Bedrock rivers are the hard points of mountain landscapes. Where a river cuts directly into rock, the pace of erosion sets the tempo for the entire drainage network, controlling how quickly mountains decay and how climate and tectonic signals propagate downstream. Sediment plays a double role in this process: a blanket of gravel can shield the bed from erosion, while individual grains act as tools that abrade and batter the rock where it is exposed. Yet in deep, fast canyons the bed is invisible at most flows, so almost nothing has been known about how much sediment these systems actually store, or where and when they store it.

The Fraser River offered an ideal natural laboratory. Between Soda Creek and Yale, the river runs through roughly 375 kilometres of bedrock-influenced canyon, with banks that are about 26 percent bedrock-bound on both sides and another 29 percent constrained by rock on one side. The team surveyed nine canyons chosen to span the full range of width, depth, slope and velocity found along the entire canyon, using a Norbit iWBMS echosounder operated at 400 kilohertz, paired with an inertial motion sensor and high-precision GNSS positioning that delivered vertical accuracy of roughly 15 millimetres. The resulting digital elevation models, built at half-metre resolution, extended bank to bank in most reaches.

The timing of the study proved extraordinary. The first survey, completed on 10 to 12 November 2021, captured the riverbed just days before an extreme atmospheric river struck southwestern British Columbia on 15 November. That storm triggered flooding and a cascade of landslides and debris flows that delivered enormous volumes of coarse sediment into the canyon during an otherwise low-flow period. Three subsequent surveys — in April 2022, August and September 2022, and August 2023 — bracketed the 2022 snowmelt freshet and a full annual cycle, allowing the researchers to watch the aftermath of the sediment pulse unfold canyon by canyon.

By differencing successive bathymetric maps, the team identified patches of erosion and deposition and integrated the elevation changes over each patch area to compute volume changes. The statistics are striking: the median elevation change per patch was only about 0.7 metres, but individual patches ranged up to roughly 85,900 square metres in area, and patch volumes spanned from about 4 to more than 325,000 cubic metres. Crucially, the pattern of change depended on channel morphology. Constriction-pool-widenings — reaches where a narrow throat forces plunging flow that scours a deep pool before the channel opens out — tended to show large, channel-spanning patches of coherent change. Rapids, with their shallow, chaotic flow, displayed a patchwork of alternating erosion and deposition both along and across the channel. The single surveyed overfall, Hells Gate Canyon, showed almost no change at all, likely because fast shallow flow upstream and an impinging jet downstream prevent sediment from settling anywhere.

Perhaps the most surprising result is that the footprint of sediment storage barely changed even as the volume of storage swung dramatically. The total mapped patch area remained nearly constant between the 2022 freshet and the following annual cycle, yet the net volume of sediment within those patches inverted: canyons that lost sediment during the freshet gained it the next year, and vice versa. Nearly all of the change was accommodated by vertical thickening and thinning of alluvial deposits rather than by expansion of bedrock exposure. This means substantial storage changes can occur with no meaningful change in how much of the bed is covered — a direct challenge to cover models that treat sediment cover as a simple function of storage volume relative to bedrock roughness.

The researchers describe the routing behaviour they observed as sediment staging. Following the 2021 supply event, sediment was first deposited near its entry points, then released and passed downstream as the initial deposits eroded. At Black Canyon, a depositional patch that formed during the 2022 freshet switched to erosion the following year, while Alexandra Canyon downstream did the opposite, eroding first and then aggrading as it received sediment from upstream. Along the Yale Rapids to Lady Franklin Rock reach, three wide sections — including the aptly named Deadmans Eddy — deposited sediment in the winter after the storm, eroded during the freshet, and then began aggrading again, with the erosional signal propagating downstream from one widening to the next like a slow-moving wave.

The net flux divergence calculations quantify this relay. Between November 2021 and April 2022, the surveyed reaches accumulated roughly 509,000 cubic metres of sediment. During the 2022 freshet the system flipped, eroding approximately 764,000 cubic metres over four months, before returning to net deposition of about 148,000 cubic metres over the following year. The greatest changes in flux divergence occurred in the same reaches each time, confirming that morphology fixes the locations of storage while supply events dictate whether those locations fill or empty. Unlike sand-bed rivers where sediment waves migrate as dunes and bars, canyon sediment appears to move as discrete pulses that transfer abruptly from one storage site to the next, dispersing in fast shallow rapids before reconcentrating in deep pools and wide sections.

The implications reach well beyond the Fraser. Sediment cover in bedrock rivers modulates incision rates, and incision rates in turn govern the long-term evolution of mountain river networks and the delivery of sediment to lowland environments. The observation that lateral roughness — variation in channel width — matters more than vertical bedrock roughness in controlling storage suggests that models of bedrock erosion need to incorporate realistic width variability, which can locally vary by a factor of two or more. It also suggests that the timing of hillslope inputs relative to peak flow, rather than the annual hydrograph alone, determines whether an erosive or depositional signal propagates downstream. For a river system increasingly battered by atmospheric rivers and landslides, knowing where canyons stash their sediment — and when they let it go — may prove essential for predicting everything from salmon habitat dynamics to flood hazards in the decades ahead.

Subject of Research: Sediment storage and routing dynamics in bedrock river canyons of the Fraser River, British Columbia

Article Title: Sediment storage and routing in bedrock canyons

Article References: Ross, C. B. A., Carr, J. C., Larimer, J. E., Hurson, M., Sklar, L. S., Wright, M., Viner, N., & Venditti, J. G. (2026). Sediment storage and routing in bedrock canyons. Earth Surface Dynamics, 14(4), 553-573. https://doi.org/10.5194/esurf-14-553-2026

Image Credits: AI Generated

DOI: 10.5194/esurf-14-553-2026

Keywords: bedrock rivers, sediment storage, Fraser Canyon, multibeam bathymetry, atmospheric river, channel morphology, sediment routing, river incision, geomorphology, debris flows, constriction-pool-widening, British Columbia