Sediment transport drives tidewater glacier periodicity

ARTICLE DOI: 10.1038/s41467-017-00095-5 OPEN Sediment transport drives tidewater glacier periodicity Douglas Brinkerhoff1, Martin Truffer1 & Andy Aschwanden 1 Most of Earth’s glaciers are retreating, but some tidewater glaciers are advancing despite increasing temperatures and contrary to their neighbors. This can be explained by the coupling of ice and sediment dynamics: a shoal forms at the glacier terminus, reducing ice discharge and causing advance towards an unstable configuration followed by abrupt retreat, in a process known as the tidewater glacier cycle. Here we use a numerical model calibrated with observations to show that interactions between ice flow, glacial erosion, and sediment transport drive these cycles, which occur independent of climate variations. Water availability controls cycle period and amplitude, and enhanced melt from future warming could trigger advance even in glaciers that are steady or retreating, complicating interpretations of glacier response to climate change. The resulting shifts in sediment and meltwater delivery from changes in glacier configuration may impact interpretations of marine sediments, fjord geochemistry, and marine ecosystems. 1 Geophysical Institute, University of Alaska Fairbanks, Fairbanks, AK 99775, USA. Correspondence and requests for materials should be addressed to D.B. (email: ) NATURE COMMUNICATIONS | 8: 90 | DOI: 10.1038/s41467-017-00095-5 | www.nature.com/naturecommunications 1 ARTICLE D NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-00095-5 Results Calibration. Our model’s physical configuration is inspired by Taku Glacier, with a mean length of ~50 km, and a maximum elevation of 2200 m. Figure 2 illustrates the geometry, climate, and sediment transport rate 100 years after the grounding line reached its minimum following model initialization. Mean bedrock erosion rates were tuned to match the tectonic uplift rate, as well as observations from coastal Alaska16. Fluvial erosion rates were calibrated with repeat radar measurements from Taku Glacier17. Fluvial deposition rates were derived from theoretical settling velocities of sediment samples taken from boreholes at a b Taku 1933 2016 Columbia 1985 2016 m 19 k 2 primary ways. First, we consider sediment transport due to glaciofluvial rather than deformational processes15, and we include a physically consistent representation of both erosion and depositional processes using mass conservation13. We find that that this coupled model, when calibrated with observations, can produce all phases of the TGC without externally driven changes in climate and with a period and amplitude in general agreement with observations from the geologic record. These temporal and spatial scales are strongly influenced by the availability of meltwater, which drives the rate of sediment transport and subsequently controling the rate of glacier advance. This dependence is sufficiently strong that a change in meltwater availability due to climate warming may (perhaps counter-intuitively) trigger advance in tidewater glaciers that are currently stable or retreating. Although necessarily a simplification of the complete physical processes governing tidewater glacier dynamics, our results provide a basis for assessing how such natural variability in sedimentation and meltwater regimes could impact fjord ecosystems, interpretation of marine sedimentary records, and predictions of ice volume change. m 5k espite a globally consistent trend of glacier mass loss1, ~1/3 of Alaska’s tidewater glaciers are advancing2. This trend shows little spatial consistency, suggesting a dynamical rather than climate mechanism is responsible. Tidewater glaciers differ from their terrestrial terminating cousins because mass loss occurs not only by surface melt, but also by calving and ablation at the oceanic boundary. The mass budget of such glaciers is therefore sensitively dependent on conditions at the marine terminus such as water temperature, motion, and depth. One example of this dependence, first observed at Columbia Glacier3, is that the presence of a terminal shoal can lead to a considerable terminus advance relative to a position in absence of sedimentation. Such glaciers also tend to be unstable: ice flux from calving increases with terminus depth, and on a retrograde slope (such as the upstream side of a moraine) this positive feedback can lead to catastrophic retreat4, 5. The coupling of nonlinear ice dynamics and sediment transport leads to a continuum of dynamical behavior known as the tidewater glacier cycle (TGC), which can be described in four archetypal phases6. In the advancing stage, development and advection of a shoal at the front reduces calving flux, causing glacier thickening and advance. The shoal may be subaqueous (e.g., Hubbard Glacier, advancing 35 m per year7) or subaerial (e.g., Taku Glacier, advancing 10 m per year8, Fig. 1). Eventually, the glacier enters an extended phase, in which the balance of accumulation and ablation halts advance (e.g., Brady Glacier9). A glacier enters the retreating phase when the glacier can no longer maintain sufficient thickness to remain grounded on the shoal, and the associated reduction in basal drag leads to retreat into progressively deeper water, triggering the instability described above (e.g., Columbia Glacier, which began such a retreat in 1985 that continues presently6, Fig. 1). Ungrounding results either from the ice thinning or from the bed lowering due to erosion or mass wasting. Retreat ends when the glacier approaches the terminus position it would assume in the absence of sedimentation, and the terminus effectively re-grounds on bedrock. This retracted state is usually many kilometers shorter than the advanced terminus position. A sediment shoal may then rebuild and the cycle begin again. The time scales of the TGC have been inferred from a combination of direct observation, radiocarbon dating, and tree-ring analysis at several glaciers in Alaska, and cycle periods range from a few hundred years10 to a few thousand11. Zero-dimensional modeling efforts have shown that sediment deposited at the glacier front can initiate advance through the calving reduction mechanism defined above12–14, even when highly simplified models of ice dynamics (e.g., volume-area scaling) and sediment transport are used. Such models are also capable of simulating the retreat phase, although the degree of hysteresis between the time scales of advance and retreat is not explicitly captured. Models were inconsistent about whether an external perturbation is required to initiate retreat. Models designed to simulate the TGC in temperate glaciers have invoked either glaciofluvial erosion13 or sediment deformation12 as a mechanism for transporting the terminal shoal. More physically complete models have been applied to the problem of coupled ice flow and sediment deformation, the primary mechanism of sediment transport for the relatively dry marine ice (...truncated)