THE IMPACT OF
DROPWINDSONDES ON OPERATIONAL HURRICANE TRACK FORECAST MODELS
Principal Investigator:
James L. Franklin
Collaborating scientist(s):
Sim D. Aberson
Objective:
To evaluate the impact of special environmental observations on the objective
hurricane track forecast guidance available to the National Hurricane Center.
Background:
Recent research with dynamical and statistical models has shown that improved
prediction of hurricane motion requires more information on the hurricane's
large-scale environmental wind and height patterns than is typically available.
While satellites can provide some wind data using cloud-tracking procedures in
the upper and lower troposphere, the middle levels are frequently almost void of
observations. Operational models used for hurricane track forecasting need
information from throughout the troposphere to be initialized. As a result,
operational models may fail to predict important changes of storm speed or
direction due to inadequate initial data, rather than inadequate physics of the
prediction models.
HRD's "Synoptic Flow" experiments use dropwindsondes to gather vertical
profiles of wind, temperature, and humidity within 1,000 km of hurricanes. The
experiment is typically conducted over the data-sparse oceanic regions of the
western Atlantic or Gulf of Mexico roughly 48-72 hours before the projected
landfall of a mature hurricane on the coast of the United States. The
dropwindsondes define the hurricane's surrounding large-scale flow, particularly
in the 400-700 mb middle tropospheric layer - the layer most directly related to
tropical cyclone motion.
Analysis Methodology:
From 1982-1996, the two NOAA WP-3D aircraft were equipped with Omega
dropwindsonde (ODW) systems. Using the historical sample of Synoptic Flow
experiments conducted by HRD from 1982-1993, an assessment was made of the
impact of ODW data on three dynamical hurricane track models providing
operational guidance to the forecasters at NHC: HRD's barotropic VICBAR model,
NCEP's global spectral model, and the GFDL hurricane model. The overall impact
of the ODWs was obtained by averaging the forecasts from the three models to
form a "consensus" forecast (CON3). CON3 forecasts were obtained by running the
forecast models with and then without the ODW observations.
Accomplishments:
In this historical sample of cases, the ODW observations accounted for
statistically significant reductions in 12-60 h track errors in the CON3
forecasts. The error reductions, which ranged from 16%-30%, are at least as
large as the accumulated improvement in operational forecasts attained over the
last 20-25 years. (FIGURE 1) shows the
mean forecast
errors with and without the ODW data, relative to forecast errors from a
climatology and persistence model (CLIPER). This work is described in the
following publication:
Burpee, R. W., J. L. Franklin, S. J. Lord, R. E. Tuleya, and S.
D. Aberson, 1996: The impact of Omega dropwindsondes on operational
hurricane track forecast models. Bull. Amer. Meteor. Soc., 77,
925-933.
More recently, the Director of NHC requested additional Synoptic Flow
experiments for Hurricane Felix on 15 August 1995, and for Hurricane Edouard on
30 August 1996. Both of these were single-aircraft experiments with a limited
number of ODWs. In the Felix case, the ODWs were responsible for
small improvements in the GFDL model at 24 h, after which the ODWs slightly
degraded the GFDL forecast. There is some early indication that the ODWs may
have helped reduce the scatter in the numerical forecast guidance for Edouard.
Plans:
In 1996, testing began on a completely new digital design of dropwindsonde based
on GPS navigation. In 1997, the GPS dropwindsonde replaces the ODW systems on
both WP-3D aircraft. Evaluation of the new GPS dropwindsonde is described in a
separate project report. In 1997 NOAA will add a new Gulfstream IV (G-IV) jet
aircraft for operational dropwindsonde flights in the environments of potential
landfalling hurricanes. HRD has been assigned responsibility for designing the
flight tracks for these operational missions, and for providing real-time
quality-control of the dropwindsonde data on the G-IV. We will continue to
evaluate the impact of the dropwindsonde data as additional missions are
conducted.
Related reference:
Franklin, J. L., and M. DeMaria, 1992: The impact of Omega
dropwindsonde observations on barotropic hurricane track forecasts. Mon.
Wea. Rev., 120, 381-391.
Click here to return to the HRD
project overview page.
Click here to
return to the Advance
Short-Term Warning & Forcast Services page.
Last modified: 23 June 1997.
Please report any problems to the HRD webmaster.