A Data-Driven Perspective on ENSO Diversity - Impacts, Definition, and Forecasting

Abstract

El Niño Southern Oscillation (ENSO) is the dominant mode of interannual variability of the global climate and is characterized by anomalously warm (El Niño) and cold (La Niña) sea surface temperatures (SST) in the tropical Pacific. El Niño and La Niña exhibit a large event-to-event variation in terms of temperature intensity, spatial pattern, and temporal evolution, known as ENSO diversity. ENSO diversity is commonly described by two distinct types — Eastern Pacific (EP) and Central Pacific (CP), based on the location of peak SST anomalies — exhibiting different impacts on weather conditions worldwide, also called teleconnections. While the coupled atmosphere-ocean feedback processes of ENSO are known, the mechanisms contributing to its diversity are not clear.

This thesis introduces data-driven approaches to model various aspects of ENSO diversity, assess ENSOs global impacts, refine its definition, and improve its forecasting accuracy. My contribution is three-fold:

i) I introduce a novel tool to visualize teleconnections of ENSO diversity worldwide, suggesting that EP El Niño events mainly impact surface temperatures in the tropics whereas CP El Niño events exhibit only minor impacts on temperature changes.

ii) Studying the impacts of El Niño events revealed inconsistencies between conventional definitions of ENSO diversity. Consequently, I propose that ENSO diversity should be defined as a continuous phenomenon, rather than the binary separation into CP and EP events. This perspective allows for a more nuanced estimation of onset dynamics and low-frequency changes of ENSO.

iii) I propose a hybrid model for ENSO forecasting, that exhibits skillful forecasts up to 18-months with uncertainty estimates. The combination of linear model and recurrent neural network is data efficient and enables interpretable analysis, highlighting potential mechanisms of ENSO diversity.

With anthropogenic climate change projected to intensify El Niño events, this work contributes to enhancing our understanding and predictive capabilities of ENSO diversity, which is crucial for agriculture, energy production, and disaster mitigation.