Seasonal-to-Decadal Predictability of Arctic Sea Ice

点击翻译按钮获取中文摘要

With the opening of the summer pack ice in 2007, there has been an increased interest in short-term, seasonal and decadal forecasting or projections of sea ice conditions. Methods used in the community to study the predictability of the system on these time scales include coupled ice-ocean models forced with atmospheric fields from reanalyses (with or without data assimilation), fully coupled General Circulation Models with initial conditions derived from observations, or statistical models and heuristic or mixed approaches. For short-term forecasting (1-2 days), thermodynamic effects are less important and fully coupled models are not necessary. In this case, forecasting skill relies on the specification of accurate initial conditions, surface atmospheric stresses and an accurate representation of sea ice and surface ocean physics. For seasonal forecasting, accurate initial conditions are again crucial along with a good representation of ice-ocean-atmosphere physics and feedbacks. For decadal projections, the sensitivity to initial conditions is lost, and predictive skill relies on an accurate representation of the ice-ocean-atmosphere physics, as well as accurate representation of internal variability and external forcing such as CO2, ozone, aerosol, solar irradiance at the top-of-the-atmosphere, etc. The long-term objectives of this research program are to improve short-term to decadal forecasts or projections of sea ice in the Arctic Ocean. The specific objectives of the proposal are to improve sea ice physics (drift and deformation); to quantify potential predictability from observed quantities and reanalysis data (ice speed, concentration, thickness and lower/upper atmospheric circulation); to identify key metrics responsible for predictability in observations and assess whether those metrics are correctly simulated in Global Climate Models (pointing to area of model improvements and improved forecasts); to quantify potential improvements in predictability associated with ocean heat flux (both lateral in the Bering Strait and Barents Sea Opening and vertical via convection and Ekman dynamics). The completion of this work will lead to a better understanding of forecast skill and potential predictability. The novelty in this work is that it assesses predictability directly from observations. This is crucial given that small biases in the simulated mean state of models can lead to significant loss of predictability. This research program will place Canada in a unique position to improve sea ice forecasts for local populations and industry, allowing for safer operations in an Arctic, that is becoming increasingly accessible to navigation. It will also reduce uncertainties in the timeline for when these changes are more likely to occur.