A detailed survey of the Fraser River in Canada challenges preconceptions about how river water flows. The results call for a re-evaluation of how waterways carve through bedrock to form canyons. See Letter p.534
Anyone who has ever hiked beside, fished in or floated down a bedrock river knows the beauty of the canyons that they create. But how exactly does a river incise through bedrock and cut a deep canyon? Venditti et al.1 explore this question on page 534 of this issue, with their report of an extensive set of data on river-flow velocity through the canyons of the Fraser River in western Canada (Fig. 1). Their observations challenge standard assumptions about water-flow patterns and suggest a mechanism for how bedrock canyons evolve in space and time.
Although researchers have long hypothesized2 about the processes that control bedrock incision, it was only in the past 20 years that bedrock rivers gained prominence in the scientific literature. A big reason for this is practical. Scientists and engineers tend to divide rivers into two types: bedrock and alluvial. Bedrock rivers cut directly into bedrock, and are often found in actively uplifting mountainous terrain and sparsely inhabited areas. By contrast, alluvial rivers are usually found in less mountainous — and therefore more habitable — terrain, where people build infrastructure that they want to protect. Understandably, extensive literature exists on the morphology and behaviour of alluvial rivers.
It is tempting to try to fit the behaviour of bedrock rivers into the mould of alluvial rivers, but the two differ in many respects. Venditti and colleagues highlight one example of this: flow patterns in narrow bedrock canyons can act both to locally increase incision rates into bedrock and to maintain nearly vertical bedrock walls through canyons.
To understand Venditti and colleagues' results, we must step back and review some fundamental ideas about flow velocity in rivers. Any student of fluid dynamics learns about the 'law of the wall', which states that the average velocity of turbulent flows increases logarithmically with distance away from a wall — or, for rivers, from the bed of the river channel3 (Fig. 2). This 'law' is ingrained in how scientists and engineers think about, sample and manage rivers. For example, the US Geological Survey has developed widely used methods for measuring the average velocity of rivers on the basis of the 'typical' logarithmic velocity profile4. But not all velocity profiles are logarithmic. Venditti and colleagues' data illustrate important, previously unobserved aspects of velocity patterns in bedrock canyons that could influence how such canyons are built and maintained.
The authors used a boat-mounted acoustic Doppler current profiler (ADCP) to survey extensive reaches of the Fraser Canyon, which consists of 42 individual bedrock canyons. An ADCP uses the Doppler shift in multiple sound-wave beams to measure flow velocities from the water surface to the near-bottom. Originally deployed by oceanographers, this tool is now becoming more widely used by scientists studying rivers. In contrast to standard flow meters, which make velocity measurements one point at a time, ADCPs provide data on the entire velocity profile — not only saving time, but also providing a more detailed view of cross-sectional and longitudinal flow patterns.
Venditti et al. observe that a core of high-velocity flow is diverted downward from near the water surface towards the bed of the channel at places where the river enters a narrow bedrock canyon (Fig. 2). This high-velocity core follows the channel bed downstream into a plunge pool (a locally deep part of the bed), where it then dissipates. Crucially, the shear stress (force per unit area) that the flow exerts on the bed is proportional to the square of the vertical gradient in velocity. A high-velocity core near the bed of the channel leads to locally higher shear stress — much higher than would be expected from a logarithmic profile — and suggests that bedrock-incision rates will be relatively high at this point.
Where the canyon deepens downstream, the high-velocity core dissipates, so that shear stress is lower and incision rates are also probably lower. This downstream change in incision rates could lead to increased erosion upstream of the plunge pool, or progressive deepening of the canyon upstream. Furthermore, Venditti and colleagues observed secondary cross-channel circulation patterns where the fast surface flow plunges downward: the downward flow in the centreline causes upward flow along the bedrock banks. Such upward flow might undercut the banks, leading to their collapse and maintaining nearly vertical bedrock-canyon walls. The researchers suggest that their findings are relevant beyond their study area, because similar bedrock-canyon shapes are observed in other systems, and experimental data5 exhibit similar flow structures to those in the Fraser Canyon.
So what does this mean for our understanding of the long-term evolution of bedrock-river valleys? A convenient way to study the evolution of bedrock rivers and the mountainous topography that they shape is to use computational models. These models must calculate the evolution of topography over million-year timescales, and are limited to spatial resolutions of the order of 100 to 1,000 metres by currently available computational capabilities. So not only is the resolution of models wider than the width of a typical bedrock canyon, it is also longer than the flow patterns observed by Venditti and co-workers (the patterns occur over channel lengths of the order of 10–100 m). Computational limitations therefore mean that the spatial (and temporal) scale of the authors' observations are too small to be incorporated into current models that calculate incision over the timescales associated with bedrock-canyon formation. Nevertheless, this glimpse into flow through bedrock canyons will spur many ideas about how the processes of bedrock incision are affected by often-overlooked water-flow patterns, and will certainly be relevant for models of water flow that operate on much shorter timescales, such as days or months.
Venditti, J. G. et al. Nature 513, 534–537 (2014).
Gilbert, G. K. Geology of the Henry Mountains, Utah (US Gov. Printing Office, 1877).
Von Kármán, T. Nachrichten von der Gesellschaft der Wissenschaften zu Göttingen, Mathematisch-Physikalische Klasse 58–76 (1930).
Rantz, S. E. US Geol. Surv. Wat. Supply Pap. 2175 (1982).
MacVicar, B. J. & Rennie, C. D. Wat. Resour. Res. 48, W02503 (2012).