The models used here have been developed using data through July, and thus apply to forecasts only for the balance of the 1997 storm season (August through November).

Forecast methodology for basin-wide activity

The two components of hurricane activity that we examine are the seasonal number of hurricanes (H) and seasonal number of intense hurricanes (I). To predict H we use ordinary least squares (OLS) regression to estimate the number of tropical-only hurricanes H_t (Elsner et al. 1996, Hess et al. 1995), to which we add a seasonal average number of baroclinically-influenced hurricanes H_b adjusted conditionally on the number of tropical-only hurricanes. The conditional adjustment amounts to removing the variance of H_b explained by H_t. This is accomplished by using a Poisson model to predict Hb from H _t. H _t and H _b are weakly negatively correlated.

The OLS regression uses three predictors, as originally suggested by Gray et al. (1993). This prediction model can be expressed as:

H = H _t + H _b

where H_t = b₀ + b₁x₁ + b₂x₂ + b₃x₃

and where the b's are coefficients on the predictors x.

The three predictors include (1) a 1-month forward extrapolation of the 30 mb zonal winds at 10N (Q30, a measure of the QBO phase), (2) Jun-Jul average rainfall anomalies (expressed in standard deviations) for the West Sahel region in Africa (RS), and (3) the past Aug-Nov average rainfall standardized anomalies in the Gulf of Guinea region (RG). The direction of the relationships are: positive correlations with the two rainfalls, and with the 30 mb zonal wind.

We also use a Poisson regression to estimate the number of intense hurricanes (3 or more on the Saffir/Simpson scale), as detailed in Elsner and Schmertmann (1993):

I = exp (g₀ + g₁x₁ + g₂x₂ + g₃x₃ + g₄x₄)

The predictors used to forecasts intense hurricanes are identical to those used to predict tropical-only hurricanes, except that the 50 mb zonal wind (Q50) is also included to provide a more complete description of the QBO state. It was determined, based on statistical tests, that the shear parameter used in the early December forecast model adds nothing significant to the early August model. The December 1993 issue of this Bulletin briefly summarizes the reasoning behind the beneficial use of the Poisson as compared to OLS regression, particularly for relatively infrequent events such as intense hurricanes.

Both the hurricane and intense hurricane models have been skill-evaluated using a hold-one-out cross-validation strategy. The correlation between actual and predicted number of hurricanes is 0.61 with a mean absolute error of 0.88 storms. For the number of intense hurricanes the correlation between actual and predicted is 0.84, with a mean absolute error of 0.74 storms.
 
 

Forecast methodology for sub-basin activity

In addition to basin-wide activity we predict activity in three sub-basins of the Atlantic including the Caribbean Sea, the Gulf of Mexico, and the Southeast U.S. Coast--Cape Hatteras south to Key West (Lehmil-ler et al. 1996). We use logistic regression to predict hurricane landfalls along the Southeast coast and the presence or absence of intense hurricanes in the Gulf and Caribbean. As with the approach taken for basin-wide activity, we express the sub-basin forecasts in terms of probabilities.

Logistic regression is a statistical model used to predict events in a yes/no framework by estimating coefficients for several predictor variables. Here we use a maximum likelihood technique to obtain the coefficients. A logical regression can be expressed as

exp (a₀ + a₁x₁ + a₂x₂ + ....... + a_px_p)

prob(yes) = -------------------------------------------------

1 + exp (a₀ + a₁x₁ + a₂x₂ + ....... + a_px_p)

where the a's are the coefficients on the p predictors x.

Predictors for these models include those used for basin-wide activity, with the addition of several others. For example, vertical shear is computed from July averaged winds at Miami (MIASHEAR) using 00 and 12 UTC rawinsonde observations. The shear is the Euclidean distance between the 700 and 200 mb wind components: [ (u₇₀₀-u₂₀₀)² + (v₇₀₀-v₂₀₀)² ]^½. Additionally, we use July averaged sea level pressures (SLPs) for stations along the U.S. East Coast. One predictor is the SLP for Cape Hatteras (SLPHAT) alone, and the other is the four-station mean July SLP using Miami, Charleston, Cape Hatteras and Boston (SLPMEAN). Normalized SLP difference between Darwin and Tahiti (SOI) is used as an expression of the state of the ENSO. Additionally, we include the zonal wind anomaly (ZWA) from upper tropospheric (12 km) zonal winds averaged over June and July (Gray et al. 1996).

Prediction error for the logistic models expressed as a cross-validated accuracy ratio ranges from 81.4% (compared with a climatology of 58.1%) for the U.S. Southeast coastal hurricane landfall model, to 78.3% (climatology of 52.2%) for the Gulf of Mexico intense hurricane model, to 82.6% (climatology of 52.2) for the Caribbean intense hurricane model. These skill estimates are based on data from 1950 to 1995.

Predictions for the 1997 Atlantic Hurricane Season

Predictor variables and regression coefficients for the models are shown in Table 1, and 1997 forecast probabilities are given in Tables 2 and 3. Despite the early season flurry of activity in the western Atlantic and northern Gulf of Mexico, the models are consistent with the early December forecast of near to below average activity for 1997. Specifically, the models call for two tropical-only hurricanes and two intense hurricanes yet for 1997. There is about a 45% chance of seeing less than two intense hurricanes. Two tropical cyclones have reached hurricane strength thus far (Bill and Danny), both of which were baroclinically-initiated (Kimberlain 1996). The average number of baroclin-ically-enhanced (non-tropical-only) hurricanes in a season is 2.9. We thus expect to see 3 more hurricanes in the Atlantic during 1997.

The forecast models for sub-basin activity are quite interesting. The probability of a landfall along the Southeast coast is less than 5%. The probability of a Caribbean intense hurricane is below the long-term average at 31%, but the probability of a Gulf of Mexico intense hurricane is high at 87%. In summary, though overall activity is expected to be below average, it is likely that a portion of the Gulf Coast will be threatened by a formidable hurricane this season.

Table 1. Model specifications for total Atlantic hurricane activity. The predictor symbols are defined in the text.

Predictor	1997  Value	Ht  (bi)	I  (gi)	IGulf  (ai)	ICarib  (ai)	HSE  (ai)
constant	-	3.61	1.05	1.26	0.59	-2.00
Q50	-1 m/s	-	0.0159	0.145	-	-
Q30	-4 m/s	0.0366	0.0083	-	0.0532	-
QDIFF	3 m/s	-	-	-	-	0.172
RS	-.57 sd	1.34	0.337	-	1.304	1.697
RG	-.41 sd	2.14	0.489	-	-	-
ZWA	1.5 sd	-	-	-0.409	-0.278	-
SOI	-1.9 sd	-	-	-0.752	-	-
MIAshear	.18 sd	-	-	-	-	1.326
SLPhat	-1.1 sd	-	-	-	-	1.697
SLPmean	-1.1 sd	-	-	-	-	-0.659

 
 
 
 
 
 
 
 
 

Table 3. Probabilities of a hurricane landfall along the Southeast coast and the probabilities of the presence or absence of an intense hurricane in the Gulf and Caribbean for 1997. H denotes hurricanes, I intense hurricanes. The climatological probabilities are shown.

	Sub-basin	1997 Probability	Climatol.  Proba-bility
H	SE Coast	.027	.488
I	Gulf	.873	.478
I	Caribbean	.314	.478

 
 
 
 
 
 
 

Acknowledgments: We are grateful to John A. Knaff and William Thorson for their help in data preparation. Partial support for this work came from the Risk Prediction Initiative and the National Science Foundation.

Elsner, J.B. and C.P. Schmertmann, 1993: Improving extended-range seasonal predictions of intense Atlantic hurricane activity. Wea. Forecasting, 8, 345-351.

Elsner, J.B., G.S. Lehmiller and T.B. Kimberlain, 1996: Objective classification of Atlantic basin hurricanes. Int. J. Climatol., 9, 2880-2889.

Gray, W.M., C.W. Landsea, P.W. Mielke, and K.J. Berry, 1993: Predicting Atlantic seasonal tropical cyclone activity by 1 August. Wea. Forecasting, 8, 73-86.

Gray, W.M., C.W. Landsea, P.W. Mielke, K.J. Berry and J.A. Knaff, 1996: LAD multiple linear regression forecasts of Atlantic tropical storm activity for 1996. Experimental Long-Lead Forecast Bulletin, 5 (no. 3), 29-31.

Hess, J.C., J.B. Elsner and N.E. LaSeur, 1995: Improving seasonal hurricane predictions for the Atlantic Basin. Wea. Forecasting, 10, 425-432.

Kimberlain, T.B., 1996: Baroclinically-Initiated Hurricanes of the North Atlantic Basin. MS Thesis. Available from the Florida State University, Dept. of Meteorology, Tallahassee, FL 32306-3034.

Lehmiller, G.S., T.B. Kimberlain and J.B. Elsner, 1997: Seasonal prediction models for North Atlantic basin hurricane location. Mon. Wea. Rev., 125, 1780-1791.