We use some simple indices of surface air temperature patterns, including the global-mean, the land-ocean contrast, the meridional gradient, the interhemispheric contrast, and the magnitude of the seasonal cycle, to describe global climate variability and change. These indices are associated with dynamical factors determining aspects of the large-scale atmospheric circulation. They contain information independent of the variations of the global-mean temperature for natural climate variations. These indices also represent the main features of the modelled surface temperature response to increasing greenhouse gases in the atmosphere. Hence, they should have a coherent response for greenhouse climate change.

We use global instrumental observations, palaeoclimate reconstructions from proxy data, and simulations from 4 different coupled ocean-atmosphere climate models to investigate the correlation structure of these indices on interannual and decadal timescales.

The variability and correlation structure of the indices from long control model simulations compares well with that from observations and palaeodata. Examination of the observed correlation structure between the indices on decadal timescales shows significant recent changes that are consistent with forced climate variations shown in the greenhouse climate change model simulations. This suggests that the changes in the correlation structure between these indices can be used an indicator of climate change.