Probabilistic skill in ensemble seasonal forecasts

Operational seasonal forecasting centres employ simulation models to make probability forecasts of future conditions on seasonal to annual lead times. Skill in such forecasts is reflected in the information they add to purely empirical statistical models, or to earlier versions of simulation models. An evaluation of seasonal probability forecasts from the DEMETER and the ENSEMBLES multi-model ensemble experiments is presented. Two particular regions are considered (Nino3.4 in the Pacific and Main Development Region in the Atlantic); these regions were chosen before any spatial distribution of skill were examined. The ENSEMBLES models are found to have skill against the climatological distribution on seasonal time scales; for models in ENSEMBLES which have a clearly defined predecessor model in DEMETER the improvement from DEMETER to ENSEMBLES is discussed. Due to the long lead times of the forecasts and the evolution of observation technology, the forecast-outcome archive for seasonal forecast evaluation is small; arguably evaluation data for seasonal forecasting will always be precious. 22 Issues of information contamination from in-sample evaluation are discussed, impacts (both positive and negative) of variations in cross-validation protocol are demonstrated. Other difficulties due to the small forecast-outcome archive are identified. The claim that the multi-model ensemble provides a �better� probability forecast than the best single model is examined and challenged. Significant forecast information beyond the climatological distribution is also found in a probability forecast based on persistence. On seasonal time scales, the ENSEMBLES simulation based probability forecasts add significantly more information to empirical probability forecasts than on decadal scales. It is suggested most skilful operational seasonal forecasts available would meld information both from simulation models and empirical models.