2005a: WFO winter weather products specification. NWS Instruction 10–513, 53 pp. [Available online at http://www.weather.gov/directives/sym/pd01005013curr.pdf.
NWS, 2005b: Central Region winter weather products specification. NWS Central Region Suppl. 02-2003, 5 pp. [Available online at http://www.nws.noaa.gov/directives/sym/pd01005013c022003curr.pdf.
NWS, 2007: Winter weather watch/warning/advisory procedures and thresholds. NWS Eastern Region Supplement 02-2003, 6 pp. [Available online at http://www.nws.noaa.gov/directives/sym/pd01005013e022003curr.pdf.].
Parker, D. J., and Neal J. , 1990: Evaluating the performance of flood warning systems. Hazards and the Communication of Risk, J. W. Handmer and E. C. Penning Rowsell, Eds., Gower Technical, 137–156.
Peacock, W. G., Brody S. D. , and Highfield W. , 2005: Hurricane risk perceptions among Florida’s single family homeowners. Landscape Urban Planning, 73 , 120–135.
Peterson, D., and Perry R. W. , 1999: The impacts of disaster exercises upon participants. Int. J. Disaster Prevention Manage., 8 , 241–254.
Peterson, R. A., 2000: Constructing Effective Questionnaires. Sage, 152 pp.
Rauber, R. M., Olthoff L. S. , Ramamurthy M. K. , Miller D. , and Kunkel K. E. , 2001: A synoptic weather pattern and sounding-based climatology of freezing precipitation in the United States east of the Rocky Mountains. J. Appl. Meteor., 40 , 1724–1747.
Reaugh, T., and Mowry L. , 2005: Partners in communication: National Weather Service and local media working together to save lives during the Michiana tornado outbreak of 24 October 2001. Natl. Wea. Dig., 29 (12) 33–46.
Redmond, J. W., 1995: NOAA Weather Radio as an emergency communication vehicle in west Tennessee. Wea. Forecasting, 10 , 485–497.
Ryerson, C. C., and Ramsay A. C. , 2007: Quantitative ice accretion information from the Automated Surface Observing System. J. Appl. Meteor. Climatol., 46 , 1423–1437.
Siegrist, M., and Gutscher H. , 2006: Flooding risks: A comparison of lay people’s perceptions and expert’s assessments in Switzerland. Risk Anal., 26 , 971–979.
Slovic, P., 1987: Perception of risk. Science, 236 , 280–285.
Smith, K., 2004: Environmental Hazards: Assessing Risk and Reducing Disaster. 4th ed. Routledge, 432 pp.
Troutman, T., Smith R. , and Rose M. , 2001: Situation specific call-to-action statements. NOAA Tech. Memo. NWS SR-215. [Available online at http://www.srh.noaa.gov/ssd/techmemo/sr215.htm.].
APPENDIX
Selected Survey Questions
The original survey included thirty-nine questions; selected questions are listed below. Demographic questions and those not discussed in the article are not included. Parenthetical statements describe the choices available to respondents.
How seriously do each of the following hazards threaten your CWA? (The following hazards were listed in a table: chemical spill, flooding, hurricane winds, ice storm, snowstorm, storm surge, terrorism, tornado, wildfire, and other—please specify. (Respondents could select from a scale of “0/not applicable” to “5/very seriously.”)
On average, how many ice storms occur each year in your CWA? (Respondents could select one of the following choices: more than 5, 3–5, 1 or 2, 0.5 (about 1 every 2 yr), 0.2 (about 1 every 5 yr), or never.)
This general statement prefaced the next group of questions concerning specific products: “For the questions below, assume that you expect an ice storm to begin in 72 hours, and the weather before it begins will be fair.”
What is the criterion(-a) would be used in deciding whether or not to issue an SPECIAL STATEMENT or HAZARDOUS WEATHER OUTLOOK highlighting the storm? (Respondents were asked to list up to three criteria.)
Assume the criterion(-a) will be met. How far in advance would you issue the product listed above that highlights the storm? (Respondents were asked to select a twelve hour time period; periods included 0–12 h, 12–24 h, etc., up to 60–72 h.)
The previous two questions were repeated, with the same format, for “watch,” “advisory,” and “warning.”
What would the title of the WARNING product be? (Respondents could select one of the following choices: winter storm warning, winter weather advisory, ice storm warning, or other.)
Does your office have previously created special impact statements in regards to ice storms?
(If yes) What organization(s) created these statements? [Respondents could select any answers that applied: my NWSFO (when I was here), my NWSFO (before I came here), another NWSFO (write identifier, if known), NWS regional office (write regional office, if known), NWS headquarters, emergency managers, department of transportation, or unknown.]
(If yes) When were these statements first created? (Respondents were asked to write how many years previous, if known.)
Does your office directly contact people outside the office to warn of ice storms?
(If yes) Which of the following groups do you or employees of your office contact directly? (Respondents could select any answers that applied: local political leaders, local emergency managers, state political leaders, state emergency managers, local media (newspapers, radio, television), utility companies, other: please specify).
Who decides when outsiders should be called? [Respondents could select one of the following choices: WCM, MIC, SOO, lead forecaster, regional office, other: please specify, combination: please specify.]
Are specific criteria (e.g., expected accumulation, time of year, etc.) used to decide when outsiders are contacted? (If yes, respondents were asked to list the criteria.)
During your annual Winter Weather Awareness week, approximately what percentage of information disseminated to the public pertains to ice storms exclusively? (Respondents could select one of the following choices: 0%–20%, 21%–40%, 41%–60%, 61%–80%, 81%–100%, or “My office does not have a Winter Weather Awareness week.”)
Does your office conduct educational programs, such as workshops or conferences, that include a discussion of ice storms?
(If yes) Which statement best describes how your educational programs include a discussion of ice storms? (Respondents could select one of the following choices: “We have a program exclusively about ice storms,” “Ice storms are part of a program about all types of winter weather,” “Ice storms are part of a general program about all types of severe weather.”)
Does your office annually mail out information about ice storms (winter weather) to select groups?
(If yes) Which of the following groups receive these mailings? [Respondents could select any answers that applied: emergency managers, other government officials, media, skywarn/co-op observers, members of the general public, other.)
(If yes) How many people are on your mailing list (total for all groups)? (Respondents could select one of the following choices: 1–100, 101–500, 501–1000, 1001–2500, more than 2500.)
Does your office hold workshops that discuss ice storms (winter weather)?
(If yes) Which of the following groups are invited to attend? (Respondents could select any answers that applied: emergency managers, other government officials, media, skywarn/co-op observers, utility companies, members of the general public, other.)
(If yes) How many workshops are held, on average, annually? (Respondents could select one of the following choices: less than 1, 1–4, 5–9, 10 or more.)
(if yes) What is the median attendance at the workshop(s)? (Respondents could select one of the following choices: 0–5, 6–10, 11–20, 21–30, 31 or more.)
Map of the number of times each state experienced losses from ice storm catastrophes from 1949 through 2000. Neither AK nor HI had any losses during this time. Map based on data published by Changnon (2003).
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
Map showing NWS regions within the continental United States.
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
Map indicating areas served by each NWSFO in the continental United States. Hatching indicates places outside the study area.
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
Map of subregions used to represent data geographically. Solid black lines indicate NWS region boundaries. Unless noted otherwise, white areas were not surveyed.
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
Maps showing (a) the average absolute rating of the ice storm hazard and (b) the average rank of ice storms relative to other hazards, by state group. White areas were not surveyed or had no response. Labels in (b) indicate the highest-ranked hazard in each subregion.
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
An example of a hazardous weather outlook, which merely suggests that a winter storm is possible.
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
An example of a winter storm watch for an ice storm. This statement includes details about the weather scenario, describes the potential for harm, and suggests a course of action for readers.
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
Dot plot illustrating when participants (by region) would issue a winter storm watch for a hypothetical ice storm, assuming foreknowledge.
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
An example of an ice storm warning. This statement has a relatively short discussion of the weather scenario (cf. Fig. 7), but it includes explicit information about how much ice is expected and what impacts the ice will cause. It also has multiple call-to-action statements (cf. Fig. 7).
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
Dot plots illustrating how far in advance of an ice storm WCMs would issue (a) a warning or (b) an advisory, assuming foreknowledge.
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
Map showing states with winter awareness events in 2006. All but five states had similar events in 2004 (see comments and notes). Sources: NWS (2004) and Magnus (2006).
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
Ice storm warnings from three adjacent NWSFOs affected by the same ice storm. Only portions discussing potential impacts are shown here. NWS text messages are typically in all capital letters and frequently use ellipses.
Citation: Weather and Forecasting 24, 1; 10.1175/2008WAF2007111.1
Demographic information about the study participants.
WCM seriousness ratings of various hazards (maximum seriousness = 5).
Summary of responses regarding contact with outsiders.
Median attendance at workshops with ice storm information.
Since the time of the survey, the Eastern Region has eliminated the use of special weather statements (SPSs) in this capacity (NWS 2007). Eastern Region policy now agrees with those already in force in the Southern and Central Regions. Thus, respondents now would only issue advisories or discuss the storm in HWOs 72 h in advance.
It is also worth noting that at least one reviewer strongly disagreed with this finding. Perhaps in the time since the survey was administered, more offices have developed these statements.
One reviewer commented that NWS offices are increasingly involved in regional conferences, often in partnership with local American Meteorological Society (AMS) and National Weather Association (NWA) chapters. Unfortunately, the survey did not ask about these activities.
