Characterizing and Understanding Large-Scale Wave Propagation in the Atmosphere through Graphs of 3D Information Flow

Abstract:

Causal discovery seeks to discover potential cause-effect relationships from observational data. Here we adopt the idea of interpreting large-scale atmospheric dynamical processes, particularly those tied to propagation of large-scale waves, as information flow around the globe, which can then be calculated using causal discovery methods. We apply a well-established causal discovery algorithm - based on constraint-based structure learning of probabilistic graphical models - toward 51 years of 6-hourly, atmospheric isobaric-level geopotential height data to construct the first-ever graphs of 3D information flow in the atmosphere. These graphs are created globally for different seasons and their connection to phase/energy propagation of atmospheric waves are investigated. Specifically, we examine the information flows 1) in the topical region that represent horizontal and vertical propagations of Kelvin and Rossby-gravity waves whose associated momentum transfer are known to play a key role in the Quasi-Biennial Oscillation (QBO), and 2) in the northern extratropics that represent propagations of planetary-scale waves whose heat/momentum fluxes are responsible for vacillations in the polar stratospheric vortex and occurrences of extreme events such as the stratospheric sudden warming. The sensitivity of the constructed graphs of 3D information flow to data resolution and pre-processing methods (e.g., spatial and temporal filtering) will be discussed.