A traceable hierarchy of global atmosphere models at resolutions ranging from typical climate (120km) through to global forecasting (25km) has been developed, in order to study the impact of improved representation of small scale processes on the mean climate, its variability and extremes. A subset of these models have also been used for coupled climate modelling.

Important aspects of climate variability are found to be better represented at higher resolutions. Atmospheric blocking over the North Atlantic and Europe is significantly improved, associated with improved simulation in both atmosphere and ocean. El Niņo-Southern Oscillation variability is also enhanced in both spatial structure and in variability.

Tropical cyclone performance in the Atlantic is greatly improved at higher atmospheric resolutions, in terms of track positions, intensity and interannual variability.

Given this improvement in the simulation of present-day variability, it is highly relevant to see what impact this might have on future climate projections. To this end, atmosphere-only integrations at 120km, 60km and 40km resolution, using SSTs and sea-ice representative of 2100, have been analysed to determine how model resolution may affect simulated climate variability in the future.