This is project meant to support the Smart-TURB ERC AdG (2021-2026) on Machine Learning applications to Eulerian and Lagrangian Turbulence.
The main goal is to build a state-of-the-art data-base of turbulent data from different important geophysical applications, either flows under rotation and with natural convection to be used to train, validate and benchmark data-driven approaches for turbulent data assimilation, super-resolution, and feature rankings. Computational Fluid Dynamics offers a unique environment where to test ML tools, proposing at the same time high quality and high quantity of data. One of the main goal is to build an open database to become the reference in the field where to develop and benchmarks new applications. The main advantage of turbulence is the possibility to make the ground truth dataset as complex as needed, by increasing the scale separation between large scale and small scale features and by increasing the degree of non-Gaussianity of the fields.