Canyon shape and erosion dynamics governed by channel-hillslope feedbacks

Abstract

Geologists frequently debate the origin of iconic river canyons, as well as the extent to which they record climatic and tectonic signals. Fluvial and hillslope processes work in concert to control canyon evolution; rivers both set the boundary conditions for adjoining hillslopes and respond to delivery of hillslope-derived sediment. But, what happens when canyon walls deliver boulders that are too large for a river to carry? River canyons commonly host large blocks of rock derived from resistant hillslope lithologies. Blocks have recently been shown to control the shapes of hillslopes and channels by inhibiting sediment transport and bedrock erosion. Here, we developed the first process-based model for canyon evolution that incorporates the roles of blocks in both hillslope and channel processes. Our model reveals that two-way negative channel-hillslope feedbacks driven by block delivery to the river result in characteristic plan-view and cross-sectional river canyon forms. Internal negative feedbacks strongly reduce the rate at which erosional signals pass through landscapes, leading to persistent local unsteadiness even under steady tectonic and climatic forcing. Surprisingly, while the presence of blocks in the channel initially slows incision rates, the subsequent removal of blocks from the oversteepened channel substantially increases incision rates. This interplay between channel and hillslope dynamics results in highly variable long-term erosion rates. These autogenic channel-hillslope dynamics can mask external signals, such as changes in rock uplift rate, complicating the interpretation of landscape morphology and erosion histories. INTRODUCTION River canyons, steep-walled valleys often developed in bedrock, evolve through a combination of deepening by river incision and widening by hillslope processes. Considerable effort has been expended on establishing the timing and mechanisms of canyon evolution (e.g., Cook et al., 2009; Schildgen et al., 2009; Flowers and Farley, 2012), with a focus on understanding landscape response to climatic and tectonic forcing. The traditional view of canyon erosion holds that river incision, driven by tectonics, climatic perturbations, or changes in substrate erodibility, lowers the canyon bottom. Adjacent hillslopes then respond to river incision by rockfall, landsliding, and/or diffusive sediment transport (e.g., Mudd and Furbish, 2007; Gallen et al., 2011). Under the assumption that sediment delivered to the channel is mobile, patterns and time scales of river incision control hillslope form and dominate canyon evolution. The majority of prior work on canyon development has embraced this view and drawn conclusions about the timing of canyon evolution under the assumption that hillslopes respond passively to river incision. However, canyon-confined rivers do not operate in isolation from their adjacent hillslopes (Egholm et al., 2013; Attal et al., 2015; Shobe et al., 2016, 2018; Bennett et al., 2016; Golly et al., 2017; DiBiase et al., 2018). Hillslopes make up the majority of canyon plan-view area and are often the primary source of sediment to the rivers. Steep canyon walls with substantial bare bedrock exposure and sufficient fracture density commonly release large pieces of rock (with diameters of several meters) into the channel (Howard and Selby, 1994; Glade et al., 2017; DiBiase et al., 2018; Glade and Anderson, 2018). Large grain delivery to rivers can inhibit incision over large spatial and temporal scales (Shobe et al., 2016, 2018), even damming rivers for short periods of time (Korup et al., 2006; Ouimet et al., 2007; Castleton et al., 2016). We propose that slowing or cessation of river incision must then influence the hillslopes by reducing the rate of hillslope steepening. Block delivery therefore acts as a negative feedback on both river and hillslope erosion. To constrain canyon evolution rates and process dynamics, it is critical to understand the interactions between canyon-confined rivers and their adjacent hillslopes. We developed a numerical model that explicitly treats block dynamics both on hillslopes and in channels. We then examined the influence of these negative channel-hillslope feedbacks on river canyon evolution, with two guiding questions: (1) Are negative channel-hillslope feedbacks necessary and sufficient to explain the cross section and planform shapes of natural canyons? (2) How do these feedbacks affect long-term erosion dynamics in river canyons responding to base-level fall? Block delivery is likely important in any blockproducing landscape. However, as a simplified test case, we focused on river canyons in layered rock (Figs. 1 and 2) in which a resistant cap rock (e.g., sandstone) overlies softer rock (e.g., shale). In this simplified geologic setting, the cap rock acts as a line source of blocks, with block size dictated by cap-rock thickness and joint spacing. The softer, underlying layer produces soil but no blocks. Canyons developed in layered rock often exhibit key morphologic features, such as a characteristic bell-shaped planform during transient response to base-level fall (Figs. 1A and 1B), block-mantled channels, and steep hillslopes (Figs. 1C–1F). Here, we explore how block delivery feedbacks influence canyon form and evolution. CITATION: Glade. R.C., Shobe, C.M., Anderson, R.S., and Tucker, G.E., 2019, Canyon shape and erosion dynamics governed by channel-hillslope feedbacks: Geology, v. 47, p. 650–654, https:// doi .org /10 .1130 /G46219.1 *These authors contributed equally to this work. †E-mails: rachel .glade@ colorado .edu; charles .shobe@ colorado .edu. Manuscript received 13 December 2018 Revised manuscript received 14 April 2019 Manuscript accepted 15 April 2019 https://doi.org/10.1130/G46219.1 © 2019 The Authors. Gold Open Access: This paper is published under the terms of the CC-BY license. Published online 2 May 2019 Downloaded from https://pubs.geoscienceworld.org/gsa/geology/article-pdf/47/7/650/4775270/650.pdf by guest on 06 July 2019 Geological Society of America | GEOLOGY | Volume 47 | Number 7 | www.gsapubs.org 651 CONCEPTUAL MODEL Several processes (e.g., landsliding, debris flows) may influence channel-hillslope coupling. We focused on the delivery of large blocks from hillslopes to channels (Fig. 2), which we hypothesized would control river canyon form. During the early stages of canyon evolution, channel incision causes failure of the cap rock (Ward et al., 2011), which delivers large blocks to the hillslopes and channel. The presence of blocks on the hillslopes inhibits soil erosion, stalling subsequent block release from the cap rock (Glade et al., 2017; Glade and Anderson, 2018). Blocks in the channel, if they are too large to be transported, reduce the river incision rate by armoring the bed and increasing hydraulic roughness (Shobe et al., 2016, 2018). Prolonged inhibition of river incision reduces the rate of hillslope steepening and hence the rate of block delivery to the channel. Thus, as long as hillslopes supply blocks to the channel (Fig. 2), erosion rates both in the channel and on the hillslopes are expected to be highly variable in time. Eventually, the hillslopes retreat far enough from the channel that blocks weather during hillslope transport to a size at which they no longer inhibit river incision (Fig. 2). From then on, the channel lowers at a rate that is unaffected by hillslope-derived blocks. NUMERICAL MODELING METHODS We tested the conceptual model shown in Figure 2 with a series of numerical experiments coupling models for channel (Shobe et al., 2016) and hillslope (Glade et al., 2017) evolution that incorporate the effects of large blocks of rock (see the GSA Data Repository1). The model domain, designed to represent the layered landscapes in Figures 1 and 2, consists of a horizontal resistant layer of rock overlying softer, more-erodible rock (the domain is 2 km wide × 1 km long, with 5 m resolution). A channel of uniform initial slope, forced with a constant base-level fall rate at its downstream end, incises the weaker, underlying rock. The channel permanently occupies the center of the model domain and has a constant width and discharge. The rest of the model domain operates under hillslope process laws (Glade et al., 2017; Glade and Anderson, 2018). We used this model to investigate erosion dynamics and time scales in a river canyon in which blocks released from the resistant cap rock cause interactions that govern both hillslope and channel evoluCanyonlands National Park, Utah Gorges du Tarn, France Great Escarpment, South Africa Columbia River Basalt, Oregon 0 20 30