Categorical Climate Forecasts through Optimal Combination of Multiple GCM Ensembles

摘要

A Bayesian methodology is used to assess the information content of categorical, probabilistic forecasts of seasonal-mean precipitation derived from General Circulation Model (GCM) forecast ensembles. A “prior” forecast given by climatological probabilities of each category is combined with categorical probabilistic forecasts derived from several GCM ensembles to develop posterior categorical probabilities. The combination algorithm assigns a weight to each particular model forecast and to the climatological forecast. Several improvements are described, that produce more robust results when many models are combined together.