On Dynamic Predictables methods . . .

     Dynamic Predictables uses two primary methods to predict future climate events.  The first is based upon analytic methods and is discussed below.  The second, analogous, method is used when deemed appropriate to complement the first method but is not addressed here.

     Dynamic Predictables has developed proprietary software that analyzes and predicts future climatic behavior based on past historical data. The programs employ both a regional thermodynamic model together with a unique predictive algorithm to achieve a high degree of prediction accuracy.

     The thermodynamic model was developed initially to explain the results of a study on global circulation models done at SUNY-Stony Brook by S. Hameed, R.G. Currie, and H. LaGrone (Int. Jour. Climatology, 15, pp.852-871, 1995). The authors pointed out the oscillatory nature of world wide pressure variations as revealed in the simulation models (OS2, OS6; also actual physical data—private communication at the time with the authors). These oscillations are fundamental to an understanding of climatic variations on a sub-regional to continental basis. The oscillatory nature of these variations allows them to be used as broad based climate predictors. In addition, they can be subtracted from the historic data to yield a climate residual.

     The climate residual data are then analyzed to determine which components of this residual data are in fact predictable by the methods utilized. The program then combines both the thermodynamic model results (the primary predictive model) with those from the residual data (the secondary model) to yield an estimate of the future behavior of the climatic variable.   The primary thermodynamic model is also known as either the Seasonal Model or First Stage Model and its output is the basis for DynaPred's mid-month prior release across the 344 NCDC Climate Divisions of the continental US.

     Spatial resolution is site specific or aggregated regional based upon appropriate length (45 years or more monthly data) and reasonable quality weather observation records. Most climate analysis has been based on monthly time-step data, but time scales on the order of days can be used.   

     Oregon Climate Division #1 (Coastal) precipitation provides an example relating DynaPred's method to nature's observed elements in the early 2000s.  The prediction's leading dynamic factors are the strong seasonal in the primary model combined with high secondary model contributions from planet Earth's Chandler Wobble (near 15 months) and what has been called the Quasi-Triennial Oscillation (QTO, near 36 months) in equatorial regions.

May 2, 2002   

Gregg Suhler, Doug O'Brien at Dynamic Predictables  URL:  http://www.dynapred.com