Uncovering a remarkably universal spring predictability barrier in global climate models
Key Points & Overview
- A simple model that uses sea ice volume to predict sea ice area in the Arctic is employed to evaluate Arctic sea-ice predictability across global climate models
- Nearly all global climate models exhibit a spring predictability barrier for regional forecasts of Arctic sea ice
- Sea ice forecasts for the marginal seas of the Arctic basin that are initialized prior to June 1 will have substantially less skill than forecasts initialized afterwards
The Arctic has undergone substantial changes over the last few decades. Near-surface air temperatures have risen at approximately twice the rate of the global mean. In terms of sea ice, there has been a sharp decline in its summertime coverage and thinning across all months. There has also been a substantial loss of multi-year sea ice. Though most of these changes are striking emblems of anthropogenic climate change, they have also fueled considerable interest in the predictability of Arctic sea ice. Given the fundamental role that sea ice plays in the climate system, terrestrial and marine ecosystems, and human population, skillful forecasts of Arctic sea ice are valuable to a broad range of stakeholders.
Recent work assessing sea ice predictability has shown that there is a barrier of prediction skill in the springtime that causes Arctic sea ice forecasts initialized prior to May to be less skillful than forecasts initialized afterwards. Yet, this spring predictability barrier has only been identified in two global climate models. A key question that has remained unanswered is: does this spring predicability barrier exist across other global climate models?
In a recent paper published in Geophysical Research Letters, I argue, along with Mitchell Bushuk and Michael Winton (who are scientists at the Geophysical Fluid Dynamics Laboratory (GFDL) in Princeton, New Jersey), that this spring predicability barrier is universal across global climate models. We first show that the dominant source of summertime prediction skill for Arctic sea ice results from sea ice volume. Since these two quantities can be easily calculated, this comparison allows us to assess summer sea ice predictability across all global climate models using a simple linear regression model. An assessment of many different Arctic regions shows that substantial decorrelation occurs as sea ice volume transitions from June to May in each global climate model.
This suggests forecasts of sea ice in the marginal seas of the Arctic basin that are initialized prior to June 1 will have substantially less skill than forecasts initialized afterwards. Perhaps most notably is that these regions encompass most viable summer shipping routes in the Arctic, suggesting this spring predictability barrier with limit accurate forecasts for end users.