Stochastic Descriptions of Atmospheric Variability over the Sub-Polar North Atlantic

We present an approach for creating a computationally efficient stochastic weather generator for the sub-polar North Atlantic. The long term aims of this work are to provide ensembles of pseudo atmospheric data for applied regional studies of ocean response to modes of atmospheric forcing. The method propagates regional circulation patterns, conditioned on the state of large-scale interannaul weather regimes. So as not to smooth out important variability, residuals effects are included and propagated by a lagged regression model with added noise components. The resulting simulations express the range of the major regional patterns of atmospheric variability and typical time scales. The modelling approach allows us to document notable shifts in short term regional dynamics associated with larger-scale/longer-term trends. As such, it presents an analysis of some properties of the dominate regional weather regimes.