Caroline Boess : February 8, 2007

Using model reduction techniques within Incremental 4-Dimensional Variational Data Assimilation.

Abstract

Data assimilation forms an important component of all numerical weather prediction systems. Incremental 4D-Var is a method of data assimilation that requires the minimisation of a series of simplified cost functions. These simplified functions are derived from a spatial truncation of the full system being approximated. In this talk a new method for deriving these simplified problems, based on model reduction, is proposed. It will be shown how this method can be combined with incremental 4D-Var. Numerical experiments are used to illustrate the superior performance to standard truncation methods.