The Relative Impacts of Initialization and Climate Forcing in Coupled Ice Sheet‐Ocean Modeling: Application to Pope, Smith, and Kohler Glaciers

Abstract

Coupled ice sheet‐ocean models are beginning to be used to study the response of ice sheets to ocean warming. Initializing an ice‐ocean model is challenging and can introduce nonphysical transients, and the extent to which such transients can affect model projections is unclear. We use a synchronously‐coupled ice‐ocean model to investigate evolution of Pope, Smith and Kohler Glaciers, West Antarctica, over the next half‐century. Two methods of initialization are used: In one, the ice‐sheet model is constrained with observed velocities in its initial state; in another, the model is constrained with both velocities and grounded thinning rates over a 4‐year period. Each method is applied to two basal sliding laws. For each resulting initialization, two climate scenarios are considered: one where ocean conditions during the initialization period persist indefinitely, and one where the ocean is in a permanent “warm” state. At first, model runs initialized with thinning data exhibit volume loss rates much closer to observed values than those initialized with velocity only, but after 1–2 decades, the forcing primarily determines rates of volume loss and grounding line retreat. Such behavior is seen for both basal sliding laws, although volume loss rates differ quantitatively. Under the “warm” scenario, a grounding line retreat of ∼30 km is simulated for Smith and Kohler, although variation in total retreat due to initialization is nearly as large as that due to forcing. Furthermore it is questionable whether retreat will continue due to narrowing of submarine troughs and limiting of heat transport by bathymetric obstacles.

A framework for time-dependent Ice Sheet Uncertainty Quantification, applied to three West Antarctic ice streams

“…4.2). This effect may be less important for ice streams strongly coupled to ocean forcing (Lilien et al, 2019;Goldberg and Holland, 2022), but could be more influential for unstable margins (Joughin et al, 2014). We use such difference to quantify (in a robust way) the variance that projections of VAF are expected to have after 40 years and identify prior strengths that can reproduce this expected variability (see Sect.…”

“…1, covers part of the Amundsen Sea Embayment (ASE) in West Antarctica and includes three ice streams: Pope, Smith and Kohler Glaciers (PSK), as well as, the Dotson and Crosson ice shelves. PSK glaciers have exhibited some of the highest retreat rates in Antarctica throughout the satellite observing record, with their grounding lines receding over 30 km in recent decades (Scheuchl et al, 2016;Goldberg and Holland, 2022). Their catchment can potentially contribute up to 6 cm to the global mean sea level (Morlighem et al, 2020), double the global mean sea level contribution of the inventory of Earth's mountain glaciers (when excluding the Antarctic and Greenland periphery, Hock et al, 2023).…”

“…Their catchment can potentially contribute up to 6 cm to the global mean sea level (Morlighem et al, 2020), double the global mean sea level contribution of the inventory of Earth's mountain glaciers (when excluding the Antarctic and Greenland periphery, Hock et al, 2023). A complete collapse of the ice shelves in this area would likely lead to accelerated mass loss from adjacent ice streams, including Thwaites Glacier (Goldberg and Holland, 2022). Previous modelling studies have shown that past and future retreat of these glaciers is strongly tied to ocean-forced melting, but that the method of calibration may affect projected rates of ice loss as well (Lilien et al, 2019;Goldberg and Holland, 2022).…”

“…A complete collapse of the ice shelves in this area would likely lead to accelerated mass loss from adjacent ice streams, including Thwaites Glacier (Goldberg and Holland, 2022). Previous modelling studies have shown that past and future retreat of these glaciers is strongly tied to ocean-forced melting, but that the method of calibration may affect projected rates of ice loss as well (Lilien et al, 2019;Goldberg and Holland, 2022). As such, and due to the vast quantity of data available for this region, we choose this area to test our model error framework in a realistic setting.…”

“…The initial guess for β, β bgd , is generated from the temperature dataset of Pattyn (2010). Based on coupled ice sheet-ocean modelling for the region (Goldberg and Holland, 2022), spatially uniform melt parameters of M max = 30 m yr −1 and z th = 600 m were chosen.…”