The North Atlantic Oscillation (NAO) is an important mode of winter atmospheric variability associated with changes in the North Atlantic stormtrack and the climates of Europe and eastern North America. New generations of climate models must be able to realistically simulate the NAO and its corresponding oceanic variations on all timescales if, for example, we are to be able to distinguish anthropogenic climate change from natural climate variability.

Until recently, the NAO has generally been validated in models on relatively short timescales. Such timescales are predominantly associated with atmospheric internal variability and there is little guarantee that the NAO will be represented well at longer (multiannual to multidecadal) timescales when ocean-atmosphere interactions become more important. Model validation of NAO variability at these longer timescales is urgently required but it is more difficult because the instrumental period is not sufficiently long to capture the full range of possible coupled variability.

We have developed a diagnostic package that focuses on the mechanisms behind the variability. By concentrating on mechanisms, which are operative at much shorter timescales, the length of the observational record may not be such a constraint. One part of the package uses lagged maximal covariance analysis to identify temperature patterns in the ocean that lag or lead the NAO. Results show that the model does a reasonable job of simulating the direct oceanic response to the NAO. However, preliminary results reveal a poorer model NAO response to North Atlantic sea-surface temperatures (SSTs). A thorough picture will be presented of the coupled variability associated with the NAO, including thermodynamic, stormtrack and oceanic circulation variability.