Intelligent Earth Seminar: Tim Tang (Oxford)

11 December 11:00
Earth Sciences: Lecture Theatre

Tim Tang is currently working as an Eric and Wendy Schmidt AI in Science Postdoctoral Fellow and Course Director of Intelligent Earth CDT at the University of Oxford. He graduated with a BEng from the University of Nottingham coming top of each year. He moved straight to a DPhil at Oxford publishing 5 journal papers. He won first place in the Osborne Reynolds Day competition for the UK's best DPhil student in fluid mechanics. His thesis covered the analysis of field data, experiments, numerical modelling using two different models and the application of data science methods to ocean engineering problems. His current research focuses on extreme events in fluid mechanics with an emphasis on machine learning, including data-driven predictions on extreme waves and extreme structural loading, considering leading order physics such as nonlinear wave dynamics and instabilities, breaking waves, and long-short wave interaction.

What makes a wave break? How machine learning can shed light on the underlying physics of breaking waves

 

A wide class of supervised machine learning methods are known to be excellent at modelling complex systems empirically. These empirical models, however, usually provide only limited physical explanations about the underlying systems. Instead, so-called “knowledge discovery” methods can be used to explore the governing equations that describe observed phenomena. In this talk, I will focus on how we can use such methods to explore the underlying physics and also model a commonly observed yet not fully understood phenomenon — the breaking of water waves.
 
In our work, we use symbolic regression to explore the equation that describes wave breaking evolution from a dataset of in silico waves generated using extremely expensive methods (a typical 2D wave with a few breaking periods takes the university’s cluster around two days to simulate). Our work discovers a new boundary equation which provides a reduced order description on how the surface elevation (i.e. the water-air interface) evolves forward in time, including the time period when the wave breaks, which has defied traditional approaches to this problem. Unlike empirical models where the underlying dynamics are hidden in huge numbers of highly tuned parameter values, the physical meaning of each term of our discovered equation can be revealed successfully through mathematical derivation or simulation. Our equation suggests a new characteristic of breaking waves in deep water – a decoupling between the water-air interface and the fluid velocities. It also hints at much cheaper ways to computationally simulate breaking waves, which we are currently working on.