Implications of seasonality and asymmetry for ENSO’s predictability and future changes

Abstract

The El Niño - Southern Oscillation (ENSO), a coupled ocean-atmosphere phenomenon in the tropical Pacific, is the dominant year-to-year driver of variations in Earth’s climate. Its impacts on temperature and precipitation are felt globally, and particularly in the tropics, where a large fraction of the world’s population lives. While a century of research since the discovery of the Southern Oscillation has advanced understanding of ENSO, there is little consensus about how it will change in the rest of the 21st century, owing to disagreement between different state-of-the-art climate models. In an effort to understand this disagreement, and improve understanding of ENSO’s future changes, this thesis investigates aspects of two ENSO features – seasonality and asymmetry – whose projected changes haven’t been well-studied, despite their prominence in observations. First, we evaluate an alternative framework – based on the Koopman operator – for representing asymmetry and seasonality in a highly idealized, conceptual model of ENSO. While conceptual ENSO models are widely used to diagnose ENSO stability in observations and global climate models, they rely on nonlinear parameterizations which may misrepresent the processes giving rise to ENSO asymmetry. We show that the Koopman-based approach, based on finding a nonlinear transformation which makes ENSO look symmetric, can reproduce the most salient aspects of ENSO’s evolution asymmetry, including the faster decay of El Niños during boreal spring and the higher frequency of multi-year La Niñas. Next, we study projected changes in ENSO’s seasonality and asymmetry in a global climate model over the course of the 21st century. We find that changes in asymmetry contribute significantly to the late-21st century decrease in ENSO amplitude projected by the model, with a weak increase in La Niña intensity opposing a much larger decrease in El Niño intensity. We show that this change is consistent with the projected weakening of the equatorial Pacific’s zonal temperature gradient, a change which limits the potential intensity of El Niños, but not La Niñas. Finally, we switch our focus to precipitation variability in the U.S. Midwest, and in particular, the ocean-to-land atmospheric moisture transport which fuels this variability. Oceanic moisture sources (vs. those from land) have become more important in recent decades, and we quantify the relative contributions from Pacific and Atlantic moisture sources and how they vary seasonally and over time. We find that an intensification of the Great Plains Low-Level Jet is correlated with an increase in the fraction of Midwest rainfall that originates from the ocean (vs. from land surfaces). While this effect is observed on interannual timescales and likely modulated by ENSO, it is most pronounced on synoptic timescales, when upper-level disturbances amplify the low-level jet and moisture transport to the Midwest; thus here we investigate it independent of ENSO. Overall, this thesis highlights the potential importance of nonlinearity and seasonality in ENSO’s future changes and suggests one alternative framework for their representation in conceptual models.