Wave inhibition by sea ice enables trans-Atlantic ice rafting of debris during Heinrich events

Highlights

• Iceberg drift and decay is modeled for the last glacial maximum and modern climate.

• Modeled meltwater decreases faster with distance from source than proxy data suggest.

• Iceberg melt may stabilize water column, allowing for a seasonal local sea ice cover.

• Presence of protective sea ice around icebergs slows down wave-driven iceberg decay.

• 4 months of local sea ice produces good fit between model results and proxy records.

Abstract

The last glacial period was punctuated by episodes of massive iceberg calving from the Laurentide Ice Sheet, called Heinrich events, which are identified by layers of ice-rafted debris (IRD) in ocean sediment cores from the North Atlantic. The thickness of these IRD layers declines more gradually with distance from the iceberg sources than would be expected based on present-day iceberg drift and decay. Here we model icebergs as passive Lagrangian particles driven by ocean currents, winds, and sea surface temperatures. The icebergs are released in a comprehensive climate model simulation of the last glacial maximum (LGM), as well as a simulation of the modern climate. The two simulated climates result in qualitatively similar distributions of iceberg meltwater and hence debris, with the colder temperatures of the LGM having only a relatively small effect on meltwater spread. In both scenarios, meltwater flux falls off rapidly with zonal distance from the source, in contrast with the more uniform spread of IRD in sediment cores. To address this discrepancy, we propose a physical mechanism that could have prolonged the lifetime of icebergs during Heinrich events. The mechanism involves a surface layer of cold and fresh meltwater formed from, and retained around, large densely packed armadas of icebergs. This leads to wintertime sea ice formation even in relatively low latitudes. The sea ice in turn shields the icebergs from wave erosion, which is the main source of iceberg ablation. We find that sea ice could plausibly have formed around the icebergs during four months each winter. Allowing for four months of sea ice in the model results in a simulated IRD distribution which approximately agrees with the distribution of IRD in sediment cores.

Introduction

Layers of sand found in ocean sediment cores throughout much of the North Atlantic indicate several widespread events during the last glacial period. The sand in these layers is too coarse to have been carried by winds or currents, and it is generally believed that this sand was rafted by icebergs during episodes of massive calving from the Laurentide Ice Sheet, called Heinrich events (Heinrich, 1988; Broecker, 1994; Hemming, 2004; Rhodes et al., 2015). These ice-rafted debris (IRD) layers are particularly pronounced in the latitude range ${40}^{\circ }$N–${55}^{\circ }$N, which is sometimes referred to as the “IRD belt” (Fig. 1). Large volumes of freshwater rich with debris are expected to have been released from icebergs to produce the observed IRD layers (Dowdeswell et al., 1995; Hemming, 2004; Levine and Bigg, 2008; Roberts et al., 2014), with estimated ice discharges up to 100 times greater than that from present-day Greenland (Hemming, 2004). Such freshwater fluxes have been found in models to cause the Atlantic Meridional Overturning Circulation (AMOC) to weaken, which leads to reduced poleward heat transport and regional cooling (Broecker et al., 1985; Manabe and Stouffer, 1997; Levine and Bigg, 2008; Otto-Bliesner and Brady, 2010). Hence investigating the distribution of meltwater and IRD during Heinrich events could inform projections of future climate change in scenarios involving substantial discharges of icebergs from the Greenland Ice Sheet.

The thickness of IRD layers provides an indication of iceberg meltwater release and drift tracks during the Heinrich events. Recent studies have investigated this in coarse-resolution climate models, and they found a persistent mismatch between modeling results and IRD thickness in sediment cores (Jongma et al., 2013; Roberts et al., 2014). Specifically, the models tend to simulate a rapid decline of meltwater input from west to east across the North Atlantic which resembles the distribution of modern iceberg sightings (International Ice Patrol, 2009), implying a similar decline in IRD layer thickness (Jongma et al., 2013; Roberts et al., 2014). Ocean sediment cores, by contrast, show a more gradual decrease from west to east (Hemming, 2004). Here we investigate this mismatch with a Lagrangian iceberg drift and decay model forced by output from a comprehensive global climate model (GCM) simulation. In contrast to previous studies, we use a higher resolution (∼1^∘) GCM, but the icebergs we simulate are non-interactive, behaving as passive tracers in the climate system.