Statistical simulation of the influence of the NAO on European winter surface temperatures: Applications to phenological modeling

We develop a modeling framework to investigate the influence of the North Atlantic Oscillation (NAO) on phenological variability in Europe through its influence on the distribution of wintertime synoptic-scale surface temperature variability. The approach employs an eigendecomposition of NCEP daily winter surface temperature estimates from the latter twentieth century to represent the spatial structure in the surface temperature field. The subset of statistically significant principal components are modeled as first-order autoregressive AR(1) processes, while the residual variance is modeled as spatially uncorrelated AR(1) noise. For those principal component time series that exhibit a statistically significant seasonal relationship with the NAO index, the parameters of the AR(1) model are conditioned on the phase ("high," "neutral," or "low") of the NAO. This allows for realistic simulations of synoptic scale surface temperature variability over Europe as it is influenced by the NAO index. The model is applied to the simulation of trends in growing degree days (GDD) over Europe where simulated GDD variations are shown to agree well with growing degrees days from the data and evidence from available phenological records. Preliminary application of this model to a climate change scenario involving an increasing NAO 50 years into the future suggests the potential for a continued advancement of the start of the growing season.