Output details

Estimating parameters in stochastic systems : a variational Bayesian approach

<25> This paper presents the culmination of 5 years of international collaboration developing a new method for data assimilation that allows efficient estimation of the probability distribution of the state of a dynamical system given incomplete observations. The method is unique and allows estimation of parameters in the model and model error. The work has been widely presented at national and international events including a national meeting of the Royal Meteorological Society on Stochastic weather prediction. The techniques developed in this paper are being discussed with Neill Bowler (neill.bowler@metoffice.gov.uk) at the UK Met Office for operational consideration.