Conclusions and Recommendations
A number of issues and recommendations can be gleaned from the observations made by the postdisaster study team that covered the effects of Hurricane Hugo on the Virgin Islands and Puerto Rico.
Hurricane Hugo was the most damaging hurricane in U.S. history before Andrew. Latest estimates place the actual property losses in Puerto Rico and the U.S. Virgin Islands alone at over $3 billion, and over $10 billion when the impact on the Carolinas is considered. Hugo was a classic Cape Verde hurricane, and reached record-tying intensity on September 15, 1989, when NOAA WP3D reconnaissance aircraft measured 918 mb central pressure and 165 knots (189.75 mph) sustained flight-level winds (at 1,500 ft) as the storm was approaching the Leeward Islands. The importance of aerial reconnaissance to hurricane monitoring was vividly demonstrated by the fact that, even though an aircraft experienced some extreme turbulence and lost one engine during the initial penetration into Hugo, valuable measurements were relayed to the NHC in real time that confirmed the hurricane's growing strength and threat to the Caribbean islands. Moreover, satellite estimates made just prior to the aircraft's encounter with Hugo were indicating a much weaker hurricane, with central pressures some 30 mb higher and much lower peak wind speeds than were actually measured.
It is imperative that NOAA and the USAF continue to provide coordinated aerial reconnaissance and monitoring of Atlantic hurricanes and relay critical data in real time to NHC forecasters. The aging NOAA aircraft and associated
instrumentation need to be modernized and, in some cases, replaced for safety and data-gathering efficiency. An independent oversight panel should be established to determine the research and operational priorities of these upgrades.
The official forecast errors for Hugo by NHC forecasters were lower than average for Atlantic/Caribbean hurricanes over the past 10 years, only 65 NM for 24-hour forecasts. On the other hand, there was a significant left-bias to the official track forecasts and, therefore, poor forecasts of Hugo's turn toward the northwest as it approached St. Croix late on September 17. Moreover, after the storm brushed the northeast coast of Puerto Rico early on September 18, great dispersion developed in the various predicted track models available to the NHC forecasters when Hugo was north of Puerto Rico. Forecasted landfall locations at 48 to 72 hours ranged from the Florida Keys northward to Cape Hatteras—a few model runs had the storm recurving entirely out to sea. Overall, the standby, revised NHC83 (a dynamical/ statistical model) produced the best forecast tracks out to 48 hours.
A much greater level of effort and associated resources is needed for the development of a consistent hurricane-prediction model, capable of more accurate prediction of tracks and intensity changes out to at least 72 hours. Currently, no operational models are available for predicting hurricane-intensity changes, which became even more important as Hugo approached the U.S. mainland.
There was a lack of accurate surface-wind-speed data in the aftermath of Hugo's passage through the Caribbean. What is more, the lack of upper-air observations over the entire Caribbean limited the performance of all forecast models in use at the time. Today, ther are fewer sites taking two upper-air radiosonde measurements per day than there were 20 years ago. Many of the federal and state agencies had anemometers, but most of them had no recording equipment or backup power, or had to be abandoned because of damage to offices. Few records
survived, and many of the instruments were damaged or destroyed by flying debris. The only sites with verifiable records were NWS and FAA/LLWAS (windshear) anemometers at San Juan International Airport and Roosevelt Roads Naval Station. Not one hard, verifiable record has been obtained anywhere in the U.S. Virgin Islands. Consequently, probable maximum sustained speeds and gusts for surface winds in the U.S. Virgin Islands and other key locations in Puerto Rico had to be estimated from the aircraft reconnaissance winds taken, for the most part, at 10,000 ft and from the postdisaster study team 's aerial and surface damage surveys. The highest winds from this assessment were at St. Croix (110 knots [127 mph] sustained, 135 knots [155 mph] peak gusts), followed by Culebra, Vieques, and St. Thomas.
There must be a reinvigoration of the surface-observing network, especially for rugged wind/pressure instrumentation, in the Caribbean islands. Technology is available now at moderate cost to do the job. In addition, the countries involved (perhaps with assistance from the United States and the Commonwealth of Puerto Rico) should develop a capability for rapid deployment of additional surface sensors in advance of approaching hurricanes to capture the wind and other important meteorological “footprints” crucial for improved definition of hurricane structure and the surface-wind field. Finally, there is no substitute for adequate upper-air data, which is needed to define the hurricane's steering flow and for input into prediction models. The team urges that the Caribbean islands, again with assistance from the United States and international organizations such as the WMO, strengthen their conventional upper-air rawinsonde networks with additional sites. At the same time, recognizing the labor-intensive nature of taking balloonborne soundings, the study team urges that early application of Wind Profiler technology be sought to provide continuous vertical-wind profiles for improving the analysis of the hurricane atmospheric environment and other tropical weather disturbances. Over the open ocean, north and east of Puerto Rico, additional data on steering flow could be obtained from instruments released by reconnaissance aircraft. Advances in both geosynchronous and polar-orbiting satellite technology, such as improved microwave soundings, may also alleviate this problem.
COMMUNICATION WITH THE NEWS MEDIA
In the weeks and months following Hurricane Hugo, there were numerous reports by highly reputable newspapers and magazines of wind speeds well in excess of 174 knots (200 mph). These reports have been shown to have no basis in fact, but
many residents and some public officials in the affected areas tend to believe the speeds were as high as or higher than those reported by the news media. Because the actions taken by building officials following extreme natural events such as Hugo are influenced to some degree by their perceptions of those events, it is imperative that these authorities be provided with accurate information.
If the public believes that the wind speeds were as high as reported, building owners will feel that their structures performed very well or were lost in an extremely rare event. These owners will see no need to improve wind resistance, while the reality will be that their buildings barely met—or did not meet—contemporary structural design requirements. Local building codes and construction practices that could be updated or replaced will be left in place, setting the stage for future disasters.
Conversely, if building officials believe that building codes should provide for wind speeds such as those reported by the news media, structures will be grossly overdesigned, resulting in a substantial and needless waste of private and public resources. In the case of St. Croix, loads based on wind speeds reported by the news media and speeds that some residents believe occurred are three to four times the load associated with current accepted design practice.
The meteorological community must accept some responsibility for the lack of knowledge about wind in the news media that covers natural disasters such as Hugo. An effort must be made to inform reporters and journalists about the nature of wind and its effects on buildings and other structures. Certain instances of inaccurate reporting encountered after Hurricane Hugo could have been avoided if the reporter had been provided with better information and explanation.
WIND MODELS RELATING AIRCRAFT-MEASURED WINDS TO SURFACE WINDS
One of the most vexing problems identified by the team during the post-storm investigation was the lack of surface-wind data with which to calibrate observed damage and to assess the performance of structures under extreme loads. The team found that current theory provides only crude, sometimes inconsistent, estimates of surface winds when extrapolating aircraft-measured winds at 10,000 ft downward to the surface.
Better wind models should be developed for relating aircraft-measured winds aloft to surface winds as measured by standard anemometers at 10-m mast heights with proper exposure. At the same time, instrumentation and techniques need to be developed for remote measurements of surface wind and other important variables, such as sea state and water temperatures. A promising development is the Fast-Scanning Microwave Radiometer (FSMR), as developed and tested over the past several years on the NOAA research aircraft by Black and Swift (1987, 1989). This remote-sensing device looks downward from the aircraft, and by directly measuring changes in sea state and associated emissivity, has produced accurate, reliable surface-wind-speed estimates that compare very favorably with measurements obtained with air-dropped buoys and island winds. The FSMR should be installed on all hurricane reconnaissance aircraft.
ALERT AUTOMATIC RAIN GAGE NETWORK AND BACKUP POWER SUPPLY FOR WATER REGULATORY STRUCTURES
The hydrology of Hugo was well defined only over Puerto Rico, where conventional rain gages were augmented with data from a special ALERT network. Whereas the ALERT network (mostly situated over the western two-thirds of the island) indicated peak rainfall of 9 to 10 inches over 48 hours, the rainfall maximum for the storm was situated on the windward side of El Yunque, the highest mountain on the island, where rainfall totals of 12 to 14 inches were found. Only a few rain gages survived Hugo's stronger, damaging winds over the Virgin Islands, but storm totals appear to have been somewhat less severe, about 8 to 10 inches. The only comprehensive, qualitative real-time data on the precipitation was from NWS radar coverage over Puerto Rico and the Virgin Islands. General flooding over Puerto Rico was at the 10-year return frequency, and many streams were near or at bankful. Flash flooding primarily occured in eastern Puerto Rico; in this area, it was confined to creeks and small streams. More serious, however, was the urban flooding in San Juan and its environs, especially the low-lying areas where residents are accustomed to seeing short-term street flooding whenever heavy summer thundershowers drop 1 to 3 inches of rain in a few hours. Such urban flooding in Hugo was aggravated and prolonged by the heavier rains, leaves and brush clogging the street drains, and the absence of backup power for pumps in low-lying areas.
A major problem, which could have become a catastrophe, was the inability of workers to activate the floodgates at El Carraizo Dam, which provides the main water supply for the city of San Juan. No backup motors were available; therefore, even though the reservoirs were filled to capacity, the 300,000 residents of San Juan
suffered for 9 days without any drinking water other than that provided by dairy milk tank trucks. Moreover, known deficiencies in the maintenance of El Carraizo dam were documented over 5 years prior to Hurricane Hugo.
The ALERT automatic rain gage network started over Puerto Rico by a cooperative effort among the local NWS office and other agencies should be expanded to include the Virgin Islands, with adequate provision for real-time communication, data display, and equipment capable of withstanding winds of 100 to 200 knots (115 to 230 mph). The ALERT rain gages and, in some cases, associated river gages should be coordinated with the installation of the NEXRAD radar at San Juan. The comprehensive quantitative rainfall measurements can be made in association with the rain gages by radar. This would facilitate flood warnings, provide information about flooding, and hasten the recovery from hurricane conditions. Hugo was not a particularly “wet” hurricane, but other hurricanes with more precipitation can be expected.
Measures for the strengthening, proper maintenance, and provision of backup motors for El Carraizo Dam must be given immediate attention, with clear lines of responsibility established within the commonwealth. In addition, the problem at El Carraizo probably reflects a problem throughout the commonwealth. Therefore, attention should be given to all reservoirs. Plans for more expeditious relief of chronic, known areas of urban flooding in San Juan need to be formulated and resources for implementing them identified.
EBS NETWORK, SHELTER READINESS, AND INTERGOVERNMENTAL RESPONSE
The team found that the WSFO performed well with regard to emergency planning and response. Watches and warnings issued by NHC were widely disseminated by Puerto Rican broadcast and print media, well in advance of Hugo's arrival. This success was clearly the result of exceptionally well-planned preparations initiated by MIC Israel Matos, with numerous conferences and public meetings beginning the previous winter and spring. These preparations were further accentuated by the near-approach and hurricane watches issued for Hurricane Dean a few months prior to Hugo. An important finding was that the local NWS preparations actions were properly enmeshed with other disaster-related organizations within the commonwealth. They used the same evacuation plan as the commonwealth and local municipal civil defense authorities. The NWS office in San
Juan had a mass-media-dedicated telephone line that facilitated frequent contacts for information updates and a mass media officer for handling these contacts. The NWS modernization planning now in progress should take this success into account for its future office staffing profiles.
The team found that the overall evacuation efforts by civil defense authorities were successful. This success can be attributed to prior planning, organization, and coordination among civic authorities called for in an evacuation study that was carried out by the U.S. Army Corps of Engineers and funded by FEMA (Newsome, 1990). The evacuation program was very effective and utilized the SLOSH storm-surge model outputs to systematically plan evacuation routes and timing. The results of this evacuation study were used by both NWS and the civil defense officials to time the evacuation and determine optimal, safe routes from the population centers. A principal negative finding was a failure of the sheltering phase during the evacuation: (1) shelters were not open on time and/or lacked staff and adequate provisions such as food, cooking facilities, and water; (2) there was a notable failure of intergovernmental cooperation and coordination during the entire sheltering process; and (3) during the recovery period, many schools used as shelters continued sheltering people too long, interrupting the school year.
The EBS network needs to be extended throughout Puerto Rico and the U.S. Virgin Islands. In addition, the new NOAA Weather Wire, which is used to provide fail-safe, satellite communications of hurricane and other severe weather information, should be expanded to cover this entire region as soon as possible.
The team identified an urgent need for a regional census of shelters in Puerto Rico to determine both their effectiveness during various hazards and their shortcomings (e.g., lack of food-preparation facilities).
The practice of using school buildings and civic centers as shelters because they “look well built” should be stopped. Structurally sound shelters should be identified and their wind resistance evaluated by professional wind and structural engineers. Care must also be given to allow adequate elevation and distance inland in those low-lying or other areas susceptible to storm surges. Puerto Rican schools are often cinder-block construction, and these perform poorly in storms. The flat roofs were slow to drain during Hugo's heavy rains.
There needs to be a followup social science analysis of the Commonwealth of Puerto Rico's political system, within which the successes and failures described above for emergency planning, preparedness, and sheltering are embedded for disaster-related programs. This analysis should examine the interaction of disaster programs and aspects of Puerto Rican society, culture, and politics. Such a study should provide answers to the question of why some programs succeed and others fail, and how to adjust disaster-related programs and policies to increase their overall effectiveness.
TECHNOLOGY TRANSFER TO IMPROVE BUILDING CODES AND CONSTRUCTION PRACTICES
Building codes and construction practices in areas affected by Hurricane Hugo need to be evaluated in the light of loading conditions that equalled or exceeded the design loads. Because the assessment of structural performance, and thus the adequacy of local codes and practices, is so highly dependent upon accurate wind-speed information, it is important that these post-storm assessments be carried out by competent and experienced wind and structural engineers.
Single-family homes suffered the greatest proportion of severe damage from Hugo's onslaught in this area. Many homes were built without regard to existing code requirements. Most important, the team found extensive damage to “do-it-yourself” types of wood construction. There were heavy losses of corrugated roofing, windows, and doors. On the other hand, reinforced concrete dwellings in Puerto Rico performed well. On St. Croix, new construction showed considerable variation in performance.
The team urges that a concerted technology effort be made by appropriate federal, commonwealth, and U.S. Virgin Island governments to provide economical, state-of-the-art design criteria and detailing practice for low-income housing in the areas impacted by Hugo.
COASTAL ZONE MANAGEMENT PLAN
The shoreline response to Hurricane Hugo was investigated by the team through direct inspection of the Puerto Rican coastline, from the west of San Juan around to the southeast coast; this area fell south of the eye passage. Peak storm-surge heights were experienced along the northeast coast, just east of San Juan, at 4 to 6 ft, perhaps as high as 8 ft. Surge heights over the Virgin Islands were somewhat less, about 3 to 5 ft. In fact, the storm surge over Puerto Rico was much less than what Hugo later produced over South Carolina because the narrow, steep shelf around Puerto Rico gives a small envelope of shallow bathymetry for surge buildup; the Carolina coasts have a wide continental shelf, which gives a broad area of water to be “pushed up”. The most severe surge damage was along the northeast and north coast of Puerto Rico, from San Juan to Fajardo. In general, developments
on rocky coastlines did well if situated above surge levels. Low elevations were most susceptible to both surge and wave damage, but not shoreline erosion as on sandy shorelines. There was a surprisingly large amount of sand overwash--up to 2 m or more--in the Pinones area east of San Juan. Moreover, the inland extent of the overwash area was most pronounced where streets were perpendicular to the shoreline. There was considerable damage to both public and private beaches and coastal developments along the entire northeast coast of Puerto Rico, especially sidewalks, seawalls, paved roads, and many small structures.
Direct-wave attack and storm-wave overwash were the principal forces of erosion impacting the shorelines of Puerto Rico and the Virgin Islands. There were no startling observations in this area. Storm-surge levels were modest, as can be expected from almost any strength hurricane because of the very narrow and steep shelf surrounding the Puerto Rico-Virgin Islands Platform. Low storm-surge levels likewise imply few problems with storm-surge ebbs, the return of storm waters to the sea after forcing winds have passed or have reversed. In South Carolina, where surges were greater than 19 ft in places, storm-surge ebb caused extensive scour, undermining of roads and utilities, destruction of buildings and seawalls, and offshore transport of structures and debris.
Helping to reduce the impact of Hugo was the rocky nature of much of the Puerto Rico and Virgin Islands shoreline. Rocky shorelines imply no “erosion,” but direct-wave attack on low-elevation structures will be (and was in Hugo) important.
Highrise condominium and hotel development continues to crowd the Puerto Rico shoreline. With the possible acceleration in the rise of sea-level, the shoreline will continue to migrate landward, and the beaches will be moved back until they run headlong into developed areas. Already a crisis-based response to shoreline erosion has resulted in the armoring of much of the developed shoreline of Puerto Rico. Beaches in front of walls have largely disappeared.
In order to save the recreational beach, either the buildings must be moved back from the shoreline or Puerto Rico must begin a beach-replenishment program. Beach replenishment is expensive and not economically practical in the long term. That leaves retreat as the only sound shoreline-management option. Whatever the response, new development should be situated well back from the beach and, where possible, at high elevation.
Beach-resource management is particularly important for current beach residents, as well as future developers. Some of the coastal impacts of Hugo uncovered by the team have left the area more vulnerable to beach erosion and structural damage from the more common winter storms over the southwestern Atlantic. Overwash sand should be returned to the beaches, not carted away inland, as was documented in some locales. Dunes should be replaced as protective barriers, not bulldozed. Road geometry should be redesigned so as to avoid roads
perpendicular to the shoreline, which accentuate inland transport of sand overwash. Future developments should be steered toward higher-elevation, rocky shorelines and discouraged in exposed, flat areas close to the water. Finally, the hazards and tradeoffs in hurricane-prone coastal areas must be cogently communicated to government policymakers and planners at all levels.
Puerto Rico needs a unified plan and approach to coastal zone management; indeed, it should make a concerted effort to protect its valuable resources of beautiful recreational beaches. The Pinones area is particularly important, because it commonly suffers the greatest amount of overwash. It is a largely undeveloped area, mostly swamp, and has only one road for access—Route 187. There is a lot of interest by developers and the present governor in developing this area. Not only would that destroy a valuable natural resource, but it also would be a dangerous place to develop because of evacuation and post-storm access problems caused by overwashing of the low-lying Route 187.