Event

Data-driven Surrogate Modeling of Turbulent Flows in the Atmosphere and Ocean

Dr. Murali Gopalakrishnan Meena

Abstract: Modeling turbulent flows is one of the most challenging problems in science, attributed to the highly nonlinear multi-scale behavior of the system.  Numerically resolving the various scales of flow phenomena in practical turbulent problems in nature requires high spatial and temporal resolution.  These numerical simulations demand extreme computational resources when computed for the entire earth over climate time scales.  In the age where data has a dominant influence on how complex systems in science and engineering are analyzed, data-driven methodologies have evolved, leading to tremendous potential in tackling such challenging problems.

This seminar will cover research work utilizing machine learning to model canonical turbulent flows in the atmosphere and ocean. Neural networks are used to formulate surrogate models for capturing the subgrid-scale effects in idealized atmospheric turbulence.  Time series prediction using various neural networks is also explored to formulate closure modeling of stratified turbulent flows found in the ocean.  These modeling frameworks take advantage of the unique dimensionality reduction enabled by such flows.  Special emphasis will be placed on the lessons learned and challenges involved with effective modeling of such complex systems using machine learning tools.

Speaker’s Bio: Muralikrishnan (Murali) Gopalakrishnan Meena is a Computational Scientist in the Advanced Computing for Life Sciences & Engineering group at the National Center for Computational Sciences at the Oak Ridge National Laboratory (ORNL). He obtained his Ph.D. in Mechanical Engineering from the University of California, Los Angeles in 2020.  His Ph.D. research focused on using network (graph) theory to characterize, model, and control vortical interactions in turbulence and wake flows. Murali joined ORNL in 2020 as a Postdoctoral Research Associate.  His research was focused on using machine learning to formulate subgrid-scale turbulence closures for atmospheric flows.  He is currently continuing this work and additionally working with the users of the Oak Ridge Leadership Computing Facility on projects involving complex fluid flows and turbulence.  Additionally, he also uses various data-driven techniques to characterize and model complex systems in physical and biological sciences.

Last Updated: January 31, 2023 - 10:32 am