DST Index Prediction using Joint and Dual Unscented Kalman Filter

B. Kaewkham-ai (Thailand) and R.F. Harrison (UK)

Keywords

joint estimation; dual estimation; unscented Kalman filter; Dst index

Abstract

This paper presents a short-term prediction of the distur bance storm time (Dst) index using unscented Kalman fil ter. Joint and dual estimation methods are studied to exam ine an improvement of Dst index prediction by estimating model parameters and updating recursively. Comparison between these techniques and a fixed model parameter pre diction are made in terms of root mean square error (rmse). It is found that joint and dual estimation methods give less rmse than state estimation alone for all Dst range, whereas state estimation alone shows better performance than joint and dual estimation for Dst below −80 nT.

Important Links:



Go Back