Using TIGGE Data to Analyze Features of Time Series of Forecast Uncertainty

Gaining flow-dependent estimates of the forecast uncertainty and the forecast probability distribution is the main motivation for ensemble forecasting systems (EFSs), yet the nature of the flow dependence and the processes that modulate this dependence are not well understood. Here, we characterize long time series of EFS forecast variance from the THORPEX Interactive Grand Global Ensemble (TIGGE) data archive. Emphasis is placed on several features including the possible presence of certain quasi-periodic components, intraseasonal “ramp-ups” in the variance, and episodes of notable volatility accentuated by large swings in the variance. To further understand these features, the time series of variance is regressed against standard flow indices such as the Arctic Oscillation (AO) index and the Madden-Julien Oscillation (MJO) index. Additionally, case studies are presented of some notable high-variance episodes in the time series.