An Investigation of the Dynamics of Forecast Errors and Its Representation in the THORPEX Interactive Grand Global Ensemble (TIGGE)

Abstract

We employ local linear diagnostics to investigate the efficiency of an ensemble in capturing the space and magnitude of forecast uncertainties. In this study, we use ensemble forecast data from the leading NWP centers included in the THORPEX Interactive Grand Global Ensemble (TIGGE).We find that the operational ensembles included in the TIGGE data set are highly efficient in capturing the space and magnitude of forecast uncertainties past about 72 hour forecast lead time. We also find that our diagnostics are able to provide valuable information on the performance of the ensembles in respect to model setup, such as ensemble generation techniques and model error parameterizations. Lastly we investigate the geographical distribution of ensemble performance for select operational centers. We utilize components of the eddy kinetic energy equation to find how transient processes at the synoptic scale might affect our diagnostics.