Observed Nondiffusive Dynamics in Large-Scale Atmospheric Flow

Abstract

The dynamics of large-scale atmospheric flow is characterized by both noisy, diffusive behavior, stemming mainly from small-scale, high-frequency atmospheric processes and by more structured, nondiffusive aspects, coming from, for example, the internal nonlinear dynamics of the flow system. In this study an attempt is made to isolate the latter influence from the former in an analysis of observational data of the atmosphere system during winter. The data are projected onto the unit sphere in a reduced three-dimensional phase space of dominant variability patterns. The probability density function associated with the data on the unit sphere shows clear maxima that correspond with flow regimes also found in previous studies. Concepts from the theory of Markov chains and phase-space partitions are used to bring out the nondiffusive, conservative dynamics of the large-scale atmospheric flow. By inspecting the asymmetries of transitions between different phase-space cells, a preferred, closed path, with a preferred direction, over the unit sphere of the reduced phase space is detected. This path, or cycle, connects episodes of zonal and blocked flow in the Atlantic sector. It is reminiscent of the remnants of a heteroclinic cycle found to guide regime transitions in a model study by the author.