VISDEM: Variational Inference in Stochastic Dynamic Environmental Models

Responsible KEG Investigator:
Dr Dan Cornford

Project Collaborator(s):

  • Met Office
  • University of Southampton
  • Surrey University
Type of Project: Standard Grant

Funder: EPSRC

Date Commissioned: 03/2006
Date Completed: 02/2010

Project Summary
This project will define the next generation of data assimilation methods and enable us to also make inference about hyper-parameters in the stochastic forcing that is used to represent model error. The project revolves around the variational treatment of stochstic differential equations and their approximate solution using a class of stochastic processes known as Gaussian processes. The project invloves 3 universities, led by Aston and includes the Met Office as a partner.
Project web page