Summary
Reconstructing past climate states requires the identification of the mean and the variability of the atmosphere and, once interpreted as a response, the type of the underlying atmospheric forcing. Adjoint modelling diagnostics based on a primitive equations simplified global circulation model (GCM) is used to analyze the climate mean induced by a stormtrack and the variability of a low frequency pattern resembling the North Atlantic Oscillation (NAO). First, given the climate mean, the associated location, shape and intensity of the forcing can be successfully determined so that, rerunning the model with the reconstructed forcing, the original climate variability is retained. Second, given the climate variability represented by a daily NAO-type index, its temperature forcing on short timescales is diagnosed in terms of temperature sensitivity patterns at different time lags. These two feasibility studies of adjoint technique applications demonstrate their potential usefulness in climate and palaeoclimate diagnostics.