Observed nonmodal growth of the Pacific-North American teleconnection pattern

Abstract

A linear-stochastic model is applied to the 10-day low-pass streamfunction field at 300, 500, and 850 mb for 40 winter seasons of Northern Hemisphere NCEP-NCAR reanalysis data. The linear operator is derived from the observed multilevel covariances, allowing for statistical representation of nonlinear processes. While all empirical normal modes of the system are decaying, increase in the streamfunction variance is possible through nonmodal growth. When the evolution of the streamfunction field following the optimal perturbation is predicted, the Pacific-North American teleconnection pattern (PNA) is found to be the most probable state of the atmosphere. Sixty-eight percent (70%) of positive (negative) PNA events are found to follow high projections onto the leading optimal, suggesting the PNA arises through constructive interference between the decaying modes and may be treated as a linear response to Gaussian white noise stochastic forcing. Implications for PNA timescale and onset mechanisms are also discussed.