A probabilistic framework for quantifying the role of anthropogenic climateforcing in marine-terminating glacier retreats

Many marine-terminating outlet glaciers have retreated rapidly in recent decades, but these changes have not been formally attributed to anthropogenic climate change. A key challenge for such an attribution assessment is that if glacier termini are sufficiently perturbed from bathymetric highs, ice-dynamic feedbacks can cause rapid retreat even without further climate forcing. In the presence of internal climate variability, attribution thus depends on understanding whether (or how frequently) these rapid retreats could be triggered by climatic noise alone.

We present simulations with idealized glaciers to analyze glacier variability in the presence of topographic thresholds, and to demonstrate a framework for attribution. We find that when termini are positioned near bed peaks in a noisy climate, rapid retreat is a stochastic phenomenon. We therefore assess the likelihood of rapid retreat, using ensembles of many simulations with independent realizations of random climate variability. Synthetic experiments show that century-scale climate trends substantially increase the likelihood of retreat. The strength of this effect is related to the timescales over which ice dynamics integrate forcing, implying that the time of onset of anthropogenic forcing is a key factor to constrain for attribution studies. We close by discussing broader considerations for framing attribution studies on marine-terminating glacier retreat, and ongoing work towards applying this framework to glaciers in Greenland.