Graduate Student / Postdoc Seminar

Data Assimilation in High Dimensions

Speaker: David Kelly

Location: Warren Weaver Hall 1302

Date: Friday, February 12, 2016, 1 p.m.

Synopsis:

This will be an introduction to data assimilation and in particular to methods that are useful for high dimensional forecasting problems. Data assimilation describes any method of using observational data in order to guide model predictions in the right direction. The most prominent application is numerical weather prediction, where forecasts are obtained by blending ocean-atmosphere models with partial observed data in order to reduce prediction uncertainty. We will discuss the basic mathematical ideas (Bayes' theorem, Kalman filters), the problem of high dimensionality and how methods can be successful in high dimensions.