Extended-range arctic sea ice forecast with convolutional long short-Term memory networks

Yang Liu*, Laurens Bogaardt, Jisk Attema, Wilco Hazeleger

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review

Abstract

Operational Arctic sea ice forecasts are of crucial importance to science and to society in the Arctic region. Currently, statistical and numerical climate models are widely used to generate the Arctic sea ice forecasts at weather time scales. Numerical models require near-real-Time input of relevant environmental conditions consistent with the model equations and they are computationally expensive. In this study, we propose a deep learning approach, namely convolutional long short-Term memory networks (ConvLSTM), to forecast sea ice in the Barents Sea at weather to subseasonal time scales. This is an unsupervised learning approach. It makes use of historical records and it exploits the covariances between different variables, including spatial and temporal relations. With input fields from reanalysis data, we demonstrate that ConvLSTM is able to learn the variability of the Arctic sea ice and can forecast regional sea ice concentration skillfully at weekly to monthly time scales. It preserves the physical consistency between predictors and predictands, and generally outperforms forecasts with climatology, persistence, and a statistical model. Based on the known sources of predictability, sensitivity tests with different climate fields as input for learning were performed. The impact of different predictors on the quality of the forecasts are evaluated and we demonstrate that the surface energy budget components have a large impact on the predictability of sea ice at weather time scales. This method is a promising way to enhance operational Arctic sea ice forecasting in the near future.

Original languageEnglish
Pages (from-to)1673-1693
Number of pages21
JournalMonthly Weather Review
Volume149
Issue number6
DOIs
Publication statusPublished - 1 Jun 2021

Keywords

  • Deep learning
  • Machine learning
  • Sea ice
  • Statistical forecasting

Fingerprint

Dive into the research topics of 'Extended-range arctic sea ice forecast with convolutional long short-Term memory networks'. Together they form a unique fingerprint.

Cite this