Spatiotemporal Clustering of Preferred Trajectories of the Euro-Atlantic Summer Circulation

Abstract We present a novel approach to studying the regime behavior of the Euro-Atlantic summer circulation by analyzing its spatiotemporal preferred trajectories. Our approach is inspired by the dynamical system concept of unstable periodic orbits (UPOs) and motivated by the potential practical utility in modeling the circulation as a sequence of short-lived but well-defined trajectories. Here, we identify the dominant spatiotemporal preferred trajectories in a large ensemble (2000-yr) of simulated present-day climate data with a multistage clustering algorithm. Unlike conventional regime definitions based solely on spatial patterns, our method also explicitly takes into account the temporal trajectory through phase space. We find that 13 spatiotemporal clusters together capture over 80% of the circulation dynamics over the Euro-Atlantic domain. We distinguish between clusters with a quasistationary tendency and clusters with a transient tendency. Markov transition probabilities between the clusters reveal that the circulation tends to alternate between quasistationary and transient episodes, instead of transitioning between quasistationary clusters directly. We show that traversing the phase space between quasistationary blocking and southerly shifted zonal jet states takes at least 10–15 days and that trajectories tend to stay close to these states in phase space for prolonged periods. Taken together, our results demonstrate that the spatiotemporal clusters capture a diverse range of well-known circulation regimes while also revealing more nuanced behavioral characteristics of the Euro-Atlantic summer circulation.