Data-driven models (DDMs) learn from large datasets to make predictions, but data limitations affect reliability. Data assimilation (DA) improves accuracy by combining real-world observations with computational models. This research explores how DA enhances DDMs despite limited data. We propose an algorithm using DA to refine DDM training iteratively. This work has broad implications for fields like meteorology, engineering, and environmental science, where accurate prediction is critical.

We propose Ensemble Model Predictive Control (EnMPC), a novel method that improves control of complex systems like weather by integrating control theory with data assimilation. Unlike traditional methods, which are computationally expensive, EnMPC uses ensemble simulations to efficiently handle uncertainties and optimize solutions. This approach reduces computational cost while maintaining accuracy, making it a promising step toward real-world applications in dynamic system control.

Climate prediction is challenging due to systematic errors in traditional climate models. We addressed this by training a machine learning model to correct these errors and then integrating it with the traditional climate model to form an AI-physics hybrid model. Our study demonstrates that the hybrid model outperforms the original climate model on both short-term and long-term predictions of the atmosphere and ocean.

We study time series of solar wind large-scale structures (magnetic clouds, sheaths, slow and fast streams). These have profound importance for causing disturbances in the heliospheric conditions and driving space weather on Earth. The used techniques include methods deriving from information theory to determine entropy and complexity. We find that all of these techniques show stochastic fluctuations, but magnetic clouds stand out due to their coherent magnetic field

The Madden-Julian Oscillation (MJO) is a large-scale tropical wave of enhanced and suppressed rainfalls, slowly moving eastward at the equator, influencing the weather and climate globally. We study the MJO using a simplified model designed to capture its large-scale features. We introduce new, more realistic, inputs into the model, show that this enhanced model successfully replicates key characteristics of the MJO, and identify some of its limitations.

Particle-tracking simulations compute how ocean currents transport material. However, initialising these simulations is often ad-hoc. Here, we explore how two different strategies (releasing particles over space or over time) compare. Specifically, we compare the variability in particle trajectories to the variability of particles computed in a 50-member ensemble simulation. We find that releasing the particles over 20 weeks gives variability that is most like that in the ensemble.

In recent years, there has been more and more floods due to intense precipitation, such as the July 2021 event in Belgium. Predicting precipitation is a difficult task, even for the next hours. This study focuses on a tool to assess whether a given situation is stable or not (i.e. is likely to stay as it is or could evolve in an unpredictable manner).

Turbulence is found in many systems in geophysics and beyond. We concentrate on normalized moments of complex fields (velocity or magnetic field) and study their scaling using either numerical simulations run for long times or theoretical models of intermittent flows. We confirm previous findings and show that in fact only the strongest intermittent structure counts for which a parabolic law is recovered. More computations will be performed, likely using simplified models.