The second trilogy, below, this time without Rudebusch, also treats Arctic sea ice forecasting, but from a real-time, fixed-target (September), perspective. All of the work came under the umbrella of the Arctic Research Consortium of the United States (ARCUS), and its Sea Ice Prediction Network (SIPN), which oversaw the Sea Ice Outlook (SIO).
More on ARCUS in the next post. But for now let's look at the second trilogy.
The basic feature-engineered real-time forecasting approach is proposed in
Diebold, F.X. and Gobel, M. (2022), “A Benchmark Model for Fixed-Target Arctic Sea Ice Forecasting,” Economics Letters, 215, 110478,
and it is compared to a feature-engineered real-time machine learning approach in
Diebold, F.X., Goebel, M., and Goulet Coulombe, P. (2023), “Assessing and Comparing Fixed-Target Forecasts of Arctic Sea Ice: Glide Charts for Feature-Engineered Linear Regression and Machine Learning Models,” Energy Economics, 124, 106833.
The trilogy culminates with a 61-author (must be real science!) assessment of June-September real-time Arctic sea ice forecasting across many models and years, including the above Diebold et al approach:
Bushuk, M., et al. (2024), “Predicting September Arctic Sea Ice: A Multi-Model Seasonal Skill Comparison,” Bulletin of the American Meteorological Society, 105, 1170-1203.
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.