General Circulation Models (GCMs) are fast becoming the principal tool with which various aspects of the global climate system are studied. Particular applications include seasonal forecasting and climate change projections. However, in order to use this information appropriately the GCMs need to be rigorously verified and an assessment of their capability noted.

Daily output from three GCM (HadCM3, CSIRO Mk2, and ECHAM4) climate change simulations are available for a thirty year control period of 1961-1990, and the last thirty years of the transient simulation spanning 2070-2099. In each case the transient simulation is forced according to the SRES A2 emissions scenario - representing a moderately high growth scenario of greenhouse gas concentration in the atmosphere.

This paper addresses two fundamental aspects of GCM modelling. The first is the characteristics of the daily circulation. Questions related to the models' simulation of daily synoptic archetypes in terms of placement, intensity and frequency are answered using self-organizing-map technology.

The second aspect of GCM modelling is how well the models reproduce the global energy cycle. Atmospheric general circulation is essentially driven by energy exchange and errors in modelled energy fields can lead to systematic biases in model simulations. Information on such model shortcomings can be used to help improve some of the systematic biases prevalent today.