Experimental evidence for fl uvial bedrock incision by suspended and bedload sediment Joel S. Scheingross1*, Fanny Brun1,2, Daniel Y. Lo1, Khadijah Omerdin1, and Michael P. Lamb1 1Division of Geological and Planetary Sciences, California Institute of Technology, Pasadena, California 91125, USA 2Geosciences Department, Ecole Normale Supérieure, 24 rue Lhomond, 75005 Paris, France ABSTRACT regime, the saltation-abrasion model was re- Fluvial bedrock incision sets the pace of landscape evolution and can be dominated by cast (by Lamb et al., 2008) in terms of near-bed abrasion from impacting particles. Existing bedrock incision models diverge on the ability of sediment concentration rather than particle hop sediment to erode within the suspension regime, leading to competing predictions of lowland lengths (herein referred to as the total-load mod- river erosion rates, knickpoint formation and evolution, and the transient response of orogens el). The saltation-abrasion and total-load mod- to external forcing. We present controlled abrasion mill experiments designed to test fl uvial els produce similar results for erosion within incision models in the bedload and suspension regimes by varying sediment size while holding the bedload regime, but within the suspension fi xed hydraulics, sediment load, and substrate strength. Measurable erosion occurred within regime the total-load model predicts nonzero the suspension regime, and erosion rates agree with a mechanistic incision theory for erosion erosion rates that increase with increasing fl uid by mixed suspended and bedload sediment. Our experimental results indicate that suspen- bed stress, leading to contrasting predictions sion-regime erosion can dominate channel incision during large fl oods and in steep channels, for landscape evolution, especially during large with signifi cant implications for the pace of landscape evolution. fl oods and in steep channels where bed sedi- ment is suspended. INTRODUCTION change (Crosby et al., 2007; Gasparini et al., Laboratory experiments offer a means to test River incision into bedrock controls the fl ux 2007), the preservation of relief in tectonically the validity of existing bedrock-erosion theories of sediment to basins, links hillslopes to chan- inactive mountain ranges over much longer under controlled conditions that are otherwise nels, and dictates the rate at which landscapes time scales than with stream-power modeling diffi cult to achieve in natural rivers. Previous evolve (e.g., Whipple et al., 2013). Bedrock (Egholm et al., 2013), and the formation of land- experimental work suggests that channel-bed incision theory allows predictions of fl uvial re- forms that do not arise in stream-power model- erosion in the suspension regime is possible sponse to external perturbations, and the most ing, such as permanent fl uvial hanging valleys (Sklar and Dietrich, 2001; Cornell, 2007; Chat- commonly used models assume that erosion is (Crosby et al., 2007) and static knickpoints that anantavet et al., 2010), but experiments have not proportional to stream power or bed shear stress can grow infi nitely in height (Sklar and Dietrich, been conducted that allow full testing of existing (e.g., Howard and Kerby, 1983). Such models 2008). In addition, in sand- and silt-bedded riv- models within the suspension regime. Herein have been widely used in landscape evolution ers and deltas where the majority of bed sedi- we present results from controlled abrasion mill modeling (e.g., Tucker and Slingerland, 1994), ment is transported in suspension during fl oods, experiments and fi nd signifi cant rates of erosion as well as in studies examining feedbacks be- the saltation-abrasion model predicts zero ero- within the suspension regime, in agreement with tween climate, tectonics, and topography (e.g., sion, counter to stream-power predictions and the total-load erosion model; these results have Willett, 1999). However, stream-power models fi eld observations of fl uvial incision into con- important implications for landscape evolution. do not explicitly capture the physical processes solidated sediment (Nittrouer et al., 2011; Shaw of river erosion (i.e., the coupling of fl uid fl ow, et al., 2013). EXPERIMENTAL SETUP sediment transport, and channel erosion), limit- Differences between the saltation-abrasion In natural river channels, erosion rates are ing their predictive ability. and stream-power models arise, in part, because likely infl uenced by multiple sediment sizes in An alternative approach is to more directly the saltation-abrasion model assumes an infi nite transport, complex bed topography, and jointed account for processes eroding rock. The salta- hop length for particles transported within the rock that may promote plucking (e.g., Hancock tion-abrasion model (Sklar and Dietrich, 2004) suspension regime, such that particles are as- et al., 1998). Our goal is not to reproduce this predicts river-bed abrasion from single-sized sumed not to impact the bed and erosion rates complexity, but rather to test the competing sediment transported in bedload over a planar are predicted to be zero (Sklar and Dietrich, predictions of the saltation-abrasion and total- bed, and several of its basic tenets have been 2004, 2006). The transition from the bedload re- load erosion models under the simplest possible confi rmed in laboratory and fi eld settings (e.g., gime to the suspension regime is often defi ned as scenarios and in accordance with inherent as- Sklar and Dietrich, 2001; Johnson and Whipple, the point in which bed shear velocity, u (a fl uid sumptions in the models, including single-sized * 2010). This has led the model, and other similar turbulence proxy), surpasses particle terminal sediment, and a planar river bed of massive, models (e.g., Turowski et al., 2007), to be wide- settling velocity, ws (Bagnold, 1966; McLean, unjointed rock. Testing existing models under ly adopted in predicting reach-scale erosion 1992), such that turbulence strongly infl uences these simplifi ed conditions is important because (e.g., Cook et al., 2012), river-profi le evolution particle trajectories. In the suspension regime, such baseline tests have yet to be performed, (e.g., Crosby et al., 2007), and landscape evolu- some particles are advected high into the water and the existing theories are widely applied to tion (e.g., Egholm et al., 2013). The saltation- column by turbulence (i.e., the suspended load); natural landscapes and used in landscape evolu- abrasion model differs from the stream-power however, the largest concentration of particles is tion simulations despite these assumptions (e.g., model in important and sometimes counterin- still near the bed (Rouse, 1937) where particles Cook et al., 2012; Egholm et al., 2013). tuitive ways. For example, the saltation-abra- impact the bed via rolling, sliding, and saltation To explore bedrock erosion rates over a wide sion model predicts decreased erosion rates for (i.e., bedload), and there is active exchange of range of transport conditions, we conducted heightened bed shear stresses, leading to slower particles between the bedload layer and sus- experiments in abrasion mills (Fig. 1) identi- transient river network response to base-level pended load above (e.g., McLean, 1992; Garcia cal to those used by Sklar and Dietrich (2001) and Parker, 1993). To account for erosion due in their study of erosion rates in the bedload *E-mail: [email protected]. to particle-bed impacts within the suspension regime. In abrasion mills, suspension of sedi- GEOLOGY, June 2014; v. 42; no. 6; p. 1–4; Data Repository item 2014185 | doi:10.1130/G35432.1 | Published online XX Month 2014 GEOLOGY© 2014 Geological | June Society 2014 | ofwww.gsapubs.org America. For permission to copy, contact Copyright Permissions, GSA, or [email protected]. 1 the suspension regime, because of smaller par- A D = 0.46 mm Motor ticle mass and fall velocity. Erosion rates should 10-1 D = 0.75 mm also approach zero with decreasing grain size as D = 1.2 mm impacts become viscously damped for particle D = 2.0 mm Stokes numbers (St, a nondimensional number D = 2.4 mm that weights the kinetic energy of particle im- 17 cm pacts to the fl uid viscosity) below ~10–100 (Jo- seph et al., 2001). -2 To achieve measurable erosion rates, we used Height above bed (m) 10 low-tensile-strength (σ = 0.32 MPa) polyure- T 0 20 40 60 80 100 120 thane foam as a highly erodible bedrock simu- 49 cm Sediment concentration (g/L) lant rather than natural rock. Tests show that foam follows the same erosion-rate scaling re- Figure 2. Rouse sediment concentration profi les (dashed and solid lines) for different lationship with tensile strength as observed by Suspended grain diameters (D) with β = 2 (β is a dimen- sediment Sklar and Dietrich (2001) for rock and concrete sionless constant weighting the diffusivities sampling tubes (see the Data Repository, and Fig. DR1 therein), of sediment relative to fl uid momentum), for 5.5 cm allowing our results to be properly scaled to total sediment load of 70 g. Symbols corre- natural rock. spond to mean of sediment concentration measurements (n = 3); x- and y-error bars Foam Disc For each experiment, we secured a 38-mm- represent geometric standard deviation of 20 cm thick foam disc to the base of the abrasion mill, measurements and radius of sampling tub- loaded the mill with siliciclastic, well-sorted, ing (3 mm), respectively. B D = 1.2 mm subangular to subrounded sediment, and fi lled u* / ws = 1.3 the mill to a depth of 49 cm with water. A pro- peller induced fl ow and sediment transport, and following Lamb et al., 2008), β is a dimen- experiments were run long enough for measur- sionless constant weighting the diffusivities of able wear of the foam disc by either volume loss sediment relative to fl uid momentum, and κ = 3 cm (using a submillimeter-precision laser scanner) 0.41 is von Karman’s constant.