Abstract
In recent decades, warming temperatures have caused sharp reductions in the volume of sea ice in the Arctic Ocean. Predicting changes in Arctic sea ice thickness is vital in a changing Arctic for making decisions about shipping and resource management in the region. We propose a statistical spatio-temporal two-stage model for sea ice thickness and use it to generate probabilistic forecasts up to three months into the future. Our approach combines a contour model to predict the ice-covered region with a Gaussian random field to model ice thickness conditional on the ice-covered region. Using the most complete estimates of sea ice thickness currently available, we apply our method to forecast Arctic sea ice thickness. Point predictions and prediction intervals from our model offer comparable accuracy and improved calibration compared with existing forecasts. We show that existing forecasts produced by ensembles of deterministic dynamic models can have large errors and poor calibration. We also show that our statistical model can generate good forecasts of aggregate quantities such as overall and regional sea ice volume. Supplementary materials accompanying this paper appear on-line.
Highlights
Over the past few decades, the sea ice covering much of the Arctic Ocean has retreated dramatically
In the first stage of our model, we propose a spatio-temporal extension of the contour model introduced by Director et al (2021) to forecast which locations will be covered by sea ice in a given month
For each validation month and lead time, we compare our statistical model-based forecasts with the European Centre for Medium-Range Weather Forecasts (ECMWF) ensemble forecast, a climatological forecast, and a forecast based on a non-spatial conditional thickness model
Summary
Over the past few decades, the sea ice covering much of the Arctic Ocean has retreated dramatically. We seek probabilistic forecasts of thickness that are marginally calibrated at individual location/time pairs, and accurately estimate spatial and temporal correlations in forecast errors Accounting for these correlations is crucial for making probabilistic forecasts of aggregate quantities like sea ice volume. More flexible models will be better able to account for this complexity, but in general will demand added computational resources To address these challenges, we propose a probabilistic, spatio-temporal two-stage thickness (TST) model. We carry out approximate Bayesian inference for both stages of the model using integrated nested Laplace approximations (Rue et al 2009) This yields probabilistic forecasts of both sea ice thickness at individual locations and times as well as aggregate quantities like sea ice volume.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
