HURRICANES AND TROPICAL METEOROLOGY

Widespread damage to roofs and windows (F1) of residences in Miami's southern suburbs after Andrew

When Hurricane Andrew of 1992 came booming ashore in South Florida, it caused the United States' most expensive natural disaster-- $27 billion worth of property destroyed. Loss of life was another matter. In Miami-Dade County, Andrew killed only 15 people by direct force of wind and water. Human casualties were so light because the storm caused little flooding in populated areas and the National Hurricane Center forecast the storm track accurately. Although hurricanes are inevitable on the East Coast of North America, correct characterization of the phenomenon on all scales enables men and women to prepare effectively. Accurate forecasts of individual events give people time and motivate them to act. The climatology of the threat's occurrence and severity is the key to wise policies implemented long before the event. Intelligent responses in the years, months, days, hours and even minutes before hurricane landfall can limit human and material losses.

Impacts

The simplest characterization of hurricane intensity is embodied in the Saffir-Simpson scale: from Category 1 ---barely a hurricane--- to Category 5 ---the worst imaginable. "Major Hurricanes" are those in Categories 3, 4, and 5 with winds stronger than 110 miles per hour equivalent to 100 kt or 50 m s^-1. Category 5 hurricanes are the most extreme and also the most rare. Only two, the 1935 Labor Day Storm and Camille in 1969 are recorded to have struck the United States. Andrew, at the very top of Category 4 was the third strongest U.S. landfall, and the second strongest on the mainland, given that the 1935 storm hit the Florida Keys.

In the 20^th century, U.S. hurricanes destroyed > $73 billion in property, not corrected for inflation. During the 70-year period from 1925 through 1995, the toll was $61 billion. If the damage from historical hurricanes is normalized for inflation, increased population, and greater individual wealth (Pielke and C. Landsea 1998), the estimate of total damage for the shorter period is $340 billion, equivalent to an average annual loss of $5 billion. During these 70 years, 244 landfalls occurred. The average landfall would have resulted in $1.5 billion in damage with today's prices and costal development. But the average doesn't tell the story. Major hurricanes accounted for 80% of the normalized damage, although they represented only 20% of occurrence.

The 1995 through 1999 seasons inclusive have been the five most active in the > 100-year quantitative climatology. Historically, hurricane landfalls on the U.S. east coast were common during the 1940s through the mid 1960s. In the 1970s and 1980s, landfalls were few. Now activity appears to have returned to the high level that characterized the immediate post-World War II period. These fluctuations in activity are most pronounced for major hurricanes. They also correlate with the observed "North Atlantic Mode", a coherent, multidecadal fluctuation of global sea-surface temperatures. During the active portion of the long-term record, Atlantic Sea-Surface Temperature (SST) anomalies in tropics and high latitudes were warm, and conversely. If the hurricane climatology and the Multi-Decadal Mode prove to be reliable guides, we may expect the first decade or two of the 21^st Century to produce as many of the most damaging major hurricanes annually as the last 5 years have.

In terms of hurricane-related mortality, the 20^th Century started badly. In Galveston Texas, on a single windy Saturday night, 9 September 1900, the "Great Hurricane" washed > 6,000 souls to their deaths. The total mortality for the century was just a bit more than twice this figure, 13,306 U.S. residents. During the first three decades of the century, the average annual loss of life was 329, or discounting the Galveston tragedy, 129. In the forty years from 1930 through 1969, it was 70. Since 1969, the average annual loss of life has been < 20, notwithstanding a 10-fold increase in coastal population from 1930.

The reason for the dramatic reduction has been effective warnings and timely evacuation from coastal areas inundated by storm surge. Invariably, large loss of life in hurricanes before 1970 stemmed from wind-driven flooding. Since 1970 drowning from inland flooding caused by torrential hurricane rains has come to predominate. Experience shows that when storm surge (or wind for that matter) completely flattens buildings, about 10% of the people present die. Evacuation insures that virtually nobody is present. On the other hand, extreme ( > 30 cm in 6 hr) rainfall places a much larger population at individually smaller risk. For example, Hurricane Floyd, the deadliest U.S. hurricane since Agnes of 1973, killed 49 of the more than 5 million people that it affected in the North Carolina Coastal Plain and Piedmont, for an average risk of dying of less than one in a hundred thousand. People in the developing world, who generally do not have the benefit from farsighted land-use policies or effective building standards, are at greater risk. The tragedy of Hurricane Mitch of 1998, which took 10,000 lives in Honduras, demonstrates that among the many bad consequences of poverty is vulnerability to natural disaster.

Aircraft

Tropical cyclones are ideal subjects for study from instrumented aircraft. The vortex core is relatively small--only a few hundred kilometers across. An airplane can traverse it 5-10 times in the course of a flight lasting 6-8 hours. The great rotational inertia of the swirling wind means that the balanced vortex changes slowly during the time that an airplane can remain on station. Expendable probes can report atmospheric or oceanic conditions as they drop from flight level. The size of the core is comparable with the range of 5 or 10 cm wavelength search radars. Although aircraft radars are relatively low powered and have small antennas, the ability to move through the storm enables them to observe storms in detail. The foregoing advantages compound for aircraft equipped with Doppler radars, particularly so if they can fly coordinated patterns in pairs to produce true dual-Doppler winds. NOAA's aircraft operations center flies two WP-3D turboprop that represent a unique scientific resource and are the mainstay of HRD's annual campaign of airborne hurricane observations. Starting in 1997, a third aircraft, a Gulfstream IVSP jet entered the inventory. This airplane is dedicated to synoptic surveillance to obtain observations in the environment around hurricanes to improve operational track forecasts.

Warnings are expensive. The economic cost of raising a warning, in terms of lost productivity, safeguarding homes and industries, evacuation of aircraft and vessels, and canceled beachfront vacations averages $0.5 to 1.0 M per mile. The cost depends upon the amount of coastal development. If a major city, such as Miami or New Orleans, lies in the warning area, cost escalates dramatically. Sometimes, as happened in Hurricane Emily of 1993, models and observations combine to give forecasters particularly clear insight into the meteorological situation, so that they can exclude large sections of coastline, perhaps hundreds of miles long, from the warning area and save the economy as much as $100M. When, on the other hand, as happened in Hurricane Floyd of 1999, the storm track remains offshore running parallel with the coast, warnings extend far beyond the area affected, and overwarning costs are immense.

As a problem in fluid mechanics, the dominant factor in hurricane motion is the surrounding wind. In effect, the storm is carried downstream by the average environmental wind. It also "propagates" westward and poleward because it is on a rotating, spherical Earth. Local vertical is more nearly parallel with the planet's axis of rotation in high latitudes. "Beta effect" propagation arises from the poleward increase of the planet's rotation around local vertical. Air that moves around the the vortex center with the storm circulation tends to keep a constant value of the sum of the planetary and relative vorticity. Vorticity in the context is the tendency of the air to swirl around a vertical axis. It is the stuff of vorticies.

An anticyclonic asymmetry thus forms on the northeast side of the vortex and a cyclonic asymmetry on the southwest side. This happens because where the planetary vorticity is larger the relative vorticty must be smaller, and conversely. These "beta" gyres induce a northwestward flow across the vortex center that pushes the storm northwestward. Hurricane motion is 80 to 90% due to large scale winds and 10 to 20% due to propagation. Nonetheless, at a typical speed of 1 m s^-1, the motion due to propagation is comparable with typical forecast errors.

Over the last 30 years, the decrease in forecast error has averaged about 1% per year , largely because of improved computer models. In fact, the models are now better than the knowledge of the "initial condition," the state of the atmosphere at the start of a prediction calculation. Research at AOML has demonstrated that dropsonde observations from aircraft flying around hurricanes can lead to another 15% reduction in the error. If the more accurate forecasts translate into reduced warning areas, savings on the order of $10 to 20M in overwarning costs per hurricane landfall are possible. This dramatic result recognizes that hurricane motion is controlled by the state of the surrounding atmosphere, so that forecasts based upon accurate and timely measurements of that state are themselves more accurate.

In these "synoptic surveillance" experiments, the atmospheric state is measured by dropwindsondes deployed by the NOAA research aircraft. Sondes are released from an altitude of 5 to 7 kilometers. They measure pressure, wind, temperature, and humidity as they fall to the sea on parachutes. Release patterns encircle the storm and reach more than 1,000 kilometers from the center. Aboard the aircraft, meteorologists process the data and transmit it to the National Centers for Environmental Prediction, where it is analyzed and used to initialize numerical models. The cost of the flight operations is about $130,000--about one percent of the potential savings from reduced overwarning.

More widespread and efficient observations of this kind are the primary reason that NOAA commissioned its new Gulfstream IV jet that has flown synoptic surveillance missions operationally since 1997. The jet procurement included the new GPS dropsonde instruments themselves and new faster in-flight processing and data-visualization software. These improvements are important contributors to the accelerated decrease in forecast errors in the late 1990s.

Dynamics

Hurricanes are nearly circular, warm-core vortices, characteristically 1000 km in radius. Strong winds and precipitation are concentrated near, but not at, their centers. Coincident with strongest winds, convective updrafts, fed by inward spiraling surface winds, release latent heat drawn from the ocean to power the storm. The clear eye, 15-30 km in radius, contains the axis of vortex rotation and is surrounded by this ring of strong winds. The eye is characterized by the warmest temperatures in the vortex, low humidity, low pressure, in extreme cases > 10% below that in the undisturbed tropical atmosphere, and calm winds at the very center. The hurricane's eye presents one of the truly magnificent vistas offered to human sight, and the hurricane's workings delight the mind as an intricate and subtle problem in hydrodynamics.

A typical radar image shows the essential structure of a hurricane: a clear eye enclosed by a ring of clouds, which is in turn surrounded by inward spiraling bands of convection. These features are common to all tropical cyclones. Hurricanes take their circular shape from the orbits of air moving in gradient balance around the low atmospheric pressure at the vortex center.

A cross section of a hurricane in the radius-height plane shows that the primary swirling flow is maintained by a radial and vertical secondary circulation. Under the influence of friction, the counter-clockwise spiraling air near the surface converges toward the lower pressure and rises around the eye in cumulus clouds. Although the inflowing air loses angular momentum to the sea, it gains heat stored as water vapor. The vapor condenses in the updrafts, releasing the stored heat and causing intense precipitation. This latent heat release is the power source for the storm. At 2-6 km altitude the updrafts entrain additional air, which is replaced by horizontal convergence that supplies both mass and angular momentum needed to spin up the wind. The updrafts may rise as high as 16 kilometers where the air flows outward to the environment. The source of the storm's energy is heat drawn from the warm sea surface at 28-29°C and returned to the surrounding upper atmosphere at -70°C.

In the center of the hurricane is the cloud-free eye. The clouds that enclose the eye form the eyewall. These clouds draw air from the eye at low levels, causing descent, drying, and warming inside the eye. As the air warms, it becomes less dense so surface pressure must fall. The clouds also draw air inward from outside the eye, thus concentrating the counter-clockwise rotation and increasing the swirling winds. In a hurricane, the strongest winds are near the surface and just outside the eyewall. An unanticipated use of the new GPS sondes was direct observation of low-level wind jets at the inner edge of the eyewall.

Hurricane Opal's northward careen across the Gulf of Mexico, showing schematically the storm's rapid deepening as a result of interaction with a "digging" mid-latitude trough and it's passage over the deep pool of warm water in an eddy spun off from the Gulf Stream.

As the experience in Hurricane Opal of 1995 illustrates, intensity predictions compare poorly to the skill exhibited by track forecasts. Forecasters at the National Hurricane Center use a half dozen models to guide track forecasts. For intensity forecasting, only one model has been available (until recently) and that was based on climatology and persistence. These rudimentary tools reflect the challenge of forecasting intensity. It is a difficult problem and it is the focus of continuing research. AOML scientists recently developed a more complete statistical hurricane intensity prediction model that includes the effects of variations in sea-surface temperature and upper tropospheric turbulent transports. Although this model is available in real time and has had some success, there is still much to be learned before accurate intensity forecasts become routine.

A typical hurricane intensifies slowly, remaining in Category 1 or reaching 2 or even 3 before it runs ashore or drifts north out of the tropics. The strongest hurricanes, such as Andrew in 1992, intensify rapidly and go from Category 1 or 2 to Category 4 or 5 in just a day or two. The process that initiates rapid intensification may begin with atmospheric waves that form on the hurricane vortex at altitudes of greater than 12 kilometers (where the new jet will fly). These waves result from interaction with nearby low-pressure systems. The waves, through a complicated chain of cause and effect, intensify the cumulus clouds that are involved in the energy conversion mechanisms of a hurricane.

Shear of the environmental wind appears to be the mechanism by which atmospheric teleconnections modulate Atlantic hurricane activity and is generally thought to have a significant role in day-to-day intensity changes of individual storms. Airborne Doppler and reflectivity radar data collected during AOML research flights into Hurricane Olivia show that an environmental shear > 10 m s^-1 imposes a wavenumber-one structure on the eyewall convection. Individual cells form ~45 to the right of the down-shear direction, reach maturity with reflectivities > 45 dBZ on the left side of the shear vector, and have largely rained out by the time they detach from the eyewall as they advect back to the right side of the shear. Similar observationally-based synthesis point the way to a more accurate understanding of rapid intensification and ultimately to better forecasting of the intensity of the most dangerous hurricanes.

AOML research has produced accurate, timely maps of surface winds in hurricanes. These maps, which are based upon aircraft observations, are invaluable for preparation of forecasts before landfall and for damage assessment after landfall. The maps are now prepared routinely whenever a hurricane threatens land. A particularly important application of these products is the rapid identification of the most severely devastated areas; this enables disaster response teams to reach them quickly. The same techniques are now being applied for reconstruction of historical storms.

The Stepped Frequency Microwave Radiometer (SMRF) is a promising tool for direct measurement of surface winds. This instrument has flown aboard research aircraft for more than 20 years and has finally matured enough to transition to operations. It works by sensing the increase in apparent microwave brightness temperature of the water surface caused by more foam as the wind increases. Because rain in the air column between the airplane and the surface also produces a frequency-dependant contribution to the upwelling microwave radiation, the radiometer needs to look at several different frequencies to correct for the precipitation effect and to deduce rainfall rate as a byproduct.

Modification?

Can human intervention diminish the force of a hurricane? From the mid-1960s through the early 1980s NOAA actively pursued Project STORMFURY, a program of experimental hurricane modification. The general strategy was to reduce the intensity of the storm by cloud seeding. The seeding, it was argued, would stimulate the formation of a new eyewall that would surround the existing eyewall. The new eyewall would contract, strangling the old eyewall and reducing the intensity of the hurricane. However, research carried out at AOML showed clearly that these "concentric eyewalls" happened often in unmodified hurricanes, thus casting doubt on the seemingly positive results of seeding in earlier experimentation. Hurricane Luis provides an example of this behavior. Moreover, observations showed that hurricanes contain little of the supercooled water necessary for cloud seeding to work.

The American Meteorological Society policy statement on planned and inadvertent weather modification, dated October 2, 1998, indicates, "There is no sound physical hypothesis for the modification of hurricanes, tornadoes, or damaging winds in general, and no related scientific experimentation has been conducted in the past 20 years." In the absence of a sound hypothesis, no Federal agencies are presently doing, or planning, research on hurricane modification.

Some techniques besides seeding clouds that have been considered over the years include: cooling the ocean with cryogenic material or icebergs, retardation of surface evaporation with monomolecular films, changing the radiational balance in the hurricane environment by absorption of sunlight with carbon black, blowing the hurricane apart with hydrogen bombs, injecting air into the center with a huge maneuverable tube to raise the central pressure, and blowing the storm away from land with windmills. As carefully reasoned as some of these suggestions are, they all fall short of the mark because they fail to appreciate the size and power of tropical cyclones. For example, when hurricane Andrew struck South Florida in 1992, the eye and eyewall devastated a swath 20 miles wide. The heat energy released around the eye was 5,000 times the combined heat and electrical power generation of the Turkey Point nuclear power plant over which the eye passed. Better building codes, wiser land use, and more accurate forecasts seem prosaic compared with environmental mega engineering but they are a great deal cheaper and have overwhelmingly favorable cost-benefit ratios.

Tropical cyclones' impact society through damage to property, expense of safeguarding life and property, human mortality, and the expense of the research and forecasting enterprise. Is the cost of the enterprise justified by reductions in hurricane effects?

Peilke and Landsea's normalization of historical damage provides a reliable estimate of expected annual property loss, $5B. This figure is larger than the $1-3B in damage experienced during most years because a few extreme events figure disproportionally in the total. A single $100B event once a century increases the average annual toll by $1B, as might a single $10B event once a decade.

Preparation costs average between $0.5 and $1.0M per mile of shoreline warned. On average, warnings extend 300-400 miles and are raised three times each season. The total warning and emergency response cost is 3x350x$0.75= $787.5M per year.

During the last thirty hurricane seasons of the 20th century, 587 U.S. residents died in hurricanes, or 19.6 annually. In policy calculations, the economic impact of a human death is usually reckoned at $5ñ10M. A dollar value assigned to human deaths is only the beginning of their impact. One must be wary of arguments that might lead to sacrifice of lives to save money. With this qualification, mortality typically contributes 19.6x$7.5M = $146.8M to hurricanes' impact.

The cost of the forecast enterprise is dominated by satellites and the expense of launching them, figured at either a single geostationary satellite or two polar orbiters annually, about $190M. The annual budgets for hurricane work at the Hurricane Center, Hurricane Research Division, Geophysical Fluid Dynamics Laboratory, Environmental Monitoring Center, and other research and forecasting organizations appear to total less than $25M, including aircraft operations.

Thus, a reasonable estimate of the annual average total costs that hurricanes impose on the U.S. economy is just under $6M. What would this impact be without forecasting and emergency preparation?

The amount of damage prevented by warnings is poorly defined. Estimates range from 10% to 50%. The low estimate $500M is equivalent to moderate damage to 10,000 upper middle class suburban homes or destruction of a half dozen first-line commercial airliners. A great deal of expensive property is mobile, and protection of windows and doors raises the wind speed threshold for major damage substantially. The 10% figure seems low. Probably the best, though still conservative, estimate of prevented damage is 20% or $1.0M

During the 40 years between 1930 and 1959, inclusive, 70 people a year died in the U.S. hurricanes. Since 1950, the midpoint of this interval, population in the 109 coastal counties between Texas and Virginia had increased by a factor of 3.4. If the forecaster's art had not improved during the last century, expected annual mortality to this larger population from hurricanes would be 238, with an economic impact of $1785M. By this estimate the stakes of the annual hurricane game are $8.9B, of which $2.8B, or 31%, is prevented through an investment of < $0.22B in forecasting and research, for a greater than twelve to one favorable benefit to cost ratio. The prevented impact is dominated by reduced mortality.

Click here to look at Climate Processes.
Click here to go to the Hurricane Research Division's home page.

Last modified: Tuesday, 24-Jan-00 11:15:10 EDT.
Please report any problems to webmaster.