摘要：Due to the abundance of observational datasets collected since the onset of its retreat (c. 1983), Columbia Glacier, Alaska, provides an exciting modeling target. We perform Monte Carlo simulations of the form and flow of Columbia Glacier, using a 1-D (depth-integrated) flowline model, over a wide range of parameter values and forcings. An ensemble filter is imposed following spin-up to ensure that only simulations that accurately reproduce observed pre-retreat glacier geometry are retained; all other simulations are discarded. The selected ensemble of simulations reasonably reproduces numerous highly transient post-retreat observed datasets. The selected ensemble mean projection suggests that Columbia Glacier will achieve a new dynamic equilibrium (i.e. "stable") ice geometry c. 2020, at which time iceberg calving rate will have returned to approximately pre-retreat values. Comparison of the observed 1957 and 2007 glacier geometries with the projected 2100 glacier geometry suggests that Columbia Glacier had already discharged ~82% of its projected 1957–2100 sea level rise contribution by 2007. This case study therefore highlights the difficulties associated with the future extrapolation of observed glacier mass loss rates that are dominated by iceberg calving.
