Assimilating the ICE‐6G_C Reconstruction of the Latest Quaternary Ice Age Cycle Into Numerical Simulations of the Laurentide and Fennoscandian Ice Sheets

Abstract

We analyze the effects of nudging 100 kyr numerical simulations of the Laurentide and Fennoscandian ice sheets toward the glacial isostatic adjustment‐based (GIA‐based) ICE‐6G_C reconstruction of the most recent ice age cycle. Starting with the ice physics approximations of the PISM ice sheet model and the SeaRISE simulation protocols, we incorporate nudging at characteristic time scales, τf, through anomalous mass balance terms in the ice mass conservation equation. As should be expected, these mass balances exhibit physically unrealistic details arising from pure GIA‐based reconstruction geometry when nudging is very strong (τf=20 years for North America), while weakly nudged (τf=1,000 years) solutions deviate from ICE‐6G_C sufficiently to degrade its observational fit quality. For reasonable intermediate time scales (τf=100 years and 200 years), we perturbatively analyze nudged ice dynamics as a superposition of “leading‐order smoothing” that diffuses ICE‐6G_C in a physically and observationally consistent manner and “higher‐order” deviations arising, for instance, from biases in the time dependence of surface climate boundary conditions. Based upon the relative deviations between respective nudged simulations in which these biases follow surface temperature from ice cores and eustatic sea level from marine sediment cores, we compute “ice core climate adjustments” that suggest how local paleoclimate observations may be applied to the systematic refinement of ICE‐6G_C. Our results are consistent with a growing body of evidence suggesting that the geographical origins of Meltwater Pulse 1B (MWP1b) may lie primarily in North America as opposed to Antarctica (as reconstructed in ICE‐6G_C).

Simulating the Early Holocene demise of the Laurentide Ice Sheet with BISICLES (public trunk revision 3298)

“…Ice thickness in this reconstruction is known to be inconsistent with fundamental ice sheet dynamics (Stuhne and Peltier, 2017) and is also likely inconsistent with the early Holocene climate simulated by HadCM3 because the climate and ice sheet were not generated in fully (i.e. two-way) coupled framework (e.g.…”

A Simple Model for Deglacial Meltwater Pulses

“…The amplitude and timing of meltwater pulses simulated with complex 3‐D ice sheet models (Abe‐Ouchi et al, ; Gregoire et al, ; Heinemann et al, ) differ significantly between models and often are in disagreement with observationally‐derived sea level records. In models that are parametrically tuned to match records of sea level and ice sheet extent (Gregoire et al, ; Stuhne & Peltier, ; Tarasov et al, ), the agreement is better (by design), but the particular physical process controlling the characteristics of meltwater pulses is unclear. In this section, we show how the parameters and initial conditions in our simple model set the amplitude and timing of meltwater pulses.…”

“…Gregoire et al () first showed that simulated intense surface melting in the ice saddle region between the Laurentide and Cordilleran ice sheets is associated with a brief acceleration in deglacial sea level rise. Subsequent studies (Abe‐Ouchi et al, ; Gregoire et al, ; Heinemann et al, ; Stuhne & Peltier, ; Tarasov et al, ) have simulated similar ice saddle collapse events during the deglaciation of the Laurentide and Cordilleran ice sheets, though the timing, duration, and amplitude of the associated meltwater pulses vary significantly between models and do not always match observational constraints. Nonetheless, the broad similarities of simulated meltwater pulses across these complicated models indicate that there are fundamental causes of meltwater pulses that are captured even in coarse ice sheet models.…”

A Simple Model for Deglacial Meltwater Pulses

“…The amplitude and timing of meltwater pulses simulated with complex 3D ice sheet models [Gregoire et al, 2012;Abe-Ouchi et al, 2013;Heinemann et al, 2014] differ significantly between models, and often are in disagreement with observationally-determined sea level records. In models that are parametrically tuned to match records of sea level and ice sheet extent [Tarasov et al, 2012;Gregoire et al, 2016;Stuhne and Peltier , 2017], the agreement is better (by design), but the particular physical processes controlling the characteristics of meltwater pulses is unclear. In this section, we show how the parameters and initial conditions in our simple model set the amplitude and timing of meltwater pulses.…”

“…Gregoire et al [2012] first showed that simulated intense surface melting in the ice saddle region between the Laurentide and Cordilleran ice sheets is associated with a brief acceleration in deglacial sea level rise. Subsequent studies [Tarasov et al, 2012;Abe-Ouchi et al, 2013;Heinemann et al, 2014;Gregoire et al, 2016;Stuhne and Peltier , 2017] have simulated similar ice saddle collapse events during the deglaciation of the Laurentide and Cordilleran ice sheets, though the timing, duration and amplitude of the associated meltwater pulses varies significantly between models and does not always match observational constraints. Nonetheless, the broad similarities of simulated meltwater pulses across these complicated models indicate that there are fundamental causes of meltwater pulses that are captured even in coarse ice sheet models.…”