ACPD

Atmospheric Chemistry and Physics Discussions

ACPD

1680-7375

Copernicus GmbH

Göttingen, Germany

10.5194/acpd-4-3653-2004

Using an extended Kalman filter learning algorithm for feed-forward neural networks to describe tracer correlations

Lary

D. J.

¹ ² ³ Mussa

H. Y.

Global Modelling and Assimilation Office, NASA Goddard Space Flight Center, USA

GEST at the University of Maryland Baltimore County, MD, USA

Unilever Cambridge Centre, Department of Chemistry, University of Cambridge, United Kingdom

30 06 2004

4 3 3653 3667

This is an open-access article ditributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

This article is available from http://www.atmos-chem-phys-discuss.net/4/3653/2004/acpd-4-3653-2004.html

The full text article is available as a PDF file from http://www.atmos-chem-phys-discuss.net/4/3653/2004/acpd-4-3653-2004.pdf

In this study a new extended Kalman filter (EKF) learning algorithm for feed-forward neural networks (FFN) is used. With the EKF approach, the training of the FFN can be seen as state estimation for a non-linear stationary process. The EKF method gives excellent convergence performances provided that there is enough computer core memory and that the machine precision is high. Neural networks are ideally suited to describe the spatial and temporal dependence of tracer-tracer correlations. The neural network performs well even in regions where the correlations are less compact and normally a family of correlation curves would be required. For example, the CH4-N2O correlation can be well described using a neural network trained with the latitude, pressure, time of year, and CH4 volume mixing ratio (v.m.r.). The neural network was able to reproduce the CH4-N2O correlation with a correlation coefficient between simulated and training values of 0.9997. The neural network Fortran code used is available for download.