Causal maps versus correlation maps: visual analysis of tropical-extratropical atmospheric teleconnections using causal discovery

Abstract

<p>In climatology, correlation maps are often used to study the relationships between one 1D time series and a (spatiotemporal) 2D or even 3D field. However, correlation measures do not necessarily capture causal relationships and similarities in correlation maps obtained from different indices may appear if the set of indices contains correlated variables. Causal discovery tools such as the Peter and Clark – Momentary conditional independence (PCMCI) algorithm can help in disentangling spurious from causal links in both linear and nonlinear frameworks. In the linear case considered in the present work, PCMCI extends standard correlation analysis by removing the confounding effects of autocorrelation, indirect links and common drivers. Combining PCMCI and Causal Effect Networks on a 2D field helps identifying, and subsequently discarding the spurious correlations and thereby allows to retain only the causal links. The resulting visualization technique is referred to as a “causal map”.</p><p>In this presentation, we illustrate the application of causal maps in combination with maximum covariance analysis to assess how tropical convection interacts with mid-latitude circulation during boreal summer at different intraseasonal timescales. The obtained causal maps reveal the dominant patterns of interaction and highlight specific mid-latitude regions that are most strongly connected to tropical convection. In general, the identified causal teleconnection patterns are only mildly affected by ENSO variability and the tropical-mid-latitude linkages remain similar under different types of ENSO phases. Still, La Niña strengthens the South Asian monsoon generating a stronger response in the mid-latitudes, while during El Niño periods, the western North Pacific summer monsoon pattern is reinforced. Our study paves the way for a process-based validation of boreal summer teleconnections in (sub-)seasonal forecast models and climate models and therefore provides important clues towards improved sub-seasonal and climate projections.</p><p> </p><p>Reference: G. Di Capua, J. Runge, R.V. Donner, B. van den Hurk, A.G. Turner, R. Vellore, R. Krishnan, D. Coumou: Dominant patterns of interaction between the tropics and mid-latitudes in boreal summer: Causal relationships and the role of time-scales. Weather and Climate Dynamics, 1, 519-539 (2020)</p>