Due to the rapid decline in Arctic sea ice extent and volume over the past decade, there has been a growing focus on developing capabilities for prediction. NOAAÎ“Ã‡Ã–s Arctic Action Plan calls for an improvement in sea ice predictability ranging from the short term (e.g. daily and weekly) to seasonal to decadal time-scales. Such an effort is particularly important in light of the large variability seen in annual sea ice minima. To gain predictive skill, one must gain an understanding of sea ice variability and the coupled terrestrial, ocean and atmospheric systems that influence this variability.
We propose to advance the understanding of Arctic sea ice variability and predictability by investigating several interrelated items that have been largely overlooked, but that we hypothesize will give further insight to the seasonal fate of sea ice. These items include (a) the influence of spring and early summer snow cover over Northern Hemisphere lands, (b) the atmospheric circulation patterns that favor ice melt, their precursors and mechanisms by which the atmosphere interacts with snow to impact the sea ice and (c) the ability of models to capture observed relationships. We will also explore additional relationships between the sea ice melt season and quantities such as melt onset date, atmospheric moisture content, and winter ice dynamics
.Speculation regarding relationships between terrestrial snow and sea ice dates back at least 20 years, but there has been no systematic investigation of mechanisms relating the two quantities at regional scales. To fill this gap, we will apply innovative tools such as complex network analysis, which can provide insight into the spatial relationships between various nodes (in this case, gridboxes of snow and ice cover) of a network (the snow and sea ice system). Using such techniques we will search for predictive power amongst snow variables, as well as a multitude of information sources such as atmospheric circulation patterns or ice melt onset date. NOAA climate data records and reanalysis products (e.g. NOAAÎ“Ã‡Ã–s CFSR) will be used in our analysis. We also propose to gain an understanding of model abilities, particularly the CMIP5 models and the NOAA Climate Forecast System (CFS), by determining whether they can reproduce observed linkages. Model deficiencies can point to structural issues, which in turn can lead to improvements. Results from the SEARCH Sea Ice Outlook (SIO; now part of the new Sea Ice Prediction Network (SIPN) led by PI J. Stroeve) indicate that there is much room for improvement within current modeling and prediction systems.
This work is directly responsive to this NOAA proposal call in that it seeks to understand where predictability can arise from and how that understanding may be applied in a forecasting context. Greater understanding of sea ice within the Arctic system is one of the goals of NOAAÎ“Ã‡Ã–s Next-Generation Strategic Plan and its central mission to understand and predict changes in climate, weather, oceans and coasts.