Incoprea Theory
Volume IV · 2026

Theory 14/Cryosphere & Ice /Glaciology/3 min

Ice Loss Is Episodic

Ice loss is modelled as a smooth function of temperature. The record shows steps.

The claim

Glacier and ice-sheet mass loss happens in discrete drainage switches rather than continuously, so seasonal and annual loss should be predictable from the timing of switches and not from the integrated melt curve.

01The evidence

  • The subglacial drainage system has been observed to grow and collapse abruptly, with each transition changing basal motion, which is the mechanism an episodic account requires.Kennicott Glacier hydrology
  • Subglacial water pressure changes are discontinuous in time, which means the forcing of basal motion is not a smooth function of surface melt.Water pressure overview

02The tension

Mass balance is computed by integrating surface melt and runoff over the season, which produces a smooth curve. Observations show step changes, and the steps are treated as noise around the trend rather than as the signal. This matters because projections inherit the smoothness assumption.

03The theory

A drainage system is bistable: distributed and channelised states are both stable, and the transition between them is fast. Each transition changes effective pressure, and therefore sliding, abruptly.

If loss is the sum of these switches, then the annual total is set by how many switches occurred and when, not by the integral of melt. Two years with identical melt can differ in loss if the switches are timed differently.

The prediction is that loss should be predictable from switch timing, and that the distribution of annual loss should be heavy-tailed because switches are threshold events. It also predicts that a warming climate will change the frequency of switches rather than simply scaling the melt curve.

This matters for projections because a model that smooths the switches will underestimate variance and mis-time the largest loss years, which are the ones that matter for coastal planning.

04What this is not

Integrated mass-balance models

Those models integrate melt and runoff, producing smooth loss. This treats loss as a sum of discrete transitions, which predicts heavy tails and a different response to warming.

05Predictions

  • P1Annual mass loss should be predictable from the number and timing of drainage switches.
  • P2The distribution of annual loss should be heavy-tailed.
  • P3Warming should change switch frequency more than it changes the melt integral.

06What would refute it

Loss tracking integrated melt with no step signature would restore the smooth model.

07What it would change

It would change how ice loss is projected and how its uncertainty is reported.

08Evidence base

  1. Growth and collapse of the subglacial drainage system, Kennicott Glacierhttps://scholar.google.be/citations?view_op=view_citation&hl=fr&user=MIBKAWwAAAAJ&citation_for_view=MIBKAWwAAAAJ:u-x6o8ySG0sC
  2. Water pressure — overviewhttps://www.sciencedirect.com/topics/earth-and-planetary-sciences/water-pressure