Notice of Intent to Change the NCEP GFDL Hurricane Forecast System
Brief description of change(s):
Major upgrades to the GFDL Hurricane Prediction System have been made
and tested for implementation for the ?03 hurricane Season. The most
significant upgrades are the replacement of the deep convection and
boundary layer physics with the physics from the GFS along with an
increase in vertical levels. An improved initialization of the mass
field is also being implemented in addition to an expansion of the
ocean coupling and an initialization of the Gulf Stream. A list of all
upgrades to the GFDL system can be found at:
http://www.emc.ncep.noaa.gov/gfdl_2003_upgrades.pdf
Reason for Change(s):
GFDL physics has not been upgraded for many years, e.g. model deep
convection was the moist convective adjustment scheme (Kurihara) The
implementation of the GFS deep convection, e.g. SAS, along with an
increased vertical resolution from 18 to 42 levels and the
implementation of the GFS boundary layer will provide a more realistic
interaction of the hurricane core circulation with the environment.
These upgrades are important to improve overall track forecasts as well
as improve intensity forecasts in highly sheared environments and
scenarios conducive to intensification. Also, upgrades to the GFDL
ocean model are being implemented to expand the model air-sea coupling
and improve intensity (and track) forecasts in cases where SST?s
changes are important.
These upgrades are a part of a plan to improve the overall performance
of the GFDL and to establish the GFDL as a benchmark in track
performance in anticipation of transition to the Hurricane WRF (HWRF)
system. The HWRF is expected to become operational in 2006.
Schedule for change:
Final testing completed [date]: March 2003
Expected implementation [date]: May 15, 2003
Description of testing:
The upgrades were extensively tested for both the 2001 and 2002
hurricane seasons. The results of the reruns and testing can be found
at:
Statistical summaries [give Web site]:
http://www.emc.ncep.noaa.gov/gfdl_2003_upgrades.pdf
Forecast examples [give Web site]:
http://www.emc.ncep.noaa.gov/gfdl_2003_upgrades.pdf
Anticipated impact on forecasts:
Improvements to both the track and intensity forecasts were anticipated
with the upgraded GFS physics and increased vertical resolution. These
upgrades are necessary to simulate the important role of the sub cloud
layer in providing the deep convection with an improved vertical flux of
heat and moisture. The improved representation of deep convection is
essential to simulate and maintain the vertical structure of the
hurricane through evolutionary stages of the interaction of the vortex
with the environment.
Summary of results:
In 2001: track error was reduced ~15% for the 2-3 day forecasts in
both the Atlantic and Pacific Basins.
In 2002: improvements in the Atlantic track forecasts in the 3-5 day
forecasts, ranged from 8% at day 3 to nearly 35% at 4-5 days.
For both seasons:
- elimination of the long standing northward bias was eliminated in EPAC
forecasts along the
Mexican coastline
- skill for intensity forecasts was noted for particular case studies
Further documentation available [give Web site]:
http://www.emc.ncep.noaa.gov/gfdl_2003_upgrades.pdf
Human point of contact for further information: Morris.Bender@xxxxxxxx
Field Evaluation:
Evaluation of all results were carried out by TPC hurricane
specialists.
Approvals:
NCEP Director: L. Uccellini Date: March 31, 2003