contributed by John A. Knaff¹ and Christopher W. Landsea²

To provide a baseline of skill in seasonal ENSO forecasting, a multiple regression has been used to take best advantage of CLImatology, PERsistence and trend of initial conditions - the ENSO-CLIPER (Knaff and Landsea 1997). This replace simple persistence as a skill threshold. "Skill" is then redefined as the ability to outforecast theENSO-CLIPER - a more difficult task.

This statistical prediction method is based entirely on the optimal combination of persistence, month-to-month trend of initial conditions and climatology. Multiple least squares regression is employed to test a total of fourteen possible predictors for the selection of the best predictors, based upon 1950-1994 developmental data. A range of zero to four predictors were chosen in developing twelve separate regression models, developed separately for each initial calendar month. The predictands to be forecast include the Southern Oscillation Index (SOI) and the Niño 1+2, Niño 3, Niño 4 and Niño 3.4 SST indices for the equatorial eastern and central Pacific at lead times ranging from zero seasons (0-2 months) through seven seasons (21-23 months). Though hindcast ability is strongly seasonally dependent, substantial improvement is achieved over simple persistence wherein largest gains occur for two to seven season (6 to 23 months) lead times. The ENSO-CLIPER model thus not only offers a baseline "no-skill" forecast of ENSO variability, but a practical forecast based upon the CLIPER premise.

The regression design called leaps and bounds (IMSL 1987) is used to develop optimal models (the best subsets of a prescribed number of predictors). Predictors include 1, 3 or 5 month averages of initial predictor anomalies as well as their recent trends. Predictors are the predictands themselves at earlier times. Some limits on predictor selection were imposed to reduce overfitting (Aczel 1989). The skill is degraded from dependent sample results to reflect estimated independent forecast skill following Davis (1979) and Shapiro (1984). Final skill estimates reflect levels comparable to those of more sophisticated statistical and dynamical models. More details about the ENSO-CLIPER model, including its skill and its predictor selection rules, are given in the June 1997 issue of this Bulletin (p. 55). A copy of Knaff and Landsea (1997) as well as future monthly ENSO-CLIPER forecasts are available at the Web site:http://tropical.atmos.colostate.edu/~knaff. The program to run ENSO-CLIPER is also available upon request.

Employing the chosen predictors in the ENSO-CLIPER model on a 1 June 1998 initialization date yields forecasts for Jun-Jul-Aug 1998 (zero season lead) out through Mar-Apr-May 2000 (seven season lead). Results for just the Niño 3.4 region SST and the SOI are shown in Fig. 1 and Fig 2.. These forecasts indicate that the end of the strong El Niño event may occur during Jun-Jul-Aug 1998 (-0.16 ^oC in Niño 3.4 and -0.12 SOI), possibly followed directly by a moderate La Niña event peaking this Northern Hemisphere autumn and winter (-0.84 ^oC Niño 3.4 SSTs and +0.97 SOI, both during Dec-Jan-Feb 1998-99). For even longer lead times, near average ENSO conditions are forecasted from Jun-Jul-Aug 1999 through Mar-Apr-May 2000. For the short leads, these forecasts are based upon an about equal contribution from the initial (Mar-Apr-May 1998) conditions as well as the trends of the anomalies. The moderate La Niña event forecasted is due to the large cooling trends outweighing the current warm conditions that exist. The near average values predicted at the long leads are due to a near cancellation from these two factors, weighted oppositely in sign to that of the short term leads.

ENSO-CLIPER predictions made over the last several seasons have verified reasonable well (Table 1). While the strong El Niño event during Sep-Oct-Nov 1997 and Dec-Jan-Feb 1997-98 was predicted several months in advance, the magnitude was underestimated except at the zero season lead. Additionally, the cooling likely to begin occurring in Jun-Jul-Aug 1998 has been suggested to happen consistently as early as June 1997.

Acknowledgments: The authors wish to thank William Gray, Tony Barnston, John Sheaffer, Dave Enfield Dennis Mayer, Barb Brumit, Amie Hedstrom, Bill Thorson and Rick Taft for all their help and comments concerning this work. The lead author is being supported by NOAA under contract NA37RJ0202 (William Gray, PI) with supplemental support given by NSF under contracts ATM-9417563 (William Gray, PI). The second author was funded through the 1995-96 NOAA Postdoctoral Program in Climate and Global Change.

References:

Aczel, A. D., 1989: Complete Business Statistics. RichardD. Irwin, Inc., 1056 pp.

Davis, R. E., 1979: A search for short range climate productivity. Dyn. Atmos. Oceans, 3, 485--497.

IMSL, 1987: FORTRAN subroutines for statistical analysis. International Mathematical & Statistical FORTRAN Library, 1232 pp.

Knaff, J. A. and C. W. Landsea, 1997: An El Niño-Southern Oscillation CLImatology and PERsistence (CLIPER) Forecasting Scheme. Wea. Forecasting, 12, 633-652.

Shapiro, L. J., 1984: Sampling errors in statistical models of tropical cyclone motion: A comparison of predictor screening and EOF techniques. Mon. Wea. Rev., 112, 1378--1388.

Figure Captions

Figure 1: Forecast of Niño 3.4 using data available though 1 June 1998. Forecasts are valid for June-Aug. (JJA) 1998, Sept.-Nov. (SON) 1998, Dec.-Feb. (DJF) 1998-99. Mar.-May (MAM) 1999, JJA 1999, SON 1999, DJF 1999-2000, and MAM 2000. Actual numerical forecasts values for these times are shown on the figure along with the estimated RMSE in vertical bars. These anomalies are based on a 1950-79 mean.

Figure2: Same as figure 1 except for SOI.