|Date:||Wed, March 15, 2017|
|Place:||Research I Seminar Room|
Abstract: This is a simple introduction to the idea of stochastic superparametrization given in the papers of I. Grooms, A.J. Majda, and K.S. Smith (2013, 2014, and 2015). The goal is to improve a quasigeostrophic coarse grid model and make it to be more realistic and closer to fine grid models with resolved mesoscale eddies. Roughly speaking, the idea is to take coarse equations, add perturbations to the coarse variables, write equations for the pertubations, split this last equation into a linear and non-linear part; the linear (expected to be computationally inexpensive, even admits explicit analysis) part is used for the simulation of perturbations between coarse time steps, and the non-linear part is used as a perturbation of the coarse equation. All discrepancies in this scheme are muffled by a random source with some tuning parameters...