Hurricane Forecasting in the United States
An Information Statement of the American Meteorological Society
(Adopted by AMS Council on 14 March 2007) Bull. Amer. Met. Soc., 88
The devastating hurricane seasons of recent years have highlighted the importance of hurricane observing, forecasting, preparedness, and assessment of longer-term risks. Without advances in all of these areas, damages will increase, and the drains on government budgets and our communities could become unsustainable. The goal of the meteorological community is to reduce the loss of life and vulnerability to hurricane-related hazards through improved observing and forecasting.
Hurricanes belong to a class of storms referred to generically as tropical cyclones. Tropical cyclones are warm-core, non-frontal, large-scale cyclones, originating over tropical or subtropical waters, with organized deep convection (thunderstorm activity) and a closed surface wind circulation about a well-defined center. The term hurricane is applied when a tropical cyclone’s maximum sustained surface wind (the 1-min mean wind at a height of 33 ft or 10 m) is at least 64 kt (74 mph or 33 m s-1). In the remainder of this statement, the terms hurricane and tropical cyclone will be used interchangeably.
The extensive damage and loss of life caused by tropical cyclones result from four primary hazards: storm surge (a local abnormal rise in coastal sea level), strong winds, heavy rains and their associated freshwater flooding, and hurricane-spawned tornadoes. While hurricanes are most hazardous in coastal regions, their weakening, moisture-laden circulations can produce extensive, damaging floods and tornadoes hundreds of miles inland days after the winds have subsided.
The largest losses of life in hurricanes result from storm surges, which are rapid rises of sea water and associated flooding that occur when hurricanes make landfall. Storm surge height can exceed 20 ft (6 m) when strong hurricanes strike a coastline with shallow water offshore. In recent decades, large losses of life due to storm surge had become less frequent; however, the rapid growth of U. S. coastal populations and related infrastructure, and the increasing complexity of evacuation have led to increased vulnerability of coastal communities. This vulnerability was tragically illustrated by 2005’s Hurricane Katrina, which by some estimates killed more than 1700 people in Louisiana and Mississippi. Improved building codes in hurricane-prone regions have greatly reduced fatalities from wind damage, but many fatalities continue to result from tropical cyclone–induced inland flooding.
Tropical cyclone-caused damage in the continental United States averages $10–11 billion (in 2007 dollars) annually. This cost is increasing due to growing population, wealth, and development in the vulnerable coastal zones.
Because tropical cyclones spend much of their existence over remote oceanic areas, they are predominantly observed using remote-sensing techniques, including satellites and radars. The advent of operationally available geostationary satellite surveillance in 1970 has greatly improved our ability to detect tropical cyclones, some of which would inevitably have otherwise escaped notice. However, traditional satellite-derived position estimates may have errors of tens of miles when the circulation center is obscured by clouds, and satellite-derived intensity estimates may have errors of tens of knots. Passive microwave imagery can provide radar-like depictions of storm structure to aid in the analysis of tropical cyclones, but these data are neither as timely nor as frequent as they need to be. Scatterometers such as the QuikSCAT provide surface wind measurements over large oceanic swaths, but these measurements are significantly degraded by rain and generally cannot resolve the maximum winds in moderate to strong hurricanes. Unfortunately, as discussed in the 2007 report of the National Research Council Earth Science and Applications from Space: National Imperatives for the Next Decade and Beyond, (available online at http://books.nap.edu/catalog/11820.html), even this recent progress in remote sensing of tropical cyclones from space may not be maintained.
When tropical cyclones threaten land areas, Air Force Reserve C-130 and NOAA P-3 reconnaissance aircraft are used to gather in situ observations of storm location, central pressure, and winds. Surface winds are generally estimated as a fraction of the winds observed at flight level, or through direct measurement by GPS dropwindsondes. Continuous surface wind measurements along the flight track can be obtained from the stepped-frequency microwave radiometer (SFMR), which is already installed on the P-3s and will soon to be added to the C-130 fleet. The SFMR is expected to significantly improve the accuracy of hurricane intensity estimates (currently believed to be near 10%) as well as analysis of the extent of damaging winds.
During the past two decades there has been a tremendous increase in the quality and quantity of remotely sensed wind and thermodynamic data over the formerly data-sparse oceans; however, the hurricane and its environment are still under-sampled. When tropical cyclones threaten land, steering currents surrounding the storm can be measured by dropwindsondes, released from the Gulfstream-IV jet, resulting in track forecast improvements in some models of up to 15%.
a. Track forecasting
The tropical cyclone track forecast is the foundation of the forecast and warning process; accurate prediction of other storm parameters is of little value if the track forecast is greatly in error. Fortunately, track forecasts have been steadily improving. For the 5-yr period 2001–2005, the NOAA National Hurricane Center (NHC) track forecast errors averaged 65 nautical miles (120 km) for the 24-h forecast and 118 nautical miles (269 km) for the 48‑h forecast. NHC track forecast errors are now roughly half of what they were in 1990. During the two very active Atlantic hurricane seasons of 2004 and 2005, NHC 12–72 h forecast accuracy was at or near record levels. As a result of these improvements, official track forecasts were recently extended from three days out to five days, and the current 5‑day forecasts are as accurate as the 3-day forecasts were 15 years ago. Much of this progress over the past two decades is attributable to improvements in global and regional numerical models, the explosion in the amount of remotely sensed data over the oceans, and advances in the assimilation of all data types.
In spite of improvements in track forecast accuracy, forecast uncertainty still necessitates the issuance of hurricane warnings for relatively large coastal areas. During the period 2000–2005, the average length of coastline under a hurricane warning in the United States was 275 n mi (510 km). This represents a substantial decrease from the preceding decade, during which the average warning length was 395 n mi (730 km), and appears to represent a reversal of an earlier trend toward larger warning areas. Even so, only about one‑quarter of an average hurricane warning area experiences hurricane conditions. Continued improvements in track forecast accuracy could decrease the amount of “over warning,” but this must be carefully balanced with the priority to safeguard lives and property. The continuing increase in coastal population and development, without concurrent expansion of evacuation routes, has resulted in longer evacuation clearance times. As a result, the NHC has put greater emphasis on enhancing the lead time of warnings, increasing lead time by more than 50% since the 1960s. However, Hurricane Katrina reminds us that even with excellent track forecasts, the United States remains vulnerable to large losses of life from hurricanes.
b. Intensity, size, and structure forecasting
For the 5-yr period 2001–2005, NHC intensity forecast errors averaged 10 knots (5 m s-1) for the 24‑hour forecast and 14 knots (7 m s-1) for the 48‑hour forecast. In contrast to the improvements noted above for track, mean intensity errors have not changed significantly during the past 30 years. Furthermore, these average statistics obscure the large errors that typically occur when storms strengthen (or weaken) rapidly. Unexpected rapid increases in strength close to landfall can result in communities being under prepared. Changes in the distribution of damaging winds, particularly those that result from concentric eyewall cycles or extratropical transition are extremely difficult to anticipate. These changes in storm structure can significantly affect storm impacts. The inability to anticipate these changes is of great concern.
c. Rainfall forecasting
Accurate prediction of rainfall from landfalling tropical cyclones remains elusive. Complicating factors include terrain effects, the low inherent predictability of convective rainfall, and uncertainties in the track and structure forecasts. While mesoscale and neural network statistical forecast models are showing promise for improvements in precipitation forecasts, operational objective techniques are still in their infancy. Recent advances in validation techniques are an important first step in improving rainfall prediction.
d. Storm surge hazard
Given a sufficiently accurate forecast of the hurricane’s track and surface wind structure, as well as accurate topographic and bathymetric data, numerical models can accurately predict storm surge inundation. However, because of uncertainty in tropical cyclone forecasts, evacuation decisions cannot be made based on individual runs of a storm surge model. Instead, a “maximum envelope of water” is estimated from a suite of possible tracks. An alternative to this traditional approach, presently under development, uses ensembles to generate a probabilistic assessment of surge risk. The action of waves, a significant contributor to surge damage, is not as well understood as it needs to be.
e. Forecast responsibility within the National Weather Service
The National Hurricane Center has overall responsibility for issuing track and intensity forecasts for Atlantic and eastern North Pacific tropical cyclones; the Central Pacific Hurricane Center is responsible for central Pacific cyclones. However, the Nation’s hurricane warning program is a collaborative one in which other national and local National Weather Service (NWS) offices play an important role. Forecasters at local NWS Weather Forecast Offices use their knowledge of the regional meteorology, oceanography, topography, and population demographics to provide specific information about the expected wind speed and direction, rainfall, storm surge, and hurricane-spawned tornadoes to local emergency management offices and the general public. In addition, the Hydrometeorological Prediction Center (HPC) provides guidance on the potential for heavy rainfall both to the local NWS offices and directly to emergency management agencies, while guidance of severe local weather, such as tornadoes, is provided by the Storm Prediction Center (SPC). The River Forecast Center (RFC) utilizes the hurricane rainfall predictions within the context of their hydrodynamical models to forecast river heights and the possibility of flooding.
f. Seasonal hurricane forecasting
Predictions of seasonal hurricane activity in the Atlantic basin have demonstrated forecast skill since the mid-1980s; however it is not yet possible to confidently predict seasonal activity for smaller regions or landfalls. Forecasts of seasonal hurricane activity issued at the start of the tropical cyclone season have been able to anticipate a substantial amount of the variance of hurricane frequencies, duration, and intensities. However, these forecasts have diminished skill when issued several months before the beginning of the season, primarily because of the low amount of skill in predicting the El Niño–Southern Oscillation and the associated changes in the atmosphere and ocean at long lead times. For the climate change and hurricane issue, refer to the World Meteorological Organization (WMO) statement on this topic recently endorsed by AMS. (http://www.ametsoc.org/policy/amsstatements_inforce.html).
Challenges for the next decade
A number of factors are presently limiting the accuracy or utility of tropical cyclone forecasts. Improvements in intensity forecasts would benefit from observational systems (e.g., Doppler radars) that can effectively observe the details of the storm core, improved data assimilation techniques, improved modeling capabilities on both the global and regional scale, as well as improved prediction of ocean–atmosphere interactions. Currently, official tropical cyclone forecasts of track, intensity, and structure are not issued until a tropical cyclone exists, thus providing little advance notice for tropical cyclones that form near the coast and quickly head toward land. Fortunately, recent theoretical and modeling advances show promise for advances in genesis forecasting in the next few years. More effective remote sensing of ocean surface winds is essential to improve the analysis, forecasting, and initial detection of incipient tropical cyclones. Better quantification of forecast uncertainty would enable users to make more informed decisions. Ensemble forecasts (multiple simulations with different forecast models and/or a model with slightly different starting conditions) provide a starting point for quantifying forecast uncertainty. The new NHC wind speed probability product is a significant step in that direction.
The National Research Council Report noted above indicates significant challenges associated with difficulties in the nation’s Earth satellite systems, including the NOAA operational satellites and the NASA research satellites. A significant part of the improvement in hurricane track forecasts over the past 20 years had been the greater availability of satellite observations and their effective use in numerical models. It is critical to continue and improve the satellite observations of winds, temperatures, water vapor, sea surface temperature and other variables in the vicinity of tropical cyclones.
The ultimate goal of hurricane monitoring and forecasting is to prevent loss of life and to reduce vulnerability to hurricane-related hazards. The 2005 Atlantic season was a powerful reminder of that vulnerability. Evacuation times for some communities exceed expectations for present and projected forecast accuracy, and some of the potentially most difficult evacuation problems have not been tested for generations. Thus, while the primary focus of the meteorological community must always be on a concerted scientific effort to improve forecasts, greater resilience to hurricanes requires effective engagement of other disciplines including engineering, ecology, biology, the social, behavioral, and economic sciences, and public policy. Further development of community awareness and preparedness programs through a comprehensive framework is also essential for ensuring public understanding of hurricane threat and the ability to take appropriate action to mitigate the loss of life and property that links the entire process — from data collection to forecast to communication of the societal impact.
[This statement is considered in force until March 2012 unless superseded by a new statement issued by the AMS Council before this date.]