摘要：This study introduces a Bayesian logistic regression framework that is capable of providing skillful probabilistic forecasts of Arctic sea ice cover, along with quantifying the attendant uncertainties. The presence or absence of ice (absence defined as ice concentration below 15%) is modeled using a categorical regression model, with atmospheric, oceanic, and sea ice covariates at 1‐ to 7‐month lead times. The model parameters are estimated in a Bayesian framework, thus enabling the posterior predictive probabilities of the minimum sea ice cover and parametric uncertainty quantification. The model is fitted and validated to September minimum sea ice cover data from 1980 through 2018. Results show overall skillful forecasts of the minimum sea ice cover at all lead times, with higher skills at shorter lead times, along with a direct measure of forecast uncertainty to aide in assessing the reliability. Plain Language Abstract Every summer, sea ice in the Arctic undergoes melt and retreat, allowing access to otherwise difficult to reach areas. This has sparked growing interest in short‐ and long‐term forecasting of summer sea ice to assist in planning and preparation of logistically intensive Arctic expeditions. Currently, forecasts more than 3 months in advance tend to be less skillful than forecasts made less than 3 months in advance. This study presents a novel approach to seasonal probabilistic forecasts of the minimum September sea ice cover through regression analysis, relating minimum summer sea ice to winter and spring sea ice, atmospheric, and oceanic conditions. We use “skill scores” to evaluate how well our forecasts perform in a variety of circumstances. We find that this method is able to skillfully predict up to 7 months early the probability that sea ice will be present across the entire Arctic Ocean at the summer minimum. This means that stakeholders interested in access to the Arctic Ocean during summer can have reliable long‐term forecasts to aide in planning and preparation.
