[Previous Article] [Next Article]

Predictions of North Atlantic Tropical Storm Activity for 1998,

and Verification of our 1997 Predictions

contributed by James Elsner¹ and Todd Kimberlain²

¹Department of Meteorology, Florida State University, Tallahassee, Florida

²Department of Atmospheric Science, Colorado State University, Fort Collins, Colorado

The two components of hurricane activity that we have studied are the seasonal numbers of hurricanes (H) and seasonal numbers of intense hurricanes (I). To predict the number of hurricanes we use ordinary least squares regression to estimate the number of tropical-only hurricanes, to which we add a constant seasonal average number of baroclinically-influenced hurricanes. The regression used for prediction of tropical type hurricanes uses five predictors, as described in Hess et al. (1995). These are the same as the basic, original predictors used by Gray et al. (1992). They include three QBO elements (50 and 30 mb zonal winds, and their shear) and two African rainfall predictors (Gulf of Guinea and West Sahel). The direction of the relation-ships with storm activity are: positive with the African rainfalls and the 50 mb zonal wind, negatively with the shear, and only weakly with the 30 mb zonal wind when all terms are present. The baroclinically-influenced hurricanes are not associated with these tropical-hurricane predictors; hence the constant adjustment to account for them in the total prediction.

To predict the number of intense hurricanes (3 or more on the Saffir/Simpson scale) we use a nonlinear Poisson regression with maximum likelihood criterion. Details of this model are given in Elsner and Schmertmann (1993). The same five predictors used for tropical-only hurricanes are used for intense hurricanes. The November 1993 issue of this Bulletin briefly summarizes the reasoning behind the beneficial use of the Poisson for intense hurricane prediction. The Poisson equation looks like:

₅

I = exp(_o + _i x_i) for 5 predictors

^I=1

Thus, in the Poisson the right hand side of a standard least squares regression equation is exponentiated.

In addition to basin-wide activity we are now predicting activity in four sub-basins of the Atlantic including the Caribbean Sea, the Gulf of Mexico, the Southeast U.S. Coast (Cape Hatteras south to Key West) and the Northeast U.S. coast (Cape Hatteras north to the Canadian border) (Lehmiller et al. 1997). We use logistic regression to predict hurricane landfalls along the Northeast and Southeast coasts 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 (₀ + ₁x₁ + ₂x₂ + ....... + _px_p)

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

1 + exp (₀ + ₁x₁ + ₂x₂ + ....... + _px_p)

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

Unfortunately at present, for the early December forecasts the only model with significant skill above climatology is the model for Caribbean hurricanes. The Caribbean hurricane model uses only the two rainfall parameters (Gulf of Guinea and Sahel) as predictors.

Skills of the two prediction models for basin-wide storm activity were estimated using cross-validation over 45 years. Results indicate a mean absolute error of 1.5 hurricanes for the least squares regression model, compared to 1.7 for climatology forecasts. For the Poisson model forecasts for the number of intense hur-ricanes the mean absolute error is 1.1, compared to 1.3 for climatology. Based on 1950-95 data, the prediction error for the logistic model for Caribbean hurricanes, expressed as a cross-validated accuracy ratio, is 80.4% compared with a climatology ratio of 58.7%.

Using the least squares regression, 0.31 tropical-only hurricanes are forecast for the 1998 season. To this we add the seasonal average number of baroclinic-ally-influenced hurricanes (2.85) and round the sum (3.16) to the nearest whole number to get a forecast of 3 hurricanes for 1998. Using the Poisson model, the intense hurricane model forecast is presented below in the form of estimated probabilities for each possible number of intense hurricanes, for 1997 (in retrospect) and for 1998.

No. of Intense Hurricanes	0	1	2	3	4	5+	Mean
1997 Forecast Probability	.110	.243	.268	.197	.109	.073	2.21
1998 Forecast Probability	.355	.368	.190	.066	.017	.004	1.04

As shown in the table above, the 1998 probabilities are considerably different from 1997 probabilities indicating a decreased likelihood of intense hurricane activity for 1998. More specifically, the Poisson model estimates that there is a 91% chance of observing less than three intense hurricanes during 1998. The logistic model indicates only a 17.6% chance of observing at least one hurricane for 1998 in the Caribbean portion of the North Atlantic basin.

As was the case this year, it is possible that the current warm ENSO will have an inhibiting influence on next year's North Atlantic hurricane activity. At the present time, however, there is too much uncertainty concerning ENSO's future state to guess at its influence. Most models are tending toward a neutral or cold ENSO by early summer, which may gradually diminish the negative effect on hurricane activity. These hurricane forecasts will be updated prior to the start of the 1998 hurricane season when we will have a much better idea of the ENSO conditions.

The following table shows the regression coefficients for the equation for the number of tropical-only hurricanes (least squares regression; first column), the number of intense hurricanes (Poisson model; second column), the number of hurricanes in the Caribbean Basin (logistic regression model; third column), and the current predictor data used to form all three of the 1998 forecasts (last column). The forecast models were estimated using 1950-96 data (except for tropical-only hurricanes, which use 1950-97).

Predictor Term in Equation	Regres Coef (Trop-Only Hurricanes)	Poisson Coef (Intense Hurricanes)	Logistic Coeff (Caribbean Hurricanes)	This year's Predictor Value
Constant	4.283	1.250	1.123	---
Aug-Nov '97 Gulf of Guinea rainfall	2.783	0.614	3.440	-0.52 sd
Aug-Sep '97 West Sahel rainfall	0.752	0.304	1.201	-0.73 sd
50mb zonal wind, 10ON, fcst Sep '98	0.018	0.034	---	-14 ms-1
30mb zonal wind, 10ON, fcst Sep '98	-0.015	-0.015	---	-31 ms-1
Absolute shear from two above winds	-0.129	-0.039	---	17 ms-1

Verification of 1997 Hurricane Season Forecasts

Using the above statistical models, the following forecasts were made for the 1997 hurricane season, compared with the observations and their percentage of the 1986-1996 average:

FOR 1997 SEASON	Forecast: Early Dec '96	Forecast: Early Aug '97	Observed (% avg)
Hurricanes	6	5	3 (60%)
Intense Hurr	2	2	1 (53%)

In 1997, both total hurricane activity and intense hurricane activity were below the long-term averages of 5.0 and 1.9, respectively. Although we did not antici-pate the degree to which the season would be below average, our models indicated that activity would be substantially less than the activity seen during the previous two years. The early December Poisson model estimated a 24.3% chance of exactly one intense hurricane for 1997. Our issuance in early December of only a 31% chance of a Caribbean hurricane (compared to a 1950-1995 climatology of approximately 50%) turned out be accurate as there were no hurricanes in the Caribbean. Taken together, our early December forecasts year are judged to be somewhat successful.

Our early August sub-basin forecasts also proved to be as accurate as can be expected based on our hindcast evaluations. Only the prediction of above normal probability of an intense hurricane in the Gulf of Mexico was wrong. It should be mentioned, however, that conditions were favorable for intense hurricane development over the extreme eastern North Pacific (EPAC) adjacent to the Gulf of Mexico, and several vigorous hurricanes did develop in this region. In particular, Pauline was EPAC hurricane that ap-proached to within a few hundred miles of the Gulf of Mexico as an intense hurricane. We hypothesize that the same environmental conditions that proved conducive for intense hurricane genesis in the EPAC were slightly displaced from the Gulf of Mexico; thus it is fortuitous that no intense hurricanes formed.

Overall, our location models performed as ex-pected. Each sub-basin forecast model has an expected accuracy of 80% (Lehmiller et al. 1997). In particular, the Caribbean sub-basin model worked well because west African rainfall was below average. For the 1997 season, three of our four models provided accurate predictions, yielding an overall observed accuracy of 75%, close to what can be expected.

Category	Prediction	Observation
Hurricane, Carib	no	no
Intense Hurr, Carib	no	no
Intense Hurr, Gulf	yes	no
Hurricane, Southeast	no	no

The 1997 season reversed the trend of the much above activity from the previous two years, especially in regard to tropical-only hurricanes (Kimberlain and Elsner 1998). It should be mentioned that our models indicated that this very active period would end with the 1997 season. In particular, only one of the three hurricanes of 1997 was tropical-only (Erika). Both hurricanes Bill and Danny were baroclinically-initiated. Bill originated from a tropical upper-tropospheric trough (TUTT), and Danny from a middle latitude mesoscale convective system (MCS). This is typical of a warm ENSO year, as it is likely that the current strong ENSO event acted to inhibit tropical-only hurricane formation during 1997. The dearth of hurricanes during 1997 is reminiscent of several years during the early 1990s, except for the very early start to the 1997 season: Two of the three hurricanes formed before August 1, resulting in exceptional inactivity during the peak of the season.

.

Acknowledgments: Partial support for this work came from the Risk Prediction Initiative (RPI) of the Bermuda Biological Station for Research (BBSR) and from the National Science Foundation (NSF).

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

Gray, W.M., C.W. Landsea, P.W. Mielke, and K.J. Berry, 1992: Predicting Atlantic seasonal hurricane activity 6-11 months in advance. Wea. Forecasting, 7, 440-455.

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. and J. B. Elsner, 1998: The 1995 and 1996 North Atlantic hurricane seasons: A return of the tropical-only hurricane. J. Climate, 11, in press.

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

[Previous Article] [Next Article]