The Earth's climate system is characterized by the intricate interplay of atmospheric and oceanic processes evolving at various timescales, exhibiting complex behaviors and nonlinear interactions. Gaining a deeper insight into the underlying dynamics of this system is crucial for understanding the physical origins of weather and climate variability, as well as for predicting climate trends and extreme events. However, this task poses significant challenges, as traditional theoretical approaches alone often fall short in capturing the full extent of these complexities. To address these challenges, data-driven methods have increasingly become indispensable tools in the study of oceanic and atmospheric dynamics.
Over the past few decades, the application of data-driven approaches has led to substantial advancements in our understanding of climate systems. Linear techniques such as normal modes, wave analysis, and Fourier methods, have long been employed to extract relevant spatiotemporal features and identify key climate modes. Furthermore, empirical dynamical methods, such as Linear Inverse Models (LIMs), have proven invaluable for the study and prediction of climate phenomena like the El Niño-Southern Oscillation (ENSO).
In recent years, the advent of non-linear data-driven methodologies has opened new avenues in the field. Techniques such as transfer operators, including Koopman mode decomposition, and various machine learning approaches have significantly broadened the scope of what can be achieved in the analysis and forecasting of climate dynamics. These methods offer potential to uncover complex patterns, improve climate predictability, and develop more accurate reduced-order models that capture the essence of the underlying dynamical processes, holding great potential for enhancing our understanding of complex atmospheric and oceanic climate processes.
This session aims to bring together researchers at the forefront of applying data-driven methods to study oceanic and atmospheric dynamical systems. We invite contributions that explore the application of these methodologies in various aspects of climate science, including (but not limited to) the following topics:
- Climate Predictability and Forecasting
- Spatiotemporal Feature Extraction
- Climate Mode Identification
- Climate Network Analysis
- Exploration of Climate Attractors
- Development of Reduced-order Models
- Extreme Event Analysis