Quantifying periglacial erosion: insights on a ... - Wiley Online Library

EARTH SURFACE PROCESSES AND LANDFORMS Earth Surf. Process. Landforms 34, 2008–2022 (2009) Copyright © 2009 John Wiley & Sons, Ltd. Published online 13 November 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/esp.1885

Quantifying periglacial erosion: insights on a glacial sediment budget, Matanuska Glacier, Alaska Colin R. O’Farrell,1 Arjun M. Heimsath,2* Daniel E. Lawson,3 Laura M. Jorgensen,4 Edward B. Evenson,5 Grahame Larson6 and Jon Denner7 1 Pioneer Natural Resources, 1401 17th St., Suite 1200, Denver, CO 80202, USA 2 School of Earth and Space Exploration, Arizona State University, Tempe, AZ 85287, USA 3 CRREL, 72 Lyme Rd., Hanover, NH 03755, USA 4 The Shaw Group, 7604 Technology Way #300, Denver, CO 80237, USA 5 Dept. of Earth and Environmental Sciences, Lehigh University, Bethlehem, PA 18015, USA 6 Dept. of Geological Sciences, Michigan State University, East Lansing, MI 48824, USA 7 USGS, Federal Building, Montpelier, VT 05601, USA Received 7 March 2008; Revised 25 June 2009; Accepted 6 July 2009 *Correspondence to: A.M. Heimsath, School of Earth and Space Exploration, Arizona State University, Tempe, AZ 85287, USA. Email: [email protected];

ABSTRACT: Glacial erosion rates are estimated to be among the highest in the world. Few studies have attempted, however, to quantify the flux of sediment from the periglacial landscape to a glacier. Here, erosion rates from the nonglacial landscape above the Matanuska Glacier, Alaska are presented and compare with an 8-yr record of proglacial suspended sediment yield. Non-glacial lowering rates range from 1·8 ± 0·5 mm yr−1 to 8·5 ± 3·4 mm yr−1 from estimates of rock fall and debris-flow fan volumes. An average erosion rate of 0·08 ± 0·04 mm yr−1 from eight convex-up ridge crests was determined using in situ produced cosmogenic 10 Be. Extrapolating these rates, based on landscape morphometry, to the Matanuska basin (58% ice-cover), it was found that nonglacial processes account for an annual sediment flux of 2·3 ± 1·0 × 106 t. Suspended sediment data for 8 years and an assumed bedload to estimate the annual sediment yield at the Matanuska terminus to be 2·9 ± 1·0 × 106 t, corresponding to an erosion rate of 1·8 ± 0·6 mm yr−1: nonglacial sources therefore account for 80 ± 45% of the proglacial yield. A similar set of analyses were used for a small tributary sub-basin (32% ice-cover) to determine an erosion rate of 12·1 ± 6·9 mm yr−1, based on proglacial sediment yield, with the nonglacial sediment flux equal to 10 ± 7% of the proglacial yield. It is suggested that erosion rates by nonglacial processes are similar to inferred subglacial rates, such that the ice-free regions of a glaciated landscape contribute significantly to the glacial sediment budget. The similar magnitude of nonglacial and glacial rates implies that partially glaciated landscapes will respond rapidly to changes in climate and base level through a rapid nonglacial response to glacially driven incision. Copyright © 2009 John Wiley & Sons, Ltd. KEYWORDS: periglacial erosion; sediment yield; glaciation;

10

Be; cosmogenic nuclides

Introduction Glaciated landscapes are among the most topographically dramatic in the world. Horns, cirques, arêtes, moraines, and expanses of glacial drift are all testaments to the erosive power of glaciers. Glacially dominated landscapes also have some of the highest erosion rates found worldwide and have been studied extensively in this context. For example, glaciated basins in Alaska produce an order of magnitude more suspended sediment than do unglaciated basins, on average (Parks and Madison, 1984). Attempts to quantify glacial erosion have typically focused on sediment yield from proglacial rivers and streams (see review in Hallet et al., 1996) because these fluvial systems integrate all the fluxes providing sediments to the glacial terminus. Proglacial sediment yield

does not, however, necessarily directly correlate with sediments eroded by glacial processes alone and the complexity of quantifying glacial sediment budgets is widely recognized and articulated (Church and Ryder, 1972; Warburton, 1990; Harbor and Warburton, 1993; Delmas et al., 2009). We focus here on the periglacial processes eroding bedrock-dominated slopes above glaciers that play a critical role in shaping the landscape (André, 1997; Zhang et al., 2001; Pan et al., 2003), and note the difference between ‘periglacial’ and ‘paraglacial’ as articulated by Church and Ryder (1972) and Ballantyne (2002). In short, periglacial refers to regions subjected to intense freeze-thaw conditions, but not covered in perennial ice (i.e. nonglacial), while paraglacial refers to nonglacial processes that occur as a result of previous glaciation. In addition to the uncertainty on periglacial sediment inputs to a

QUANTIFYING PERIGLACIAL EROSION

glacial system, debate also continues as to whether glacial (Harbor and Warburton, 1993; Hallet et al., 1996; Kirkbride and Matthews, 1997; Koppes and Hallet, 2002) or fluvial and related processes (Evenson and Clinch, 1987; Hicks et al., 1990; Summerfield and Kirkbride, 1992; Hebdon et al., 1997; Meigs et al., 2006) are the chief control on proglacial sediment yield. The relative contributions of these processes remain largely unresolved (Brocklehurst and Whipple, 2002), despite being critical to understanding the role glaciation plays in mountain landscape evolution (Spotila et al., 2004). We present a first-order approximation of periglacial sources of sediment to the proglacial sediment yield of the Matanuska Glacier, Alaska. In particular, we quantify the erosional processes that are active on oversteepened valley walls and unglaciated ridge crests in the upper reaches of the glaciated basin using a combination of field measurements. These include colluvial and alluvial sedimentary slope deposits, a single melt season of suspended sediment measurements from tributary glacial and nonglacial streams, and cosmogenic nuclide (10Be) erosion rates from ridge crests. We compare these field measurements of nonglacial erosion and sediment yield from a small glacial stream to proglacial sediment yield measurements that span an 8-yr period at the terminus of the Matanuska Glacier.

Field Site The Matanuska basin is located ~160 km NE of Anchorage in the Chugach Mountains of Alaska and covers an area of 600 km2, with 58% ice cover (Figure 1). The Matanuska Glacier has its highest source at an elevation of 3500 m and flows north (inland) through steep terrain to a terminus at an elevation of 500 m (Williams and Ferrians, 1961). Although stable for most of the last 200 years, the terminus has thinned and retreated ~10–30 m yr−1 during the past decade (Lawson, unpublished data; Larson et al., 2003). Sedimentation and erosion processes at the ice margin are well documented and sediment transport to the ice margin in subglacial streams and basal ice have been quantified (Lawson and Kulla, 1978; Lawson, 1979; Arcone et al., 1995; Alley et al., 1998; Lawson et al., 1998). Our study of periglacial erosion rates focused on a tributary valley on the east side of the Matanuska Glacier, referred to here as Sheep Valley (Figure 1). The Border Ranges Fault, a right lateral fault that cuts perpendicular to the flow of the Matanuska Glacier, bisects Sheep Valley (Figure 1). The fault juxtaposes Late Cretaceous foliated meta-sedimentary rocks (phyllite) on the south side of the fault with Jurassic gabbroic rocks on the north side of the fault (Winkler, 1992), and the importance of this lithologic boundary is shown below in our discussion. The fault geometry thus divides Sheep Valley along its axis from near the confluence with the Matanuska Glacier to the middle of our field area. The fault then strikes eastward, dividing Sheep Valley perpendicular to its axis (Figure 1). A small, unnamed glacier (referred to herein as Sheep Glacier) is at the head of Sheep Valley, covering about 32% of its sub-basin. Based on a clearly visible trimline, Sheep Glacier was previously tributary to the Matanuska Glacier, although we cannot be certain of the exact timing of this confluence. With the retreat of Sheep Glacier, alpine catchments have progressively deposited sediment onto the former subglacial surface and into the proglacial stream. We utilize the juxtaposition of different lithologies created by deposition from these alpine catchments onto the surface previously occupied by Sheep Glacier as part of our methods. Copyright © 2009 John Wiley & Sons, Ltd.

2009

Conceptual framework and methods Erosion rates are typically calculated using the suspended sediment yield of proglacial streams, but they have the potential to underestimate true erosion rates if supraglacial, englacial and basal debris contributions as well as subglacial sediment transport by deformation (Alley et al., 1997) are not included. Due to their integrative nature, suspended sediment yields alone cannot resolve patterns of sediment movement within the glacial system. Notably, suspended sediment yields cannot distinguish between sediment being actively eroded by glacial processes from sediment being evacuated from storage reservoirs beneath the ice or adjacent to active channels entering the glacier from lateral sources (Warburton, 1990; Harbor and Warburton, 1993). Quantifying the upland periglacial sediment fluxes is thus a critical part of furthering our understanding of the various fluxes within the glacial sediment budget. To help open the conceptual black box of erosional processes contributing sediment to a glacier we focus on periglacial processes within the context of a glacial sediment budget (Warburton, 1990; Harbor and Warburton, 1993; Slaymaker et al., 2003; Delmas et al., 2009). A simplified conceptual schematic of the Matanuska Glacier basin’s sediment budget over Holocene timescales identifies the fluxes we have attempted to quantify, as well as the fluxes that we have not quantified (Figure 2). The unglaciated upland areas above glaciers deliver sediment to the glacial system, either to the ice surface or along lateral margins where it enters via tributary streams (Heimsath and McGlynn, 2008). This sediment is transported by ice along supraglacial, englacial, or basal pathways, or within subglacial streams (Lawson, 1993). Sediments emerge at the terminus of the glacier in the subglacial drainage system, or are released from englacial and supraglacial transport by ablation of the ice. We used two methodologies to quantify nonglacial erosion rates and deliberately do not attempt to distinguish between periglacial and paraglacial processes, as defined above. First, we measured the volume of alluvial and colluvial fans deposited at the base of catchments tributary to Sheep Valley. Half of the fans are sourced from steep catchments adjacent to and above Sheep Glacier, while the remaining are deposited by lower gradient catchments located down-valley from Sheep Glacier’s present-day terminus (Figure 1). The landscape may readily be divided on the basis of gradient and overall landscape morphology into what we term the ‘valley fan’ landscape type and the ‘supraglacial fan’ landscape type (Figure 3). We determined representative landscape lowering rates for the two different landscape types by measuring colluvial fan volumes and contributing areas, and estimating the fan ages. Second, we selected representative unglaciated ridge crests from the upland area above the Matanuska Glacier and measured cosmogenic 10 Be concentrations from surficial samples of bedrock (Figures 1 and 4 show sample locations) to determine point-specific erosion rates for landscapes eroding by steady-state processes like grain–grain spallation and shallow freeze–thaw leading to talus production. We then applied the corresponding measurements of the calculated erosion rates to the appropriate portions of the unglaciated landscape using representative hillslope morphometry, and estimated the basin-wide, nonglacial sediment flux. This estimated nonglacial sediment flux is then compared with suspended sediment measurements and bedload estimates from the terminus of the Matanuska Glacier.

Alluvial and colluvial fans Using field surveys, we calculated hillslope erosion rates, E (L T−1), from volumetric estimates of alluvial and colluvial Earth Surf. Process. Landforms, Vol. 34, 2008–2022 (2009) DOI: 10.1002/esp

2010

C.R. O’FARRELL ET AL.

Figure 1. (a) Location of the study area at the Matanuska Glacier, south-central Alaska. (b) The main shaded relief image delineates the extent of the Matanuska basin and shows the topography of the basin, the Border Ranges Fault, the South Glacial Stream (SGS) gaging station, and cosmogenic sampling points (e.g. MAT 15). The inset contour map shows the Sheep Valley study area with glaciers, colluvial fans, streams, and the Sheep Creek gaging station identified. Elevation data for both the main image and inset contours is based on USGS NED data (60 m resolution).

Figure 2. Conceptual sediment budget schematic for the Matanuska Glacier basin. The schematic guides our conceptual understanding of the sediment budget at the Matanuska and serves as the framework for our attempts to quantify various sediment fluxes. Quantified fluxes are shaded green; assumed fluxes are shaded yellow; fluxes not addressed are white. Sediment pathways are indicated by arrows, along with the process driving the flux (where possible). Finally, the budget is zoned (Upland, Glacier, Proglacial) to indicate where in the landscape a given process is taking place. This figure is available in colour online at www. interscience.wiley.com/journal/espl Copyright © 2009 John Wiley & Sons, Ltd.

Earth Surf. Process. Landforms, Vol. 34, 2008–2022 (2009) DOI: 10.1002/esp


2011

Figure 3. (a) Classification of the Matanuska basin into supraglacial fan type (red shading) and valley fan type (green shading) landscapes on the basis of landscape morphometry. Supraglacial fan type landscapes are assigned a lowering rate of 1·8 ± 0·5 mm yr−1, calculated from the surveyed supraglacial fans at Sheep Glacier. Valley fan type landscapes are assigned a lowering rate of 8·5 ± 3·4 mm yr−1, representing an average lowering rate calculated from the Camp, Dry, and Laura Fans. (b) Photograph of the valley fan area shows the gentle gradient associated with this landscape type. (c) Photograph of a supraglacial fan type area shows steep slopes and active colluvial deposition onto the glacier. This figure is available in colour online at www.interscience.wiley.com/journal/espl

Figure 4. Contour map of the Matanuska Glacier basin, with convex ridge top sampling points for cosmogenic 10Be marked by filled circles and labeled with sample number (e.g. MAT 27). Inset photographs show sampling sites as indicated. This figure is available in colour online at www. interscience.wiley.com/journal/espl

deposits of fans on the surface of Sheep Glacier as well as down valley of the glacier (Church and Ryder, 1972; Reneau et al., 1989; Bertran, 2004): ρ  Af ∗ Hf ∗  s   ρr  E= Αc ∗ Tf Copyright © 2009 John Wiley & Sons, Ltd.

(1)

where Af is the area of the fan (L2); Hf is the average fan thickness (L); ρs and ρr are the bulk densities of the colluvial fan deposit and bedrock, respectively (M L−3); Ac is the drainage area contributing to the fan (L2); and Tf is the time since fan deposition began (T). Equation (1) is only valid for fan deposits that represent complete retention of sediment from the contributing area. The set of supraglacial fans deposited onto the surface of Sheep Glacier satisfy this condition, as they were Earth Surf. Process. Landforms, Vol. 34, 2008–2022 (2009) DOI: 10.1002/esp

2012

C.R. O’FARRELL ET AL. (a)

n p OFa Cam R

P*

Dry Fan Laura Fan N

K

C

M

B

D

A*

ne orai pM m a C

L Q

I*

J E

F

G*

H*

(b)

Figure 5. (a) Photograph of valley fans in Sheep Valley, looking north with Laura, Dry, and Camp Fans labeled. Soil pit locations are indicated by letters A through R. Soil pits that encountered subglacial material are marked with an asterisk (*). The large Sheep Glacier end moraine on which we camped is also identified as Camp Moraine. (b) False-color aerial photograph of the valley fans, with fan boundaries delineated by dashed lines. Posited Sheep Glacier terminal moraines are outlined in solid lines, with posited moraine ages indicated. Posited moraine ages are based on chronologies of nearby glaciers by Wiles and Calkin (1994). A large slump feature instigated by cutbank erosion of Sheep Creek is also indicated. This figure is available in colour online at www.interscience.wiley.com/journal/espl Table I. Uncertain Parameters Uncertainty (%)

Basis

Supraglacial Fans Contributing Area Fan Area Fan Thickness Age

10% 20% 20% 0%

Standard map analysis uncertainty Photogrammetric analysis of fit between solid wedge approximation and fan shape Natural variability observed in fan thickness (1-sigma) Observation of spring snow layer provides timing of initial deposition

Valley Fans Contributing Area Surveyed Fan Sections Truncated Wedge Approximations Age

10% 30% 50% 30%

Standard map analysis uncertainty Uncertainty in closing field surveys Conservative estimate of projecting field surveys Variability in age of moraine-forming intervals (Wiles and Calkin, 1994)

Proglacial Sediment Yield Annual Matanuska Sediment Yield

28%

Inter-annual variability on 8-year record (1-sigma)

Parameter

deposited in the year preceding our field investigations and did not show any evidence of sediment loss. Truncation of the fans located in the unglaciated portion of Sheep Valley obviously occurred, as indicated by the steep fan faces above Sheep Creek’s floodplain (Figure 5a) and we account for this loss of material as described below. Uncertainties on each of the variables in Equation (1) were propagated using standard error propagation techniques; Table I details the uncertainty values for these variables and describes the relative uncertainty on each term.

Valley fans Fans on the north side of the study valley were deposited following retreat of Sheep Glacier from a Little Ice Age (LIA) Copyright © 2009 John Wiley & Sons, Ltd.

maximum over the last several hundred years (Figure 5b). We estimate the recent history of Sheep Glacier based on glacial chronology studies at analogous sites in the nearby mountains of South-Central Alaska. Padginton (1993) estimated that equilibrium line altitudes (ELA) of land-terminating glaciers in the southern Kenai Mountains were depressed 100 to 150 m during the Little Ice Age. Wiles and Calkin (1994) present high-resolution glacial chronologies for 16 land-terminating and seven tidewater glaciers in the Kenai Mountains, with a focus on the Little Ice Age. The LIA chronologies of the Wosnesenski, Dinglestadt, Goat, and Nuka Glaciers were chosen as the best analogs for Sheep Glacier given their basin morphology and position relative to the drainage divide. Analysis of stereophoto imagery of Sheep Valley indicates the presence of a moraine feature downvalley of Laura Fan Earth Surf. Process. Landforms, Vol. 34, 2008–2022 (2009) DOI: 10.1002/esp


(Figure 5b). The elevation of this feature is consistent with a 200 m depression of Sheep Glacier’s ELA from its present location. No other potential moraine features are evident further down valley of this feature and we assumed that this feature represents Sheep Glacier’s LIA terminal moraine. Ages assigned to moraine features and the retreat rates assigned to Sheep Glacier are consistent with Wiles and Calkin (1994, Figures. 21 and 22). Initial fan deposition was assumed to occur when the glacier terminus retreated past a given fan’s catchment mouth. The timing of the postulated scenario is not constrained by field data and directly influences fan ages and thus calculated fan-based erosion rates. The variability in moraine-forming intervals among the land-terminating glaciers suggests an uncertainty of 30% on fan age (Wiles and Calkin, 1994). We used a differential Trimble GPS-Total Station to produce a digital elevation model (DEM; 10 m resolution) for each of the three valley fans. The Border Ranges Fault juxtaposes the foliated metamorphic rocks of the glaciated upper reaches of Sheep Valley with the massive gabbroic rocks of the catchments sourcing the valley fans (Figure 1). The Sheep Glacier effectively erodes, transports, and deposits sediments from the headwaters of the basin to the lower reaches of Sheep Valley such that the presence of foliated rocks from the south side of the fault was used as a key indicator of glacial influence down valley. We dug soil pits on the fans (points A through R of Figure 5a) to identify the transition from massive igneous (native – fan) to phyllite (exotic – glacial) lithology. This transition was considered to mark the base of each fan deposit and provided the lower surface over which to compute fan isopachs. In some cases, we were unable to reach the base of a fan deposit and could only determine a lower bound on fan volume. This uncertainty in sediment depth contributes to the propagated uncertainty. Some of the fans have lost material to lateral erosion by Sheep Glacier’s proglacial stream or to incision and remobilization by fan source streams. In the case of material lost to lateral erosion, a sharp break in slope is evident on a downgradient topographic profile of Laura Fan (Figure 6a) and Dry Fan. We approximated the amount of sediment lost to erosion as a wedge extending from the base of the fan face to the stream channel and running along the length of the fan face

(a)

2013

(shown schematically in Figure 6b). The angle of the ‘wedgetop’ is parallel to the gradient of the fan surface at its distal end (Figure 6a). We then added the volume of these missing wedges to the volumes generated from our digital survey and fan depth measurements to determine the total fan volume. The determination of the fan volumes is the largest source of uncertainty in calculating erosion rates from fans. Observations of the contact between fan deposits and subglacial deposits are sparse and made determination of the predepositional surface difficult at times. We have good control on fan thickness where deposits pinch out to subglacial valley walls, and at the distal end of the fans where soil pits have established the presence of subglacial deposits. In between, the subglacial surface is mostly unconstrained and thus fan thickness could vary by several meters or more. For example, soil pits Q and K at the exposed base of Laura Fan (Figure 5a) failed to encounter subglacial material, and provide only a lower bound on fan thickness of 22 m and 15 m, respectively. Additional volumetric uncertainties result from our attempt to account for material lost to stream erosion at the surface and at the toe of the fan, resulting in conservative uncertainty of 30% to fan volumes determined from survey data and 50% to volumes determined by wedge reconstruction of missing material (Table I). We delineated catchment areas using the National Elevation Dataset (USGS, 2004, 30 m resolution) using standard GIS software. We subsequently applied the average of the three Valley Fan erosion rates from Equation (1) to calculate sediment yield from all of the gently sloping terrain north of the Border Range Fault. To calculate a lowering rate for bedrock from colluvial deposits, we estimated a regional value for the ρ bulk density ratio ( s ) in Equation (1). The valley fans are ρr primarily composed of debris flow deposits; several debris flow lobes are shown on Dry Fan in the foreground of Figure 5a. Numerous debris flow levees also exist on other portions of the fans but are obscured by vegetation. Iverson and Vallance’s (2001) review of flume experiments and field observations of debris flows indicates that the bulk density ratio for debris flow deposits ranges from 0·3 (pyroclastic flows on Mt St Helens) to 0·6 (USGS flume flows), with a mode of 0·5. We use the modal value for the bulk density ratio.

(b)

Missing material approximated as wedge

Figure 6. (a) Down-fan profile (x, z) of Laura Fan, from its apex at the catchment mouth to its terminus at the Sheep Glacier Creek floodplain. (b) Schematic of wedge approximation for missing fan material at the terminus of Laura Fan. The wedge extends along the fan face (assumed to have been truncated by Sheep Glacier Creek) and out to the stream channel. Copyright © 2009 John Wiley & Sons, Ltd.


2014


Supraglacial fans The steep (45° mean) catchments above Sheep Glacier deposit colluvial fans on the glacier’s surface (Figure 7a). Field observations indicate that rockfall, rockslide, and debris flow are the primary sources for these fans. Colluvial deposits surveyed in August, 2004 overlaid snow that had not undergone complete melt–freeze metamorphosis, indicating that snow deposition must have occurred in late spring (Figure 7b). These scree deposits were roughly uniform mantles of gravel to cobble sized clasts and showed no evidence of large block failures or landslides occurring in the source area. The latespring snow underlying the colluvial fans provided a constraint on the beginning of fan deposition. Given that the majority of freeze–thaw contributing to rockfall occurs in spring and early summer (Matsuoka and Sakai, 1999), we assume that our mid-August survey captured one year of colluvial fan deposition from steep periglacial catchments, and that the calculated erosion rates were approximating annual scree generation from the catchment areas. Imagery of the Matanuska basin taken in April 2003 (15 months prior to our field survey) shows that colluvial fans are common features on the surface of Sheep Glacier and the Matanuska Glacier (Digital Globe, 2003). We surveyed three fans and approximated them as triangular solids to calculate their volumes. Distances and angles were determined in the field by tape and compass survey, with GPS control points at triangle apices. We dug soil pits in the fans to determine sediment thickness. Catchment areas were determined using photogrammetric methods combined with a DEM (USGS, 2004). To capture an erosion rate applicable to a larger contributing area, we combined fan volumes determined from our survey data with photogrammetry to assign volumes to the remaining fans (Figure 7a). We then divided the sediment volume by the entire upslope contributing area to determine an average erosion rate for the entire steep slope pictured in Figure 7a. The fans studied do not appear to include any re-worked lateral moraine material, nor do the analogous fans pictured in Figure 3c. Such an interaction

between periglacial mass movements and moraines is, however, certainly possible and would serve to mobilize formerly stored moraine sediment onto the glacier surface for transport. Uncertainties in our supraglacial fan erosion rates result from propagation of uncertainty in each term of Equation (1). Examination of the fit between our triangle approximations and actual fan areas suggests that the uncertainty in fan area is about 20% (Figure 7). Measured variability in colluvium depth suggests an average uncertainty in fan thickness of 20%. We assume an uncertainty of 10% in the ArcGIS determinations of catchment area, resulting mainly from errors in delineation of source areas. The age of each fan is well-defined by the underlying snow horizon and therefore we assume no uncertainty in assigning fan ages. We applied the erosion rates calculated for each fan type to the rest of the unglaciated portions of the Matanuska basin on the basis of landscape morphology (Figure 3). Supraglacial fan type is characterized by steep, narrow catchments dominated by freeze–thaw and rockslide processes (Figure 3c). Valley Fan type represents catchment areas with gentler slopes dominated by hillslope and fluvial processes (Figure 3b). The Matanuska watershed was classified manually by overlaying a grid of land-surface gradient (from NED data) on the USGS topographic maps of the basin. The supraglacial fan type landscape covers an area of 188 km2 whereas the Valley Fan type landscape covers an area of 63 km2.

Cosmogenic nuclides We collected bedrock samples for analysis of in situ produced cosmogenic 10Be concentration at locations along the crests of ridges that remained ice-free during the Last Glacial Maximum (Manley and Kauffman, 2002). Our collection strategy followed the extensive work on quantifying soil production and point-specific erosion rates done by Heimsath (2006). Sample locations were chosen carefully to avoid areas subject to episodic mass-wasting events and none of the slopes that

Figure 7. (a) Photograph of annually deposited supraglacial colluvial fans and source catchments at Sheep Glacier, with tape-and-compass surveys overlaid. Tape-and-compass survey shows the triangular solid approximation of the fans used for calculation of fan volume. (b) Oblique photograph of several adjacent supraglacial fans, showing a late-spring snow layer overlain by colluvium from the steep (45°) catchments above. This figure is available in colour online at www.interscience.wiley.com/journal/espl Copyright © 2009 John Wiley & Sons, Ltd.



we worked on showed any evidence of recent shallow landsliding or other nonsteady-state erosional processes. The ridge crests pictured in Figure 4 show the typical morphology of ridge top sampling sites. Nuclide concentrations extracted from quartz-bearing rocks and sediments depend on nuclide production and decay rates, as well as the erosion rate of the sampled surface, such that they reflect the exposure history of the sample (Lal, 1991; Nishiizumi et al., 1993; Bierman, 1994; Gosse and Phillips, 2001). For relatively smooth, convex-up ridge crests that are arguably eroding by steady-state diffusionlike processes (e.g. freeze–thaw, ravel) the nuclide concentrations at the surface record long-term (~10 kyr) background rates of nonglacial erosion. We followed standard and extensively applied quartz separation and analytical procedures in the processing of the samples for this study (Gosse and Phillips, 2001). Results from these samples yielded such low erosion rates in comparison with our fan-based methodology that separating the narrow convex-up ridge crests from the rest of the landscape in the DEM yielded no change in calculate erosion for the entire region. We therefore only report these rates and discuss them in the context of our other results.

Suspended sediment and discharge The primary meltwater discharge of the Matanuska Glacier occurs along its western margin (Figure 1). Subglacial conduits feed multiple vents and fountains along the ice edge, forming larger channels that drain from the moraine area into the Matanuska River (Lawson, 1979; Lawson et al., 1998; Ensminger et al., 1999). A single gage, indicated as SGS on Figure 1, captures the overwhelming majority of the proglacial drainage. At times, the percentage of total proglacial drainage at SGS may be as little as 80% with ice flow and reorganization of the subglacial drainage system, but this is the exception and we use the more typical value of 90% total discharge for our calculations. Suspended sediment concentration and discharge during the melt season (typically June through September) have been measured at SGS since 1995. Discharge measurements were calculated from a rating curve using stage levels taken at 10 min intervals, while suspended sediment samples were taken every 1–2 h by an ISCO© automated stream sampler. Suspended sediment concentrations were measured using standard filtration techniques. Suspended sediment flux is calculated as the product of suspended sediment concentration and the average discharge over the 1–2 h interval. Pearce et al. (2003) studied two of the numerous vents where meltwater is discharged under pressure from the subglacial drainage system and did not find a significant amount of bed load in those upwellings. The overdeepenings that occur within the lower reaches of the terminal lobe (Arcone et al., 1995; Lawson et al., 1998) are probably responsible for trapping the coarse material that is transported in the glacial system. With the extensive efforts extended in quantifying contributions of bedload in mountainous rivers (Warburton, 1992) and the relatively wide range of estimates for this contribution (Pratt-Sitaula et al., 2007), it is unreasonable to ignore bedload in the proglacial sediment flux of the Matanuska system. Measured bedload contributions to total proglacial load vary, with Church (1972) finding percentages varying from 20 to 80% on the Lewis River, Canada, Østrem et al. (1973) finding percentages varying from 20 to 50% at snouts of 5 Norwegian glaciers, and Bogen (1989) finding percentages varying from ~30 to 60% at Nigardsbreen, Norway. Many studies have used 50% suspended load and 50% bed load, but these studies have often been undertaken in areas Copyright © 2009 John Wiley & Sons, Ltd.

2015

where the bedrock is crystalline or relatively resistant to abrasion. The Matanuska Glacier’s basin is composed of metamorphic rocks of varying grades (Winkler, 1992), some of which are highly susceptible to abrasion. At the Matanuska terminus, the grain-size distribution of debris in basal ice and moraine sediments consists of 28 to 43% clasts greater than 1 mm, which would be carried as bed load in a stream. Recent work does, however, suggest that bedload contributes about 35% to the proglacial sediment yield due to the breakdown of the larger clasts into suspended load (Waterson, 2003). We account for the effects of coarse material storage on the overall sediment flux from the upper reaches of the Matanuska basin and assume a first-order ratio of 1 : 1 ratio for bedload to suspended load (i.e. 50% bedload contribution) in a system where bedload appears to be trapped beneath the glacier near the terminus (Gomez, 1987; Lawson, 1993). To test the sensitivity of our results to bedload contribution assumptions, we also report sediment yields, lowering rates, and nonglacial contribution percentages for the end-member bedload contribution cases from the literature.

Sheep Creek The stage and suspended sediment concentration of the Sheep Creek were monitored at a gage installed near the valley fans (Figure 1). The gage has an overall contributing area of 30·4 km2, with 28% ice cover, that includes the Sheep Glacier (16·7 km2 with 32% ice cover) and two smaller unnamed basins with alpine glaciers. Proglacial streams from the three sub-basins join Sheep Creek approximately 300 m upstream from the gage. Stage was measured automatically with a pressure transducer every 15 min over a 7-week period from June 27 through August 13, 2004. Water samples (0·75 l) were collected by an ISCO© automated water sampler every 2–6 h. We calculated discharge using the stage data and a rating curve determined from our cross-sectional survey of Sheep Creek on August 7, 2004. We could not resurvey the crosssection at another time during the melt season because of safety concerns. Sediment concentration was calculated using standard filtration techniques and calculation of sediment yield for the period of record follows the procedures we followed for the SGS gage. We also surveyed the cross-section of the east tributary, west tributary, and main stem of Sheep Creek near the stream confluence, 300 m upstream from the gage to determine the relative contribution of each of the three sub-basins to the flow recorded at the gage. Each survey consisted of a cross-sectional topographic profile and a series of stream velocity measurements using floating sticks over a measured reach. Surveys yielded discharges (Q) for the Sheep Glacier stream of 2·5 ± 0·8 m3 s−1, 0·4 ± 0·1 m3 s−1 for the eastern tributary, and 0·7 ± 0·2 m3 s−1 for the western tributary, indicating that Sheep Glacier’s proglacial stream accounts for 69 ± 23% of flow at the gage. Field logistics precluded a more complete sampling scheme to determine the suspended sediment contribution of these tributaries and so we assign suspended sediment and discharge quantities to the Sheep Glacier proglacial stream based on this 69 ± 23% amount. Our period of record for Sheep Creek spans only part of the melt season, from late June to mid-August. In 2004, this same period of record at the Matanuska terminus accounted for 53% of the annual discharge. We use this ratio to extrapolate the Sheep Creek mid-summer discharge and suspended sediment data to the whole year. Bed load was not measured for Sheep Creek but, based on published estimates (Hallet et al., 1996), we assume 50% bed load contribution to the proglacial Earth Surf. Process. Landforms, Vol. 34, 2008–2022 (2009) DOI: 10.1002/esp

2016


sediment yield of Sheep Glacier (Gomez, 1987; Lawson, 1993). Ignoring bedload for Sheep Creek is unreasonable given our observations of active bedload transport even during the relatively low stages of our field season, but measuring precisely bedload was beyond the scope of this project.

Results Fan deposits The volume of sediment in the three valley fans ranges over an order of magnitude (Table II). The overall upslope contributing area for the three fans is 5·5 ± 1·1 km2. Given the Sheep Glacier retreat scenario depicted by Figure 5b, we calculate an average erosion rate of 8·5 ± 3·4 mm yr−1 for the valley fans during the last 150–200 years (Equation (1)). Applying this average lowering rate to the 66 km2 of unglaciated terrain classified as ‘Valley Fan’ type (Figure 3), yields an annual periglacial sediment flux of 1·4 ± 0·6 × 106 t for this area of the Matanuska watershed (Table III). Supraglacial fans (Mat Fans 1, 2, 3 of Figure 7) had relatively similar volumes of debris relative to their contributing drainage area and therefore yielded roughly similar rates of erosion. The spatially averaged erosion rate for the upslope area contributing sediment to the supraglacial fans on Sheep Glacier is 1·8 ± 0·5 mm yr−1 (Table III). Application of this lowering rate to the 188 km2 of unglaciated terrain classified as ‘Supraglacial Fan’ type (Figure 3), yields an annual nonglacial sediment flux from this portion of the landscape within the Matanuska basin of 0·89 ± 0·40 × 106 t.

Table II. Fan survey results

Feature Valley Fans MatFan 1 MatFan 2 MatFan 3

Age (yrs)

Fan Volume (m3)

Drainage Area (km2)

1 1 1

520 ± 150 390 ± 110 400 ± 110

0·015 ± 0·002 0·097 ± 0·010 0·095 ± 0·010

301680 ± 90500 1519460 ± 298240 4546800 ± 924740

0·065 ± 0·006 0·510 ± 0·051 4·478 ± 0·448

Supraglacial Fans Camp Fan 188 ± 56 Dry Fan 174 ± 52 Laura Fan 163 ± 49

Cosmogenic nuclide analyses Eight samples of intact bedrock or weathered bedrock at the ground surface yield point-specific erosion rates based on the concentrations of in situ produced 10Be. Samples span the field area (Figures 1 and 4) and are collected across the Border Ranges Fault to span bedrock types for the field area. Local morphology for the sample locations is in all cases convex-up with no evidence of extensive soil cover or non-steady-state erosional processes being active. Local slope varied from zero to 33° and erosion rates show no correlation with slope (Table IV). Rates vary by a factor of three, from about 0·05 mm yr−1 to about 0·15 mm yr−1, and showed no systematic variation with any estimated or measured morphometric parameter. The average of all eight samples is 0·08 ± 0·01 mm yr−1. We attribute this relative lack of variation to our selection of erosionally stable (i.e. showing no gullying, rilling, landsliding) ridge crests that would capture the background nonglacial erosion rate of the landscape. These long-term (~10 kyr) erosion rates from the ice-free parts of the Matanuska basin provide a comparison with the proglacial and periglacial rates that are valid over annual to 100-year timescales. Given the sampling locations, the 10Be-derived erosion rates also provide insight into the partitioning of erosional processes for the landscape, which we discuss below.

Matanuska Glacier proglacial sediment yield Daily average discharge data from the SGS gage at the Matanuska Glacier terminus show maxima in mid-summer that gradually decrease through the remainder of the melt season. A plot of annual suspended sediment yield versus annual discharge for the period of 1997–2004 shows significant inter-annual variability (Figure 8). We use, therefore, the 8-year average annual suspended sediment yield of 1·4 ± 0·4 × 106 t as our representative value. Uncertainty on suspended sediment yield represents 1-sigma standard deviation on the range of inter-annual variability over our period of record. Inclusion of bed load (50% bed load to 50% suspended load) yields a basin-wide sediment yield of 2·9 ± 1·0 × 106 t. The suspended sediment yield corresponds to a basin-wide lowering rate of 0·9 ± 0·3 mm yr−1, while the total estimate of sediment yield, including bed load, corresponds to a rate of 1·8 ± 0·6 mm yr−1 (ρ = 2650 kg m−3), for the basin area of 600 km2. Measured values for bedload contribution to the total proglacial flux range from 20% to 80% of the total flux (Church,

Table III. Summary of Lowering Rates and Sediment Yields

Summed Valley Fans Summed Supraglacial Fans Sheep Nonglacial Total Sheep Suspended Proglacial Sheep Glacier Total (20% bedload) Sheep Glacier Total (35% bedload) Sheep Glacier Total (50% bedload) Sheep Glacier Total (80% bedload) Total Nonglacial Matanuska Suspended Proglacial Matanuska Glacier Total (20% bedload) Matanuska Glacier Total (35% bedload) Matanuska Glacier Total (50% bedload) Matanuska Glacier Total (80% bedload) Copyright © 2009 John Wiley & Sons, Ltd.

Contributing Area (km2)

Erosion Rate (mm yr-1)

± ± ± ± ± ± ± ± ± ± ± ± ± ±

8·5 ± 3·4 1·8 ± 0·5 N/A 6·1 ± 2·0 7·5 ± 4·3 9·3 ± 5·3 12·1 ± 6·9 30·2 ± 17·2 N/A 0·9 ± 0·3 1·1 ± 0·4 1·4 ± 0·5 1·8 ± 0·6 4·5 ± 1·4

5·5 0·2 5·1 16·7 16·7 16·7 16·7 16·7 252 600 600 600 600 600

1·1 0·04 1·01 3·34 3·34 3·34 3·34 3·34 50·4 120 120 120 120 120

Annual Sediment Yield (tonnes) 1·4E+06 8·9E+05 5·4E+04 2·7E+05 3·4E+05 4·1E+05 5·4E+05 1·4E+06 2·3E+06 1·4E+06 1·8E+06 2·2E+06 2·9E+06 7·2E+06

± ± ± ± ± ± ± ± ± ± ± ± ± ±

6·0E+05 4·0E+05 2·0E+04 8·3E+04 1·9E+05 2·3E+05 3·0E+05 7·5E+05 1·0E+06 4·0E+05 6·2E+05 7·7E+05 1·0E+06 2·5E+06

% of Total

16% 13% 10% 4%

± ± ± ±

11% 9% 7% 3%

128% 104% 80% 32%

± ± ± ±

71% 58% 45% 18%



2017

Periglacial processes

Figure 8. Hysteresis plot of annual suspended sediment flux versus annual discharge for proglacial drainage from the SGS gage at the Matanuska Glacier terminus for the period 1997–2004 (filled circles). The 8-year average is also shown (open triangle).

1972). Corresponding end-member estimates of sediment yield range from 7·2 ± 2·5 × 106 t to 1·8 ± 0·6 × 106 t, with basin-wide lowering rates of 4·5 ± 1·4 mm yr−1 and 1·1 ± 0·4 mm yr−1, respectively.

Sheep Glacier sediment yield The recorded and extrapolated discharge and suspended sediment load for the 2004 melt season at the Sheep Creek gage yields a total annual suspended sediment flux of 0·39 ± 0·12 × 106 t. Accounting for bedload and the contributions of tributaries to Sheep Creek, we calculate an annual proglacial sediment yield from Sheep Glacier of 0·54 ± 0·30 × 106 t in 2004. This sediment yield corresponds to a basin-wide lowering rate of 12·1 ± 6·9 mm yr−1 (ρ = 2650 kg m−3, basin area of 16·7 km2). Corresponding end-member estimates of sediment yield range from 1·4 ± 0·8 × 106 t to 0·3 ± 0·2 × 106 t, with basin-wide lowering rates of 30·2 ± 17·2 mm yr−1 and 7·5 ± 4·3 mm yr−1, respectively.

Discussion Calculated periglacial erosion rates are indicative of differences in factors driving erosion throughout the unglaciated landscape. Position relative to glacial ice, lithology, timescale of measurement, and difference in dominant process all play roles in the differences that we observe in periglacial erosion rates. Comparison of summed periglacial fluxes with the Matanuska Glacier’s total proglacial sediment flux suggests that periglacial fluxes are equivalent to 80 ± 45% of the total proglacial sediment yield, providing an important estimate for the hillslope contributions that have previously been relatively unknown (Warburton, 1990). Given the uncertainties in our first-order estimates, we suggest that the contributions of sediment from glacial and nonglacial sources are similar in magnitude. Copyright © 2009 John Wiley & Sons, Ltd.

The relative magnitude of lowering rates calculated for the Supraglacial and Valley Fan landscapes (1·8 ± 0·5 mm yr−1 and 8·5 ± 3·4 mm yr−1, respectively) are due to differences in glacial state, lithology, and timescale of processes captured by the fan deposit. We suggest that the greater erosion rates calculated for the Valley Fan landscape are a consequence of lowered base level after glacial retreat, similar to the review of Ballantyne (2002) for paraglacial landscapes. The retreat of Sheep Glacier from its confluence with the Matanuska (likely at Last Glacial Maximum) and subsequent Little Ice Age advance/retreat have served to depress local base level and effectively oversteepen the walls of Sheep Valley. The conceptual model of the paraglacial sediment cycle presented by Church and Slaymaker (1989, Figure 2) is based on their findings for recently deglaciated catchments in British Columbia and is consistent with our results showing the highest erosion rates in recently deglaciated upland landscapes. In this framework, the Supraglacial Fan landscape has not yet reached the peak of the upland sediment yield curve associated with deglaciation while the Valley Fan landscape is at or beyond this peak. This difference may be most clearly represented by the ‘missing material’ shown in Figure 6b, where the missing material from this valley fan is precisely the type of increased sediment yield captured by the peak in the Church and Slaymaker (1989) conceptual figure, and discussed further in Slaymaker (2003) and Delmas et al. (2009). Lithologic differences between the two landscape types may also play a role in the approximately four-fold difference in calculated erosion rates. The catchments sourcing the supraglacial fans and much of the upper Matanuska basin are composed of medium-grade metamorphic phyllite. The Valley Fan source catchments are located within a zone of intensely sheared gabbroic rocks (Winkler, 1992) and exhibit transport limited morphology, as their channels and upper reaches are choked with colluvium and debris flow deposits (e.g. Dry Fan catchment in Figure 5a). Although our cosmogenic nuclide measurements do not show clear differences in the long-term attrition rates attributable to these two bedrock types, it is likely that the differences in material strength would contribute to the erosional processes and rates active across these very different rock types. The fact that the morphology changes dramatically from Supraglacial Fan type (Figure 7a) to Valley Fan type (Figure 6b) across the lithologic boundary supports the role of material properties in helping set the differences in basin scale erosion rates. The supraglacial erosion rate is based on measurements of one year’s rockfall, a high-frequency, low magnitude process, supported by our observations of the deposits, while the Valley Fan erosion rate includes events spanning at least the last 100 years. The longer period of record for the valley fans may allow them to integrate higher magnitude episodic processes not reflected by the supraglacial fans. For example, in the nearby Chugach-St. Elias Mountains of southern Alaska, Arsenault and Meigs (2005) have shown that bedrock-seated landslides alone account for 0·48 mm yr−1 of bedrock lowering in a basin of ~ 19 km2. Their results suggest an average recurrence interval of 55 years for supraglacial landslides that contribute rock to the glacier surface, indicating that erosion rates for the periglacial landscape based solely on shorttimescale, annual features may be underestimated. Previous work in mountainous terrain has also suggested that measurements of short-term sediment yield may underestimate longerterm erosion rates (Kirchner et al., 2001). Thus, our supraglacial fan-based rates may represent a lower bound on nonglacial contributions from this portion of the landscape, significantly Earth Surf. Process. Landforms, Vol. 34, 2008–2022 (2009) DOI: 10.1002/esp

2018


Table IV. Summary of Cosmogenic Lowering Rates derived from in-situ

Latitude MAT MAT MAT MAT MAT MAT MAT MAT

1a 5 15 21 22 25 27 42

61° 61° 61° 61° 61° 61° 61° 61°

45·156′ 45·312′ 45·419′ 40·926′ 40·910′ 40·693′ 40·650′ 45·800′

Longitude 147° 147° 147° 147° 147° 147° 147° 147°

36·030′ 35·955′ 34·286′ 26·410′ 26·760′ 25·660′ 26·110′ 37·924′

10

Be

Elevation (m)

Slope (deg)

10 Be atoms/gm

σ(10Be)+ atoms/gm

corr* P0

1021 1045 1305 1570 1478 1450 1440 1242

16 16 0 0 20 25 25 33

133494·5 157460·6 201466·7 158458 181695·3 69826·2 84486·6 195833·2

3129·7 3786·4 4820·9 10216·2 4431·9 1748·4 2063·5 3821·6

12·3 12·6 15·7 19·5 17·9 17·3 17·2 14·3

Erosion Rate mm yr-1 0·058 0·050 0·048 0·077 0·061 0·155 0·127 0·045

± ± ± ± ± ± ± ±

0·004 0·004 0·004 0·009 0·005 0·012 0·009 0·003

* Corrected 10Be production rate based on elevation, slope and latitude, e.g. Gosse and Phillips (2001). Uncertainty based on propagated analytical from all lab measurements.

+

underestimating the long-term erosion rates that include large landslides or ridge-lowering blockfall events. Erosion rates for sampled convex ridge tops in the Matanuska basin average 0·08 ± 0·04 mm yr−1 (Table IV), an order of magnitude lower than rates calculated from proglacial suspended sediment and two orders of magnitude lower than rates calculated from fan volumes. This difference illustrates the relative importance of the various erosional processes acting across the basin, compared with the stable ridge crests, and is consistent with previous work indicating that short-term estimates of erosion from sediment yield in glaciated basins of the Chugach are an order of magnitude higher than longterm exhumation rates (Spotila et al., 2004), possibly due to the paraglacial effect during present-day glacial retreat. In that case, the paraglacial effect results in the oversteepened slopes adjacent to the channel and the significantly increased sediment delivery to the channel. The difference in rates and processes active between the ridge crests and the entire basin also supports the modeling results of Dadson and Church (2005) that were calibrated with field data from British Columbia, Canada. Relatively high erosion rates calculated from the valley and supraglacial fans allow us to infer that the unglaciated portion of the landscape responds rapidly to glacially driven incision. The efficacy of this rapid response accounts for the absence of a deeply incised valley confined to the central axis of the Matanuska basin. The rockfall, debris flow and fluvial processes producing the valley and supraglacial fans are apparently highly dominant compared with the steady-state processes acting to lower the convex ridge crests. The disparity between cosmogenic and fan-based rates also indicates that high rates of erosion in gullies and catchments do not appear to drive erosion on the ridges. The differential erosion rates calculated between ridge tops and catchments imply relief development of 168 ± 68 m in the valley catchments and 34 ± 10 m in the supraglacial catchments over the last 20 ka. Such a pattern of relief development is consistent with the present-day observed morphology of deep catchments in the Valley Fan landscape and shallow gullies with small interfluves in the supraglacial landscape. Second-order catchments within the Laura Fan drainage have an average depth of 166 ± 14 m when measured from interfluve to interfluve whereas supraglacial fan catchments have an average depth of 34 ± 10 m. Figure 9 demonstrates the derivation of these catchment depths between interfluves. Interestingly, both Heimsath et al. (1999) and O’Farrell et al. (2007) measured higher erosion rates for whole catchments compared with ridge tops. Such agreement across field areas suggests that point-specific erosion rate measurements using cosmogenic nuclides on stable ridges may underestimate basin scale erosion rates that Copyright © 2009 John Wiley & Sons, Ltd.

include ridge-lowering mass-wasting events that happen infrequently.

Proglacial versus periglacial rates Glacial sediment yields and lowering rates provide a context for comparing our results with other Alaskan glaciers (Hallet et al., 1996; Koppes and Hallet, 2002). Our estimate of 1·8 ± 0·6 mm yr−1 for the erosion rate of the Matanuska basin from proglacial sediment yield is on the lower end of the range of reported values – values that vary by an order of magnitude. The relative magnitude of periglacial, proglacial suspended, and total glacial sediment flux indicates that there is significant sediment stored in the Matanuska basin. Total periglacial flux is composed of 68% from the Valley Fan landscape and 32% from the supraglacial landscape and exceeds the proglacial suspended sediment flux by 60%. The magnitude of nonglacial sediment yield is 80 ± 45% of the total glacial sediment flux when a 50% bedload contribution is included. The magnitude of nonglacial sediment yield ranges from 32 ± 18% to 128 + 71% of the total glacial sediment flux for bedload contributions of 20% and 80%, respectively. Subglacial storage primarily consists of coarse clasts, as do the majority of upland fluxes; these upland fluxes are not presently being delivered to the glacier terminus and the proglacial Matanuska River. The fans we surveyed are an obvious sediment reservoir and are ubiquitous throughout the Matanuska basin. Pearce et al. (2003) posited storage of coarse sediment in subglacial reservoirs as a possible explanation for the absence of bed load in subglacial vent discharge at the Matanuska terminus. The role of such stored sediment in the total sediment budget of glaciated regions is reasonably well known and broadly articulated (Church and Ryder, 1972; Warburton, 1990; Slaymaker, 2003; Delmas et al., 2009), but rarely considered in assessing the rates of subglacial erosion based on proglacial sediment yield or depositional sequences such as in fjords of tidewater glaciers. Alley et al. (2003) have shown that down-glacier bed slopes modulate glacial sediment production by developing equilibrium, long bed profiles about 50% steeper than the overlying ice-surface slope. Achievement of this slope strongly limits additional erosion. Perturbations in the ice slope may shift the subglacial regime from one of erosion to sedimentation. Because of this tendency to equilibrium long-bed profiles in subglacial basins, the release of these sediments to proglacial and distal sedimentary environments over the long term will require continued down-cutting of proglacial regions or tectonically-driven tilting and uplift (Meigs and Sauber 2000; Sheaf et al. 2003). The Matanuska Glacier has developed Earth Surf. Process. Landforms, Vol. 34, 2008–2022 (2009) DOI: 10.1002/esp


2019

Figure 9. (a) Topographic map and photograph of the Laura Fan catchment, with annotations showing the measured depth of several channels from interfluve to interfluve, with both first-order and second-order channels represented. Channel depth is considered to be the maximum vertical distance between a line of equal elevation pegged to two interfluve crests and the channel bottom. (b) Topographic map and photograph of the supraglacial fan catchment above Sheep Glacier, with annotations showing the measured depth of several channels from interfluve to interfluve. Only first-order channels exist in this region. Channel depth is considered to be the maximum vertical distance between a line of equal elevation pegged to two interfluve crests and the channel bottom. This figure is available in colour online at www.interscience.wiley.com/journal/espl

overdeepenings within its bed (Lawson et al., 1998; Arcone et al., 1995) that may now be limiting further erosion and inducing deposition and sediment storage. Glaciofluvial erosion is the primary means of removal of subglacial sediment from temperate glaciers, but if the slope out of an overdeepening is too steep relative to the ice-surface slope, then the pressure dependence of the melting point causes subglacial channels to freeze closed, greatly decreasing sediment flux, and armoring the glacier bed with till that blocks further erosion. Over sufficiently long times, the bed profiles of terminal regions of highly erosive glaciers will be controlled by the ability of proglacial regions to down cut. The proglacial region beyond the Matanuska terminus has remained relatively stable with little down cutting by the main river evident over the last 100 years or more, and this has created a basin(s) within which sediments are now trapped as the ice thins, thereby limiting sediment yield. The upper area of the Sheep Glacier basin is essentially a closed system with well-defined boundaries and reasonably well-documented periglacial and proglacial fluxes; it provides an interesting counterpoint to the trends observed in the larger Matanuska basin. Dimensions differ significantly from the Matanuska, as does the percentage of ice cover (32% of a 16·7 km2 basin versus 58% of a 600 km2 basin). We surmise that the Sheep Glacier has been rapidly retreating over the last several decades based on the extensive area of loose, unsorted and unvegetated gravels and diamictons located immediately down-valley of the glacier (Figure 10). The 2004 sediment yield of Sheep Glacier corresponds to a basin-wide lowering rate of 12·1 ± 6·9 mm yr−1. This erosion rate is ~7 times greater than that calculated for the largely stable Matanuska Glacier. The percentage of sediment contribution from nonglacial sources is 10 ± 7% for Sheep Glacier, considerably less than for the much larger Matanuska Glacier and consistent with periglacial inputs being overwhelmed by increased glacial erosion and/or release of stored sediment during Sheep Copyright © 2009 John Wiley & Sons, Ltd.

Lateral Erosion by Sheep Creek Slump of unvegetated subglacial material Increased sediment in Sheep Creek

Figure 10. Photograph of the loose, unvegetated glacial drift immediately downvalley of Sheep Glacier and interpreted to be evidence of its recent retreat. The sediments considered to be sourced from the glacier’s bed (not including moraines) are outlined by a dashed line. Recent slumps of this material into Sheep Creek are highlighted and identified. This figure is available in colour online at www. interscience.wiley.com/journal/espl

Glacier’s rapid retreat. The magnitude of nonglacial sediment yield ranges from 4 ± 3% to 16 + 11% of the total glacial sediment flux for bedload contributions of 20% and 80%, respectively. Increased sediment yield from a retreating glacier is consistent with findings elsewhere in Alaska (Hallet et al., 1996; Koppes and Hallet, 2002). Increased subglacial erosion at the Sheep Glacier could be accomplished by greater subglacial abrasion associated with an increase in the mass of ice transiting the glacial system (Koppes and Hallet, 2002) or an abundance of glacial meltwater, the primary mode by which Earth Surf. Process. Landforms, Vol. 34, 2008–2022 (2009) DOI: 10.1002/esp

2020


sediment is transported to the glacier terminus (Hallet et al., 1996). We do not have data on the bed configuration, but the fact that water discharges through a tunnel exiting at ground level suggests that at least near the margin, no overdeepening is present. We suggest that increased sediment yield is due at least in part to mobilization of unconsolidated sediments previously stored subglacially and those entering paraglacially (Church and Ryder, 1972; Meigs et al., 2006). These sediments could have been stored in subglacial pockets not fully tapped by the subglacial drainage network, similar to the Matanuska. Subaerial exposure has the potential to increase the mobility of these sediments via unrestricted lateral erosion of proglacial streams and overland flow from precipitation and snowmelt. The subaerial exposure of these deposits is a direct consequence of Sheep Glacier’s retreat and has the potential to contribute a large quantity of sediment to the proglacial stream. Figure 10 shows the recently deglaciated Sheep Glacier foreland and identifies a recent slump that delivered sediment to Sheep Creek as an example of such a process. The difference in the nonglacial contribution between the Sheep and Matanuska glaciers suggests that the rate and style of recession of a glacier may greatly impact its sediment budget, with mobilization of subglacially stored sediment during rapid retreat potentially overwhelming other sources of sediment (Harbor and Warburton, 1993). The role of the overdeepening in the Matanuska Glacier as a sediment trap is an important control as the glacier slowly thins and recedes. Given the potential magnitude of these contributions, as suggested by Sheep Glacier, subglacially stored sediment is a critical part of the glacial sediment budget and needs to be more fully quantified and considered when comparing apparent sediment yields and erosion rates of glacial systems. The Matanuska and Sheep basins also provide examples of the variable importance of paraglacial erosion as a function of glacial ice state, both between basins and at varying points within Sheep Valley. The variability in ice state means that the fluxes quantified plot along different parts of the conceptual paraglacial sediment cycle curve suggested by Church and Slaymaker (1989, Figure 2) and Slaymaker (2003, Figure 5). Comparison with a completely unglaciated catchment nearby would be useful to constrain background rates of erosion without any ice influence. Inputting such field-based quantification of rates into a landscape evolution model incorporating the different erosional processes would help resolve the long-term effect, over many glacial cycles, of such differences in observed rates.

Glacier basin suggest that nonglacial processes contribute 80 ± 45% and 10 ± 7% of the proglacial sediment yield, respectively. Differences between the two glaciers may be due to differences in apparent rates of retreat, supporting the concept that retreating glaciers contribute large sediment yields due in part to the mobilization of stored sediment. The disagreement between our cosmogenic lowering rates and the rates determined from the colluvial fans is likely to be a function of differences in dominant process at different locations relative to glacier surface. Areas sampled for 10Be were on gently sloped ridge tops, separated from the immediate influence of glacier erosion, whereas the colluvial fan sediment volumes were a function of slope processes operating over a much larger area, integrating sub-basins below the ridge tops that were being eroded by landsliding and extensive scree production processes. By lowering base level and debuttressing adjacent valley slopes, glaciers set the pace of erosion in upland landscapes in a manner similar to rapidly incising rivers (Meigs and Sauber, 2000; Dadson et al., 2003). In contrast, long-term rates in glaciated mountains may be modulated by a coupling between tectonic uplift and climate (Spotila et al., 2004). Our findings suggest that glaciers are likely to drive significant erosion through slope adjustments, not solely through subglacial mechanisms, and that subglacial and nonglacial erosional processes are similar in magnitude. However, subglacial erosion rates and sediment yield may also differ over the short and long term in response to changes in the down glacier bed profile and perturbations in ice surface slope modulating erosion, sedimentation and storage of sediment at the bed (Alley et al., 2003). Sediment yields will respond to changes in ice dynamics and the configuration of the glacier bed, while also responding to the current state of the proglacial environment and ongoing erosion or sedimentation activity. Building upon the contrasts between the Sheep and Matanuska glaciers in this study, we suggest that further work at the Matanuska Glacier focused on understanding subglacial sediment storage patterns could improve our understanding of the importance of glaciers as erosional agents. A study of several nearby, unglaciated basins would also provide an excellent baseline for exploring background rates of alpine erosion under similar climatic and tectonic regimes where the slopes are not influenced by glacial ice. Similarly, a more complete analysis of the sediment budget throughout the Matanuska basin, including more extensive use of cosmogenic nuclides, will help resolve how nonglacial processes contribute sediment to large glacier systems.

Conclusions

Acknowledgements—We thank Bill and Kelly Stevenson for accommodation, bush plane flights, and help with general field logistics. Funding for this work was from NSF-EAR-9909335 and EAR-0239655 to AMH, Geological Society of America, the Cold Regions Research and Engineering Laboratory (CRREL), the Villhjálmur Stefánsson Fellowship for Arctic Studies, the Richter Memorial Trust, and the American Alpine Club. Extensive reviews, including one from P Jansson on an earlier version, and editorial suggestions from S Lane have significantly helped this manuscript.

Relatively high erosion rates in the unglaciated parts of the Matanuska basin indicate that valley walls respond rapidly to glacially-driven incision. The ~7-fold difference in erosion rates between the two fan types is likely driven by hillslope debuttressing and increased stream incision in the formerly glaciated Valley Fan landscape following ice retreat. The importance of ice in setting base level is consistent with the results of Parks and Madison (1984), which suggest that unglaciated basins produce considerably less sediment than basins with even 10% ice cover. Lithologic differences between the weak, sheared gabbronite of the valley fans and the more resistant metamorphic phyllite of the supraglacial fan landscape may also play a role in the differences observed. Nonglacial contributions to the sediment budget of valley glaciers are significant and contribute to the high erosion rates in partly glaciated, mountainous terrain. Application of fanbased erosion rates to the whole Matanuska basin and Sheep Copyright © 2009 John Wiley & Sons, Ltd.

References Alley RB, Cuffey KM, Evenson EB, Lawson DE, Strasser JC, Larson GJ. 1997. How glaciers entrain and transport basal sediment: Physical constraints. Quaternary Science Reviews 16: 1027–1038. Alley RB, Lawson DE, Evenson EB, Strasser JF, Larson GJ. 1998. Glaciohydraulic supercooling: a freeze-on mechanism to create stratified, debris-rich basal ice. Journal of Glaciology 44: 563–569. Alley RB, Lawson DE, Larson GL, Evenson EB, Baker GS. 2003. Stabilizing feedbacks in glacier-bed erosion. Nature 424: 758–760. Earth Surf. Process. Landforms, Vol. 34, 2008–2022 (2009) DOI: 10.1002/esp

QUANTIFYING PERIGLACIAL EROSION André MF. 1997. Holocene rockwall retreat in Svalbard: a triple-rate evolution. Earth Surface Processes and Landforms 22: 423–440. Arcone SA, Lawson DE, Delaney AJ. 1995. Short-pulse radar wavelet recovery and resolution of dielectric contrasts within englacial and basal ice of Matanuska Glacier, Alaska, USA. Journal of Glaciology: 41: 68–86. Arsenault AM, Meigs AJ. 2005. Contribution of Deep-Seated Bedrock Landslides to Erosion of a Glaciated Basin in Southern Alaska. Earth Surface Processes and Landforms 30: 1111–1126. Ballantyne CK. 2002. Paraglacial geomorphology. Quaternary Science Reviews 21: 1935–2017. Bertran P. 2004. Soil erosion in small catchments of the Quercy region (southwestern France) during the Holocene. The Holocene 14: 597–606. Bierman PR. 1994. Using in situ produced cosmogenic isotopes to estimate rates of landscape evolution; a review from the geomorphic perspective. Journal of Geophysical Research, B, Solid Earth and Planets 99: 13885–13896. Bogen J. 1989. Glacial sediment production and development of hydro-electric power in glacierized areas. Annals of Glaciology 13: 6–11. Brocklehurst SH, Whipple KX. 2002. Glacial erosion and relief production in the Eastern Sierra Nevada, California. Geomorphology 42: 1–24. Church M. 1972. Baffin Island sandurs: a study in Arctic fluvial processes. Geological Survey Canada Bulletin 216. Church M, Ryder JM. 1972. Paraglacial sedimentation: consideration of fluvial processes conditioned by glaciation. Geological Society of America Bulletin 83: 3059–3072. Church M, Slaymaker O. 1989. Disequilibrium of Holocene sediment yield in glaciated British Columbia. Nature 337: 452–454. Dadson SJ, Hovius N, Chen H, Dade WB, Hsieh M, Willet S, Hu J, Horng M, Chen M, Stark CP, Lague D, Lin J. 2003. Links between erosion, runoff variability and seismicity in the Taiwan orogen. Nature 426: 648–651. Dadson SJ, Church M. 2005. Postglacial topographic evolution of glaciated valleys: a stochastic landscape evolution model. Earth Surface Processes and Landforms 30: 1387–1403. Delmas M, Calvet M, Gunnell Y. 2009. Variability of Quaternary glacial erosion rates – a global perspective with special reference to the Eastern Pyrenees. Quaternary Science Reviews 28: 484–498. Digital Globe. 2003. Digital Globe Image Date: 2003-04-02, Catalog ID: 1010010001C4A602. Ensminger SL, Evenson EB, Alley RB, Larson GJ, Lawson DE, Strasser JC. 1999. Glacial processes, past and present: example of the dependence of ice motion on subglacial drainage system evolution, Matanuska Glacier, Alaska. In Geological Society of America Special Paper 337, Mickelson DM, Attig JW (eds); 11–22. Evenson EB, Clinch JM. 1987. Debris transport mechanisms at active alpine glacier margins: Alaskan case studies. Special paper – Geological Survey of Finland 2: 111–136. Gomez B. 1987. Bedload. in Glaciofuvial Sediment Transfer, Gurnell AM, Clark MJ (eds). John Wiley and Sons: New York; 355–376. Gosse JC, Phillips FM. 2001. Terrestrial in situ cosmogenic nuclides: theory and application. Quaternary Science Reviews 20: 1475–1560. Hallet B, Hunter L, Bogen J. 1996. Rates of erosion and sediment evacuation by glaciers: a review of field data and their implications. Global and Planetary Change 12: 213–235. Harbor J, Warburton J. 1993. Relative rates of glacial and nonglacial erosion in alpine environments. Arctic and Alpine Research 22: 26–42. Hebdon NJ, Atkinson TC, Lawson TJ, Young IR. 1997. Rate of glacial valley deepening during the late Quaternary in Assynt, Scotland. Earth Surface Processes and Landforms 22: 307–315. Heimsath AM. 2006. Eroding the land: Steady-state and stochastic rates and processes through a cosmogenic lens. In Application of Cosmogenic Nuclides to the Study of Earth Surface Processes: The Practice and the Potential, Siame LL, Bourles DL, Brown ET (eds). Geological Society of America: Boulder, CO; 111–129. Heimsath AM, Dietrich WE, Nishiizumi K, Finkel RC. 1999. Cosmogenic nuclides, topography, and the spatial variation of soil depth. Geomorphology 27: 151–172. Copyright © 2009 John Wiley & Sons, Ltd.

2021

Heimsath AM, McGlynn R. 2008. Quantifying periglacial erosion in the Nepal high Himalaya. Geomorphology 97: 5–23. Hicks DM, McSaveney MJ, Chinn TJ. 1990. Sedimentation in proglacial Ivory Lake, Southern Alps, New Zealand. Arctic and Alpine Research 22: 26–42. Iverson RM, Vallance JW. 2001. New views of granular mass flows. Geology 29: 115–118. Kirchner JW, Finkel RC, Riebe CS, Granger DE, Clayton JL, King JG, Megahan WF. 2001. Mountain erosion over 10 yr, 10 k.y., and 10 m.y. time scales. Geology 29: 591–594. Kirkbride M, Mathews D. 1997. The role of fluvial and glacial erosion in landscape evolution: the Ben Ohau Range, New Zealand. Earth Surface Processes and Landforms 22: 317–327. Koppes MN, Hallet B. 2002. Influence of rapid glacial retreat on the rate of erosion by tidewater glaciers. Geology 30: 47–50. Lal D. 1991. Cosmic ray labeling of erosion surfaces: in situ nuclide production rates and erosion models. Earth and Planetary Science Letters 104: 424–439. Larson GJ, Evenson EB, Lawson DE, Ensminger SL, Baker G, Alley RB. 2003. Glacial geology of upper Cook Inlet, Matanuska Glacier and Denali highway, Alaska. In Quaternary Geology of the United States – INQUA 2003 Field Guide, Easterbrook DJ (ed). Geological Society of America: Boulder, CO; 245–264. Lawson DE. 1979. Sedimentological analysis of the western terminus region of the Matanuska Glacier, Alaska. Report 79–9, US Army Cold Regions Research and Engineering Laboratory, Hanover, NH. Lawson DE. 1993. Glaciohydrologic and Glaciohydraulic Effects of Runoff and Sediment Yield in Glacierized Basins. Monograph 93-2, US Army Cold Regions Research and Engineering Laboratory: Hanover, NH. Lawson DE, Kulla B. 1978. An oxygen isotope investigation of the origin of the basal zone of the Matanuska Glacier, Alaska. Journal of Geology 86: 673–685. Lawson DE, Strasser JC, Evenson EB, Alley RB, Larson GJ, Arcone SA. 1998. Glaciohydraulic supercooling: a freeze-on mechanism to create stratified, debris-rich basal ice: I. Field evidence. Journal of Glaciology 44: 547–562. Manley W, Kauffman D. 2002. Alaska Paleoglacier Atlas. Available online at http://instaar.colorado.edu/QGISL/ak_paleoglacier_atlas/. Accessed February, 2005. Matsuoka N, Sakai H. 1999. Rockfall activity from an alpine cliff during thawing periods. Geomorphology 28: 309–328. Meigs AJ, Sauber J. 2000. Southern Alaska as an example of the longterm consequences of mountain building under the influence of glaciers. Quaternary Science Reviews 19: 1543–1562. Meigs AJ, Krugh WC, Davis K, Bank G. 2006, Ultra-rapid landscape response and sediment yield following glacier retreat, Icy Bay, southern Alaska. Geomorphology 78: 207–221. Nishiizumi K, Kohl CP, Arnold JR, Dorn R, Klein J, Fink D, Middleton R, Lal D. 1993. Role of in situ cosmogenic nuclides 10Be and 26Al in the study of diverse geomorphic processes. Earth Surface Processes and Landforms 18: 407–425. O’Farrell CR, Heimsath AM, and Kaste, J.M. 2007. Quantifying hillslope erosion rates and processes for a coastal California landscape over varying timescales. Earth Surface Processes and Landforms 32: 544–560. Østrem O, Ziegler T, Ekman SR. 1973. A study of sediment transport in Norwegian glacial rovers, 1969. Institute for Water Resources Pub # IWR-35, University of Alaska, Fairbanks. Padginton CH. 1993. Reconstructions of icefield outlet glaciers and Little Ice Age climatic change, southern Kenai Mountains, Alaska. MS thesis, State University of New York at Buffalo. Pan B, Burbank DW, Wang Y, Wu G, Li J, Guan Q. 2003. A 900 k.y. record of strath terrace formation during glacial–interglacial transition in northwest China. Geology 31: 957–960. Parks B, Madison RJ. 1984. Estimation of selected flow and water quality characteristics of Alaskan streams. Water Resources Investigations Report: 84-424, US Geological Survey. Pearce JT, Pazzaglia FJ, Evenson EB, Lawson DE, Alley RB, Germanoski D, Denner JD. 2003. Bedload component of glacially discharged sediment: insights from the Matanuska Glacier, Alaska. Geology 31: 7–10. Earth Surf. Process. Landforms, Vol. 34, 2008–2022 (2009) DOI: 10.1002/esp

2022


Pratt-Sitaula BA, Garde M, Burbank DW, Oskin M, Heimsath AM, Gabet E 2007. Bedload ratio, regional erosion rate, and rapid bedrock incision from Himalayan landslide-dam lake record. Quaternary Research 68: 111–120. Reneau SL, Dietrich WE, Rubin M, Donahue DJ, Jull JT. 1989. Analysis of hillslope erosion rates using dated colluvial deposits. Journal of Geology 97: 45–63. Sheaf MA, Serpa L, Pavlis TL. 2003. Exhumation rates in the St. Elias Mountains, Alaska. Tectonophysics 367: 1–11. Slaymaker O. 2003. The sediment budget as conceptual framework and management tool. Hydrobiologia 494: 71–82. Slaymaker O, Souch C, Menounos B, Filipelli G. 2003. Advances in Holocene mountain geomorphology inspired by sediment budget methodology. Geomorphology 55: 305–316. Spotila JA, Buscher JT, Meigs AJ, Reiners PW. 2004. Long-term glacial erosion of active mountain belts: Example of the Chugach-St. Elias Range, Alaska. Geology 32: 501–504. Summerfield MA, Kirkbride MP. 1992. Climate and landscape response. Nature 355: 306. United States Geological Survey. 2004. National Elevation Dataset: 2 arc-second Alaska series. EROS Data Center: Sioux Falls, SD.

Copyright © 2009 John Wiley & Sons, Ltd.

Warburton J. 1990. An alpine proglacial fluvial sediment budget. Geografiska Annaler 72A: 261–272. Warburton J. 1992. Observations of bed load transport and channel bed changes in a proglacial mountain stream. Arctic and Alpine Research 24: 195–203. Waterson, N.J. 2003. Subglacial erosion and fate of course material in the subglacial drainage system of the Matanuska Glacier, Alaska. MS thesis, Michigan State University, East Lansing. Wiles GC, Calkin PE. 1994. Late Holocene, high-resolution glacial chronologies and climate, Kenai Mountains, Alaska. Geological Society of America Bulletin 106: 281–303. Williams JR, Ferrians OJ. 1961. Late Wisconsin and recent history of the Matanuska Glacier, Alaska. Arctic 14: 82–90. Winkler GR. 1992. Geologic map and summary geochronology of the Anchorage 1° × 3° quadrangle, southern Alaska. Miscellaneous Investigations Map 1–2283. United States Geological Survey. Zhang P, Molnar P, Downs WR. 2001. Increased sedimentation rates and grain sizes 2–4 Myr ago due to the influence of climate change on erosion rates. Nature 410: 891–897.
