Analysis of an Arctic sea ice loss model in the limit of a discontinuous albedo

TL;DR

This paper investigates Arctic sea ice loss models with a focus on bifurcation and hysteresis, analyzing the effects of a discontinuous albedo feedback limit to understand model behavior and parameter influences.

Contribution

It introduces a novel analysis of an Arctic energy balance model in the discontinuous albedo limit, revealing bifurcation behaviors and parameter effects not seen in smoothed models.

Findings

Discontinuous albedo models can replicate bifurcation behaviors of smoothed models.

Uncovered parameter sets leading to transitions from perennial ice to seasonal ice-free states.

Demonstrated the relevance of sliding intervals in understanding hysteresis loops.

Abstract

As Arctic sea ice extent decreases with increasing greenhouse gases, there is a growing interest in whether there could be a bifurcation associated with its loss, and whether there is significant hysteresis associated with that bifurcation. A challenge in answering this question is that the bifurcation behavior of certain Arctic energy balance models have been shown to be sensitive to how ice-albedo feedback is parameterized. We analyze an Arctic energy balance model in the limit as a smoothing parameter associated with ice-albedo feedback tends to zero, which introduces a discontinuity boundary to the dynamical systems model. Our analysis provides a case study where we use the system in this limit to guide the investigation of bifurcation behavior of the original albedo-smoothed system. In this case study, we demonstrate that certain qualitative bifurcation behaviors of the albedo-smoothed system can have counterparts in the limit with no albedo smoothing. We use this perspective to systematically explore the parameter space of the model. For example, we uncover parameter sets for which the largest transition, with increasing greenhouse gases, is from a perennially ice-covered Arctic to a seasonally ice-free state, an unusual bifurcation scenario that persists even when albedo-smoothing is reintroduced. This analysis provides an alternative perspective on how parameters of the model affect bifurcation behavior. We expect our approach, which exploits the width of repelling sliding intervals for understanding the hysteresis loops, would carry over to other positive feedback systems with a similar natural piecewise-smooth limit, and when the feedback strength is likewise modulated with seasons or other periodic forcing.