Conclusions and Recommendations
1. Need for In-Depth Postdisaster Studies
Conclusion: Hurricane Elena struck an area that in recent years had experienced two major hurricanes (Camille in 1969 and Frederic in 1979). The hurricane threat was well recognized by the local communities, which believed they had taken steps to prepare for such events through the use of NWS forecasts, evacuation planning, and appropriate building codes. Forecasting and evacuation efforts appear to have been fairly successful, but there was a considerable amount of wind damage to buildings and other structures.
The present report has attempted a more detailed analysis of the causes of structural failures than has been attempted in the past. Since similar design conditions and building control procedures exist along hurricane-prone coasts from Texas to South Carolina (with the exception of South Florida), the conclusions drawn from such a detailed study of performance in Elena were considered to be relevant to a very large number of buildings.
Unfortunately, funds for such detailed studies are not usually available—a situation that has been strongly criticized by the engineering community, especially considering the influence that such in-depth investigations can have on the safety and reliability of future buildings. In the case of the present report, the detailed work required was undertaken outside of normal funding procedures at the initiative of the study team—clearly not an ideal process.
Recommendation: The federal or state governments, possibly supported by insurance companies and code bodies, should establish properly funded
study teams that can conduct investigations of hurricane and tornado damage. These teams should have access to structural testing facilities and wind tunnels and should operate in a manner similar to air crash investigation teams.
2. Better Wind-Speed Data
Conclusion: On the basis of aircraft-measured wind speeds and damage observations, the NWS estimated that the maximum surface wind speed in Hurricane Elena, averaged over 1 minute, was 127 mph. Analysis of surface wind speed measurements at landfall suggests that the 1-minute average probably did not exceed 100 mph over the ocean and 90 mph a short distance inland.
The NWS relies very heavily on the Saffir-Simpson scale to relate damage to wind speed. Whenever a detailed analysis of wind speed data is made, as in this case, the estimates based on surface measurements are much lower than those estimated from the Saffir-Simpson scale and by the procedures used by the NWS to convert aircraft measured wind speeds to surface wind speeds. In most cases it has been found that the maximum 1-minute sustained winds reported by the NWS are fairly close to the maximum 2-to 3-second gusts measured near the shoreline.
The speeds reported by the NWS are widely accepted by the press and the public and sometimes find their way into technical reports. Their use gives an inflated impression of the severity of a storm in relation to the conditions used for the design of buildings and other structures.
Clearly, the forecasting role of the NWS has taken precedence over its role as archivist of all important storm data. Since the NWS has a very limited number of anemometer sites along the hurricane-prone coasts, rarely attempts to collect and analyze data from other anemometers, and places its faith in the Saffir-Simpson scale, there has been an erosion of confidence in some quarters of the engineering and meteorological communities in the agency's ability to maintain accurate records of hurricanes. It should be noted that the paucity of good surface-level wind data is not peculiar to the Gulf Coast but is a problem that plagues the nation as a whole.
Much of the current difficulty stems from the lack of adequate funds to fulfill both the NWS's forecasting and record-keeping missions. Moreover, the two communities served by these missions—meteorologists and wind engineers—often have different needs and often do not communicate well with each other. Data that may be entirely adequate for meteorological purposes such as forecasting hurricane movements may be grossly inadequate for detailed engineering studies.
Nor is there much hope of significant progress in this area under the current budgetary restrictions, even though all parties agree that a reliable wind record would be of immense value. Marginal improvement may come when additional data collection stations come on line in the 1990s as part of the NWS's Automated Surface Observing System (ASOS), but the number of new data points will be very limited. (ASOS will ultimately consist of 1,500 sites around the country, each containing an anemometer and a host of other weather-observing instruments.)
The installation of NEXRAD—NWS's advanced weather radar facilities that are scheduled to be introduced in the early 1990s—should help improve NWS's wind data by providing quantitative information on boundary layer winds within 80 miles of each NEXRAD facility. However, such information is far from the direct and accurate measurement of ground-level winds needed by the engineering and meteorological communities.
Recommendation 1: Representatives of the primary users of wind-speed data and senior administrators of the NWS should meet to resolve the problems associated with obtaining accurate ground-level wind-speed data. In particular, it is recommended that NWS—or another appropriate agency, such as the National Institute of Standards and Technology—take responsibility for collecting and analyzing anemometer records. To increase the accuracy of these data, the following steps are recommended:
The ocean data buoy network should be expanded to permit adequate documentation of both hurricane wind speeds and sea conditions.
Inexpensive, portable instrumentation should be developed that could be dropped from reconnaissance aircraft to obtain sea-or ground-level winds.
Inexpensive and perhaps portable land-based anemometer stations should be located along the hurricane-prone coastline and activated during periods of high winds.
The NWS and other organizations that currently possess anemometers (such as the military or local police and fire stations) should be encouraged to standardize their anemometer height and data format to comply with the recommendation of the World Meteorological Organization.
Each NWS station should maintain a list of reliable anemometers in its area together with correction factors to convert data obtained from them to standard conditions.
Recommendation 2: The procedures used to estimate wind speeds using damage observations should be carefully reviewed by the NWS and wind engineers. Until this is done it is recommended that the wind speeds in both the Saffir-Simpson scale and the Fujita Scale be taken as 2-to 3-second gusts.
OTHER NATIONAL WEATHER SERVICE ACTIVITIES
3. Continued Cooperation between NHC and the Local and National Media
Conclusion: The local and national media provided widespread coverage of Hurricane Elena. Many of the broadcasts originated from the National Hurricane Center, with assistance from NOAA Public Affairs in coordinating the effort. This resulted in an extremely high level of awareness by the public, which, in turn, led to a rapid response during the massive evacuation.
Recommendation: The media should receive continued encouragement from the NWS to make live broadcasts from the NHC. In addition, it is imperative that NOAA Public Affairs continue to coordinate the media activities at the NHC in order to ensure smooth operations. Such NOAA assistance would be required only during significant landfalling hurricanes that will impact the United States.
4. Use of Amateur Radio and Hurricane Drills
Conclusion: Amateur radio was employed for emergency communications in many areas threatened by Elena. Valuable information concerning weather conditions, property damage, and needs for assistance were passed to the NWS and emergency management officials in a timely manner by these amateur radio operators.
Recommendation: The NWS should continue to avail itself of the skills offered by amateur radio operators in emergencies to the full extent allowed by law. In fact, the use of amateur radio communication during hurricane emergencies should be expanded to the greatest extent possible. It is also recommended that hurricane drills be held annually with the various amateur radio hurricane networks.
5. Need for Improvement of Numerical Forecast Models
Conclusion: Forecasts during Hurricane Elena derived from the three different NWS numerical forecast models showed considerable differences, especially at the time three critical changes in direction of movement of the hurricane occurred. The consequence of the different forecasts from the three numerical models was a loss of confidence by NHC forecasters.
Recommendation: The National Meteorological Center should continue to make increased efforts to improve its ability to analyze and forecast tropical cyclones. This should include improvement in accuracy as well as a single unified forecast presentation for use by NHC.
6. Continued Effective Use of Telephone Hurricane Information Service
Conclusion: The NOAA, AT&T, and NBC telephone hurricane information service, 1-900-410-NOAA, generated over 64,000 calls and 133,500 call minutes during Hurricane Elena. This provided an excellent source of information for the public during this hurricane.
Recommendation: The NOAA, AT&T, and NBC hurricane information service was useful and cost-effective and should be continued.
7. Need for Quick Information Dissemination from SLOSH Runs
Conclusion: The NWS Sea-Lake-Overland Surge from Hurricanes (SLOSH) storm-surge prediction model accurately forecasted the storm surge for this storm. The SLOSH model was used operationally during this storm to anticipate the potential impact on each section of the coast that was threatened. These scenarios proved valuable for planning purposes by local emergency management officials. However, the model is sensitive to hurricane path, and it proved difficult to quickly provide accurate information to local emergency managers.
Recommendation: NWS and FEMA should explore methods for dissemihating information from SLOSH runs quickly to local emergency managers.
THE EVACUATION PROCESS
8. Incorporation of Forecast Uncertainties in Evacuation Planning
Conclusion: Watches and warnings issued by the NHC are insufficient bases for determining the timing of evacuation notices, yet many government agencies rely upon them almost exclusively for actual decision making.
Recommendation: State and local officials need to make systematic use of threat information provided by the NHC in marine advisories. However, because of the imperfection of forecasts included in marine advisories, agencies also need to incorporate into their decision-making processes ways of recognizing the uncertainty in the forecasts (with regard to hurricane track, forward speed, and severity) and employing that information in assessing the threat.
9. Need for Multiagency Hurricane Evacuation Studies
Conclusion: The multiagency hurricane evacuation studies for both the Tampa Bay and tristate areas proved to be useful for local emergency management officials during Hurricane Elena.
Recommendation: The federal government should continue to support hurricane evacuation studies and to encourage the initiation of studies in those coastal areas where none exist.
10. Hypothetical Behavioral Assumptions Underlying Evacuation Plans
Conclusion: Behavioral assumptions underlying evacuation plans are often based upon hypothetical surveys with area residents. As demonstrated in parts of the Gulf Coast evacuated in Elena, hypothetical surveys alone are invalid for projecting how people will actually behave during hurricane threats.
Recommendation: Behavioral assumptions regarding evacuation should be based primarily upon empirical studies of actual response patterns in past hurricane threats. Generalizations about such patterns can be applied to locations where no evacuation has occurred by matching site, demographic, and response plan characteristics of individual locations to those observed in other places. Evacuation responses will vary from threat to threat in the same location, and responses documented in one evacuation should not necessarily be used for general planning. Hypothetical response surveys have extremely limited utility and should not be used as literal indications of actual response.
11. Calculation of Clearance Time in Evacuation Studies
Conclusion: Clearance times calculated in regional evacuation studies prior to Elena matched times estimated in Elena in some locations, but not in others. Actual clearance times will vary from storm to storm, as public response, actions by public officials, and other factors vary. Further, clearance time calculations are potentially of great value during response decision making but are not yet fully utilized. Clearance time calculations and shelter demand projections can vary, depending upon whether adjacent counties are evacuating.
Recommendation 1: As a part of regional evacuation studies, a variety of scenarios should be modeled—including optimistic ones—to maximize the value of these studies to state and local officials during planning and real-time decision making.
Recommendation 2: Clearance times should be incorporated systematically into structured decision systems employing marine advisory forecasts and other threat information.
Recommendation 3: Studies should vary their assumptions regarding local evacuation patterns in order to assess the impact of such variance in findings, including the shelter demand projections.
PERFORMANCE OF BUILDINGS AND OTHER STRUCTURES
12. Need for Nationally Applicable Wind-Loading Provisions
Conclusion: There has been concern about the adequacy of building codes and their enforcement with regard to wind resistance. When structural failures were not blamed on excessive—and unsubstantiated—wind speeds, the blame was usually attributed to poor enforcement. While poor enforcement may have had some bearing on the quality of construction affected by Elena, the present study has shown that much of the blame must lie with the building code itself. The failure of the Standard Building Code to incorporate adequate wind-load specifications and consistent construction requirements can be traced to the democratic manner in which code changes are made. Although this is a highly controversial issue, it would appear that the lack of a properly prepared national building code of the type used in most other developed countries has hindered the incorporation of up-to-date wind-loading provisions and consistent structural requirements.
Recommendation 1: Since a national building code is unlikely in the foreseeable future, it is recommended that all model and local building codes adopt the American National Standard A58.1 for the specification of wind and other loads.
Recommendation 2: Building codes should be checked to ensure that any prescriptive provisions regarding structural details are consistent with their wind-loading provisions. This is not the case at present with the Standard Building Code or the Uniform Building Code, although steps are being taken to remedy this.
13. Design Needs for Nonengineered Structures
Conclusion: As it stands at the moment, it is virtually impossible for a contractor or a building official to determine whether a nonengineered building is in compliance with the wind-loading provisions of the Standard Building Code. Some form of engineering input must be used in the determination of the structural form of buildings that are not professionally designed. The traditional manner of doing this is to adopt prescriptive requirements that are deemed to comply with the performance requirements of the code. Some of these are now being prepared for use with the Standard Building Code. These deemed-to-comply standards must be based on accurate wind-force determination. The Standard Building Code's present wind-loading provisions assume that all low-rise buildings are in open country away from the sea. In fact, very few nonengineered buildings in hurricane-prone areas are in such a location. Some are in open country adjacent to the sea (where the Standard Building Code underestimates the wind loads), but the majority are in forests or subdivisions (where it overestimates the wind loads).
The deemed-to-comply standards for wood-framed houses, now being issued by the Standard Building Code, are not being well received, because they require radical and expensive changes to traditional forms of construction. Such changes are certainly required for buildings located adjacent to the sea; but it was observed in Elena that houses in forests and densely packed subdivisions performed perfectly well with, at most, the addition of hurricane anchors to attach the roof to the walls. The use of ANSI A58.1 to define the wind loads would enable the exposure conditions to be considered and sensible deemed-to-comply standards to be prepared.
An alternative procedure to the use of these standards has been introduced recently that enables the structural requirements for individual wood-framed houses to be determined. It is a microcomputer-based expert system (Sparks and Singh, 1989). The program uses ANSI A58.1 to determine the wind loads on the structure and checks to see if the structure can be built in accordance with a commonly used prescriptive code (Council of American Building Officials [CABO], 1986). If there is insufficient load capacity, recommendations for ties, anchors, and shear walls are made.
Recommendation 1: Deemed-to-comply standards should be prepared for nonengineered structures using realistic wind-loading requirements and incorporated into building codes.
Recommendation 2: The use of computer-based building codes and expert systems should be explored as a means of designing and checking nonengineered structures.
14. Need for Design Checks of Professionally Designed Buildings
Conclusion: Without making any independent structural checks, building officials usually accept that professionally designed buildings are in compliance with the prevailing building code. This has led to unsatisfactory construction, particularly in low-rise buildings.
Part of this problem arises from the use of empirical design procedures for wood and masonry included in the Standard Building Code. Some of these are inconsistent with the wind-loading provisions of the code. However, in some instances, it would appear that architects or engineers have failed to recognize the effect of wind forces on the structural systems they are designing.
Recommendation: If local jurisdictions are unable or unwilling to provide building officials with the resources to make checks on the wind resistance of professionally designed structures, structural review panels or individual consultants should be employed to review the designs. This is a practice used extensively outside the United States.
15. Insurance against Wind Damage
Conclusion: It is usually assumed by insurers, the general public, and emergency management officials that the issuance of a certificate of occupancy by a building official means that a building is able to withstand, with a reasonable factor of safety, the wind loads specified in the locally enforced building code. Structural failures in storms are therefore attributed to excessive wind speeds. Hence it is reasoned that to reduce insured losses the design wind speed must be raised. Some states have attempted to raise design wind speeds that would appear to have a mean recurrence interval of more than 2,000 years. Yet, even if the design wind speed had been raised to such a level on the Mississippi and Alabama coasts following Hurricane Camille, it is unlikely that the extent and nature of the damage caused by Elena would have been significantly different from that actually encountered, since the design wind speed had very little bearing on the structural form of many of the damaged buildings. However, properly engineered buildings would have been considerably more expensive.
The Federal Flood Insurance Program has shown that insurance can have a beneficial effect on the form of construction used in an area. A similar scheme for wind insurance could have saved many of the buildings in Elena. However, wind insurance is in the hands of private companies and state catastrophe pools, which rely on the building code system to ensure structural adequacy. For certain types of buildings, this system has not performed well.
Recommendation 1: If the building code system cannot be reformed, the federal government or private insurance companies should consider imposing their own regulations or, at the very minimum, using insurance rates that reflect the risk of wind damage.
Recommendation 2: State officials and others concerned with reducing wind damage should consult wind engineering experts concerning the causes of wind damage before proposing legislation or regulations.
16. Concern about Industry Standards
Conclusion: Many industry standards adopted by reference by the Standard Building Code show a very poor appreciation for the effects of wind loads on structures. Most use factors of safety that are lower for wind loads than other loads and in some cases contain empirical design rules that are likely to result in dangerously unsafe structures, for example, the recently adopted ACI-ASCE proposed masonry standard (American Concrete Institute-American Society of Civil Engineers, 1988). (For more detailed discussion of this document see Sparks and Saffir, 1989.)
Recommendation 1: Wind-engineering experts should make a thorough review of the standards adopted by reference in building codes to check their adequacy.
Recommendation 2: Wind-engineering experts should be consulted during the preparation of standards that have an influence on wind resistance, particularly those that contain empirical design rules.
17. Concern about Using School Buildings as Shelters
Conclusion: The extremely poor performance of school buildings calls into question the frequent use of these structures as evacuation shelters.
Recommendation 1: The design of future schools should be checked for structural adequacy, and frequent inspections should be made during construction. This will require significant changes in the present inadequate control process.
Recommendation 2: Existing schools and other buildings designated as evacuation shelters should be checked to ensure that they are able to carry the wind loads specified in ANSI A58.1. Particular attention should be paid to roofing systems and their anchorage to exterior walls. This should be given high priority by emergency planning coordinators along the entire hurricane-prone coastline of the United States. If insufficient public buildings are available to act as shelters, it may be necessary to consider the use of privately owned buildings.
18. Concern about Preengineered and Masonry-Walled Buildings
Conclusion: Preengineered buildings fabricated, marketed, and erected in the ''design-build mode'' continue to present a problem in that the design and reliability of various components not supplied by the manufacturer (doors, facade, glazing, and foundation details) may be lost to the engineer who designed the building. Similar comments can be made regarding the glazing and doors of masonry-walled buildings using light roofs.
Recommendation 1: Building officials should make every attempt to ensure that all components and details of such preengineered buildings satisfy appropriate wind-load criteria. If components cannot meet the wind loading requirements, the buildings should be designed as partially enclosed.
Recommendation 2: If buildings are designed as enclosed, checks should be made to ensure that collapse will not occur if a component, such as a window, is accidentally damaged by flying debris. Particular attention should be paid to masonry-walled buildings with light roofs and preengineered metal buildings that contain few interior partitions. Such systems are sensitive to internal wind pressure, often lack structural redundancy, and are prone to progressive collapse.
19. Roof Performance
Conclusion: The performance of the roof coverings of otherwise fully engineered buildings continues to be inadequate. The inability of many coverings to maintain watertightness during storms frequently results in damage to building contents and business interruption costs. Additionally, roof coverings that separate from the structure and become airborne provide the largest source of windborne debris, which, in turn, results in an increase in risk to human safety as well as the potential for compromising other structures and property downstream.
Recommendation: Roof systems and testing procedures should be developed that reflect the high suctions experienced by low-pitched and flat roofs in hurricanes.
20. Performance of Signs and Building Appurtenances
Conclusion: Historically, building codes and standards have mostly neglected signs and building appurtenances. While the performance of highway signs has improved, such has not been the case for commercial signs adjacent to shopping malls and residential areas, where the failure of such signs provides a high hazard to life and property.
Recommendation: Code bodies should take the lead in developing more realistic provisions for signs. Additional wind tunnel testing may be necessary.