Arctic sea ice extent has a continuous decreasing trend in all months, according to the satellite record. Recently, this decreasing trend has been accelerating, especially in September (Stroeve et al., 2012). Ice-free summers might be realized as early as 2030 (Stroeve et al., 2007). Then, why does sea ice melt more and more? There are many linked processes, involving increasing GHG, and internal variability.

 

Sea ice exhibits a nonlinear response to external climate forcing. GHG forcing thins spring ice and causes earlier melt in late-spring. This leads to an enhanced ice-albedo feedback and more open water in autumn. Larger areas of open water can then warm winter atmospheric temperatures through higher heat fluxes. This, in turn, further intensifies the thinning of spring sea ice (Stroeve et al., 2012). And the loop continues… 

 

Also, changes in the atmospheric and oceanic variability can lead to an accelerating ice melt. Recent work has analyzed the role of North Atlantic Oscillation (NAO), Pacific Decadal Oscillation (PDO), as well as Arctic Dipole (DA) pattern in the recent sea ice decline (Serreze et al., 2007; Wang et al., 2009; Zhang et al., 2008). In addition to atmospheric processes, ocean heat transport from the North Atlantic and the North Pacific can change the sea ice variability (Steele et al., 1998; Polyakov et al., 2005).

 

Despite the recent research process, there are many knowledge gaps in our understanding of sea ice. Our research focuses on improving mechanistic understanding of factors leading to the recent sea ice decline.

 

The second aspect of our sea analysis research is future projections. Right now, projections are made on the basis of climate models. Unfortunately, these models are dependent, and share code, and parameterizations. However, there is a lack of statistical tools that can deal with model dependence. We have developed a novel method that can take into account all levels of model dependence (e.g., how any model depends on any cluster of other climate models). In fact, the method can be used more generally outside of the field for testing multiple non-exclusive hypotheses, with potential applications in finance, medicine, or social sciences. Our research focuses on improving future probabilistic projections of sea ice and other environmental variables using this, and other novel statistical methods.