The predictability of atmospheric blocking in global ensemble prediction systems
Institut für Meteorologie und Klimaforschung, Karlsruhe Institute of Technology
Blocking is an atmospheric phenomenon which has major implications for local weather. It can lead to extremely high or low temperatures due to the long lasting almost stationary high pressure system. Furthermore, a block can exert a strong impact on upstream, in-situ and downstream synoptic weather patterns by disrupting the midlatitude westerly flow. Thereby, blocking can be the cause for severe precipitation anomalies in cut-off lows surrounding the high pressure system. The state of the art ensemble prediction systems (EPSs) which are part of the THORPEX Interactive Grand Global Ensemble (TIGGE) perform well in predicting the frequencies of Atlantic European and Pacific blocking but have difficulties in predicting the onset and the decay of blocking. The predictability of the onset is about 1 day worse than the predictability if the blocking is already specified in the initial conditions. Furthermore, the TIGGE ensemble prediction systems (EPSs) have problems in simulating the frequencies of Greenland and Ural blocking. This study is dedicated to investigate the dynamics of selected blocking events in the medium range, the monthly, and the seasonal ECMWF EPS. Two different clustering methods based on EOF/PC analysis and one method based on ensemble analysis covariance will be applied to the EPSs. Links between different types of block formation or decay and large scale events like breaking Rossby waves or small scale diabatic processes like precipitation resulting from the advection of moist air masses will be investigated. Their roles in the life cycle of a blocking event will be compared. The same investigation methods will be applied to the TIGGE ensembles and to carry forward the results obtained for the medium range ECMWF EPS alone. Additionally, for two winter seasons and for a whole year scores and skill scores used at the ECMWF will be calculated and the predictive skill of the EPSs will be assessed with regard to blockings.
Predictability of one specific blocking event
A low anomaly correlation in terms of the 5 day forecast of the 500 hPa geopotential was associated with the decay period (at about 18 October 2011) of a blocking event in October 2011.
Low predictability at 18 October 12 UTC is also seen in the ECMWF 51 member ensemble prediction system (EPS). Both a hovmoeller plot of the ensemble spread in terms of the potential temperature on the dynamic tropopause (Fig. 1a) and a spaghetti plot of a 500 hPa geopotential height line (Fig. 1b) show uncertainty in the EPS which can be assigned to the representation of the blocking event.
Fig. 1: EPS forecast initialized on 13 October 2011 at 12 UTC. a) Hovmoeller plot of the 51 member ECMWF EPS standard deviation for up to 180 h forecast lead time in terms of the potential temperature at the dynamic tropopause (defined as PVU = 2). b) Spaghetti plot of the 120 h forecast of the 5600 gpm geopotential height.
Applying a EOF / Clusteranalysis to the 5 day forecast shows that variability can be assigned to uncertainty in the representation of the orientation, the depth and the propagation speed of the upstream trough. Furthermore, uncertainty is found in a cut-off low located over the black sea. One of the clusters wrongly predicts a distinct amount of convective and large-scale precipitation over the Mediterranean Sea and fails to predict the strong precipitation over the Black Sea.