Skill of a New Two- to Six-Week Forecast System

A new system for producing forecasts for the coming two to six weeks is demonstrating sufficient skill to contribute new capability and confidence to management of subseasonal climate variability. These forecasts are created by the World Climate Service (WCS) using both statistical and dynamical model predictors and include an optimum combination of the subseasonal forecasts of the U.S. National Weather Service (Climate Forecast System v2) and the European Centre for Medium-Range Weather Forecasts. The forecasts are available for the entire globe and for specific localities and are displayed as weekly averages in probabilistic form.

The utility of the forecasts for business decisions is demonstrated with two statistics: the fraction of forecasts that are correct (fraction correct) and the fraction of events that are predicted correctly (success ratio). In multi-year verification studies of temperature forecasts, both measures of skill are significantly higher than those of forecasts in which the terciles are selected randomly. Considerably better skill is obtained for subsets of the forecasts in which the predicted probability is high, which demonstrates that users can obtain a large likelihood of success if they are willing to wait for a confident forecast.

The WCS subseasonal forecast system includes statistical and dynamical model forecasts of major subseasonal climate phenomena, such as the MJO, high-latitude blocking, and sudden stratospheric warming events. Future development of the system will allow the user to select analogs based on these phenomena, or based on objective pattern matching, and then combine the resulting analog forecast information with the dynamical model predictions.