ENSEMBLE FORECASTING AND PREDICTION OF MODEL PERFORMANCE

Principal Investigator: Sim D. Aberson
Collaborating scientist(s):
Morris Bender (NOAA/GFDL)
Robert Tuleya (NOAA/GFDL)
Stephen Lord (NOAA/NCEP)
Objective: Evaluation of methods to improve operational forecasts of tropical cyclone motion.
Rationale: Recent years have shown a slow and steady increase in skill of operational tropical cyclone track forecasts. A number of operational track forecast models have average skill approaching 40%, yet none of them performs well consistently, and a serious challenge for operational forecasters is to choose which of the various possibilities presented by the ensemble of different models is the most likely to occur. Not only should the guidance itself be presented to forecasters, but an estimate of the confidence which should be placed in that forecast as well. Thus, a method for the Prediction of Model Performance (POMP), and methods in ensemble forecasting are studied.
Method: A method to objectively estimate the confidence in track forecasts has been developed for the ensemble of models currently in use. Data from the synoptic situation in the environment of each tropical cyclone, climatological data, and past performance of each model are analyzed from the 1989 - 1995 hurricane seasons to allow a linear discriminant analysis to differentiate between forecasts which are expected to perform well from those expected to be poor, giving the forecasters an object estimate of confidence of the individual forecasts.

In addition, ensemble forecasting of hurricane tracks using both the HRD VICBAR model and the GFDL model developed at the Geophysical Fluid Dynamics Laboratory will be used during the 1996 hurricane season to quantify the value of using the bred growing modes from the NCEP ensemble forecasting program to assess the variability of forecast motion in the models. Assuming that the ensemble perturbations quantify the local amount of uncertainty in the model analysis and forecast, use of the ensembles for the purpose of adaptive observations, making observations in certain locations likely to have impact on the tropical cyclone, will be studied (Figure 1).


Accomplishment: The linear discriminant analysis for the HRD VICBAR model has shown that individual forecasts which are expected to be poor can be separated from those expected to be good, allowing forecasters to disregard the expected-poor information and improve their forecasts. Preliminary results from the ensemble forecasting study show that there may be value in using the ensembles developed at NCEP to find problem areas in the individual guidance models.

The following manuscript has been accepted for publication:

Aberson, S. D., 1996: The prediction of the performance of a nested barotropic hurricane track forecast model. Wea. Forecasting, (accepted).


Key references:
Ooyama, K. V., 1987: Scale controlled objective analysis. Mon. Wea. Rev., 115, 2479-2506.

DeMaria, M., S. D. Aberson, K. V. Ooyama, and S. J. Lord, 1992: A nested spectral model for hurricane track forecasting. Mon. Wea. Rev., 120, 1628-1643.

Aberson, S. D., and M. DeMaria, 1994: Verification of a nested barotropic hurricane track forecast model (VICBAR). Mon. Wea. Rev., 122, 2804-2815.


Last modified: 9/11/96