May 17, 2004 — Destructive winds, storm surge and heavy rains that produce flooding make hurricanes one of nature's most extreme hazards. As hurricane season approaches, Americans who increasingly populate hurricane-prone areas must prepare themselves should these storms strike their community. Fortunately, advances by the NOAA Hurricane Research Division, part of the NOAA Atlantic Oceanographic and Meteorological Laboratory, are helping to make this happen. Using an advanced new sampling method, known as “targeted observation” strategies, the division has been able to increase the accuracy of the models that the NOAA National Hurricane Center uses to issue its official hurricane track forecasts by an average of 25 percent (over a five-day forecast period).

Frank Marks, director of the NOAA HRD in Miami, Fla., said, “Targeted observation strategies support one of NOAA’s primary mission goals — to serve society’s needs for weather and water, since it will enable both coastal residents and emergency personnel to more effectively prepare for hurricanes — thus saving lives and minimizing property damage.”

With the 2004 Atlantic hurricane season officially starting on June 1, NOAA’s use of targeted observations should better prepare citizens for any potential threats posed by the 2004 hurricane season (NOAA released its 2004 Atlantic Hurricane Season Outlook on May 17, 2004).

Each year, an average of ten tropical storms develop over the Atlantic Ocean, Caribbean Sea and Gulf of Mexico — six of which become hurricanes. Although many of these hurricanes remain over the ocean and never impact land, an average of five hurricanes strike the U.S. coastline every three years — killing approximately 50 to 100 people and causing billions of dollars worth of damage. Of these, two are considered “major" or "intense" hurricanes (a category 3 or higher storm on the Saffir-Simpson Hurricane Scale).

HRD Targeted Observations
The NOAA HRD has determined optimal use of limited flight time aboard NOAA’s Gulfstream-IV jet as it gathers data for both research and forecast purposes. This technique, known as targeted observations, is the culmination of two decades of research and collaborations within NOAA (i.e., NOAA Research, the NOAA Aircraft Operation Center and the NOAA National Weather Service), improvements to hurricane models and instrument technology, advanced high speed computing capabilities and the acquisition of NOAA’s G-IV jet aircraft. Specifically, given the limited number of observations that can be obtained over an eight hour “surveillance” mission, NOAA HRD researchers concluded that sampling in all regions around a hurricane may not yield the best overall forecast. As a result, these researchers developed strategies for determining and sampling specific targets around the hurricane where observations are most likely to improve global model hurricane track forecasts.

The process of finding these "targets" involves locating regions where even slight inaccuracies in the model initial environmental conditions can result in potential large errors in subsequent hurricane track forecasts. These inaccuracies derive from the fact that every part of the atmosphere cannot be sampled all the time. Instead researchers must estimate the current atmospheric state based on the few observations they are able to acquire. Errors in these estimates are compounded in model simulations and increase at different rates over time. The basis of the targeted observation strategy is identifying and sampling those regions where small changes in atmospheric conditions result in the greatest variance in model simulations. The result is reduced error in the final track forecast. For example, the graphic (above left) shows the variability between the different model simulations when using slightly different initial environmental conditions. Where the greatest variance occurs (black dots near orange-yellow colors) is where the observations are targeted in the G-IV surveillance missions (Click on the NOAA image for a larger view of this graphic).

The History of Improved Hurricane Track Forecasting
In 1982, the HRD and other NOAA offices began to investigate whether hurricane track forecast models would provide better forecasts if the hurricane environment were observed more accurately. Such observations are possible with small, expendable dropwindsondes (see photograph to the right) that are released from aircraft and measure vertical profiles of wind, temperature, humidity and pressure as they fall from the aircraft to the ocean surface. From 1982 to 1996, 19 such sets of observations were collected by the NOAA Lockheed WP-3D Orion aircraft, operated by the NOAA Aircraft Operation Center. These data helped to reduce significantly errors in hurricane track forecasts in three high-performing models: the NOAA National Meteorological Center (now the NOAA National Centers for Environmental Prediction) global model, the HRD VICBAR model and the NOAA Geophysical Fluid Dynamics Laboratory hurricane model.

This success led to the development of a new, highly accurate GPS dropwindsonde and NOAA’s acquisition of the G-IV jet aircraft dedicated to obtain dropwindsonde measurements in regions around hurricanes. Concurrent with these advances, NCEP improved both its global and hurricane model systems by: improving the incorporation of data into models, advancing representation of physical processes, increasing model resolution and investing in high-speed computing to support the general development of state-of-the-art models. As a result, the G-IV is now deployed in the Atlantic basin for storms that threaten the coastal areas of the continental United States, Puerto Rico, the U.S. Virgin Islands and in the Pacific to cover hurricanes threatening southern California and Hawaii.

Results from the 2002 and 2003 Hurricane Seasons
During the 2002 and 2003 hurricane seasons, a total of 32 missions — mainly in major Hurricanes Isidore, Lili, and Isabel — utilized the targeted observation sampling strategy. By incorporating the data into NOAA high performance forecast models, the accuracy of NOAA hurricane track forecasts increased by an average of 25 percent (over a five-day forecast period). Consistently achieving this level of improvement in future hurricane track forecasts (over 36 to 60-hour forecast intervals) will greatly assist in the decision making process used to issue official hurricane watches and warnings. Analysis of the last five years of missions suggests that the GPS dropwindsondes are an especially useful tool in forecasting the track of strong (category 4 and 5 on the Saffir-Simpson Hurricane Scale) and rapidly intensifying hurricanes (i.e., hurricane Opal in 1995).

Benefits to Coastal Communities
As a hurricane approaches land, successful evacuations and damage mitigation measures, based on the most accurate landfall forecasts available, can greatly reduce loss of lives and property. Recently, it has been estimated to cost up to $1 million (depending on the amount of development) to prepare each mile of coastline in a hurricane warning area. Therefore, significant value is realized if forecasters can avoid calling for unnecessary preparation when a hurricane comes close to, but does not strike a large coastal community. Specifically, one complete G-IV flight of eight hours and the required dropwindsondes cost about $40,000 — much less than the average cost of preparing just one mile of coastline for a hurricane landfall.

Furthermore, according to the recently released 2004 NOAA Statistics Book, NOAA National Weather Service forecasts, warnings and associated emergency responses result in a $3 billion savings during a typical hurricane season. Two-thirds of this savings ($2 billion) is attributed to the reduction in hurricane-related deaths, and one-third of this savings ($1 billion) is attributed to a reduction in property-related damage because of preparedness actions.

With continued research into strategies to locate and sample specific targeted regions during hurricane surveillance missions, more experience with twice daily missions, and even better methods to assimilate hurricane data into NOAA models, greater forecast improvements can be expected in the near future.

Although we may never be able to prevent hurricanes from making landfall along the coastal United States, the new technology and more complete databases developed by the NOAA HRD will yield more improved hurricane track forecasts and give America's coastal residents the time needed to safely evacuate the predicted landfall area, thus minimizing lose of life and property.

Relevant Web Sites
NOAA Hurricane Research Division

NOAA National Hurricane Center

NOAA Aircraft Operations Center

2004 NOAA Statistics Book

Hurricane Frequently Asked Questions

NOAA CREATES “HISTORICAL HURRICANES MAPPING & ANALYSIS TOOL”

NOAA’s ATLANTIC OCEANOGRAPHIC AND METEOROLOGICAL LABORATORY

NOAA's HURRICANE RESEARCH

O'DARK THIRTY Observations of NOAA Aircraft Production Controller/WP-3D Flight Engineer

Media Contact:
Jana Goldman, NOAA Research, (301) 713-2483