Extended Abstract (300K)3.5
Statistical Turbulence Prediction
Walter C. Kolczynski Jr., Penn State University, University Park, PA; and S. E. Haupt
We take a statistical approach to predicting convectively-induced turbulence by training models on each of the variables individually. We then construct ensembles of those models and creatively combine the information contained therein to produce the consensus forecast that best matches the training data. This method essentially constructs a pseudo-ensemble that can then be analyzed using methods from the vast ensemble literature.
Session 3, Third Annual AMS Artificial Intelligence Forecasting Contest: Methods and Results
Wednesday, 20 January 2010, 10:30 AM-12:00 PM, B204
Previous paper