Citation

Andersson LE, Scibilia F, Imsland L. Cold Reg. Sci. Technol. 2019; 158: 128-142.

Copyright

(Copyright © 2019, Elsevier Publishing)

DOI

10.1016/j.coldregions.2018.11.016

PMID

unavailable

Abstract

This article proposes a statistical model for short-term iceberg drift forecasts by transforming the problem of forecasting the iceberg velocity into a problem of forecasting the ocean current velocity. A Vector-autoregression model is identified using historical ocean current data as a training set. The proposed forecast scheme is tested and analysed on four real iceberg drift trajectories. Based on these results, recommendations about the forecast horizon, the filter horizon and model order are given. Moreover, it is shown that the statistical forecast approach presented in this article offers superior performance to a conventional dynamic iceberg forecast model for short-term drift forecasts.

Language: en

Keywords

Iceberg drift forecasting; Model identification; Moving horizon estimator; Multivariate empirical mode decomposition; Offshore operations; Parameter estimation