This post is also available in: עברית (Hebrew)
When the TV weatherman predicts sunshine and it rains, people get wet. When meteorologists are wrong about a hurricane, people can die. While forecasters are now able to predict where a hurricane will land, they have not been able to reliably determine how strong a storm will be.
However, Navy researchers say they have developed a computer model that can predict the intensity of hurricanes up to five days in advance. The Coupled Ocean/Atmosphere Mesoscale Prediction System-Tropical Cyclone, or COAMPS-TC, is now being used by the Navy’s Fleet Numerical Meteorology and Oceanography Centeras well as the National Oceanic and Atmospheric Administration. It was developed by the Naval Research Laboratory, Office of Naval Research and several universities.
Civilians will benefit from better storm intensity forecasts, but so will the Navy, which has a weather gap in the Pacific that COAMPS-TC can help fill.
“Unlike the Atlantic Ocean, there are no operational tropical cyclone reconnaissance flights in the Western Pacific basin, an area of strategic importance for the U.S. Navy,” said Simon Chang, a scientist with the Naval Research Laboratory.
According to Defense News COAMPS-TC, conceived in 2006, is a mesoscale model, which means it is designed for regional weather phenomena of 5 to 1,000 kilometers in length — such as hurricanes. It is a supplement to the basic COAMPS model, which has been around for about 15 years but could only predict a storm’s track. What has been missing is basic data needed to model the intensity of hurricanes (and cyclones and typhoons, which are the same thing).
“Historically, measurements at very high wind speeds near the ocean surface were almost nonexistent, because it’s such a harsh environment,” said Ronald Ferek, ONR’s program officer for COAMPS-TC. Without data of high-wind-speed conditions, meteorologists had to extrapolate from low-wind-speed models.
“They were making incorrect assumptions about the transfers of heat, moisture and momentum between the ocean and the atmosphere at very high wind speeds where they had never been observed before,” Ferek said.
But researchers gathered measurements from surface buoys, ships, land stations, commercial airliners, weather reconnaissance aircraft, drone and satellites. Chang said this enabled COAMPS-TC to better model the physics of a hurricane’s boundary layer (the area from the earth’s surface to 1 kilometer high, where turbulence dominates), stratus and convective cloud conditions, and solar and terrestrial radiation.
Where previously forecasters had relied on statistical models that used historical data to predict the intensity of a storm, COAMPS-TC is a dynamic model that can feed new data to reflect changing weather conditions as they happen. It achieved remarkable results during Hurricane Irene in 2011, producing windspeed errors of less than five knots, compared with a typical error of plus or minus 20 knots.
However, the model doesn’t always achieve such striking results because so many hurricane dynamics remain obscure. When used to predict the intensity of thousands of storms between 2010 and 2012, the mean average error of the predicted maximum wind was 12 knots at one day out, 14 knots at three days, and 17 knots at five days. But this is still an improvement over older forecasting, and it will only get better as researchers capture improved data.
While hurricanes have relinquished some of their whirling secrets, they stubbornly cling to others. Now that meteorologists have a better idea of surface conditions that determine the strength of a hurricane, they are discovering that they don’t know enough about crucial weather dynamics at high altitudes above the top of a storm.
The issue is one of aerodynamics as much as atmospherics; it is difficult to figure out conditions at 30,000 feet to 60,000 feet when WP-3D and WC-130J weather reconnaissance aircraft can’t even reach 30,000 feet. This is where the Global Hawks come in, as part of NASA’s Hurricane and Severe Storm Sentinel project. The high-altitude drones were used last year to fly above hurricanes Leslie and Nadine and release dropsondes (parachute-equipped instrument packages) that floated down through the storms.
Unfortunately for places like the American Midwest, mesoscale models such as COAMPS-TC aren’t effective at predicting tornadoes, such as those that recently devastated Oklahoma.
“Mesoscale models do not resolve the scales of tornadoes, only the environmental conditions that may be favorable to forming these storms,” Chang said. “For tornadoes, the National Weather Service warns on detection, not on forecast. Warn-on-forecast is a decade or more away.”
But Ferek predicts that COAMPS-TC is the beginning of a radical improvement in storm intensity forecasting.
“I think that in three to five years, the way we forecast hurricanes and cyclones will be totally different,” he said. “There will be a lot more detail and a lot more confidence in making predictions about the intensity of a storm.”