NOAA duo create hurricane readers

MIAMI—Joe Griffin and James Franklin do not work with big systems, big budgets or
big computer issues in the federal government. But they are responsible for a computer
system that is expected to have a big scientific and economic impact.


For their collaboration on development of a new data gathering and analysis system, the
duo last year received the Bronze Medal from the Commerce Department.


Griffin, a computer scientist, and Franklin, a meteorologist, work in the National
Oceanic and Atmospheric Administration’s Hurricane Research Division on their
brainchild, the Hurricane Analysis Processing System (HAPS). NOAA’s Atlantic
Oceanographic and Meteorological Laboratory is housed in a five-story building with thick
concrete walls on Virginia Key, an island close to Miami.


The system they developed is small by any standards, comprising just two Unix
workstations running HP-UX. One sits on a table in a dark and narrow computer room in the
lab.


But the key part of the system is the second Unix workstation. Rack-mountable and
ruggedized to withstand extreme vibration, the workstation flies on a NOAA research jet
where it gathers and processes data from enormous tropical storms in the Atlantic Ocean.


HAPS provides data for long-term research and also transmits storm data to forecasters
at the National Hurricane Center. The Hurricane Research Division and the Air Force
Reserve’s Hurricane Hunter Squadron both run missions into storms for research and
storm forecasting.


Forecasters at the hurricane center enter the data into computer models, which they use
to predict the track of an advancing storm.


Although both Griffin and Franklin have written software for HAPS, it is Griffin, 44,
who handles all the system chores for HAPS. His work on the project includes everything
from the initial proposal through acquisition, installation, maintenance, backups and,
when there’s time, programming, he said.


Franklin relies on Griffin’s software expertise to accomplish his daily task:
researching ways to improve storm track forecasts. During the hurricane season—June
to November—he designs the hurricane research flights. He spends time the day before
each mission plotting where to fly and gather data.


“This is a research place,” Griffin said. “We focus on trying to
understand things in general. Part of our function here is to develop things that get used
eventually by the hurricane center” to improve the accuracy of hurricane forecasts.


HAPS data is doing just that, but more significantly, it is giving researchers new data
and analysis tools that could lead to breakthroughs in hurricane understanding.
Researchers are looking forward to a revolution in understanding hurricane motion,
Franklin said. Already, the data has helped improve the accuracy of computer models from
16 percent to 30 percent—as much as the accumulated improvement in operational
forecasts attained over the last 20 to 25 years, he said.


Most days Griffin works regular daytime hours, but calls come in the middle of the
night summoning him to solve a problem with HAPS or one of the other systems in the lab.
“When they’re out flying and there’s a problem, they call and I try to
debug it or make suggestions to fix things,” he said.


Working in computer systems development in a small group—the Hurricane Research
Center’s staff numbers about 35—“makes my job hard, and it makes it
interesting too,” Griffin says.


He spends about 25 percent of his day developing programs; the rest is spent on system
planning, operation, integration and troubleshooting. During this hurricane season, he
expects to fly a couple of missions with HAPS. The missions can be difficult because they
take him away from his wife and two small children for as long as a week. “My boss is
encouraging me to let other people run the stuff” now that HAPS is online, he said.


The missions can consume 17-hour days. The excitement of planning the missions and
flying them continues up to about the half-way point of the mission, Franklin said,
“and then I get really tired and I’m ready to go home.”


The missions are flown aboard the $46 million NOAA Gulfstream IV research jet. The HAPS
workstation, mounted in the fuselage of the jet, gathers data from several sources. But
the most important input comes from remote sensors that the crew drops overboard into the
storm. Software that Griffin wrote ingests data from the devices, called drop windsondes.


Each windsonde is a 3-inch diameter tube about 2 feet long. The instrument packages
fall at 33.3 feet per second through the atmosphere to the ocean surface while beaming
pressure, humidity and temperature data back to HAPS at half-second intervals.


Using Global Positioning System circuits, the windsondes provide the most accurate wind
speed readings yet gathered from hurricanes.


Doppler radar on the jet can give an overall view of a storm as the jet flies at safe
altitudes, Franklin said, but the view doesn’t reach the ocean surface. The
windsondes “work all the way down to the surface, which is not something we’ve
been able to do before,” he said. “We just don’t have direct measurements
of surface wind.” Before windsondes, the data came from trained observers who
estimated wind speed by looking down at the effects on the ocean.


Aboard the jet, the scientists run quality control on the HAPS data using software that
Franklin wrote. Then HAPS formats a data message and Griffin’s programming takes over
again to send the processed data by modem to a satellite.


Before the windsonde data systems, forecasters “weren’t able to see, except
on satellite picture, what the structure of the storm looked like,” Franklin said.


The duo recall the first data gathered with HAPS on the Gulfstream jet in November
1996. Claudette, a hurricane that threatened the coast in July 1997, provided the first
test under storm conditions. After that first run, Claudette, which had been headed for
Cape Cod, Mass., veered.


Franklin and Griffin share more than a sense of humor, however. Both have experienced
hurricanes as researchers and as natives of South Florida, where powerful storms often
strike without warning.


Griffin grew up in the Redlands area south of Miami where his grandfather homesteaded
in the 1800s. Franklin, 40, was born and raised in Miami. His interest in a career
studying the storms was sparked by seeing them as a youngster.


During the last 20 years, research has annually reduced forecast error by 1 percent.
The two scientists estimate that they can improve it by 20 percent with data gathered by
HAPS. Reducing a typical warning area by just 10 percent, about 34 miles, could save more
than $45 million in five years.


Reports on the system suggest the windsondes and HAPS could help spark a
“technical and scientific revolution in hurricane forecasting,” Griffin said.
   

inside gcn

  • artificial intelligence (ktsdesign/Shutterstock.com)

    Machine learning with limited data

Reader Comments

Please post your comments here. Comments are moderated, so they may not appear immediately after submitting. We will not post comments that we consider abusive or off-topic.

Please type the letters/numbers you see above

More from 1105 Public Sector Media Group