Verification of Ensemble Forecasts - A Survey

Laurence J. Wilson

Meteorological Service of Canada, Montreal, Quebec

 

The output of an ensemble prediction system is a set of forecast values for each projection time and for each point in the model domain. The set of forecast values is intended to represent an estimate of the distribution of possible outcomes at the valid time.  The evaluation of distributions of forecast values presents challenges to forecast verification methodology and has led to the extension of the “toolbox” of verification techniques.

 

Verification methodology for ensemble forecasts is directed toward two main purposes:  The diagnostic assessment of the characteristics of the ensemble distribution, and the verification of probability forecasts derived from the ensemble.  More recently, a third purpose has emerged:  the assessment of value or potential value of probabilistic ensemble forecasts.  This paper presents a brief survey of new verification methods that have been developed specifically for application to ensemble forecasts.