need for long-term and sustained observations, contingency plans for satellite systems, and better coordination within the modeling, observational, and stakeholder communities.
The modeling panel discussion focused on the challenges in modeling and ways to improve interaction with the observational community. Cecilia Bitz of the University of Washington suggested that trying to initialize a model based entirely on observations is a current gap in our understanding. Fully coupled numerical and statistical models will need to be used for prediction, and observations are needed for initial conditions and validation of predictions. She indicated that an important path forward is to communicate the limits of predictability to stakeholders and others, and to use models to determine the most needed measurements. A focus on the coupling between ice, atmosphere, and the ocean was discussed by Elizabeth Hunke from Los Alamos National Laboratory. She indicated that better observations are needed (particularly snow on ice) to continue to make model improvements. She also noted that sea ice predictability is critically dependent on the predictability of the applied forcing and the ice equilibrium state associated with the applied forcing. The strength of feedbacks (including atmosphere and ocean fluxes) needs to be better understood.
Andrey Proshutinsky from the Woods Hole Oceanographic Institution commented on the oceans role in sea ice changes (including three components of influence: atmospheric circulation changes, heat release and ice melt, and sea ice dynamics). He noted that the modeling challenges include reanalysis and reconstruction of sea ice and ocean conditions, implementation of high-resolution models, climate and process studies based on modeling, and systematic model calibration and validation. Possible advances in modeling could address these issues. Examples include increased model resolution, development of a landfast ice model, inclusion of tidal and atmospheric pressure forcing, and improvements of data assimilation methods. Like many of the other panelists, he also mentioned the importance of collaboration with the observational community and stakeholders.
During the breakout group sessions, participants noted that there is a strong stakeholder need for both seasonal predictions (used in planning for fishing, research cruises, industry activities, etc.) and decadal predictions (used in infrastructure planning, national security planning, environmental assessments, endangered species status, etc.). The groups indicated that making predictions on the interannual time frame is particularly difficult. Suggested next steps include work on the question of atmospheric forcing (this can help bridge the gap between the seasonal and decadal timescales), and on the treatment of the ocean in models (models cannot currently resolve vertical stratification and heat fluxes). Other topics that the groups discussed include the transition from first-year ice to multiyear ice, the type of model (statistical, physical) and initializations that