A Regional Seasonal Forecast Model of Arctic Minimum Sea Ice Extent: Reflected Solar Radiation vs. Late Winter Coastal Divergence
AbstractThinning sea ice cover in the Arctic is associated with larger interannual variability in the minimum Sea Ice Extent (SIE). The current generation of forced or fully coupled models, however, have difficulty predicting SIE anomalies from the long-term trend, highlighting the need to better identify the mechanisms involved in the seasonal evolution of sea ice cover. One such mechanism is Coastal Divergence (CD), a proxy for ice thickness anomalies based on late winter ice motion, quantified using Lagrangian ice tracking. CD gains predictive skill through the positive feedback of surface albedo anomalies, mirrored in Reflected Solar Radiation (RSR), during melt season. Exploring the dynamic and thermodynamic contributions to minimum SIE predictability, RSR, initial SIE (iSIE) and CD are compared as predictors using a regional seasonal sea ice forecast model for July 1, June 1 and May 1 forecast dates for all Arctic peripheral seas. The predictive skill of June RSR anomalies mainly originates from open water fraction at the surface, i.e. June iSIE and June RSR have equal predictive skill for most seas. The finding is supported by the surprising positive correlation found between June Melt Pond Fraction (MPF) and June RSR in all peripheral seas: MPF anomalies indicate presence of ice or open water that is key to creating minimum SIE anomalies. This contradicts models that show correlation between melt onset, MPF and the minimum SIE. A hindcast model shows that for a May 1 forecast, CD anomalies have better predictive skill than RSR anomalies for most peripheral seas.