C41B-0469
Bayesian Change Point Analysis for Water Pressure Data
We use wavelets in a Bayesian context to identify changes in the pattern of data collected over time, when missing observations are present. Our work is motivated by the interest in identifying and modeling change points in the measurements of sub-glacial water pressure during melt season along the length of the Bench glacier in Alaska. This modeling will provide insights into the sub-glacial hydrology including the discharge mechanism during the melt season. A Bayesian analysis based on the empirical wavelet coefficients is used to find the change point (Oden and Lynch,1998) in 18 water pressure data sets available.We compare this method with wavelet based method suggested byWang (1995), which examines the empirical wavelet coefficients of the data at the fine scale levels. The above methods have to be adapted for accommodating missing observations, which are present in our data sets.
C41B-0470
Seasonal Acceleration of Inland ice via Along-Flow Coupling to Marginal ice
We use an ice flow model to demonstrate how flow variations initiated in the marginal zone of an ice sheet affect flow farther inland via along-flow coupling. Our findings allow for an alternate interpretation of seasonal accelerations observed near the equilibrium line of the Greenland ice sheet [Zwally et al., 2002]. We demonstrate that these observations can be explained by accelerations initiated up to 12 km closer to the margin where the ice is known (from field observations) to be 40 percent thinner, is heavily crevassed, experiences a seasonal doubling of velocity, and where the ablation rate, surface meltwater flux, and ice temperature are likely higher. Our modeling and observations suggest that conditions and processes normally found near ice sheet margins are adequate to explain the observations of Zwally et al. [2002]. In this case, we argue that seasonal acceleration may have a limited impact on ice-sheet mass balance even in the face of future climate warming.
C41B-0471
Premelting dynamics on the boundary between glaciological and hydrological systems
The effective stress N, rate of basal freeze-on (or melting) V, heat flow, and rates of frictional and viscous dissipation control how glaciers interact with sub-glacial hydrological systems at the glacier/sediment interface. Interactions between ice and porous media are well understood from studies of ground-freezing in subaerial environments. Though the overburden beneath glaciers is much larger, N is often within the typical range that is considered by models for frost-heaving behavior in fine-grained sediments. Adapting these models to the subglacial environment, thermodynamic and mechanical considerations require that a fringe of partially frozen sediment extend beneath soft-bedded glaciers when N>pf≈1.1 (Tm- Tf) MPa/°C, where Tm-Tf is larger for sediments with finer grain sizes, and is equal to the temperature drop below the pressure-melting point that is needed for ice to infiltrate the pore space. Coupling between the glaciological and hydrological systems determines the fringe thickness h (sometimes 0) as a function of the local effective stress N, the basal energy balance, and sediment properties. Typically, pf=O(104) Pa and fringe thicknesses of several decimeters to meters in scale are predicted. The rate V that water can be transported through the fringe and frozen onto or melted from the glacier base can achieve a steady-state that is in balance with the rate that latent heat is transported to or from the basal interface. This water transport can represent a primary source or sink in the hydrological system, and the latent heat transport can make a leading-order contribution to the basal energy balance. For a glacier that is perched above a given sediment, larger h is predicted with higher N and the lower frictional heat input that accompanies slower sliding rates Ws. Unsteady behavior can lead to large changes in h when there is a mismatch between the rate that latent heat can be extracted and the rate that fluid is supplied to the ice--liquid boundary. Changes in N or Ws are expected to cause transient behavior, with the freezing rate V adjusting rapidly to satisfy force-balance constraints, and h relaxing on the time scale for conduction of latent heat. This can cause the integrated pattern of sediment deformation to be distributed over a finite depth range even when shear is perfectly localized at any given instant in time. Channelized subglacial flow implies spatial variations in N that require h and V to vary laterally in a predictable fashion that might be tested against field observations. An improved understanding of the coupled glacio-hydraulic system can be achieved with a more accurate parameterization of conditions on their common boundary.
C41B-0472
Evidence for catastrophic outbursts and tunnel valley formation in northern Denmark during the last deglaciation
A series of buried valleys was mapped in northern Denmark using SkyTEM and conventional TEM. The valleys are 2-13 km long, 0.5-2 km wide and up to 170 m deep, and their distribution can be linked to former ice margins. Exploration wells were drilled in the centre and on the flanks of representative valleys in order to understand the environmental development and to establish a chronological framework for their genesis. The recovered samples were analysed using a wide range of sedimentogical and biostratigraphical proxies and the chronology is based on OSL and AMS 14C age determinations. Most of the tunnel valleys turned out to be incised into Eemian to Late Weichselian marine or lacustrine clays, and they have been refilled with silty and sandy sediments containing a mixed, redeposited Eemian and Weichselian fauna. Based on lithostatigraphic correlation and absolute datings, as well as the biostratigraphy of wells drilled through the flanks, the age of the valleys can be constrained to ~18 ka BP at which time northern Denmark was deglaciated, and they have been refilled within a few centuries in a sub- or proglacial environment. This implies that much of the meltwater from the Scandinavian Ice Sheet was drained along systems of tunnel valleys formed during catastrophic flood events and not by steady state drainage.
C41B-0473
Assessing Elevation Changes Over Large Antarctic Subglacial Lakes From ICESat Repeat Altimetry
The recently described Recovery Lakes group (~13,300 km2) and well-known Lake Vostok (~15,690 km2) are both crossed by numerous Ice, Cloud, and land Elevation Satellite (ICESat) altimetry profiles. Since those observations have begun in late 2003 and 11 repeats of each profile have now been completed, a number of smaller lakes along East and West Antarctic ice streams have also been detected and described. However, because of the size of these large lakes as well as the Recovery Lakes relationship to the acceleration of ice into the Recovery Glacier, an assessment of their lake levels has immediate relevance to broader climate studies. Our study uses the same repeat altimetry technique to assess elevation changes at these two major subglacial lake sites. Because the elevation changes in these lake basins are not expected to be as large as the changes found under ice streams, this effort will rely on the sub-3 cm shot-to-shot precision of ICESat's laser altimeter. And to compensate for inexact repeat profiles that cross the subglacial lake topography at varying locations and angles, cross track slopes will be modeled to try to detect small but significant elevation changes. Any trends at these locations can then be evaluated in terms of lake level and potential for broader climate impact.
C41B-0474
Constraints on melt-water flux through the West Greenland ice-sheet: modeling of hydro- fracture drainage of supraglacial lakes
Recent observations of rapid ice-flow acceleration following the onset of surface melting from Greenland suggest that ice-sheets may respond more rapidly and dramatically to climate change than previously believed. Melt- water may drain advectively through the ice-sheet, lubricating and warming the bed, and accelerate ice flow. Understanding the mechanisms and constraints on melt-water transport through ice-sheets is critical for incorporating these feedbacks into models of ice-sheet evolution and for estimating future rates of ice-sheet drawdown and sea-level change. We have investigated the viability of rapidly transporting melt-water from the ice- sheet's surface to its bed using a linear elastic fracture mechanics approach to model crack propagation in ice. Building on the work of Weertman (IASH, 1973), Alley et al. (Annal Glac, 2004) and van der Veen (GRL, 2007), who showed that only water filled crevasses will propagate through an ice-sheet, we model the size and shape of water-filled crevasses. In doing so, we place volumetric constraints on the amount of water necessary to keep water-filled crevasses propagating through 1-2 km of sub-freezing ice. This volume of water is then used to constrain the minimum lake sizes that can hydro-fracture to the bed. For example, in our study area there are over 1000 lakes that contain enough water (> 0.3-0.5 km in diameter) to keep a crevasse filled to the ice-bed interface (~1.5 km of ice). In addition, our calculations of the opening geometry of hydro-fractures demonstrate that the time scale for draining the largest known lakes through a single crevasse is on the order of 2-18 hrs, significantly faster than the time-scale for refreezing of crevasses in ice. These model-based results of crevasse geometry and water drainage are consistent with and supported by recent field observations of rapid lake drainage south of Jakobshavn Isbrae, Greenland.
C41B-0475
Analogue Experimental Study of Debris Entrainment by Ice Growing into Supercooled Water
Field studies have identified several processes by which glaciers and ice sheets can accumulate debris-rich basal ice. Although significant effort has gone into modeling these processes and the inferred effects on glacier dynamics, there is a dearth of laboratory investigations. In particular, in several field settings debris entrainment by frazil ice growing in glaciohydraulically supercooled water channels has been suggested as the source of the dirty basal ice. However, the physical mechanisms by which frazil ice actually traps the sediment and then incorporates it into the basal ice remain unclear. In order to clarify this process, we have frozen solutions of 140 nm-diameter silica spheres dispersed in water in a 400 micron-thick cell supercooled to between -2 and -6 C. The thinness of the cell results in an essentially two-dimensional ice fabric, thus allowing us to observe the particle distribution and movement during freezing and subsequent coarsening. During the nucleation and unstable growth of the ice we find that many particles segregate to the liquid interstices between ice dendrites. Over the next 10 seconds the inter-dendrite liquid freezes and some particles become trapped between dendrites while some are pushed ahead of the ice front. As the ice coarsens over several days, particles migrate through the ice due to temperature gradient regelation and move with the grain boundaries/veins/nodes, which evolve to accommodate the coarsening. While the physical conditions in our experiment do not correspond exactly to those of subglacial conduits, our observations of the microscopic dynamics of particles during the freezing and evolution of supercooled, sediment-laden water can help guide future experiments and provide insight into the ice and sediment textures observed in glacier basal ice.
C41B-0476
Surface Melt Area and Water Balance Modeling on the Greenland Ice Sheet 1995–2005
SnowModel was used to simulate variations in snow water equivalant, surface snow and ice melt, and other water balance components for the period 1995–2005 in Greenland, including the Greenland Ice Sheet (GrIS). Meteorological observations from 25 stations inside and outside the GrIS were used as model input. Winter and summer mass balance observations, spatial snow depth observations, and snow melt depletion curves derived from time lapse photography from the Mittivakkat and Zackenberg glacierized catchments in East Greenland were used to validate the performance of SnowModel. Model results compared well with observed values. The yearly modeled GrIS interior non-melt area differs from satellite observations by a maximum of ~68,000 km2 (or ~6%) in 2004, and the lowest uncertainties (<8,000 km2, or <1%) occur for the years with the smallest (2005) and most extensive (1996) non-melt area. The modeled inter-annual variability in non-melt area also agrees with observation records (R2 = 0.96), yielding a simulated GrIS non-melt cover of 71% for 1996 and 50% for 2005. Modeled surface melt occurred at elevations reaching 2,950 m a.s.l. for 2005. On average, the simulated non- melt area decreased ~6% from 1995 through 2005; this trend is similar to satellite observed values. An average loss in GrIS storage of -79(±98) mm w.eq. y-1 (or -153(±181) km3 y-1), and a runoff of 211(±31) mm w.eq. y-1 (or 392(±58) km3 y-1) also occurred over the period 1995–2005. Approximately 58% and 42% of the runoff occured from the GrIS western drainage and eastern drainage areas, respectively. The modeled average specific runoff from the GrIS was 6.7 l s-1 km-2 y-1, which, over the simulation period, represents a contribution of ~1 mm y-1 to global sea level rise.
C41B-0477
Ice Flow History, Basal Melt, and Hydraulic Connections in the Dome C Lake District Headwaters Over the Last Glacial Cycle Inferred from Hyrdopotential, Internal Layers, and Basal Reflectvity in Airborne Radar Sounding Data
We have tracked 11 dated isochronal layers in the vicinity of Dome C using airborne ice-penetrating radar data for the purposes of calculating vertical strain and estimating horizontal velocity. Using the calculated vertical strain and a back-stripping of the layers we then obtain paleo-accumulation rates for the last 127 k. Combining the strain rate, accumulation history with a surface temperature record from the EPICA Dome C ice core, an assumed lapse rate of 10 degrees per km of elevation and a small linear correction for horizontal advection, we are able to calculate a long-term temperature model for every record of radar data. Characteristically lenticular features in the internal layer structure, associated with accumulation highs at the upstream ends of subglacial lakes Concordia and Horseshoe, are tracked to reveal horizontal velocities of 0.5 and 0.75 m/yr respectively. Velocities in both locations appear to have decreased during glacial maxima to a value nearly half that found during the interglacials. The modern velocities are consistent with those obtained from InSAR satellite measurements. Layer-based paleo-accumulation rates indicate that the present inferred distribution of accumulation rates is broadly equal to the accumulation distribution during the Eemian interglacial. Rates during the last glacial maximum were roughly ½ to 1/3 the rates found presently. Our temperature model reveals that a significant portion of the basal ice in the Dome C region may be at or near the pressure melting point, consistent with other published results from Dome C ice Core. Widely distributed melt rates in the major topographic valleys are generally less than 1 mm/yr throughout the region with slightly higher melts in the basin draining into Vincennes Subglacial Lake. The two largest subglacial lakes within the survey, Concordia and Vincennes are both associated with enhanced basal melting on their upstream shores at rates locally greater than 2 mm/year. Although published estimates for geothermal flux are capable of explaining the behavior of ice and water in most of the area, the two melt anomalies associated with the two largest subglacial lakes require an additional source of basal heat. These results have major implications for the filling rates and histories of several of the known subglacial lakes in this region.
C41B-0478
Detection of Candidate Subglacial Lakes Using ICESat and a Newly Discovered 1000 km**2 Lake System
We examine ICESat laser altimetry over Antarctica spanning 2003-2007. Crossovers are computed between 10 fully-calibrated laser campaigns of approximately 33 days each. Scrutinizing these changes, certain locations appear to exhibit more or less consistent patterns: either elevation increases or decreases, with others exhibiting some fluctuation. Several of these spots can be correlated with the known subglacial lake inventory, and many new areas are observed near catchment boundaries throughout the continent. One newly discovered area in East Antarctica has been identified as a candidate lake, or system of subglacial lakes, affecting surface elevations across an area on the order of 1000 km**2. This distinctive area appears to cross two catchment boundaries, spanning three different drainage basins. This spot is located far enough south to be sampled by more than 30 ICESat repeat tracks. Detailed elevation histories are presented and other properties of this new lake system are investigated. Given the exceptional performance and data characteristics of ICESat (2-3 cm precision, less than 10 cm accuracy, 170 m along-track resolution, latitudes to 86 deg.), a quick method for detecting localized elevation variability, or candidate active subglacial lake regions, is via crossover computation. Crossover analyses are more efficient computationally and more accurate than repeat-track analyses which will contain some level of cross-track slope and/or interpolation errors. However, data density of repeat-track methods are nearly an order of magnitude greater than crossovers for ICESat, and therefore suspected regions can be targeted for follow-up study in smaller dedicated repeat-track analyses. The accuracy of ICESat is sufficient for direct crossover height differences to be used in the calculation of rates of elevation. We show elevation rates derived from ICESat crossovers and illustrate their spatial and temporal variability. We then define certain characteristics at the crossovers (based on elevations rates), and later repeat-track elevation profiles, which help to detect candidate subglacial lakes, based upon characteristics of known active water systems, including the newly discovered subglacial system.
C41B-0479
Estimating the Basal Sampling Area of a Borehole During the Summer-Winter Transition
We present a simple numerical model that generates the water level changes observed during the summer- winter transition of a temperate valley glacier. The model allows us to approximate the basal area sampled by a borehole during this transition. For a period spanning three years, borehole water levels were recorded continuously in 43 different boreholes drilled to the bed of the Bench Glacier, Alaska. Summer melt season water levels generally consist of large diurnal variations; winter water levels show little to no diurnal variation and are generally quite high (70-100% of ice overburden pressure). An observed style of transition from summer to winter borehole water level behavior is a repeating series of asymptotic water level rises that are abruptly terminated by rapid water level decreases. We first present a conceptual model to generate this behavior. Once a borehole becomes disconnected from the 'fast' hydrologic system, the water level rises due to ice deformation closure of the borehole and the basal zone connected to the borehole (now referred to as the borehole system); the water in the borehole system is displaced up the borehole. This rise in water level reduces the effective pressure in the borehole system, and consequently reduces the rate of deformation. The abrupt water level fall results from locally catastrophic releases of water from the borehole. Due to similar water levels at the initiation of each discharge event, and the similar water level at the termination of each discharge event, we believe the water from the basal system repeatedly escapes through the same orifice. Rates of discharge up to 10-3 m3 s-1 are observed, and the total volume discharged is as great as 0.5 m3. Inflow to the system is negligible in the late season. Our numerical model uses the Glen flow law and known parameters to generate the theoretical curves of rising water level for varying sizes of borehole systems due to ice closure. Curve fitting the observed water level rises allows for approximation the volume of water in the borehole system. This volume can then be used to approximate the basal area sampled by the borehole.
C41B-0480
Overcharging the Subglacial Hydrologic Network: Sliding at Kennicott Glacier, Alaska, When Water Inputs Exceed Outputs
Basal sliding of Kennicott Glacier, Alaska, occurs when water storage within the glacier is increasing. During the 2006 melt season, we deployed five GPS receivers on the ice surface along the glacier centerline, lake level sensors in four ice-marginal lakes, a stage sensor above the glacier outlet river, and temperature sensors on the ice surface. Outlet water chemistry was monitored with a conductivity sensor and near-daily water samples. These instruments allowed us to assess the relationship between ice motion and glacier hydrology at 1 hr or finer resolution and indicate that when water is entering the glacier faster than it drains out, ice surface velocity increases on both diurnal and seasonal timescales. The largest increase in ice surface speed occurred on a third timescale, that of the outburst of ice-dammed Hidden Creek Lake 15 km from the terminus. During the peak in rate of water storage in the glacier during the jokulhlaup, ice surface motion below the lake increased to 3.0 m/d, up to 6 times faster than non-flood times. Low solute concentrations in the river discharge from the terminus during periods of increasing storage support the view that water from the linked cavity system is prevented from leaving the subglacial hydrologic network by an over-pressurized conduit system. When water inputs exceed what conduits can transmit, the system develops backpressure, and the cavity-to-conduit hydraulic gradients expected in steady state are reversed. The pressurized conduits drive water into and increase pressure in the linked cavity system, promoting basal sliding. A simple numerical model, driven by a calculated water balance time series, predicts water partitioning between the subglacial and englacial reservoirs and ice motion at speeds that closely resemble the observed speeds. These observations and models suggest that persistent high melt rates will not sustain high sliding rates, as over time, the efficiency of the subglacial conduit system will increase. However, large discrete inputs of water to the bed can incite hydraulic transients and associated sliding, potentially explaining recent accelerations of the Greenland Ice Sheet, where rapid drainage of large melt ponds delivers water through cold polar ice.
C41B-0481
Modelling geochemical and biogeochemical reactions in subglacial environments
The determination of geochemical and biogeochemical processes in subglacial environments is critical to our understanding of processes of terrestrial chemical denudation and global biogeochemical cycles in glaciersed environments. This study examines current subglacial biogeochemical weathering models using PHREEQCi, a computer-based speciation mass-balance (SMB) model, parameterised using hydrochemical and mineralogical field data from a small Alpine glacier (Haut Glacier d'Arolla, Switzerland). The aim is to explore the utility of SMB models for identifying and quantifying subglacial biogeochemical mechanisms, and the potential of such approaches in polar and sub-Antarctic hydrological systems. The chemical evolution of meltwaters along a subglacial flow path is modelled, and the mathematical and thermodynamic robustness of current subglacial weathering models in a temporally-variable, spatially- heterogeneous hydrological and geochemical subglacial environment is tested. SMB modelling produced a broad range of weathering outcomes, but delivered no unique weathering scenario which could account for the observed changes in water chemistry. Organic carbon oxidation and sulphide oxidation by dissolved O2, coupled to carbonate dissolution and incongruent silicate dissolution could account for seasonal changes in meltwater chemistry, supporting current subglacial biogeochemical weathering models. Atmospheric CO2 was not required under any weathering scenario, and organic carbon and atmospheric CO2 dissolution was only possible in one weathering scenario, at a mass ratio of 10:1, further suggesting that CO2 driven dissolution is relatively unimportant in subglacial environments. The suggestion of current subglacial biogeochemical weathering models that Fe3+(aq) can act as a sulphide oxidising agent under anoxic conditions was also tested. Under suboxic conditions (pE ~ 0) conditions, abiotic speciation could not generate aqueous Fe3+ (aq), suggesting that abiotic processes cannot account for the generation of Fe3+. This suggests that microbial oxidation of Fe2+ to Fe3+ is required if Fe3+ is indeed an important oxidising agent in subglacial environments.