C13A-0550 INVITED
Application of an inverse method to infere the velocity pattern from isochronous layers in firn
We present a kinematic approach to find the velocity field from dated internal-layer architecture in firn. Internal layers are isochrones and the depositional age of a layer particle is treated as a tracer. The forward problem uses two-dimensional steady-state advection of age, and conservation of mass to predict layer architecture. It thus only considers the kinematic equations and does not use any dynamic equations, an advantage for the application in firn, where rheological properties vary spatially. Different combinations of constraints on horizontal or vertical velocity properties are added. The inverse problem can be formulated as the solution of underdetermined and overdetermined systems of equations. The systems are solved using singular-value decomposition, allowing analysis of the singular-value spectrum, model resolution, and data resolution. For synthetic scenarios, the solutions of the inverse problem are evaluated by comparing the velocity-field solutions with synthetic input velocity data. Compared to conventional accumulation estimates, the new approach takes lateral advection into account, enabling improved separation of spatial and temporal variations in accumulation (deduced from the vertical velocity components). We present two glaciological applications: the determination of the migration velocity of a spatially non- stationary and highly variable accumulation pattern in a dune field, and reconstruction of past accumulation with a focus on its stationarity over time.
C13A-0551
Iterative Methods for Reconstructing Glacier Basal Boundary Data
There are no direct methods to measure the boundary condition at the base of a glacier. We present here two techniques for inferring basal boundary data from surface observations in the case of flow through a transverse cross-section. Both methods are based on a scheme, known as Kolzlov-Maz'ya iteration, which involves solving well-posed forward problems. The two methods discussed are distinguished by their assumptions concerning the form of the basal boundary condition: the first is agnostic and makes no particular assumption, whereas the second assumes a specific relationship between basal stresses and velocities involving an unknown coefficient. We demonstrate the methods' effectiveness by applying them to synthetic data as well as to a real glacier, Athabasca, which has a known basal velocity distribution.
C13A-0552
Large-scale modeling of the Antarctic ice sheet using a massively-parallelized finite element model (CIELO).
We implemented a fully-three-dimensional, thermo-mechanical, finite element model of the Antarctic Ice Sheet with a spatial resolution varying from 10 km inland to 2 km along the coast on a massively-parallelized architecture named CIELO and developed at JPL. The model is based on a "Pattyn" type formulation for ice sheets, and a "MacAyeal shelf-stream" formulation for ice shelves. Both types of formulations are coupled using penalty methods, which enables a considerable reduction of the computational load. Using a simple law of basal friction (based on locally computed balanced velocities), the model is able to replicate the location and order-magnitude speed of major ice streams and ice shelves. We then coupled the model with observations of ice motion from SAR interferometry to refine the pattern of basal friction using an inverse control method (MacAyeal 1993). The result provides an excellent agreement with observations and a first complete mapping of the pattern of basal friction along the coast of Antarctica at a resolution compatible with the size of its glaciers and ice streams.
C13A-0553
Basal drag evolution of Pine Island Glacier in the wake of its retreat
Pine Island Glacier, in West Antarctica, has been retreating, thinning and accelerating steadily since at least
the 1970s. In more recent years, the glacier acceleration has exceeded values recorded in the last 3
decades.
As the glacier accelerates and retreats into deeper ground, a central question is to determine how its
force balance evolves with time, especially basal drag. Basal drag remains one of the most poorly
understood aspect of ice dynamics. As the glacier ungrounds, the resistance to flow will decrease in a way
that is
easily calculated. But it is not clear how this evolution will affect areas upstream of the zone of ungrounding,
i.e.
how the perturbations in basal resistance will propagate up glacier.
Here, we use ice velocities from 1996, 2000 and 2007 to infer the spatial and temporal variability in basal
drag of Pine Island Glacier using an inverse control method applied to the finite element model of McAyeal
(1989).
The interpretation of the results include considerations for error sources from the remote sensing data, the
inverse method, and the relevance of the model assumptions. This experiment makes it possible to examine
the
correlation between grounding line migration, glacier thinning, speed up and basal drag evolution.
This work was performed at the California Institute of Technology's Jet Propulsion Laboratory under a
contract
with the National Aeronautics and Space Administration's Cryosphere Science Program.
C13A-0554
Basal Dynamics of a Small Surge-Type Glacier Investigated Using 1-D Geophysical Inverse Modeling
As part of a study to characterize glacier response to climate in the St. Elias Mountains, Yukon Territory, Canada, the dynamics of a small surge-type valley glacier are investigated. Pole displacements were measured using kinematic global positioning system (GPS) techniques during three consecutive summer field seasons. Calculated surface velocities range from 5 to 10 m/yr over the lower 1.5 km of the 4.5 km-long glacier and increase to around 30 m/yr over the upper 2.5 km. Basal velocities are reconstructed from the measured surface velocities using a 1-D geophysical inverse model. An analytical relationship between the basal velocity, deformational velocity and surface velocity of an ice body flowing in a channel defines the forward model, which is subsequently linearized using a method of longitudinal averaging for variable ice thickness and surface slope. The inversion itself is performed using a spectral decomposition method. Flowline ice thicknesses and surface elevations are required model inputs and are measured with ice- penetrating radar and kinematic GPS surveying, respectively. Sensitivity of the inverse model to uncertain quantities such as the flow-law coefficient, the shape factor and the longitudinal averaging length is quantified through a series of tests. The inversion results show that basal motion accounts for at least 50% and up to 100% of the total surface motion along the flowline. The unexpectedly high flow speeds recorded in the upper reaches of the glacier may explain the formation of fresh crevasses, while the high proportion of basal motion is consistent with observed surface undulations. From these observations and our modeling results, we suggest that the current dynamics of this glacier may be more than a direct response to climate and that the glacier may be undergoing a slow surge. Simple hydrological models are being used to investigate whether spatial variability in the efficiency of the subglacial drainage system may be responsible for the current flow regime.
C13A-0555
Inversion of Electromagnetic Data to Obtain Ice Thicknesses at Mts. Adams, Baker and Rainier, Washington
Helicopter electromagnetic (HEM) data over Mt. Adams, Mt. Baker and Mt. Rainier volcanoes were originally collected to map altered zones for volcano hazards assessments but also can be used to obtain the thickness of ice by a numerical inverse process that includes magnetic susceptibility and dielectric properties of rocks in addition to the standard resistivity and thickness parameters. The multifrequency electromagnetic system used 5 coplanar (106400, 25400, 6121, 1515 and 388 Hz) and one coaxial (3315 Hz) coil. Transmitter and receiver coils in the electromagnetic sensor detected the electromagnetic response of the exposed and ice-covered ground at different frequencies to obtain information from different depths. A one- dimensional (1-D) program using analytic Jacobians was used to invert HEM data. The final three- dimensional model was stitched together from a series of 1-D inversions. Sensitivity tests of the responses to simple two-layer resistivity models representing volcano structure were calculated to guide the inverse process. The tests and practical noise levels showed that ppm response for a 10 ohm-m bottom layer, representing wet and altered rocks, below a very resistive, 1 M ohm-m layer, representing ice, is above the 40 ppm noise limit for up to depths of ~100 m. The results also showed that basement resistivities exceeding 1000 ohm-m below 30 m ice will not be well resolved. The effect of magnetic susceptibility on the measured Hz fields was evaluated for the case of a 30-m-thick top layer overlying a 1000 ohm-m bottom layer. The susceptibility of 0.025 SI strongly influences the in-phase component by lowering its value by a few ppm along the complete spectrum of frequencies. The in-phase and quadrature responses of the Hz component for varying relative dielectric permittivity were calculated for the same model and show that only the highest frequencies, above 10 kHz, are influenced by the dielectric properties of the model, and only for the case of high water permittivity. Based on these results, the inversion of the HEM data with the specific goal of detecting ice thickness assumed a two-layer model, with a top layer representing ice and a bottom layer, rock. Frequency-dependent dielectric constants of 3.19 as a low frequency value, 95.3 as a high frequency value, and a 7558.0 relaxation frequency measured at -1°C were used for the ice and a fixed dielectric constant of 10 for the rock. The susceptibility of the top layer was 0, while the basement susceptibility was calculated from the low- frequency (1500 Hz) negative in-phase component. Estimated ice thickness results were compared with ice thickness data from radar and well data. The inverted ice thickness agreed well with the old drillholes and one radar profile over Mt. Adams in the areas of low- resistivity (tens of ohm-m) water-saturated alteration basement. On Mt. Baker, a radar profile located just south of a flight line showed similar thickness to the HEM model. In this case, the basement resistivity was higher, around 100-200 ohm-m, in comparison to the Mt. Adams basement resistivity. For Mt. Rainier, the presence of the water below some glaciers was interpreted by the lower resistivity values, but the quality of the data was problematic, and the ice-thickness data were not reliable.
C13A-0556
Scalable Solvers for Optimal Control of a Power-Law Fluid
Inverse problems such as determining internal fields or basal boundary conditions quickly become very large when applied to large glaciers or ice sheets. Millions of degrees of freedom are required to resolve critical flow features in a three dimensional ice sheet model. Traditional inverse methods involve forming and factoring a linearized inverse operator which is a dense matrix. The memory needed to store a single 106 × 106 matrix is 8 terabytes and the CPU cost to factor such a matrix is impractical. Instead, we pose the inverse problem as PDE-constrained optimization and apply fully iterative methods. For our problems, the design and state spaces are mesh-dependent, so we seek a method which scales linearly with respect to both spaces. The recently developed Lagrange-Newton-Krylov-Schur (LNKS) method applies the well-established Newton-Krylov machinery directly to the first-order optimality condition. Unlike reduced space quasi-Newton methods, the Newton iteration displays mesh-independent quadratic convergence and the Krylov iteration will also exhibit mesh-independent convergence if an optimal preconditioner is available. Obtaining an optimal preconditioner is difficult since the Jacobian matrix has saddle point structure and is very poorly conditioned, but Schur complement preconditioning is effective and frequently optimal. Since the incompressible flow constraint also has saddle point structure, we require two levels of Schur complement preconditioning, providing a vast array of choices. For a power-law fluid, globalization by continuation in the exponent is natural but reduced space alternatives are available when the preconditioning matrix is expensive to form. We present a computational study of LNKS and reduced space methods in the context of a novel p-version finite element method which applies the full-order Jacobian matrix-free and assembles a much sparser preconditioning matrix using low-order finite elements. We devote special attention to the design of Schur complement preconditioners which are robust with respect to strength of the power-law nonlinearity since these are critical for efficient forward modeling as well.
C13A-0557
Interpolating an Ice Core Depth-Age Relationship from Sparse Data Using an Inverse Approach
Abstract Often Antarctic ice cores can be dated at only a few discrete depths; however, it is desirable to have a physically based interpolation of the depth-age relationship between the sparse dated points. Piece-wise linear or spline interpolations in the depth-age domain can introduce serious error in the implied accumulation-rate history, because they do not properly account for the variation of dynamical strain with depth due to ice flow, and they do not account for temporal changes in accumulation rate. Our forward problem uses a transient one-dimensional kinematic ice-flow model that produces a depth-age relationship, for specified histories of ice thickness, divide migration, and accumulation rate. We use an inverse method to determine a smooth accumulation-rate history that reproduces the measured depth-age data at a tolerance specified by data uncertainty. In comparison with direct interpolations in the depth-age domain, our inferred accumulation-rate history produces a more physically realistic interpolation of the measured depth-age data, without spurious accumulation-rate events. Here, we produce a continuous depth-age relation for the Siple Dome ice core from layer-counted ages and sparse ages of occluded gases (CH4 and O2), from 98ky to present. Accumulation rate and thinning function pairs vary widely depending on the surface elevation history. Using this kinematic approach, the unique accumulation rate and thinning function pair is not determined. All possible pairs, from different surface elevation and ice divide histories, produce identical depth-age relationships and layer thickness patterns.
C13A-0558
Spatially Resolved Information on Ice Dynamics From Topography on the Martian North Polar Layered Deposits
Circumpolar layered deposits on Mars consist almost entirely of water ice (Ih), but the roles of surface mass fluxes and possible ice flow in shaping those deposits remains a fundamental, open question. Topography of the layered deposits is dominated by arcuate, v-shaped troughs (with wall slopes of roughly 10 degrees) separated by smooth, low-slope topography (slopes less than roughly 1 degree). In the north, smooth, inter- trough topography is relatively common on a flat-bedded lobe known as Gemina Lingula, which bounds Chasma Boreale on the south and east. Based on prior work in Antarctica, we have applied an inverse method to interpret inter-trough topography. We found strong evidence for past ice flow on Gemina Lingula, prior to formation of the troughs, based on observations of ice-mass geometry normalized by characteristic dimensions. Solution of the inverse problem with normalized topography yields estimates of the ice-flow-law exponent, but does not yield information on absolute time-scales. Here we analyze un-normalized topography to estimate a new dynamical parameter, which relates absolute surface mass flux rates at the time of flow to near-basal temperatures. Surface mass flux rates and ice mass geometry in turn determine the time-scale of equilibration to variations in climatic forcing, for given temperatures. We find a strong, spatially coherent variation in the dynamical parameter over an area spanning roughly 400 km on Gemina Lingula. From this, we quantify the spatial variation in equilibration time-scales on Gemina Lingula, under several temperature scenarios based on variations in Martian orbital parameters over the past 5 million years.
C13A-0559
Inferring Mass-Balance Patterns and Rates from Radar-Internal Layers in Martian Ice
Past and present accumulation and ablation rates are fundamental unknowns for the Martian Polar Layered Deposits (PLD), and this information is necessary if we are to decipher the connection between climate and PLD formation, evolution, and observable structure. Internal layers in ice masses can be detected with ice- penetrating radar, and these data are currently being collected across the PLD. These layers are the primary record of relative spatial patterns of accumulation and ablation (mass balance) available. Additional rate- controlling information, such as the layer age, the ice temperature, or the ice-grain sizes and ice-crystal fabric, can be used to infer the absolute rate of mass balance. Using synthetic data, we solve four different inverse problems that all attempt to infer the relative spatial pattern and the absolute rate of mass-balance in a flowing ice mass, by assuming that different combinations of information are available. In the first inverse problem, only the shape of an internal layer is available. This problem could potentially be solved with data currently available for Mars. In the second inverse problem, only the ice-surface topography is available. The shape of the ice surface only weakly reflects smaller-scale spatial variations in mass balance, and we can recover only an average of the actual values. In the third inverse problem, we know the shape of an internal layer, and the surface topography and ice temperature when the ice was flowing, and we can successfully recover both the spatial pattern and the rate of mass balance. In the fourth inverse problem, we know the shape of an internal layer, the ice-surface topography, and ice-rheological parameters involving at least two activation energies. We can also successfully infer the spatial pattern and rate of mass balance with this information, if conditions at the time of flow were in a regime where more than one deformation process was active. If past ice flow has affected the shapes of Martian internal layers, it is necessary to use an inverse approach to infer mass-balance patterns from internal-layer shapes. We demonstrate the feasibility of recovering this valuable information.
C13A-0560
Using internal layers to infer the internal velocity field and boundary conditions of polar ice sheets
We first analytically determine the relationship between velocity field and isochrone geometry along a steady flowline of an ice sheet. The method is based upon the stream function and its vertically normalized form, the normalized stream function (NSF). We show that the slope of the isochrones is the slope of the iso-NSF lines, plus a path term which is the cumulative result of the past trajectory of the ice particles. Secondly, we show how this property can be used to infer the internal velocity field from isochrones along a flowline with known mass balance conditions. We illustrate the method on two different situations: varying basal sliding and varying velocity profile around a divide. Third and finally, we show how internal layers can be inverted to infer the surface accumulation rate and we apply this method on a flowline downstream of the EPICA Dronning Maud Land drilling site.
C13A-0561
Deriving Spatial and Temporal Statistics of Meltwater Infiltration Using a Bayesian Hierarchical Modeling Analysis of 10m Borehole Thermal Data
A snow traverse from the percolation to saturation zone of the Greenland ice sheet has obtained year long temperature records from 10m boreholes. Each borehole records a temporally and spatially dense (delta time: 15mins, delta space: 0.25m) vertical temperature distribution over the summer melt season and winter freeze-up. Analysis of this thermal data should be deterministic, except for the dominance of the breakthrough process of piping which allows surface melt water to rapidly descend to considerable depths (10m+) in cold snow. The spacing, temporal variability and depth of piping are major unknowns in this zone (which is crucial to surface water generation). With only one borehole every 10 kilometers of traverse, thermal anomalies in the data poorly constrains the problem of determining the horizontal and depth distribution of refreezing water. (Thermal anomalies may represent minor ice lensing close to the borehole or much larger refreezing further away from the borehole.) In an ideal case, with perfect and infinitely dense data, this problem is theoretically invertible in 2D, but not in 3D. However, with discrete data in space and time, and with measurement errors, any inversion is swamped by errors. In this analysis we apply a statistical Bayesian hierarchical approach coupled to forward thermal modeling to allow inversion in a statistical sense. In other, words we seek the most likely distribution of piping events in the region of the borehole that satisfy the data, as well as additional constraints. Two important constraints are: any potential refreezing distribution must itself be unbiased with respect to the borehole location, and: the net heat gained from refreezing must match the winter cooling profile.
C13A-0562
What controls dead-ice melting under different climate conditions?
In the geological record, hummocky dead-ice moraines represent the final product of the melt-out of dead- ice. Processes and rates of dead-ice melting in ice-cored moraines and at debris-covered glaciers are commonly believed to be governed by climate. Here, backwasting rates from 14 dead-ice areas are assessed in relation to mean annual air temperature, mean summer air temperature, mean annual precipitation, mean summer precipitation, and the sum of degree days > 0 deg. C. The highest correlation was found between backwasting rate and mean annual air temperature. However, the correlation between melt rates and climate parameters is low, stressing that processes and topography play a major role in governing the rates of backwasting. The rates of dead-ice melting from modern glacial environments should serve as input to de-icing models for ancient dead-ice areas in order to assess the mode and duration of deposition. A challenge for future explorations of dead-ice environments is to obtain long-term records of field-based monitoring of melt progression. Furthermore, many modern satellite-borne sensors have high potentials for recordings of multi-temporal Digital Elevation Models (DEMs) for detection and quantification of changes in dead-ice environments. However, time series of high-resolution aerial photographs remain essential for both visual inspection and high-resolution stereographic DEM production. Reference: Schomacker, A. 2008. What controls dead-ice melting under different climate conditions? Earth- Science Reviews, in press.
C13A-0563
Jakobshavn Isbrae, Greenland: DEMs, orthophotos, surface velocities, and ice loss derived from photogrammetric re-analysis of July 1985 repeat aerial photography
Jakobshavn Isbrae drains about 7 % of the Greenland Ice Sheet and is the ice sheet's largest outlet glacier. Two sets of high elevation (~13,500 m), high resolution (2 m) aerial photographs of Jakobshavn Isbrae were obtained about two weeks apart during July 1985 (Fastook et al, 1995). These historic photo sets have become increasingly important for documenting and understanding the dynamic state of this outlet stream prior to the rapid retreat and massive ice loss that began in 1998 and continues today. The original photogrammetric analysis of this imagery is summarized in Fastook et al. (1995). They derived a coarse DEM (3 km grid spacing) covering an area of approximately 100 km x 100 km by interpolating several hundred positions determined manually from block-aerial triangulation. We have re-analyzed these photos sets using digital photogrammetry (BAE Socet SetŠ) and significantly improved DEM quality and resolution (20, 50, and 100 m grids). The DEMs were in turn used to produce high quality orthophoto mosaics. Comparing our 1985 DEM to a DEM we derived from May 2006 NASA ATM measurements showed a total ice volume loss of ~ 105 km3 over the lower drainage area; almost all of this loss has occurred since 1997. Ice stream surface velocities derived from the 1985 orthomosaics showed speeds of 20 m/d on the floating tongue, diminishing to 5 m/d at 50 km further upstream. Velocities have since nearly doubled along the ice stream during its current retreat. Fastook, J.L., H.H. Brecher, and T.J. Hughes, 1995. J.of Glaciol. 11 (137), 161-173.
C13A-0564
Time-lapse photography of ice-shelf rift processes
During the 2004-2005 Antarctic field season, an automated, Iridium-connected camera system was installed
overlooking the rim-wall of a large rift on the Ross Ice Shelf to monitor processes occurring within the rift,
including:
1. enhanced snow accumulation associated with blowing snow trapped by the rift,
2. cornice formation and in-fall on the opposing rift face,
3. downward tilting of the rift-crest lip,
4. avalanche of steep snow drifts on opposing rift wall,
5. snow-apron formation,
6. slumping, tumbling and sinking of ice-melange forming the rift floor, and
7. rift-floor tectonics associated with spreading of the opposing rift walls.
http://www.thistle.org
C13A-0565
Iceberg Calving and Flow Dynamics at Helheim Glacier, East Greenland, from Time-Lapse Photography
Helheim Glacier in East Greenland is the focus of coordinated studies aimed at understanding tidewater outlet-glacier dynamics and kinematics, and their link to glacial earthquakes. As part of this effort, we installed three time-lapse cameras overlooking the calving terminus of the glacier during the Arctic summer of 2008. Images were captured every five minutes during the mostly unattended period of operation. Several interesting aspects of the glacier's behavior are observed in the image sequences, including vertical displacement of the glacier terminus by ocean tides, and very large calving events. These observations, in combination with simultaneous measurements of ice flow, ocean tides (including tsunamis) and seismic activity, contribute to our understanding of the dynamics of Helheim Glacier and the source mechanism of glacial earthquakes.
C13A-0566
Terrerstrial photogrammetry for measurements of the calving rate of Kronebreen, Svalbard
Calving is a large component in the mass budgets of tide water glaciers. Measurements of calving rates are essential for the determination of a calving law that can be used in a mass balance model of the glacier. Changes in climate may alter the ice dynamics and the melt water production, thus the calving rates may change rapidly. Despite their importance, calving processes are not well represented in numerical ice dynamic models, partially due to the difficulties and dangers of field data collection. Measurements of calving must be conducted in a safe distance to the calving front, and in this project we have applied terrestrial photogrammetry to the front of Kronebreen, Svalbard, in August-September 2007, May 2008 and August- September 2008. Repeated stereo pairs were photographed from Nielsen-fjellet mountain side over a one week period each time. Markers were placed on the strand and measured with GPS to georeference the images. In addition repeat glacier photographs were taken in August-September 2008 to get higher time- scale resolution of iceberg calving. The digital images are processsed photogrammetrically for velocity measurements and volume estimates of the calving ice loss. The velocity data are compared to ground based interferometric radar measurements conducted in the same period, showing a velocity of 2.7 m.d-1 in 2007 and 3.2 m.d-1 in August-September 2008.
C13A-0567
Automated Ground-based Time-lapse Camera Monitoring of West Greenland ice sheet outlet Glaciers: Challenges and Solutions
Monitoring Greenland outlet glaciers using remotely sensed data has drawn a great attention in earth science communities for decades and time series analysis of sensory data has provided important variability information of glacier flow by detecting speed and thickness changes, tracking features and acquiring model input. Thanks to advancements of commercial digital camera technology and increased solid state storage, we activated automatic ground-based time-lapse camera stations with high spatial/temporal resolution in west Greenland outlet and collected one-hour interval data continuous for more than one year at some but not all sites. We believe that important information of ice dynamics are contained in these data and that terrestrial mono-/stereo-photogrammetry can provide theoretical/practical fundamentals in data processing along with digital image processing techniques. Time-lapse images over periods in west Greenland indicate various phenomenon. Problematic is rain, snow, fog, shadows, freezing of water on camera enclosure window, image over-exposure, camera motion, sensor platform drift, and fox chewing of instrument cables, and the pecking of plastic window by ravens. Other problems include: feature identification, camera orientation, image registration, feature matching in image pairs, and feature tracking. Another obstacle is that non-metric digital camera contains large distortion to be compensated for precise photogrammetric use. Further, a massive number of images need to be processed in a way that is sufficiently computationally efficient. We meet these challenges by 1) identifying problems in possible photogrammetric processes, 2) categorizing them based on feasibility, and 3) clarifying limitation and alternatives, while emphasizing displacement computation and analyzing regional/temporal variability. We experiment with mono and stereo photogrammetric techniques in the aide of automatic correlation matching for efficiently handling the enormous data volumes.
C13A-0568
Geodetic Constraints on the Glacier Mass Balance Record of Place Glacier, British Columbia, Canada
Conventional (glaciological) measurements of glacier mass balance provide important records to assess the relation between climate and glacier nourishment. Inadequate sampling, interpolation errors, or systematic biases can introduce uncertainty into these records. We calculated the geodetic mass balance of Place Glacier over the periods 1965-1973, 1973-1981, 1981-1987, 1987-1993, 1993-1997 and 1997-2005 to compare with the mass balance obtained from conventional techniques. Although the cumulative balance of the geodetic (33.9 m) and traditional (33.2 m) records accord, the conventional and geodetic results differ by 15-25% for some periods. These deviations primarily arise from the use of aerial photography that was acquired at times which did not coincide with the end of the ablation season. However, the geodetic thinning rate of the glacier over the period 1973-1981 was nearly twice as large as the field-based measures despite any corrections applied to the geodetic data. Inspection of the conventional mass balance record indicates that fewer sites were used to derive glacier wide changes in mass during the 1970s. Our results bear directly on studies that utilize traditional records for glacier mass balance modeling since the period 1973-1981 represents an interval where the glacier initiated a significant transition to more negative net balance.
C13A-0569
Double Exposure: Photographing Climate Change
Double Exposure, Photographing Climate Change, is a fine-art photography exhibition that examines climate
change through the prism of melting glaciers. The photographs are twinned shots of glaciers, taken in the
mid-20th century by world-renowned photographer Brad Washburn, and in the past two years by Boston
journalist/photographer David Arnold. Arnold flew in Washburn's aerial "footprints", replicating stunning black
and white photographs, and documenting one irreversible aspect of climate change. Double Exposure is art
with a purpose. It is designed to educate, alarm and inspire its audiences. Its power lies in its beauty and the
shocking changes it has captured through a camera lens. The interpretive text, guided by numerous experts
in the fields of glaciology, global warming and geology, helps convey the message that climate change has
already forced permanent changes on the face of our planet. The traveling exhibit premiered at Boston's
Museum of Science in April and is now criss-crossing the nation. The exhibit covers changes in the 15
glaciers that have been photographed as well as related information about global warming's effect on the
planet today.
http://www.doublexposure.net
C13A-0570
Using Repeat Ground-Based, Airborne, and Space-Based Photography and Multispectral Imagery to Document the Post-Little Ice Age Behavior of Alaksan Glaciers
Alaska supports thousands of glaciers, situated on 11 mountain ranges, a large island (Kodiak), an island chain (Aleutian Islands), and an island archipelago (Alexander Archipelago). Glaciers cover about 75,000 km2 of Alaska. Individually and collectively documenting the post-Little Ice Age Behavior of these Alaskan glaciers is only possible through a comprehensive assessment of ground-based, airborne, and space-based photography, as well as space-based multispectral imagery. Alaskan glaciers have been repeatedly photographed from the ground (beginning in 1893), from the air (beginning in 1926), and from space (beginning in the early 1960s). Since 1972, all Alaskan glaciers have been sequentially imaged with space-based multispectral sensors. Analysis of this massive compellation of repeat photographs and images has been used to quantitatively and qualitatively determine the distribution, extent, and multiple decadal- scale behavior of glaciers throughout Alaska. These results have recently been published by the U.S. Geological Survey in Glaciers of Alaska, Chapter K of the Satellite Image Atlas of the Glaciers of the World, Professional Paper 1386-K. Because of the size of the area covered by glaciers in Alaska and the lack of large-scale maps of the glacierized areas, multiple-source photography and sequential satellite imagery are the only practical means of monitoring regional changes in the length, area, and volume of Alaskan glaciers in response to short- and long-term changes in the maritime and continental climates of the State. These data were coupled with a review of the literature to determine both the individual and the regional status of Alaskan glaciers and to characterize changes in thickness and terminus position of representative glaciers in each mountain range or island area. Additionally, three detailed test sites were identified where an extensive record of historic ground-based photographs provided an additional opportunity to assess sub-decadal to century-scale changes to glaciers and landscapes. These locations were Glacier Bay National Park and Preserve of the St. Elias Mountains, Kenai Fjords National Park of the Kenai Mountains, and the northwestern Prince William Sound area of the Chugach Mountains. In each, detailed assessments of this all-source data were performed for every fiord to document and understand fiord-specific glacier and landscape change and evolution. On a broad regional basis, the analysis determined that every mountain range and island area investigated can be characterized by significant glacier retreat, thinning, and (or) stagnation, especially those glaciers that end at lower elevations. Of more than 2,000 glaciers investigated, less than a dozen are currently thickening and advancing. At others, retreat that started as early as the 18th century has continued into the 21st century. At some locations, glaciers completely disappeared during the 20th century. On a local basis, variability proved to be far more complex, with adjacent fiords displaying significantly different behaviors.