Application of blocking diagnosis methods to General Circulation Models. Part II: model simulations D. Barriopedro • R. Garcı ´a-Herrera • J. F. Gonza ´lez-Rouco • R. M. Trigo Received: 24 March 2009 / Accepted: 10 February 2010 / Published online: 3 March 2010 Ó Springer-Verlag 2010 Abstract A previously defined automatic method is applied to reanalysis and present-day (1950–1989) forced simulations of the ECHO-G model in order to assess its performance in reproducing atmospheric blocking in the Northern Hemisphere. Unlike previous methodologies, critical parameters and thresholds to estimate blocking occurrence in the model are not calibrated with an observed reference, but objectively derived from the simulated cli- matology. The choice of model dependent parameters allows for an objective definition of blocking and corrects for some intrinsic model bias, the difference between model and observed thresholds providing a measure of systematic errors in the model. The model captures rea- sonably the main blocking features (location, amplitude, annual cycle and persistence) found in observations, but reveals a relative southward shift of Eurasian blocks and an overall underestimation of blocking activity, especially over the Euro-Atlantic sector. Blocking underestimation mostly arises from the model inability to generate long persistent blocks with the observed frequency. This error is mainly attributed to a bias in the basic state. The bias pattern consists of excessive zonal winds over the Euro- Atlantic sector and a southward shift at the exit zone of the jet stream extending into in the Eurasian continent, that are more prominent in cold and warm seasons and account for much of Euro-Atlantic and Eurasian blocking errors, respectively. It is shown that other widely used blocking indices or empirical observational thresholds may not give a proper account of the lack of realism in the model as compared with the proposed method. This suggests that in addition to blocking changes that could be ascribed to natural variability processes or climate change signals in the simulated climate, attention should be paid to signifi- cant departures in the diagnosis of phenomena that can also arise from an inappropriate adaptation of detection meth- ods to the climate of the model. Keywords Atmospheric blocking Á Automatic methods Á General Circulation Models Á ECHO-G model Á Climate change 1 Introduction Numerous studies have examined the ability of General Circulation Models (GCMs) to reproduce many aspects of the general circulation. Very often, applications have focused on the long-term average behaviour of the most important large scale and hemispheric modes of atmo- spheric variability (Latif et al. 2001; Cohen et al. 2005; Lucarini et al. 2007). However, the mid-latitude atmo- spheric circulation is also influenced by transient synoptic- scale systems and persistent disturbances locked in geographically recurrent regions. The analysis of these small-to-large scale phenomena on a daily basis constitutes a more stringent test to GCMs since they reflect the day-to- day behaviour of the model and result from interaction processes covering a wide range of scales, some of them at the limits of the model resolution. Thus, the diagnosis of weather systems in GCMs provides a powerful tool for: (1) model validation and improvement (e.g. D’Andrea et al. D. Barriopedro (&) Á R. M. Trigo CGUL-IDL, Faculdade de Cie ˆncias, Universidade de Lisboa, Ed. C-8, Campo Grande, 1749-016 Lisbon, Portugal e-mail: dbarriopedro@fc.ul.pt R. Garcı ´a-Herrera Á J. F. Gonza ´lez-Rouco Departamento de Fı ´sica de la Tierra II, Facultad de C.C. Fı ´sicas, Universidad Complutense de Madrid, Madrid, Spain 123 Clim Dyn (2010) 35:1393–1409 DOI 10.1007/s00382-010-0766-6