3
Preparedness and Response

Even without striking most of the locations it threatened, Hurricane Elena posed a severe test of emergency preparedness systems. In many areas, residents evacuated, returned home, and were then told to evacuate again. In addition, the area from Tampa Bay to Sarasota was evacuated for the first time in many years. This was the first major evacuation since completion of comprehensive, quantitative, regional hurricane evacuation studies in the threatened areas.

The following discussion describes the hurricane warning process in general, the information on the hurricane threat that was available to public officials during Elena, the actions taken by officials, and public response to those actions. The adequacy of prior hurricane preparedness activities and the appropriateness of actions taken during Elena are then evaluated.

THE WARNING PROCESS

With satellites, aircraft, and coastal radar, the National Hurricane Center (NHC) monitors hurricanes and weather systems having hurricane potential. Using data from these and other sources, the NHC issues forecasts of the storm's track, its speed along the track, and its strength. Specifically, forecasts are made for where and how intense the storm will be in 12, 24, 36, 48, and 72 hours. These forecasts are revised at least every 6 hours. Based on these predictions, the NHC issues a hurricane watch for roughly 300 miles of coastline when it appears that a storm will make landfall or pass close enough to the coast to cause hurricane conditions within the next 36 hours. When the storm moves within an anticipated 24 hours or less of making landfall, the hurricane watch is changed to a hurricane warning.

Two factors make it necessary for the NHC to provide additional infor-



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 3 Preparedness and Response Even without striking most of the locations it threatened, Hurricane Elena posed a severe test of emergency preparedness systems. In many areas, residents evacuated, returned home, and were then told to evacuate again. In addition, the area from Tampa Bay to Sarasota was evacuated for the first time in many years. This was the first major evacuation since completion of comprehensive, quantitative, regional hurricane evacuation studies in the threatened areas. The following discussion describes the hurricane warning process in general, the information on the hurricane threat that was available to public officials during Elena, the actions taken by officials, and public response to those actions. The adequacy of prior hurricane preparedness activities and the appropriateness of actions taken during Elena are then evaluated. THE WARNING PROCESS With satellites, aircraft, and coastal radar, the National Hurricane Center (NHC) monitors hurricanes and weather systems having hurricane potential. Using data from these and other sources, the NHC issues forecasts of the storm's track, its speed along the track, and its strength. Specifically, forecasts are made for where and how intense the storm will be in 12, 24, 36, 48, and 72 hours. These forecasts are revised at least every 6 hours. Based on these predictions, the NHC issues a hurricane watch for roughly 300 miles of coastline when it appears that a storm will make landfall or pass close enough to the coast to cause hurricane conditions within the next 36 hours. When the storm moves within an anticipated 24 hours or less of making landfall, the hurricane watch is changed to a hurricane warning. Two factors make it necessary for the NHC to provide additional infor-

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 mation before and during watches and warnings. First, emergency preparedness officials and others may need to take certain actions and precautions even before a watch is issued. Second, the science of hurricane forecasting is such that there is a great deal of error inherent in all three types of hurricane predictions: track, forward speed, and intensity. For example, the average error associated with the 24-hour position forecast is about 120 miles, and the average error of the 48-hour position forecast is more than 250 miles (Neumann and Pelissier, 1981). Error statistics are not reported for intensity predictions, but NHC forecasters say that intensity is even more difficult to anticipate than position. The size of the hurricane watch and warning areas reflects these uncertainties to some extent, but it is often necessary to shift the boundaries of these areas during a threat if earlier forecasts prove to be worse than average. A hurricane warning will sometimes be posted without having been preceded by a hurricane watch, and the time remaining before landfall when a warning is issued can be substantially less than 24 hours. In short, decision makers need more information about the impending hurricane threat than simply whether a watch or warning has been posted for their area. The principal decision makers during a hurricane threat are local and state emergency management professionals who advise elected officials in their respective jurisdictions. In most states the legal authority to order evacuation of residents is shared by the governor and local officials, such as boards of county commissioners. Through the news media the public will receive much of the same hurricane threat information that emergency management authorities do, but few people actually evacuate until explicitly advised or ordered to do so by public officials. In response to this need for threat information by decision makers and the public, the NHC issues a variety of notices. Initially, when disturbances that could develop into tropical storms are detected, such as tropical waves and depressions, the NHC typically disseminates ''special statements'' to indicate their existence. When a disturbance develops certain meteorological attributes (referred to as tropical characteristics), it is called a tropical depression, and the NHC begins making regular forecasts of where and how severe it will be in 12, 24, 36, 48, and 72 hours. (The 36-hour forecast was not disseminated prior to 1988.) The forecasts are included in notices called marine advisories. More general information describing the system's strength, current position, movement, appropriate actions for the public, and so forth are formulated into public advisories. Both marine and public advisories are transmitted by hard copy devices, for which agencies and others pay a subscription fee. If the depression achieves sustained winds of a least 40 mph, it is called a tropical storm and given a name. If the winds reach 74 mph, the storm is upgraded and called a hurricane. Marine and public advisories are continued for tropical storms and hurricanes.

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Public advisories also contain tables listing numerous coastal towns and cities and giving the probability that the center of the depression (or hurricane if it intensifies sufficiently) will pass within 65 miles of the locations. These figures are intended to give decision makers a quantitative indication of how likely the storm is to affect their communities, but also to remind people of the uncertainties in position forecasts. A storm might be predicted to strike Apalachicola, for example, and the probability for that location might be 37 percent. At the same time, however, Tampa's probability might be 18 percent, serving as a caution for decision makers in Tampa to keep their guard up. Public advisories are normally revised every 6 hours, although intermediate advisories are often issued when a storm gets closer to land or when there is a notable change in its track, forward speed, or intensity. When designated coastal areas are placed under hurricane watches and warnings, that information is included in advisories. Local offices of the National Weather Service (NWS) will issue local statements repeating part of the public advisory information, pointing out its significance for the local area, reporting local conditions such as roadway flooding, and advising appropriate actions by the public. State and local emergency management professionals receive the above information and decide what responses are appropriate, usually after consulting directly with NHC or local NWS representatives. Private meteorologists are consulted in some cases. Regional hurricane evacuation studies have estimated the time required to safely evacuate residents at risk in each coastal county during specific multicounty evacuations. When the risk is judged high enough, local elected officials or the governor issue recommendations or orders that people in specified locations evacuate. The evacuation notices are disseminated to the public by the news media (primarily radio and television) and by law enforcement personnel, firefighters, and other officials going through neighborhoods with loudspeakers and, in some cases, knocking on doors. Law enforcement personnel manage traffic flows, and Red Cross workers open public shelters in schools, churches, and shopping malls. ELENA AND THE GULF COAST'S RESPONSES At 3 p.m. on Wednesday, August 28, while over central Cuba, Elena received its name upon becoming a tropical storm. Gale warnings were posted for the Florida Keys. The situation became serious on Thursday, August 29, at 6 a.m. EDT. Elena was still a tropical storm, but intensification was expected, and a hurricane watch was issued from Grand Isle, Louisiana, to Apalachicola, Florida. The eye of the storm was forecast to reach Pensacola, Florida, in

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 just over 24 hours. Pensacola's probability of receiving hurricane conditions was given as 30 percent. A little less than 3 hours later Elena was upgraded to a hurricane, and at 9 a.m. EDT, the watch was changed to a hurricane warning. The forecast storm track was shifted westward into Mississippi, however, and the warning area extended from Morgan City, Louisiana, to Pensacola. Pensacola's probability dropped to 22 percent, compared with a high of 34 percent at Buras, Louisiana. Local and state officials in southeast Louisiana, Mississippi, Alabama, and Escambia County, Florida, issued evacuation notices for residents of highest-risk locations before or soon after noon. At midnight (12 a.m. EDT on Friday, August 30), however, the forecast track swung back to Pensacola, with landfall again expected in a little over 24 hours. Warnings were extended eastward to Apalachicola. Pensacola again had the highest probability, 27 percent, but from Buras to Panama City, Florida, probabilities ranged only from 24 to 21 percent, indicating the uncertainty in the forecast. Before noon, evacuation was ordered in Oklaloosa and Walton counties in Florida. At noon the forecast track was shifted to the east once more, this time placing the hurricane's eye between Ft. Walton Beach, Florida, and Panama City. Evacuation was begun in Bay County, Florida. Despite the forecast shifts, Elena's course over the past day or more had been consistently to the northwest and later more to the north. Friday morning, August 30, the storm slowed, which usually affords greater opportunity for a change of direction. Around 2 p.m. EDT on Friday there was evidence of movement to the east for the first time, and at 6 p.m. warnings were dropped west of Pensacola and extended to Tarpon Springs, Florida, as landfall was forecast near Steinhatchee, Florida, in 36 hours. Evacuees in southeast Louisiana, Mississippi, and Alabama were allowed to return home, while Floridians in parts of Franklin, Gulf, Wakulla, Jefferson, and Dixie counties were told to evacuate. At 6:45 p.m., a local statement from the Tampa NWS office hinted that evacuation might be necessary. At 7:45 p.m., a press release from the Florida governor's office recommended voluntary evacuation from low-lying areas as far south as Pinellas County. At 9:45 p.m. another statement indicated that residents vulnerable to expected tides should complete evacuation as soon as possible and that local officials were in the process of deciding whether to make evacuation notices voluntary or mandatory. At 11 p.m. EDT, the easterly drift became more pronounced, and the warning area was changed again, now reaching from Panama City all the way to Sarasota, Florida. Escambia, Oklaloosa, and Walton counties in Florida were dropped from the warning area. The new forecast called for landfall near Crystal River, Florida, in 18 hours. Evacuation notices soon extended through Sarasota County, and evacuees west of Panama City began to return home.

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 At 8 a.m. EDT on Saturday, August 31, Bay County was dropped from the warning area. Labor Day weekend vacationers were free to visit resort areas in Mississippi, Alabama, and most of the Florida Panhandle on Friday evening or Saturday. Most of the areas filled with visitors to near-normal levels, but Bay County probably never reached 50 percent capacity. As late as noon Saturday, landfall was still expected in the next few hours between Crystal River and Cross City, Florida, but then the storm stalled and remained more or less stationary the rest of the day and night. Around 2 a.m. Sunday, September 1, the NHC began reporting a series of very minor shifts of position to the west, too small to be interpreted as a trend. Since 8 a.m. EDT Friday, Elena's peak winds had been reported by aircraft to be 100 mph, and gale force winds had extended as much as 125 miles from the storm's center. Thus, although the hurricane still had not crossed the coastline, considerable damage had been done to coastal properties and beaches affected by Elena's outer winds and accompanying waves and tides. At 9 a.m. on Sunday, peak winds in the storm had increased to 110 mph, and for the first time Elena was labeled a major hurricane (a category 3 on the five-category Saffir-Simpson scale). By noon Sunday, winds had reached 115 mph, and the NHC was noting that those slight moves to the west could be a trend, possibly necessitating that warnings be posted again for the Florida Panhandle. At 2 p.m. EDT, these warnings were indeed reinstated for the Panhandle, as movement was reported to the west-northwest at 5 to 10 mph. The newest landfall forecast was between Pensacola, Florida, and Mobile, Alabama, and warnings reached from Sarasota, Florida, to Bay St. Louis, Mississippi. The Florida governor's office issued a mandatory evacuation order for all of the panhandle area of the state, and similar actions were taken by state and local officials in Alabama, Mississippi, and southeast Louisiana later in the day. Notices were generally more emphatic and urgent than those issued during the earlier evacuation. At 6 p.m. EDT, winds were at 125 mph, and warnings were dropped south of Yankeetown, Florida. The increase in winds from 100 to 125 mph was more significant than might be obvious. Because damage potential increases with the square of the wind velocity, the 25 mph increase represented a 56 percent increase in damage potential. Advisories were noting that tides could be 10 to 12 ft above normal at the landfall point. Although well to the east of the forecast landfall location, Apalachicola's probability reached 93 percent at 6 p.m. Sunday, because the eye of the storm was only 40 miles away. Panama City's probability was 86 percent because the storm was expected to cause hurricane conditions there, although passing offshore, and Pensacola's eventually reached 95 percent for the same reason. By midnight, the forecast track had shifted west of Mobile, and Elena kept parallel to the Florida Panhandle. By 4 a.m. EDT on Monday, September 2,

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 the eye had passed west of Pensacola, and the storm eventually made landfall near Biloxi, Mississippi, around 8 a.m. CDT, Monday. EVALUATION There were several issues bearing examination following the storm: How well did the evacuation decision-making process function during the threat? How did residents respond to being told twice in just a few days to evacuate? How did the public respond in the Tampa Bay-to-Sarasota area, where few residents had ever experienced a hurricane? How did Labor Day weekend tourists respond? How useful and accurate were the comprehensive, quantitative hurricane evacuation studies that had been prepared for the threatened areas? Emergency Response Decision Making The driving force behind most decisions to advise or order evacuation was the posting of hurricane warnings by the NHC. To overstate the case only slightly: When a warning was issued for your area, you ordered evacuation; if you were not in a warning area, you did not. While issuance of warnings is obviously correlated with other threat factors, the correlation is far from perfect. The threat varies significantly within the warning area, for example, and it might be quite reasonable for communities on the edges to respond differently from those in its center, where probabilities are higher. Responses could reasonably differ within the warning area if the time needed to effect a safe evacuation varied from site to site within the area. Nor should a hurricane warning or even a hurricane watch be necessary to prompt officials to order an evacuation. Pinellas County, Florida, is an interesting case. At 6 p.m. EDT on Sunday, September 1, warnings were extended southward to Tarpon Springs, which lies at the northern edge of Pinelias County. Most of the county was not in the warning area, and it disturbed the county officials to have the county "split," as their response plans called for coordinated countywide actions. This reaction seemed to imply that they did not feel comfortable evacuating the part of the county outside the warning area, but at the same time they did not feel comfortable not evacuating the part inside. Efforts were made to have the NHC either include the entire county in the warning or exclude Tarpon Springs. The fact is that the exact extremes of warning areas are determined somewhat arbitrarily by the NHC and should not sway response decisions as heavily as they often do. At 11 p.m. EDT, the forecast track moved south and so did the warning area (to Sarasota), so Pinelias decided to order evacuation. However, anxiety was created by the fact that the new forecast called for landfall (around

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Crystal River) in just 18 hours, and studies had indicated that 20 to 26 hours would be required to evacuate parts of the county safely. Officials in the county were surprised at the short time apparently remaining before predicted landfall, since they had not previously been under a hurricane warning or even a watch, and the 6 p.m. forecast had predicted that the storm was more than 36 hours away from landfall. Other 6 p.m. data, however, had indicated that Tampa's probability of receiving hurricane conditions within the next 24 hours or less had been given as 12 percent, compared with 19 to 23 percent at the forecast landfall point; Pinelias County was within the 70-to-80 percent probability ellipse for the position where the storm was forecast to be in 24 hours. If at 6 p.m. the warning area had been extended to Sarasota, with other threat information remaining the same, officials in Pinellas would probably have ordered evacuation at that time. In fairness to Pinelias County, its emergency management office has been recognized as a national leader in addressing this kind of decision dilemma, that is, the problem of making evacuation decisions while fully availing themselves of all the relevant information at their disposal. Pinellas officials, for example, used probability ellipses to interpret the 6 p.m. forecast, used the evacuation times calculated in the region's evacuation study, and relied (probably too heavily) on the time-until-landfall implied by the 6 p.m. forecast. Nor should it be concluded that the county's decision not to order evacuation at 6 p.m. was wrong. The real issue here is the decisionmaking process itself, which, given the pressures of the moment, often relies heavily on one or two types of threat information out of the many types available. This same problem is faced by every community and state emergency management office. In recognition of this problem, FEMA provided $100,000 in 1982 to Florida's Department of Community Affairs for a consultant to develop a computer-based hurricane decision system. The system, however, reflected only the consultant's judgments about how much risk involving the lives of Florida citizens was acceptable. The state never understood the rationale, implications, or consequences of those judgments, and today the system is nonoperational. More importantly, state officials still are prey to the same overdependence for their evacuation decisions on the NHC's issuance of hurricane warnings and forecast positions as counties are. Another observation on this matter relates to the role of the NHC in the response decision process. Few evacuation decisions are made without telephone conference calls among state and local emergency management staff with NHC staff, preferably (from emergency management's viewpoint) with the director. The NHC forecasters can then explain the reasoning behind their forecasts, the issuance of warnings for some areas and not others, and so forth. In many cases, the emergency management officials

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 are seeking the advice and approval of NHC staff. It should come as no surprise that these calls tend to reinforce the officials' reliance on warning areas, as it was the NHC staff who decided on the areas initially. Such conference calls can be very valuable to decision makers if they know the right questions to ask about forecast uncertainties, but most do not. The NHC staff is no better equipped to make judgments about acceptable risks in local communities than outside consultants. In the same vein, emergency response plans in many communities are keyed to the issuance of hurricane watches as well as hurricane warnings. The fact that, during Elena, hurricane warnings were posted for some areas without hurricane watches preceding them caught some communities off guard. The Elena experience should teach communities to interpret prewarning storm-threat information for themselves in the absence of an official watch, but the NHC should also be aware of how heavily some decision makers rely on hurricane watches and warnings. One final point concerns the availability of storm-intensity information to officials. Advisories issued by the NHC rely heavily, although not exclusively, on wind velocities recorded several thousand feet above sea level by reconnaissance aircraft, and these velocities are in some instances substantially higher than those at sea or ground-level. The NHC's problem, as indicated earlier, is that there is no uniformly applicable method for estimating ground-level winds from data obtained at higher altitudes, and ground-level observations are rarely available, particularly when a storm is still at sea. This sometimes results, however, in an exaggeration of storm intensity at ground-level and can lead to overreaction to the threat by officials. Whenever possible, it would be valuable for NHC forecasters to apprise emergency management officials and the media if there is reason to suspect that the aircraft data significantly overstate ground conditions. It might have been useful, for example, to know the significance of the fact that when the center of Elena was passing 35 miles south of Pensacola, the local NWS office was reporting maximum gusts of 93 mph, while at the same time the aircraft was reporting sustained winds near 125 mph. Given the wind profile detected by the aircraft, would 93 mph gusts 35 miles from the center at ground-level imply 125 mph sustained winds on the ground at the storm center? Multiple Evacuations Studies of past evacuations using sample surveys conducted after hurricane threats indicate that if local officials are successful in reaching the public with evacuation notices, response is excellent (Baker, 1986). On barrier islands and other high-risk areas, more than 90 percent of the residents usually leave when ordered, provided that officials are aggressive in

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 communicating the notice. Such studies were conducted after Eloise and Frederic in the northern Gulf area, and response in those storms was very good. Thus, there was no reason to expect response problems in Elena, at least in the initial evacuation. The "cry-wolf syndrome" is a concern of many preparedness officials. That is, if people evacuate during one threat and the storm misses their area, are they willing to leave during a subsequent threat? Normally, the subsequent threat is expected to be a different storm. In Elena, two distinct threats were posed by the same hurricane in areas from Louisiana to Apalachicola. The cry-wolf issue is usually overrated as it applies to hurricanes. In surveys conducted with evacuees after false alarms, 80 to 90 percent of the respondents indicate they would do the same thing again, and empirical evidence suggests they indeed do. Hurricane Diana in 1984 is the only welldocumented case of a single storm causing two evacuations in the same area, and the response in North Carolina was good. Telephone sample surveys conducted in Panama City and Panama City Beach, Florida, did not find a decrease in response from the first to the second Elena evacuation, nor did a third evacuation that occurred in November in response to Hurricane Kate (Baker, 1987). Several locations reported substantially higher public shelter use in the second evacuation than in the first. In Jackson County, Mississippi, for example, 2,500 people were sheltered the first time and 12,000 the second, although the total number of evacuees in the two evacuations is believed to have been roughly the same. It is normal for local shelter use to be higher in hurried evacuations. Vacationer Response When warnings were lifted for Louisiana, Mississippi, Alabama, and the western part of the Florida Panhandle on Friday, August 30, vacationers poured into resort communities late Friday and on Saturday morning for Labor Day weekend. Vacationers increase the population at risk dramatically in some communities, and there is an additional concern about whether vacationers will evacuate during a hurricane threat as willingly as residents. Virtually all the systematic data on public response in actual evacuations stems from studies of how residents reacted. When evacuation notices were posted Sunday afternoon, they included Labor Day visitors who had arrived only the day before and had not planned to leave until Monday. Emergency management officials throughout the area indicated, however, that vacationers left readily and promptly when orders were issued. Baldwin County, Alabama, for example, evacuated twice as many people during the second threat in half the time required in

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 the first threat. The great majority of evacuating vacationers appear to have returned home rather than go to local shelters. Public Response in the Tampa Bay-to-Sarasota Area Many meteorologists and preparedness officials alike are fond of stating that public response is a major problem in many areas because residents have no hurricane experience and therefore do not appreciate the hazard posed by hurricanes. (A variation on this theme relates to a concern over "false experience," whereby people think they have been in a hurricane when they actually have not.) The fact is that postevacuation sample surveys show that previous hurricane experience is uncorrelated with whether residents leave in subsequent threats. Admittedly, those studies typically test for individual differences within the same community, where the community as a whole often has a fair amount of experience. In the Pinellas, Hillsborough, Manatee, and Sarasota counties area, the entire region has little experience with major hurricanes, because almost all major hurricane activity in the 25 years before Elena had been in the northern and western Gulf of Mexico. Telephone sample surveys indicated, however, that in the dangerous Pinellas County beach areas, more that 90 percent of the residents evacuated (Baker, 1987). The Red Cross in Pinellas County estimated that 120,000 people went to public shelters, and officials presumed that figure represented 40 percent of the evacuees, which would have been a very high rate of public shelter use. The rate was probably overstated because of an underestimate of the total number of evacuees (officials were assuming that no one evacuated outside the areas where evacuation was ordered), and sample survey results indicated that fewer than 20 percent of the evacuees went to public shelters. The late hour of the evacuation and its apparent urgency did not provide people the luxury of traveling as far as might have been necessary for them to reach the homes of friends and relatives inland from the beach, and they might not have had time to arrange accommodations either with friends or at motels, so even 20 percent might be higher than one should usually expect in the region. Furthermore, the Tampa Bay Regional Planning Council has been extremely active in distributing public awareness information, including maps showing shelter locations, and assigning particular neighborhoods to them. Those materials might have led residents to depend on the shelters more heavily than they would have otherwise. There is some question, however, whether shelter users went to the locations assigned in the public information tabloids. Most indications are that the evacuation was very successful. In fact, it appears to have taken less time than the regional evacuation study had calculated for the area. There were, however, instances of public protest

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 following the storm, since beach residents were not allowed to return to their homes when warnings were dropped for the area until officials performed safety checks and preliminary cleanup. Use of Regional Hurricane Evacuation Studies Since 1979, the U.S. Army Corps of Engineers and FEMA have been active in providing funds for comprehensive, quantitative hurricane evacuation studies, and the state of Florida has funded updates for some of their earlier studies. Except in Louisiana, all the areas that were under warnings for Elena had been included in such studies already completed or almost completed at the time of the threat. The southeast Louisiana study was under way, and several usable components were available. These studies include quantitative estimates of the following: areas that would be inundated by various hurricane tracks and storm intensities, number of people needing to evacuate in these various scenarios, how the public would respond to evacuation, and how long it would take to safely evacuate the population at risk in various storms. Evacuation Zones Clearly, the studies were extremely useful in helping officials decide upon which areas locally needed to be evacuated during Elena. Accordingly, they were also useful in anticipating the number of people who would need to evacuate and in helping officials make resource deployment decisions. Several emergency management officials reported the usefulness of the maps and other study outputs in convincing their elected officials of the need to take various actions. Large-scale evacuation-zone maps were frequently used on television to make local residents aware of the locations being evacuated. Behavioral Analyses Behavioral studies estimate the percentage of residents who would be likely to evacuate during a hurricane, the time period over which various percentages of evacuees would be expected to leave, what percentage of evacuees would use public shelters, and so forth. In the Mississippi, Alabama, and Florida Panhandle, such estimates were made from analysis of past hurricane evacuations, establishing rules of thumb for anticipating different responses in different risk areas and in different threat scenarios. In the Tampa Bay region the figures were originally derived from sample surveys in which residents were simply asked how they would respond to evacuation notices. Having been apprised by experts in the field that survey responses

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 to such hypothetical questions have been shown to be extremely poor predictors of actual behavior during real hurricane threats, the regional planning council attempted to modify its original behavioral assumptions in 1984. A new analysis was not conducted by behavioral specialists, however. The Tampa Bay study was still deficient with respect to some of its behavioral assumptions. It assumed, for example, that only people living in areas instructed by officials to evacuate would evacuate. In fact, 20 to 40 percent of residents outside such evacuated areas also leave during many hurricane threats. That error, combined with lack of awareness of particular circumstances causing higher-than-average shelter use, led to an underestimation of shelter demand in Elena and to the erroneous conclusion that 40 percent of evacuees had gone to shelters. More generally, there is a need for behavioral analyses to recognize that public response will vary from threat to threat, to identify what will occur in various circumstances and situations, and to not make use of hypothetical response data without major adjustments. In an update of the Tampa Bay evacuation study subsequent to Elena, behavioral assumptions were revised by qualified consultants. Clearance Times Clearance time refers to how long it takes between initiation of evacuation and its completion, the latter usually referring to leaving the county of origin. Clearance times, therefore, are taken as the time needed to evacuate people. Prelandfall hazard times (the time before actual landfall during which a weather hazard exists and evacuation efforts will be hampered) are added to the clearance times to yield the total time period before expected landfall required to allow the population at risk to be safely evacuated. Officials in the Panhandle generally reported that clearance times calculated in the evacuation study conducted in their region were close to what they observed in Elena. The study was successful in pinpointing the locations of bottlenecks well inland from the coast as evacuees attempted to reach destinations like Montgomery, Alabama, and it accurately predicted the roughly 10 hours necessary for the congestion to clear. Table 3-1 was compiled by the Mobile district of the Corps of Engineers as a result of conversations with local emergency management and law enforcement officials, but was not based upon structured research. Although quantitative data of the same sort are not available at this time for the Tampa Bay area, there are reports that roads cleared much more rapidly than analyses had led officials to expect. At this time, the reasons for this are not known, and it is possible that the calculated times were not as far off as people now believe. The study had estimated that such an evacuation in Pinellas County would take 11 hours, for example, and it

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 TABLE 3-1 Evacuation Times (hours)   August 29 September 1 County Observed Calculated Observed Calculated Hancock, Mississippi 7 9.5 5 8.5 Harrison, Mississippi 11 11.0 9 11.0 Jackson, Mississippi 12 13.0 11 11.5 Mobile, Alabama 11 15.5 12 14.5 Baldwin, Alabama 16 16.0 12 15.0 Escambia, Florida 12 14.5 10 13.0 Santa Rosa, Florida 7 8.5 7 8.0 Okaloosa, Floridaa 7 14.0 13 14.0 Walton, Florida 7 9.5 10 11.5 Bay, Floridab 11 22.0 11 22.0 a Okaloosa had a partial evacuation on August 29. b Bay had partial evacuations in both threats. SOURCE: U.S. Army Corps of Engineers, Mobile district. probably took at least 8 hours, counting from the governor's recommendation at 7:45 p.m. EDT. It is likely that calculated clearance times are sometimes overstated because of planners' desires to be cautious in arriving at the estimates. Such a cautious approach may be justified, but it would prove useful to label the calculations to indicate the degree of caution inherent in the assumptions made. In northwest Florida, Alabama, and Mississippi, planners did, in fact, vary assumptions about how quickly people would respond to the threat and how many people would evacuate, and clearance times were calculated based on combinations of those variations in assumptions. Updates to the Tampa Bay study have subsequently calculated clearance times under various sets of assumptions. There are additional opportunities for such sensitivity analyses, however. In the original Tampa Bay study, analysts reduced their assumptions regarding traffic flow rates by 10 percent to account for prelandfall rain conditions and further reduced the assumed flow rates of evacuees to account for congestion due to normal ''background'' (nonevacuating) traffic. When there is no rain and little or no background traffic (as was the case at midnight), however, flow rates will be greater than those assumed. Just as behavioral responses will vary from storm to storm, so will clearance time. By the same token, the apparent accuracy of times calculated for the Mississippi, Alabama, and Florida Panhandle area should not lead one to believe that clearance times in future evacuations will always be the same as those observed in Elena.

OCR for page 27
Natural Disaster Studies: Volume Two, Hurricane Elena, Gulf Coast - August 29–September 2, 1985 Regional Boundaries The regional (multicounty) approach to hurricane evacuation studies was taken after a Lee County, Florida, study was frustrated by the fact that evacuees cross county boundaries, thereby making it difficult to consider their effect on Lee's evacuation times and patterns without considering what was occurring in the adjacent counties simultaneously. The multicounty approach solves this problem to some extent but creates two others. First, pressure is sometimes created for an entire region to act in concert. For example, disaster management officials may be more likely to order evacuation of adjacent counties at the same time, because clearance times, shelter demands, and so forth for each county were calculated with the assumption that the entire region was evacuating. This might not always be appropriate, given the variations in risk that can exist within a region. The Mississippi-Alabama-West Florida area is very linear, and road networks are such that intercounty movement of evacuees is not substantial, so individual counties need not question whether their calculated clearance times will change depending on the actions of adjacent counties. In a clustered region like Tampa Bay, however, intercounty flow is more important. Second, evacuees will often cross regional boundaries just as they cross county boundaries within a region, and not all regional plans have an adequate means of anticipating the effects of actions occurring in an adjacent region on their own evacuation times and patterns. The so-called tristate study is an example of a study that considers the effect of adjacent regions. Little flow of evacuees from southeast Louisiana into Mississippi is generally anticipated, although some evacuees from the two areas would probably merge on some of the same northerly evacuation routes after leaving their respective areas. At the eastern boundary of the Mississippi-Alabama-West Florida study area, analysts were able to include quantitatively the flows from the adjacent region, partly because the same consultant performed the analyses for both regions. As new regional studies are funded and as old ones are updated, analyses should treat the regions in a less unified and isolated manner. Clearance time sensitivities, for example, should include scenarios in which only part of a region evacuates. Similarly, the sensitivities should include scenarios in which adjacent regions do and do not evacuate.