- Reconstructing the sensitivity of past climate to forcings, and of ancient glaciers and ice sheets to this climate, can allow us to better understand the range of climate and cryosphere behavior we may see in the coming centuries. The Arctic is a region of particular importance due to its well-documented amplification of climate change, the presence of the Greenland Ice Sheet, and the acceleration of climate and cryosphere change in the past two decades. In this dissertation, I investigate the past 20,000 years in the area, focusing specifically on the last deglacial period and the late Holocene to answer the overarching question: what magnitude of forcing is required to exceed regional variability in the climate system? I address this through three sub-questions: (1) was there regional variability in retreat of the Greenland Ice Sheet during the last deglaciation, and if so what drove this variability? (2) does Arctic regional climate match the average climate history of the region during the Common Era, and (3) is ice sheet and glacier behavior in southernmost Greenland during the late Holocene synchronous across different glacier types? I use a combination of geostatistical and geochemical techniques to answer these questions, and throughout stress the power of large datasets to reveal significant climate signal within proxy noise.
To answer the first question, I present a database of all surface exposure ages and radiocarbon ages published for Greenland. I recalculate all ages using consistent production rates and scaling schemes to allow all ages to be directly compared with each other and to ice sheet models. Factor analysis of surface exposure ages reveals diachronous retreat of the ice sheet, with east Greenland deglaciating during and immediately following the end of the Younger Dryas cold period (~12.9-11.7 ka), with peaks in retreat ~13.0-11.5 ka. In contrast, terrestrial retreat of south Greenland occurs between 11.0 and 10.0 ka, and west Greenland retreats between 10.5 and 7.5 ka. This spatial variability is not present in either minimum-limiting radiocarbon ages or an ice-sheet model reconstruction, and is likely due to ocean warming forcing an early retreat in east Greenland. The model reconstruction we compare to does not include realistic ice-ocean interactions which likely account for the model-data misfit.
To answer the second two questions, I investigate climate and cryosphere interactions in the late Holocene (~4 ka to present) in the Arctic. Temperature changes in the latest Holocene are less dramatic than the deglacial temperature increase, but occur at centennial timescales relevant to human societies. To address the second question, I use a climate field reconstruction to merge geochemical climate proxies that cover the Common Era (2 ka to present) with a spatially-resolved reanalysis dataset. This approach allows me to build gridded maps of Arctic temperatures at five-year resolution covering the period 10-2010 C.E., then use these maps to investigate spatial patterns in climate, potential forcings of this climate, and timings of significant temperature shifts in individual gridcells.
Throughout the Common Era prior to recent warming, we observe significant spatial variability, particularly between the eastern and western Arctic, which are uncorrelated through most of the period. Continental areas in general show more variability than ocean regions. A preliminary variance attribution study reveals that a combination of five proposed forcings of Arctic climate (volcanic eruptions, the North Atlantic Oscillation, Atlantic meridional ocean circulation, CO2, and total solar irradiance) explain at most ~20% of climate variance during this period. Onset of Little Ice Age cooling is also spatially variable, with a range of ~600 years, and is very sensitive to the reference period used to calculate the transition to significantly colder temperatures. By contrast, recent emergence into significantly warmer temperatures is less sensitive to reference period, and shows warming significantly above background temperatures across most of the Arctic between 1850 and 1950 C.E., suggesting relatively early emergence throughout the Arctic, except in parts of the North Atlantic, Arctic Europe, and Siberia.
To answer question 3, I present a set of 51 10Be cosmogenic surface exposure ages from six sites in southernmost Greenland. Each site records the late Holocene maxima of the adjacent glacier, so the exposure dates indicate timing of retreat from the late Holocene maximum extent of the ice sheet. Here, I find substantial variability in the response of the south Greenland Ice Sheet and its surrounding glaciers and ice caps to climate. Smaller mountain glaciers and ice caps, as well as glaciers draining the piedmont-like Qassimiut lobe of the ice sheet show a potential response to a localized warming event between ~1.0-0.4 ka. However, those glaciers draining the high altitude alpine-like Julianhåb Ice Cap show greater variability, with retreat ages between 3.7 and 0.4 ka, This suggests the magnitude and duration of climate changes in the late-Holocene (before the most recent warming) were too small to force a widespread response of south Greenland’s terrestrial cryosphere.