We evaluate spatial and temporal variations in denudation of the north-central Wasatch Mountains, Utah, by determining catchment-wide denudation rates with 10Be concentrations in alluvial sediment and comparing these rates with previously published data on rock uplift and exhumation of the range. Catchments draining the range front show relatively little variation in denudation rate (0.07–0.17 mm/yr), while steeper (mean hillslope gradient >30°) catchments in the core of the range show larger variation (0.17–0.79 mm/yr). We attribute the larger spatial variation in catchment-wide denudation in the core of the range to landsliding of hillslopes at threshold gradients; faster denudation in this region may signify landscape adjustment to late Pleistocene glaciations. The mean denudation rate for all catchments (0.2 mm/yr) is generally consistent with longer-term exhumation rates derived from thermochronometers and with shorter-term vertical fault displacement rates, suggesting that denudation of the north-central Wasatch has been roughly steady, or decreasing slightly, over the past 5 m.y. Although 10Be-based catchment-wide denudation rates are sensitive to localized geomorphic processes and events, overall, they appear to reflect the larger tectonic forces that have driven denudation of the Wasatch Mountains over longer time scales.
Quantification of spatial and temporal variations in denudation of tectonically active areas can provide key insights into mountain belt evolution. For example, spatial variability in mountain denudation rates can help identify tectonically active structures and determine surface uplift rates. Temporal variability can reveal the timing of tectonic events and can clarify the relative roles that tectonics and climate play in driving rock uplift and denudation of orogens. Long temporal records of denudation also provide a means of assessing the steadiness of mountain erosion and topography through time.
Spatial variations in denudation rate can be quantified simply by collecting samples distributed across a broad region, but temporal variations are more challenging, requiring measurements over different time scales. Rates of long-term rock exhumation are often inferred from zircon fission-track (ZFT), apatite fission-track (AFT), and apatite (U-Th)/He (AHe) thermochronometers, which average denudation over relatively long time scales (105–108 yr) (e.g., Ehlers and Farley, 2003; Reiners and Brandon, 2006). Over shorter time scales (102–105 yr), denudation rates can be determined using cosmogenic nuclides such as 10Be (e.g., Gosse and Phillips, 2001). Beryllium-10 concentrations in alluvial sediment are particularly useful indicators of landscape-scale denudation because alluvial sediment is generally thought to integrate catchment-wide denudation signals (e.g., Bierman and Steig, 1996; Granger et al., 1996; Schaller et al., 2001; von Blanckenburg, 2005; Schaller and Ehlers, 2006). Spatial variations in catchment-wide denudation can be related to the proximity of tectonically active structures (Wobus et al., 2005). Furthermore, catchment-wide denudation rates derived from cosmogenic nuclides can be compared to long-term rock exhumation rates derived from thermochronometers to evaluate temporal variations in denudation (e.g., Kirchner et al., 2001; Matmon et al., 2003; Vance et al., 2003; Wittmann et al., 2007; Cyr and Granger, 2008), providing insights into mountain belt evolution. However, because these methods average denudation over different time periods, they have differing sensitivities to geomorphic processes; the degree to which these factors affect direct comparisons between geochronometers has not been fully explored.
We investigated spatial and temporal variations in denudation of the north-central Wasatch Mountains, an area with a high density and quality of existing geochronologic data. Our objectives were twofold: (1) to evaluate spatial variations in catchment-wide denudation rates derived from cosmogenic 10Be concentrations in alluvial sediment, specifically testing whether erosion rates correlate with proximity to or distance along the range front fault, and (2) to evaluate temporal variations in denudation of the Wasatch Mountains by comparing our 10Be-based catchment-wide denudation rates with previously published long-term exhumation rates derived from thermochronometers and with shorter-term vertical fault displacement rates.
GEOLOGIC SETTING OF THE NORTH-CENTRAL WASATCH MOUNTAINS
The Wasatch Mountains, reaching >3.4 km in altitude, have been uplifted in the footwall of the west-dipping Wasatch fault zone (Fig. 1), one of the most extensively studied segmented normal faults in the world. The Wasatch fault zone is an active normal fault system composed of at least six segments; the Salt Lake City and Weber segments bound our study area on the west (Fig. 1). Deformation initiated on the Wasatch fault in the past 12–17 m.y. and continues today (Friedrich et al., 2003; Machette et al., 1992). The total magnitude of exhumation in the central Wasatch adjacent to the Salt Lake City segment is estimated to be ~11 km (Ehlers et al., 2003; Parry and Bruhn, 1987). Rock uplift along the Wasatch fault occurs as tilt about a footwall hinge located ~20–25 km east of the fault (Ehlers et al., 2003). Rocks in the footwall of the Salt Lake City segment are composed primarily of granitic rocks from the Oligocene-age Little Cottonwood Stock, with lesser amounts of quartzites, shales, and siltstones of the middle Proterozoic–age Big Cottonwood Formation (Bryant, 1990; Ehlers and Chan, 1999). Rocks in the footwall of the Weber segment are mainly gneisses and granitic rocks of the middle Precambrian Farmington Canyon Complex (Davis, 1983). Quaternary fluvial, colluvial, glacial, and lacustrine deposits are present in both areas; along the Weber segment, glacial material is found only east of the range crest (Davis, 1983).
Previous studies have used ZFT, AFT, and AHe thermochronometry to study the exhumation history of the Wasatch fault footwall (Fig. 1) (Naeser et al., 1983; Kowallis et al., 1990; Ehlers et al., 2003; Armstrong et al., 2003, 2004). Coupled two-dimensional thermokinematic and age prediction models suggest exhumation rates of <0.2–0.4 mm/yr over the past 5 m.y. for most of the Wasatch front, including the Weber and northern Salt Lake City segments (Ehlers et al., 2003; Armstrong et al., 2004). These rates agree with a rate of ~0.4 mm/yr for the Weber segment determined from AFT ages (Naeser et al., 1983). The exception is the central Wasatch along the southern Salt Lake City segment, where exhumation rates are apparently more rapid, at 0.6–0.8 mm/yr (Ehlers et al., 2003; Armstrong et al., 2004). This region corresponds to the highest peak elevations and greatest topographic relief in our study area. Assuming that exhumation rates in this setting are driven in the long term by rock uplift, rates of rock uplift along the Wasatch fault appear to be similar across fault segments at <0.2–0.4 mm/yr, except for the more rapidly uplifting southern Salt Lake City segment. Spatial exhumation patterns suggest strong tectonic forcing; maximum exhumation rates occur at the range front fault and decrease linearly to 0 mm/yr at the footwall hinge (Ehlers et al., 2003).
Fault trench studies along the Salt Lake City segment suggest an average net tectonic vertical displacement rate (the relative vertical distance between the hanging wall and footwall following an earthquake) of 1.0–1.7 mm/yr over the past ~6 k.y. (Friedrich et al., 2003). Paleoseismic studies and scarp profile analyses of the Weber segment yield a vertical displacement rate of 1.2 +3.1/−0.6 mm/yr over the past ~18 k.y., where rates are highest in the center and decrease toward the northern and southern tips of the Weber segment (Nelson and Personius, 1993; Lund, 2005, and references therein). However, older offset alluvial fans suggest that the average vertical displacement rate on the Salt Lake City segment since ca. 250 ka is much slower, at 0.1–0.3 mm/yr (Machette, 1981; Machette et al., 1992), which is consistent with other estimates of late Pleistocene vertical displacement rates on Wasatch fault segments (Niemi et al., 2004) and more similar to longer-term exhumation rates.
BERYLLIUM-10–DERIVED CATCHMENT DENUDATION RATES
We determined catchment-wide denudation rates in the north-central Wasatch by measuring 10Be concentrations in alluvial sediment exiting catchments. To reduce the number of variables affecting catchment denudation rates, and thus make meaningful comparisons with longer-term rock exhumation rates and shorter-term fault displacement rates, we developed strict criteria to guide our sampling. We restricted our sampling to catchments that: (1) were of similar size; (2) had similar underlying lithologies; (3) were within areas previously sampled for thermochronometry; (4) were positioned at varying distances along and perpendicular to the Wasatch fault; and (5) were not extensively glaciated (<10% of the total catchment area) during the latest Pleistocene or Holocene (Laabs et al., 2006; Madsen and Currey, 1979; Davis, 1983). This last criterion is important because cosmogenic nuclide exposure histories from glaciated catchments likely do not satisfy steady-state denudation assumptions due to ice shielding during erosion, prolonged sediment storage, and/or fluvial reworking of glacially derived sediment (e.g., Wittmann et al., 2007).
We sampled 14 catchments that fulfilled these criteria. The catchments drain the range front along both the Weber and Salt Lake City fault segments, and also tributaries to Big Cottonwood and Little Cottonwood Canyons within the core of the central part of the range, adjacent to the Salt Lake City segment (Fig. 1). Thirteen of the sampled catchments are incised predominantly or entirely into granitic rocks of the Little Cottonwood Stock or Farmington Canyon Complex, from which most thermochronometry samples were collected (Armstrong et al., 2003, 2004); the exception, Stairs Gulch, is cut into quartzites of the Big Cottonwood Formation. Catchment relief (the difference between the minimum and maximum elevation within a catchment) ranges from 1.07 to 1.50 km (Table 1). Mean hillslope gradients range from 26° to 43°, and the steeper catchments occur in the core of the range (Table 1; Fig. 1). Based on field observations of the amount of bedrock exposed in channels, the sampled catchments appear to be either primarily detachment (weathering)–limited or intermediate between detachment-limited and transport-limited (cf. Howard et al., 1994). Annual precipitation in the study area ranges from ~500 mm/yr near the range front to ~1400 mm/yr near the range crest (Western Region Regional Climate Center, www.wrcc.dri.edu).
We collected alluvial sediment from the active channel just upstream of each catchment outlet. For the catchments draining the range front, we sampled upstream of the highest Lake Bonneville shoreline to preclude reworking of beach sediments. We processed the 2–4 mm grain size fraction for the Salt Lake City segment catchments and the 0.25–0.5 mm grain size fraction for the Weber segment catchments. Different hillslope denudation processes acting over varying time scales can contribute different sediment grain sizes to fluvial systems. Recognizing that grain size can potentially affect 10Be concentration (e.g., Belmont et al., 2007; Matmon et al., 2003; Brown et al., 1998), we processed both grain size fractions from one catchment (Rocky Mouth Creek) as a basis for comparison. As shown in Table 2, 10Be concentrations for the two size fractions from this catchment are identical within analytical uncertainty, increasing our confidence that catchment-wide denudation rates are comparable between the Salt Lake City and Weber segment catchments.
We determined 10Be production rates for each sampled catchment using a 10-m-resolution digital elevation model (DEM) of the study area and calculating local 10Be production rates, scaled for latitude, altitude, and topographic shielding, for each DEM pixel. Since the north-central Wasatch receives abundant winter snow that is highly dependent on elevation, we calculated snow shielding factors (e.g., Gosse and Phillips, 2001) for each DEM pixel using local monthly precipitation-elevation and temperature-elevation regressions (Laabs et al., 2006). Snow shielding proved to be significant for catchments in the core of the range, with snow shielding factors as low as 0.88 (Table 2). Although the climate data used to calculate snow shielding only span ≤50 yr, we do not expect large variations in snowfall for the Holocene time scales over which our denudation rates are integrated. Multiplication of production rates by snow shielding factors yielded the final catchment-wide 10Be production rates (Table 2). Our calculations of catchment-wide denudation rates assume steady, uniform erosion from all DEM pixels for which 10Be production rates were calculated. Although we did not specifically test this assumption, nonglaciated catchments with similar metrics elsewhere in the Basin and Range Province have been shown to be eroding uniformly (Stock et al., 2006).
RESULTS AND DISCUSSION
Spatial Variations in Denudation Rate
Beryllium-10–based denudation rates for the sampled catchments vary by roughly an order of magnitude, ranging from 0.07 to 0.79 mm/yr (Table 2). Spatial variation in denudation along the range front, spanning both the Salt Lake City and Weber fault segments, is relatively small; rates range from 0.07 to 0.17 mm/yr (Fig. 2). However, denudation rates from catchments within the core of the range along the Salt Lake City segment are much more variable, ranging from 0.17 to 0.79 mm/yr (Fig. 2).
Given that rock uplift rates and thermochronometer-derived exhumation rates are at maximum values adjacent to the range front fault (Armstrong et al., 2003; Ehlers et al., 2003), catchments draining the range front might be expected to display faster 10Be-based catchment-wide denudation rates than those in the core of the range. On the contrary, we find that catchments along the range front display the slowest denudation rates (Fig. 2); catchments draining the range front (n = 10) yield a mean denudation rate of 0.10 mm/yr, while catchments in the core of the range (n = 4) yield a mean denudation rate of 0.47 mm/yr (Tables 1 and 2; Fig. 2). There is only weak correlation of denudation rates with along-strike variations in fault slip rate, despite previous work suggesting that tectonic activity controls landscape development in the Wasatch footwall (Frankel and Pazzaglia, 2005). As denudation rates are not strongly correlated with proximity to, or distance along, the range front fault, localized tectonic forcing does not appear to exert strong control on the spatial distribution of catchment-wide denudation rates in this setting, at least on the Holocene time scales over which these measurements are made. Although tectonic forcing probably drives landscape-scale denudation of the Wasatch Mountains, localized geomorphic processes and events, acting in response to this forcing, appear to exert stronger first-order control on the spatial distribution of catchment-wide denudation rates over shorter time scales.
The geomorphic processes that accomplish catchment denudation are often associated with topographic metrics such as catchment area, relief, elevation, and hillslope gradient (e.g., Ahnert, 1970; Montgomery and Brandon, 2002; Roering et al., 2007). In our study area, catchment-wide denudation rates are poorly correlated with catchment relief (Fig. 3C). Denudation rates are moderately correlated with mean catchment elevation (Fig. 3B), suggesting that elevation-dependent processes (e.g., frost cracking; Hales and Roering, 2007) may be important in driving denudation in these primarily detachment-limited catchments. Catchment-wide denudation rates are also moderately correlated with mean hillslope gradient, though the correlation is valid only at lower hillslope gradients and denudation rates; for hillslope gradients >30°, denudation rates show considerable scatter (Fig. 3D). The nonlinear relationship between mean hillslope gradient and denudation rate has been observed in other data sets (e.g., Roering et al., 2001; Montgomery and Brandon, 2002; Binnie et al., 2007) and has been attributed to hillslopes reaching a threshold condition, above which mass-wasting processes dominate hillslope processes (Burbank et al., 1996). We conclude that the relatively large spatial variation in catchment-wide denudation rates in the core of the range is likely due to discrete landslide events occurring on hillslopes at threshold gradients. Landslide events are sediment point sources, so they violate assumptions of steady, uniform erosion implicit in the catchment-wide denudation rate calculations from 10Be concentrations; thus, landslides can skew short-term catchment-wide denudation rates (Niemi et al., 2005; Belmont et al., 2007). Denudation rates may be faster in catchments with smaller contributing areas (Fig. 3A) because the effects of landslides on 10Be concentrations in alluvium are proportionally greater in smaller catchments. Although the extent of landslide activity in our study area has not been quantified, we did observe isolated fresh talus deposits in those catchments draining to Little Cottonwood Canyon (Fig. 1).
Cosmogenic nuclide concentrations are generally not sensitive to glacial-interglacial fluctuations except at high (>0.5 mm/yr) denudation rates (Schaller and Ehlers, 2006), but they can record landscape adjustments to glacial erosion. Much of the core of the north-central Wasatch, including Little Cottonwood Canyon, contained large trunk glaciers during the Last Glacial Maximum (Madsen and Currey, 1979; Laabs et al., 2006). Glacial erosion in the major canyons likely deepened these canyons, inciting accelerated denudation in tributary catchments. Prominent (>10 m) knick-points are not apparent in the DEM-derived longitudinal channel profiles of tributary catchments to Little Cottonwood Canyon, but field observations indicate numerous smaller (<10 m) knickpoints, which may be significant at the scale of these catchments. Thus, the discrepancy between faster exhumation rates at the range front over long time scales, as recorded by thermochronometers, and faster denudation rates in the core of the range over shorter time scales, as recorded by cosmogenic 10Be, may relate to the lingering effects of late Pleistocene glaciations in this part of our study area. We posit that landsliding in those sampled catchments draining to Little Cottonwood Canyon (Lisa Falls, Tanner Gulch, and Coalpit Gulch) may be accelerated due to late Pleistocene glacial valley deepening.
Temporal Variations in Denudation Rate
In primarily nonglacial catchments like those sampled, channel incision sets the pace at which hillslopes are eroded, which in turn drives overall denudation of landscapes (i.e., rock exhumation). In a landscape experiencing steady denudation over long time periods (e.g., Willett and Brandon, 2002), shorter-term 10Be-based catchment-wide denudation rates should be similar to longer-term thermochronometer-derived exhumation rates. If denudation is driven primarily by rock uplift and/or base-level fall, and rock uplift and denudation are in equilibrium, catchment-wide denudation rates should also be similar to vertical fault displacement rates.
This situation appears to be borne out by our data from the north-central Wasatch. The mean denudation rate for all sampled catchments is 0.2 ± 0.2 mm/yr, time-averaged over ~5 k.y. (Table 2). This rate is consistent with exhumation rates of <0.2–0.4 mm/yr over the past 5 m.y. for most of the Wasatch front, including the Weber and northern Salt Lake City segments (Ehlers et al., 2003; Armstrong et al., 2004; Naeser et al., 1983). It is also consistent with exhumation estimates derived from fluid inclusions (<0.5 mm/yr since ca. 5 Ma; Parry and Bruhn, 1987). The mean catchment-wide denudation rate is fully consistent with late Pleistocene vertical displacement rates on the Salt Lake segment (0.1–0.3 mm/yr) and other segments (0.2–0.4 mm/yr) of the Wasatch fault (Machette et al., 1992; Mattson and Bruhn, 2001) (Fig. 2). Only at the shortest time scales is there a substantial difference between catchment-wide denudation rates and vertical displacement rates; the mean catchment erosion rate of 0.2 mm/yr is considerably slower than Holocene vertical displacement rates on the Salt Lake (1.7 mm/yr over the past ~6 k.y.) and Weber fault segments (~1.2 mm/yr since 7.9 ka; Friedrich et al., 2003; McCalpin and Nishenko, 1996). However, these displacement rates are considered to be anomalous because of clustered seismic strain release during this relatively short time interval (Chang and Smith, 2002; Friedrich et al., 2003; Niemi et al., 2004). The fact that 10Be-based catchment-wide denudation rates are roughly an order of magnitude slower than Holocene vertical fault displacement rates, and are instead similar to longer-term (~250 k.y.) displacement rates, suggests that the north-central Wasatch landscape is apparently not sensitive to short-term (103 yr) perturbations. Instead, the Wasatch Mountains are influenced by longer-term tectonic signals (cf. Whipple, 2001) and landscape response times that are greater than the recurrence interval of individual seismic events or earthquake clusters.
The similarity of denudation rates over long time scales is striking considering that there are fundamental differences in the ways in which the various geochronometers record denudation. For example, thermochronometer-derived denudation rates include uncertainties in the kinetics of He diffusion in apatite (e.g., Farley, 2000) and potentially oversimplified assumptions regarding the thermal field through which bedrock samples cool (e.g., Ehlers, 2005; Braun, 2005). The 10Be-based denudation rates include uncertainties in 10Be production rates (e.g., Gosse and Phillips, 2001) and potentially oversimplified assumptions regarding steady, uniform erosion of catchments. Averaged over long time scales, thermochronometer-based denudation rates are largely insensitive to discrete tectonic (e.g., individual earthquake), climatic (e.g., glacial-interglacial cycles), and geomorphic (e.g., individual landslide) events. In contrast, 10Be-based denudation rates may be quite sensitive to these events.
In landscapes that denude predominantly during stochastic events such as landslides, individual 10Be-based catchment-wide denudation rate measurements are more likely to be skewed away from the longer-term denudation rate as measured by thermochronometers. Thus, even in cases of erosional steady state, 10Be-based denudation rates will probably show greater scatter about the mean long-term rate. In this case, the mean catchment-wide denudation rate for our study area is generally consistent with results from other geochronometers. Direct comparisons of 10Be-based catchment-wide denudation rates with longer-term thermochronometer-based rates at the range front suggest that denudation of the Wasatch Mountains has been roughly steady, perhaps slightly decreasing along the southern Salt Lake City segment, over the past 5 m.y. Ultimately, when comparing denudation rates derived from different geochronometers, it is important to consider the underlying assumptions of the various methods, the sensitivity of those methods to individual geomorphic events, the amount of time over which denudation rates are averaged, and the number of samples necessary to adequately capture signals of mountain denudation.
Cosmogenic 10Be concentrations in alluvial sediment from the north-central Wasatch Mountains indicate catchment-wide denudation rates of 0.07–0.79 mm/yr. Denudation rates along the range front are relatively uniform, between 0.07 and 0.17 mm/yr (mean 0.10 mm/yr), while steeper catchments within the core of the range show greater variation, between 0.17 and 0.79 mm/yr, and overall faster rates (mean 0.47 mm/yr). Unlike thermochronometer-derived exhumation rates, 10Be-based catchment-wide denudation rates do not display a spatial distribution strongly influenced by localized tectonic forcing (i.e., faster denudation rates adjacent to the range front fault). Faster and more variable denudation rates in the core of the range likely relate to higher hillslope gradients that are at (or near) threshold values for landsliding. We suggest that denudation rates are higher in the core of the range because landscapes there are still adjusting to changes incurred by late Pleistocene glaciations.
The mean denudation rate for all sampled catchments (0.2 mm/yr) is generally consistent with both longer-term exhumation rates and shorter-term vertical fault displacement rates, suggesting that denudation of the north-central Wasatch has been roughly steady, or decreasing slightly, over the past 5 m.y. Because thermochronometers average denudation over long time scales, they generally record rock exhumation without regard to specific geomorphic processes, and therefore they provide relatively stable estimates of long-term erosion. In contrast, cosmogenic nuclide concentrations in alluvial sediment average denudation over shorter time scales and can be highly sensitive to localized geomorphic processes and events. We conclude that although 10Be-based catchment-wide denudation rates are sensitive to these events, taken as a whole, they appear to reflect the large-scale tectonic forces that have driven denudation of the north-central Wasatch Mountains over the past 5 m.y.
We thank Dave Whipp and Matt Densmore for field assistance, Erin Bachynski and Alicia Nobles for analytical assistance, and Frank Pazzaglia and James Dolan for discussions and support. Greg Balco wrote the Matlab code used to calculate catchment-averaged 10Be production rates. Beryllium-10 measurements were made at PRIME Lab and Lawrence Livermore National Laboratory (LLNL). We thank Arjun Heimsath, Darryl Granger, and editor Jon Pelletier for helpful comments that improved the paper. This work was supported by National Science Foundation (NSF) grant EAR-0544954 (Ehlers and Stock), an LLNL-UEPP Fellowship (Frankel), and grants from the Georgia Tech Research Foundation and the Colorado Scientific Society (Frankel).