Atmosphere Ocean Science Colloquium

Seasonal Prediction and Predictability of Arctic Sea Ice in a Dynamical Forecast System

Speaker: Mitch Bushuk, GFDL

Location: Warren Weaver Hall 1302

Date: Wednesday, October 12, 2016, 3:30 p.m.


The rapid decline at Arctic sea-ice extent (SIE) has created an increased need for skillful seasonal predictions of sea ice. Currently there is a significant gap between the potential predictability of SIE and the forecast skill of operational prediction systems. In this work, we explore avenues for closing this gap in the context of the GFDL prediction system. First, using a 700-year control integration and a suite of initialized forecast ensemble experiments, we identify sea-ice thickness and the ice-albedo feedback as key sources of predictability for summer SIE. We find that thickness anomalies are persistent on interannual timescales and, moreover, that these anomalies are enhanced over the summer months by a positive feedback the sea-ice state and surface albedo. Second, we move towards stakeholder-relevant spatial scales, investing regional SIE prediction skill in a suite of retrospective seasonal forecasts spanning 1981-2015. The regional SIE prediction skill scores are highly region and target month dependent, but generically exceed the skill of a damped persistence forecast. Notably high skill is found for winter and spring SIE in the Barents and Labrador Seas, which is partially attributable to the model's initialization and persistence of subsurface ocean temperature anomalies.