Supplemental Information -Glacier Model Glacier Model OverviewThis model uses a finite-element numerical model from (Kessler et al., 2006) where ice accumulation and movement on a given terrain surface is governed by explicit equations for ice flux and mass conservation.The mass balance is the combination of a prescribed annual accumulation and calculated annual melt.Melt is approximated using a positive degree-day method with an additional factor to account for melt from solar radiation.The model is subsequently calibrated to observed ice limits and transient scenarios are run to explore climate sensitivity and the required climate forcings needed to reconstruct Divide Ice Cap activity over the last ~2000 years.This supplement provides details on model design, parameter selection and calibration, sensitivity analysis, and characterization of uncertainty. Glacier Model Setup2.1.Terrain Production Prior to model implementation, a terrain model of the bedrock surface is required.This involves removing (to the best approximation) the modern Divide Ice Cap from an existing digital elevation product.ASTER digital elevation data from 2011 CE was used as a base for the two-dimensional terrain model, resampled to a 60 m pixel size.The ASTER data product was retrieved from the online Data Pool, courtesy of the NASA Land Processes Distributed Active Archive Center (LP DAAC), USGS/Earth Resources Observation and Science (EROS) Center, Sioux Falls, South Dakota, https://lpdaac.usgs.gov/data_access/data_pool.The resulting surface was smoothed using a 7x7 mean filter to remove artifacts in the raw data that would lead to instabilities in the model.Modern ice, including Divide Ice Cap and the ice on the surrounding summits, had to be removed to create an ice-free terrain to model upon.Using the best approximation of basal shear stress (t b ) to be ~100 kPa (Haeberli, 2016), current ice thicknesses (H) were calculated following Cuffey and Paterson (2010): 𝐻 = 𝜏 % 𝜌𝑔𝜃where r is the density of ice (917 kg m -3 ), g is gravitational acceleration (9.81 m s -2 ) and q is the surface slope of the modern ice surface.Calculated thicknesses were then subtracted from the modern terrain surface to produce an ice-free surface.Eventual model runs show that calculated and modeled modern ice thickness are the same within 10% of each other.Kessler et al. (2006) drove ice formation with a climate dictated by an equilibrium line altitude (ELA), a mass balance gradient with elevation, and a maximum positive balance (maximum accumulation).However, the overall low accumulation rates of our high-latitude site and the variable aspect of the Divide Ice Cap, which increases the influence of solar radiation during the melt season, necessitates a different approach (Benn and Evans, 2010).Ice core records, observational studies in the eastern Canadian Arctic, and previous modeling work suggest a maximum accumulation of 0.3 m water equivalent (m.w.e.) per year throughout the Mass Balance
Abstract. Records of Neoglacial glacier activity in the Arctic constructed from moraines are often incomplete due to a preservation bias toward the most extensive advance, usually the Little Ice Age. Recent warming in the Arctic has caused extensive retreat of glaciers over the past several decades, exposing preserved landscapes complete with in situ tundra plants previously entombed by ice. The radiocarbon ages of these plants define the timing of snowline depression and glacier advance across the site, in response to local summer cooling. Although most dead plants recently exposed by ice retreat are rapidly removed from the landscape by erosion, where erosive processes are unusually weak, dead plants may remain preserved on the landscape for decades. In such settings, a transect of plant radiocarbon ages can be used to construct a near-continuous chronology of past ice margin advance. Here we present radiocarbon dates from the first such transect on Baffin Island, which directly dates the advance of a small ice cap over the past two millennia. The nature of ice expansion between 20 BCE and ~1000 CE is still uncertain, but episodic advances at ~ 1000, ~ 1200, and ~ 1500 CE led to the maximum Neoglacial dimensions ~ 1900 CE. We employ a two-dimensional numerical glacier model to reconstruct the pattern of ice expansion inferred from the radiocarbon ages and to explore the sensitivity of the ice cap to temperature change. Model experiments show that at least ~ 0.44 °C of cooling over the past 2 ka is required for the ice cap to reach its 1900 margin, and that the period from ~ 1000 to 1900 CE must have been at least 0.25 °C cooler than the previous millennium; results that agree with regional climate model simulations. However, ~ 3 °C of warming since 1900 CE is required to explain retreat to its present position, and, at the same rate of warming, the ice cap will disappear before 2100 CE.
ABSTRACT Lake sediment records give valuable insight into the dynamic events that characterized the last deglaciation in Iceland. Here, we focus on the well‐dated sediment record from Hestvatn, a low‐elevation lake in south Iceland, that features six graded bedding events deposited by outburst floods from glacial lakes dammed by the decaying Iceland Ice Sheet (IIS) in the time period of the Vedde Ash and the G10ka Series tephra. Using climate proxies preserved in the sediment cores, in conjunction with regional glacial geomorphology, we reconstruct the retreat of the IIS in south Iceland, from a marine‐based glacier during the Younger Dryas to a land‐based glacier during the Preboreal. As the ice sheet margin withdrew to the central highlands, ice‐dammed lakes formed along glacier margins. The ice‐dams were occasionally breached, generating large‐scale jökulhlaups (catastrophic outburst floods) that deposited thick turbidite sequences preserved in the sediment record of Hestvatn. The high concentration of volcanic material incorporated within deglacial sediments indicates that along with IIS retreat, subglacial volcanic activity may have helped initiate some of the jökulhlaups. Onset of more stable Holocene conditions was reached after the final turbidite at ~10 ka bp , when the IIS had withdrawn from most of the highlands of Iceland.
Abstract. Records of Neoglacial glacier activity in the Arctic constructed from moraines are often incomplete due to a preservation bias toward the most extensive advance, often the Little Ice Age. Recent warming in the Arctic has caused extensive retreat of glaciers over the past several decades, exposing preserved landscapes complete with in situ tundra plants previously entombed by ice. The radiocarbon ages of these plants define the timing of snowline depression and glacier advance across the site, in response to local summer cooling. Erosion rapidly removes most dead plants that have been recently exposed by ice retreat, but where erosive processes are unusually weak, dead plants may remain preserved on the landscape for decades. In such settings, a transect of plant radiocarbon ages can be used to construct a near-continuous chronology of past ice margin advance. Here we present radiocarbon dates from the first such transect on Baffin Island, which directly dates the advance of a small ice cap over the past two millennia. The nature of ice expansion between 20 BCE and ∼ 1000 CE is still uncertain, but episodic advances at ∼ 1000 CE, ∼ 1200, and ∼ 1500 led to the maximum Neoglacial dimensions ~ 1900 CE. We employ a two-dimensional numerical glacier model calibrated using the plant radiocarbon ages ice margin chronology to assess the sensitivity of the ice cap to temperature change. Model experiments show that at least ∼ 0.44 °C of cooling over the past 2 kyr is required for the ice cap to reach its 1900 CE margin, and that the period from ∼ 1000 to 1900 CE must have been at least 0.25° C cooler than the previous millennium, results that agree with regional temperature reconstructions and climate model simulations. However, significant warming since 1900 CE is required to explain retreat to its present position, and, at the same rate of warming, the ice cap will disappear before 2100 CE.
Abstract. Strong similarities in Holocene climate reconstructions derived from multiple proxies (BSi, TOC, δ13C, C/N, MS, δ15N) preserved in sediments from both glacial and non-glacial lakes across Iceland indicate a relatively warm early-to-mid Holocene from 10 to 6 ka, overprinted with cold excursions presumably related to meltwater impact on North Atlantic circulation until 7.9 ka. Sediment in lakes from glacial catchments indicates their catchments were ice-free during this interval. Statistical treatment of the high-resolution multiproxy paleoclimate lake records shows that despite great variability in catchment characteristics, the records document more or less synchronous abrupt, cold departures as opposed to the smoothly decreasing trend in Northern Hemisphere summer insolation. Although all lake records document a decline in summer temperature through the Holocene consistent with the regular decline in summer insolation, the onset of significant summer cooling, occurs ~5 ka in high-elevation interior sites, but is variably later in sites closer to the coast, suggesting some combination of changing ocean currents and sea ice modulate the impact from decreasing summer insolation. The timing of glacier inception during the mid-Holocene is determined by the decent of the Equilibrium Line Altitude (ELA), which is dominated by the evolution of summer temperature as summer insolation declined as well as changes in sea surface temperature for glacial systems particularly in coastal settings. The glacial response to the ELA decline is also highly dependent on the local topography. The initial nucleation of Langjökull in the highlands of Iceland starting by ca 5 ka, was followed by a stepwise expansion of both Langjökull and northeast Vatnajökull between 4.5 and 4.0 ka, with a second abrupt expansion ca. 3 ka. However, the initial appearance of Drangajökull in the NW of Iceland was delayed until after 2.5 ka. All lake records reflect abrupt summer temperature and catchment disturbance at about 4.5 ka, statistically indistinguishable from the ~4.2 ka event with a second widespread abrupt disturbance centered on 3.0 ka. Both are intervals of large explosive volcanism on Iceland. The most widespread increase in glacier advance, landscape instability, and soil erosion occurred shortly after 2 ka, likely due to a complex combination of increased impact from volcanic activity, cooling climate, and increased sea ice off the coast of Iceland. All lake records indicate a strong decline in temperature ~1.5 ka, culminating during the Little Ice Age between 1300 and 1900 CE when most glaciers reached their maximum dimensions.
