Weakening of atmospheric information flow in a warming climate in the Community Climate System Model

Abstract

We introduce a new perspective of climate change by revealing the changing characteristics of atmospheric information flow in a warming climate. The key idea is to interpret large-scale atmospheric dynamical processes as information flow around the globe and to identify the pathways of this information flow using a climate network based on causal discovery and graphical models. We construct such networks using the daily geopotential height data from the Community Climate System Model Version 4.0 (CCSM4.0)'s 20 th century climate simulation and 21 st century climate projection. We show that in the CCSM4.0 model under enhanced greenhouse gases (GHGs) forcing, prominent midlatitude information pathways in the midtroposphere weaken and shift poleward, while major tropical information pathways start diminishing. Averaged over the entire Northern Hemisphere, the atmospheric information flow weakens. The implications of this weakening for the interconnectivity among different geographical locations and for the intrinsic predictability of the atmosphere are discussed. Climate Networks Based on Causal DiscoveryThe traditional idea of climate networks is to define a correlation network of nodes where each node represents a point on a global grid [Tsonis and Roebber, 2004]. Any two nodes are connected if the correlation of a specific atmospheric and/or oceanic variable measured at the two nodes is beyond a threshold [Tsonis and Roebber, 2004;Tsonis et al., 2006]. The introduction of a climate network effectively brought the vast framework of network analysis to climate science and triggered a flurry of research activity in this area, including classification of network type, analysis of network properties such as density of connections, backbone of the network, and effects of natural modes of climate variability (e.g., El Niño) on network properties [Tsonis et al., 2007;Tsonis and Swanson, 2008;Gozolchiani et al., 2008;Yamasaki et al., 2008;Donges et al., 2009;Steinhaeuser et al., 2010]. While correlation networks remain the most common type, three more definitions have recently been proposed. MI networks [Donges et al., 2009] use mutual information, rather than correlation, to decide whether two nodes should be connected. Phase synchronization networks [Yamasaki et al., 2009] use the level of robust phase synchronization between nodes as criterion for connections, and event synchronization networks [Boers et al., 2013] focus on the level of synchronization of extreme events (maxima and minima) in the nodes' time series. While correlation and MI networks emphasize instantaneous connections, synchronization networks focus on connections over time by considering temporal lags. All of the above network definitions, however, decide whether an edge exists between two nodes, X and Y, in the network based only on a test involving those two nodes. Thus, they do not truly distinguish between direct and indirect causal connections. Furthermore, the results for many of them have been described as being fairly similar to corr...

Using Causal Effect Networks to Analyze Different Arctic Drivers of Midlatitude Winter Circulation

“…This algorithm is a modified version of the Peter Spirtes and Clark Glymour (PC) algorithm (Spirtes et al 2000), which was first applied to climate research by Ebert-Uphoff and Deng (2012) to study interactions between major climate modes. Causal discovery approaches have since been used to study atmospheric flows (Deng and Ebert-Uphoff 2014), causal relationships in the Walker cell in the tropics , the monsoonal dynamics in the Pacific-Indian Ocean (Runge et al 2015), and decadal ocean circulation in the Atlantic (Schleussner et al 2014).…”

Introduction to Focus Issue: Complex network perspectives on flow systems

“…While the former study 63 presents one of the first attempts to combine causality concepts with spatially explicit climate network generation and analysis at global scale, 64,65 Tirabassi et al 66 discuss how the even more elaborated concepts of renormalized partial directed coherence and directed partial correlation can be used in a climatological context. Both measures have recently proven their potentials in neurophysiological signal analysis and are here for the first time employed to climate data in two specific case studies on ENSO-monsoon interactions as well as air-sea interactions in the South Atlantic Convergence Zone.…”

High-Dimensional Dependency Structure Learning for Physical Processes

“…The ability to infer interactions between variables from high-dimensional data sets has the potential to help geoscientists answer numerous questions critical for improved modeling and prediction capabilities for various geoscience processes. Using atmospheric science as an example, it would enable us to (1) delineate better the interactions between atmospheric disturbances of different spatial scales, which is critical for understanding the working of a weather-climate continuum; (2) develop a better understanding of the degree and spatial pattern of coupling between the top of atmosphere (TOA) radiative imbalance and surface temperatures, which provides a unique perspective of climate feedback processes; (3) identify causal pathways in the atmospheric circulation and infer how they might change under a warming climate [1]; and (4) study the dynamical processes of air-sea interaction that lead to the onset of the monsoons. These applications would contribute to both our understanding of the key processes determining the main features of the Earth's climate system and our capabilities to predict changes in this system with changing external forcing (e.g., aerosols and greenhouse gas emissions) in the near future.…”