Dissertation Abstract
Observations and modeling of marine-terminating outlet glaciers
Enderlin, Ellyn M 2013 https://sites.google.com/site/ellynenderlin/
School of Earth Sciences, The Ohio State University (United States), 131 pp.
Increased mass loss from the Greenland Ice Sheet due to rapid changes in marine-terminating outlet glacier dynamics has the potential to substantially increase sea level within this century, yet the external factors triggering these changes and the internal controls that govern the glaciers’ dynamic response to external forcing are poorly understood. Gaps in our understanding of dynamic change arise from observational limitations and the impacts of the associated uncertainty on numerical modeling results. The observational and numerical modeling studies presented herein focus on changes in dynamics for the marine-terminating outlet glaciers that drain the Greenland Ice Sheet, but the results from these studies can be applied broadly to marine-terminating outlet glaciers in other regions such as Antarctica, Alaska, and Svalbard. These analyses are presented as four separate studies.
Remotely-sensed ice surface elevations obtained for 34 marine-terminating outlet glacier in NW Greenland between 2000 and 2010 are used to improve the temporal constraints on the onset of dynamic ice loss from this region. The data indicate that the timing and magnitude of dynamic thinning varied widely between glaciers, but that 10’s of meters of thinning occurred along the majority of the outlet trunks prior to the detection of regional mass loss acceleration in 2005. The lag time between the onset of dynamic thinning and regional mass loss acceleration is likely due to the time required for dynamic changes to propagate into the ice sheet interior.
Remotely-sensed data are also used to estimate submarine melt rates beneath the floating ice tongues of 13 Greenland marine-terminating outlet glaciers on a semi-annual basis between 2000 and 2010. Melt rates ranged from 0.03 m/d to 2.98 m/d within the study period, and accounted for 5-85% of the volume lost from the floating ice tongues. Melt rates were uncorrelated with terminus retreat, glacier flow speed, and changes in ocean temperatures and no clear spatial pattern was evident in the dataset.
Using a width- and depth-integrated numerical ice flow model (i.e., 1D flowline model), the influence of glacier shape on the dynamic response of glaciers following a stress perturbation initiated at the ice-ocean boundary is examined. The modeling results indicate that for a given ice flux, wider glaciers and those that overlie deeper basal depressions tend to be closer to flotation, so that less dynamically induced thinning results in rapid, unstable grounding line retreat several years after the onset of the applied perturbation. Thus, shape differences may help explain the intra-regional variability in marine-terminating outlet glacier behavior observed throughout Greenland. Additionally, these results indicate that the glacier shape must be precisely simulated in numerical ice flow models in order to accurately predict future dynamic behavior.
The 1D flowline model is also used to analyze model sensitivity to uncertainty in the parameterizations that govern ice flow. A non-unique combination of parameter values to describe the ice rheology and basal sliding are used to reproduce similar steady-state glacier configurations. Once perturbed, however, the response of the simulated glaciers varies widely with the choice of parameter values. Taken together, these results suggest that predictions of future dynamic behavior must be accompanied by sensitivity tests that account for parameter uncertainty.
Remotely-sensed ice surface elevations obtained for 34 marine-terminating outlet glacier in NW Greenland between 2000 and 2010 are used to improve the temporal constraints on the onset of dynamic ice loss from this region. The data indicate that the timing and magnitude of dynamic thinning varied widely between glaciers, but that 10’s of meters of thinning occurred along the majority of the outlet trunks prior to the detection of regional mass loss acceleration in 2005. The lag time between the onset of dynamic thinning and regional mass loss acceleration is likely due to the time required for dynamic changes to propagate into the ice sheet interior.
Remotely-sensed data are also used to estimate submarine melt rates beneath the floating ice tongues of 13 Greenland marine-terminating outlet glaciers on a semi-annual basis between 2000 and 2010. Melt rates ranged from 0.03 m/d to 2.98 m/d within the study period, and accounted for 5-85% of the volume lost from the floating ice tongues. Melt rates were uncorrelated with terminus retreat, glacier flow speed, and changes in ocean temperatures and no clear spatial pattern was evident in the dataset.
Using a width- and depth-integrated numerical ice flow model (i.e., 1D flowline model), the influence of glacier shape on the dynamic response of glaciers following a stress perturbation initiated at the ice-ocean boundary is examined. The modeling results indicate that for a given ice flux, wider glaciers and those that overlie deeper basal depressions tend to be closer to flotation, so that less dynamically induced thinning results in rapid, unstable grounding line retreat several years after the onset of the applied perturbation. Thus, shape differences may help explain the intra-regional variability in marine-terminating outlet glacier behavior observed throughout Greenland. Additionally, these results indicate that the glacier shape must be precisely simulated in numerical ice flow models in order to accurately predict future dynamic behavior.
The 1D flowline model is also used to analyze model sensitivity to uncertainty in the parameterizations that govern ice flow. A non-unique combination of parameter values to describe the ice rheology and basal sliding are used to reproduce similar steady-state glacier configurations. Once perturbed, however, the response of the simulated glaciers varies widely with the choice of parameter values. Taken together, these results suggest that predictions of future dynamic behavior must be accompanied by sensitivity tests that account for parameter uncertainty.