Abstract
A modified ensemble Kalman filter (KF) is proposed which can enhance performance for highly non-linear prognostic models. The algorithm differs from the traditional ensemble KF by the addition of an expectation maximization step, which estimates the parameters of a Gaussian mixture model for the ensemble of forecast states. The algorithm is tested in twin experiments using a simple phytoplankton—zooplankton model.
DOI: https://doi.org/10.1111/j.1600-0870.2007.00246.x | Journal eISSN: 3035-9554
Language: English
Page range: 749 - 757
Submitted on: Nov 6, 2006
Accepted on: Mar 1, 2007
Published on: Jan 1, 2007
Published by: Stockholm University Press
In partnership with: Paradigm Publishing Services
© 2007 Keston W. Smith, published by Stockholm University Press
This work is licensed under the Creative Commons Attribution 4.0 License.
