Stochastic Transport in Fluid Dynamics

Date: 
Monday, September 18, 2017 - 3:00pm
Location: 
HP Auditorium, Soda Hall
Speaker: 
Professor Darryl Holm

Imperial College London

About: 

In next-generation weather and climate models, stochastic parameterization will likely be an important element in providing reliable estimates of model uncertainty and forecast variability.

Current practices for estimating model uncertainty and achieving realistic variability may introduce stochastic perturbations into initial conditions for computer simulations, for example. Or they may add random forcing and dissipation to existing numerical simulation models, then adjust noise parameters on an empirical basis, by trial and error.
However, a posteriori addition of stochasticity to an already tuned model to increase weather variability, for example, tends to be more valuable for ‘hindcasts’, rather than forecasts. It is generally understood in the weather and climate science community that stochasticity should be incorporated at a fundamental level within the design of physi-cal process parameterisations for improvements in the dynamical core of the simulation models.

This talk will explain what has been done recently to develop stochastic fluid dynamics in the context of geometric mechanics and homogenisation theory for fast and slow timescales. It may be interesting to a wide range of mechanical engineers, fluid dynamicists, meteorologists, and mathematicians.

Biography: 

Darryl Holm has been a Professor of Mathematics at Imperial College London since 2005. Previously, he had held a variety of research positions at Los Alamos National Laboratory for 34 years. Holm’s field of research is called Geometric Mechanics. Topics include propagation and breaking of nonlinear waves in fluids; emergent singular solutions of ideal fluid equations; and turbulence models for global ocean circulation. Recently, Holm has derived a new data-driven variational theory of Stochastic Geometric Mechanics for geophysical fluid dynamics (GFD). Holm’s approach is especially designed for data assimilation and uncertainty quantification in the ocean, atmosphere, weather and climate science.

 

 

Hosted by: Prof. Shawn C. Shadden, 5126 Etcheverry Hall, 664-9800, shadden@berkeley.edu