Ten Years after Andrew: What Should We Be Preparing for Now?

AIR Special Report August 2002 Ten Years after Andrew: What Should We Be Preparing for Now? Technical Document_HASR_0208

I. Overview Ten years ago, on August 24, 1992, Hurricane Andrew tore across the southern tip of Florida and became the costliest natural disaster in U.S. history. Within hours of Andrew s landfall, AIR, a five-year-old catastrophe modeling company, estimated that this catastrophe could result in losses exceeding $13 billion, a record for insured losses that remains unbroken to this day. Losses of this magnitude were not thought possible; it was only months later that the insurance industry had to face the reality that losses had indeed been incurred on so large a scale. Will another event take the industry by surprise? While much has been done by companies over the past 10 years to manage their exposures along the most vulnerable parts of the coast, such as south Florida, some companies may not be so well prepared for a major hurricane striking the northeastern U.S. Such an event may be as unexpected today as Hurricane Andrew was in 1992, yet it is sure to happen; the only question is when. NOAA satellite Image of Andrew approaching Florida, August 23, 1992 II. Hurricane Andrew: Portrait of a Killer Hurricane Andrew started as a tropical wave or trough in the easterly trade winds, moving from the west coast of Africa to the tropical North Atlantic on August 14, 1992. A cyclonic rotation developed while passing south of the Cape Verde Islands. By August 17, it became the first Atlantic tropical storm of the 1992 season. Over the next four days steering currents gradually slowed Andrew and shifted it to a northwesterly course. On August 22, Andrew s course had altered towards the Bahamas and Florida and the storm achieved hurricane strength. Within 36 hours Andrew reached a peak intensity 2002 AIR Worldwide Corporation SPECIAL REPORT 2

between Saffir-Simpson categories 4 and 5, and caused an estimated quarter-billion dollars in damage in the Bahamas. At about 4:00 a.m., August 24, the eye of Hurricane Andrew passed over Elliot Key on the eastern edge of Biscayne Bay. The Fowley Rocks Buoy, located just to the east, recorded sustained winds of 141 mph (227 kph) with a peak gust of 169 mph (272 kph) as the eye wall passed. Data transmission ceased after that reading. As the eye moved over Biscayne Bay, storm surges occurred from Turkey Point in the south to as far north as Miami. The highest recorded surge was 16.9 feet (5 m) at the Burger King International Headquarters on the western coast of the bay. NOAA radar composite of Andrew making landfall August 24, 1992, at Dade County, Florida Andrew s eye made landfall just east of Homestead Air Force Base at about 5:00 a.m. The eye was about 15 miles in diameter and central pressure had fallen almost 45 millibars to an estimated 922 mb, the third lowest pressure for a hurricane at landfall in the U.S. in this century. Just before they were destroyed, instruments at the National Hurricane Center (NHC) in Coral Gables, at the northern edge of the eyewall, recorded a maximum sustained wind of 138 mph (222 kph), with a peak gust of 164 mph (264 kph). The NHC announced on August 21, 2002, a full decade after Andrew s landfall in South Florida, that it has reclassified the storm as a category 5 hurricane, up from Category 4. This makes Andrew only the third Category 5 hurricane to strike the continental U.S. this century. 2002 AIR Worldwide Corporation SPECIAL REPORT 3

Area of Dade County where Andrew caused greatest damage (above) and devastation in Homestead (left) For the next three hours, Andrew destroyed or heavily damaged buildings from Kendall, and South Miami Heights in the north, through Homestead and Florida City in the south. By 8:00 a.m. Andrew had torn through the Everglades National Forest to Big Lostman s Bay on Florida s west coast, devastating the wetland ecology. Andrew briefly reintensified over the Gulf of Mexico and approached Louisiana. It spawned an isolated F-3 tornado on August 25, which left a path of damage 9 miles long and 150 yards wide through Laplace, LA. Later that day Andrew came ashore near Burns Point, LA. Total damage in Louisiana has been estimated at $1 billion. Andrew destroyed over 25,000 homes in Dade County and damaged 100,000 more. About 90% of all mobile homes in south Dade County were totally destroyed according to the Dade County Grand Jury, and that figure rose to 99% in Homestead. The loss of life was fortunately small: 26 deaths were directly attributed to Andrew, with an additional indirect toll of about 65. It took several months for the insurance industry to calculate the actual losses from this event. The final tally, as issued by Property Claim Services, was $15.5 billion. 2002 AIR Worldwide Corporation SPECIAL REPORT 4

III. What Will Be the Next Event to Take the Industry by Surprise? The magnitude of losses caused by Hurricane Andrew took the industry completely by surprise, despite the fact that the AIR hurricane model had been producing estimates of potential Florida hurricane losses of this magnitude, and even greater, years before. Today, catastrophe modeling is the global standard technology for catastrophe risk assessment. But are companies reaping the full benefits of what this technology has to offer? Do the lulls between catastrophe events still lead to industry complacency? Are there potential natural catastrophes that will, once again, catch the industry off guard? AIR estimates that a strong Category 3 hurricane striking the northeastern U.S. could cause insured damages twice as high as Hurricane Andrew s 1992 record loss. Hurricanes making landfall in the Northeast U.S., 1900-2001 Since 1900, 11 hurricanes have made a direct hit on the coastline between New Jersey and Maine. By far the most severe was a strong Category 3 hurricane that hit Suffolk County, Long Island on September 21, 1938 and continued through Connecticut, Massachusetts, and into Vermont. Maximum sustained winds of 121 mph (195 kph), with 2002 AIR Worldwide Corporation SPECIAL REPORT 5

peak gusts of 184 mph (299 kph), and a central pressure reading of 946 mb were recorded. Even though the affected area was far less populated than it is today, the cost of the Great New England Hurricane of 1938 was immense for the times. Close to 9,000 buildings were destroyed and 15,000 more were damaged. More than 600 people died. Significant structural damage occurred far inland, as the photo taken in Worcester, Massachusetts, some 140 miles from point of landfall, shows. Damage caused by the Great New England Hurricane of 1938. (Photos courtesy of NOAA) So much damage occurred inland because the storm was traveling at a forward speed of 50 miles per hour, covering 600 miles in just 12 hours. The time since landfall was insufficient for the eye to fill. Furthermore the faster the forward speed, the faster are the winds on the right-hand side of the eye an effect known as asymmetry. 2002 AIR Worldwide Corporation SPECIAL REPORT 6

The AIR hurricane model estimates that, were this storm to recur today, losses could exceed $28 billion. Were the 1938 Great New England hurricane to recur today, insured losses could exceed $28 billion The ZIP Code distribution of losses illustrated in the figure above shows that the damages from another Great New England Hurricane would not be confined to coastal areas. The 1938 storm traveled at 50 mph through Connecticut and Massachusetts without much diminution in intensity. Furthermore, because hurricanes in the Northern Hemisphere have their strongest winds on the right-hand side of the storm track, significant losses could be expected in heavily populated regions of Rhode Island, Massachusetts, and southern New Hampshire. Despite the place of prominence the 1938 storm still holds in the record books, perhaps the least anticipated and prepared for scenario today is that of a major hurricane striking the heavily populated regions of the Northeast. This region is often given less attention when assessing exposure to hurricane risk due to the relative infrequency of such events. The brevity of reliable instrumental records hampers our current understanding of historical hurricane activity in the Northeast. When it comes to research on the frequency of major hurricanes, the limitations of the data are even more acute because of the relative rarity of such events. 2002 AIR Worldwide Corporation SPECIAL REPORT 7

However, historical accounts (newspapers, letters, journals, etc.) and ship logs indicate that the northeast has experienced at least one major hurricane comparable to the 1938 storm each century. More recently, scientific researchers have used a so-called proxy approach to reconstruct prehistoric records of intense hurricanes. The essence of such an approach is to locate the geologic signals of tropical cyclones in certain preserved physical materials and then to interpret the stratigraphic record using known relationships between tropical cyclones and their impact on the environment. Among these signals are the sedimentary records of overwash deposits, recorded effects of wind on vegetation and the isotopic signature of rainfall. For example, Dr. Jeffrey Donnelly of the Woods Hole Oceanographic Institution and his colleagues at Brown University have studied sedimentary records in several sites along the northeast coast. The working hypothesis is that major hurricanes can create significant storm surges (> 3 m). Surges of such magnitude can overtop barrier islands, which separate beaches from inland marshes, lakes and lagoons. The storm surge transports sediments, such as sand from beaches, into back-barrier marshes, lakes, and lagoons. As storms wane, the transported sediments are left behind and these overwash deposits are preserved as the barrier island migrates over marsh or lagoon deposits in a regime of rising sea level. By examining vibracores (core collection of soft marine sediments), Donnelly and his colleagues have found stratigraphic evidence of major hurricane landfalls. These and other proxy studies indicate that the annual probability of an intense (Category 3 or higher) storm affecting the northeastern United States is about 0.9%, or roughly one storm every 100 years. Major storms will affect the northeast again. It is not a question of whether large losses can occur in this region, but rather when. It should also be emphasized that the Great New England Hurricane of 1938 does not represent the largest loss scenario. More severe storms and therefore larger losses are possible. Hurricane Bob, a weak Category 2 storm, wreaked havoc in Wareham, Massachusetts on August 19, 1991 2002 AIR Worldwide Corporation SPECIAL REPORT 8

IV. Conclusion In 1992, and within hours of landfall, AIR estimated losses for Hurricane Andrew at a then unprecedented level. While Andrew continues to hold the record for insured losses from a natural catastrophe event, other entirely plausible scenarios could result in much greater losses. What is the next event that we should be preparing for? AIR catastrophe loss estimation technology enables companies to anticipate the likelihood and severity of extreme events so that they can adequately prepare for their financial impact. AIR models produce catalogs of thousands of simulated potential future events to ensure that clients are provided with scenarios representing the entire spectrum of potential catastrophe experience not just events that result in losses of average probability but also those that make up the tail of the statistical distribution. V. About AIR Worldwide AIR Worldwide (AIR) pioneered the development and application of catastrophe loss estimation technology a technology that provided companies, for the first time, valuable tools to assess and manage their catastrophe risk. Today, AIR provides clients with a full suite of integrated products for underwriting, pricing, portfolio management, risk transfer and financing. AIR has developed models to estimate potential catastrophe losses from all major natural hazards, including hurricanes, earthquakes, extratropical cyclones, tornadoes, hailstorms and flood, for more than 35 countries throughout North and South America, the Caribbean, Europe and the Pacific Rim. AIR is also a leading provider of weather risk management services. In response to the increasing demand for accurate weather and climate information for evaluating new weather risk management opportunities, AIR provides weather data and climate forecasts for the weather risk management market around the world. AIR has created a broad range of software solutions to serve the diverse needs of our clients, among them, CLASIC/2, CATRADER, and CATMAP /2. Web-based applications include ALERT, AIRWeather ClimateCast and AIRProfiler. To find out more about AIR products and services, please visit our website at www.air-worldwide.com. For more details on how your company can take advantage of this leading edge technology, please contact: AIR Worldwide Corporation 101 Huntington Avenue Boston, MA 02199 617.267.6645 2002 AIR Worldwide Corporation SPECIAL REPORT 9