Collaborative Research: Advancing predictive understanding of summertime Arctic sea ice cover

Understanding how to predict changes in the Arctic environment, especially sea ice variations, is crucial because these changes have big impacts on economies and societies both locally and globally. This project focuses on developing new ways to forecast Arctic sea ice during summer when sea ice is melting and reaches its minimum, looking at time periods from a few weeks to an entire season. It also aims to determine how far into the future these predictions can be made. The amount of summer sea ice in the Arctic is influenced by many factors, ranging from daily weather changes to long-term shifts in global wind patterns, affected by slowly changing ocean temperatures around the world. However, the understanding of how these factors interact is still limited. This limitation comes from the short duration of reliable satellite data monitoring and the complexity of the connections between these elements, which are challenging for traditional climate models or simple statistics to interpret. This project will use advanced and sophisticated machine learning methods to potentially improve predictions of Arctic summer sea ice. Additionally, the project will provide college students with opportunities to learn across different polar and climate science topics, leveraging the resources and expertise of the participating institutions. Preconditioning of sea ice before the summer months has long been recognized as a vital predictor of September's Arctic sea ice extent. The dynamic interactions between ice, ocean, and atmosphere are also major contributors to the changes observed in summer sea ice. The researchers will examine the impacts from external climate components and how they interact with the persistent local conditions before the summer season, which has not been fully considered in previous studies. This project will develop models of regional Arctic sea ice coverage based on a diverse array of observational data at a global scale, integrated by an advanced machine learning method. This approach aims to capture the complex, non-linear variations in both local and remote influences across timescales in a global context. The investigators will conduct a series of meticulously designed reforecast experiments to isolate and quantify the influence of various physical drivers on summertime Arctic ice within the predictive framework. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.