Reading a valley, Part 2: How can you retrace a river’s footsteps?

For geoscientists, lacking direct experience is a common quandary (and it should be noted, a happy one!) One great workaround: the thought experiment. Back in 1909, the eminent geographer William Morris Davis used his powers of imagination to visualize how the Connecticut River might have carved its valley and left terraces. Here’s a beautiful sketch from his writings:


River terraces in Massachusetts, from W. M. Davis’ Geographical Essays.

Davis surmised that the Connecticut River, migrating to and fro about its valley, left terraces as it gradually cut downward.  So was his guess plausible?

Nowadays Davis is better known for his ideas on whole-landscape evolution, decades before the plate tectonics revolution. In contrast, Davis’ hypothesis for “accidental” terrace formation popped up only now and then in the intervening decades and largely went untested. In the meantime, a different explanation gained popularity.

River terraces  — those remnant valley bottoms perched along river valleys all over the world — are eroded into old river sediments or bedrock, and are often interpreted as the signature of a river’s response to climate change.  The argument goes like this: if tectonic uplift pushes the landscape up at a steady rate, then a river steadily cuts downward to keep pace. In this quasi-equilibrium state, the river never pauses its downward carving. But if an outside force disturbs the river from equilibrium — for instance, by changing the supply of water and sediment–then the river may temporarily pause its downward journey before catching up with a rapid pulse of downcutting. Terrace surfaces would be eroded during the pauses, and the terraces would be stranded when the river cut downward.

Davis’ hypothesis is that rivers make terraces without pausing their downcutting, but instead by the unsteady lateral shifting of the river during downcutting. This distinction gets to the heart of reading the terrace record: if rivers in equilibrium also make terraces, then terraces may say nothing about regional changes in climate or tectonics. Instead, terraces would simply be the accidental remnants of wandering rivers.

So how might we distinguish between these competing arguments, given that most of the world’s terraces formed long before today?

One way is to see if the numbers add up: for example, how fast would a meandering river need to move in order to make an “accidental” terrace? To test this, my work unfolded in two steps. First, I translated Davis’ thought experiment to a simplified mathematical model, and then to computer code. With a flurry of for-loops and deluge of data structures, a Caltech supercomputer named for a turn-of-the century exploration ship tested hundreds of scenarios for valley evolution. Here’s a movie of the model in action:

A movie of valley evolution from our new paper in JGR-Earth Surface.

The top part shows a meandering river (in blue) flowing from left to right, viewed from above. Initially, the river buts up against bedrock walls: the topographic cross section in the lower part shows shows bedrock (black) and sediment (white), with the channel showing up as a narrow notch in the center of the cross section. As the clock zooms forward through 25,000 years, the meander bends migrate freely through the sediment and deform as they meet resistant bedrock. The uneven pattern of lateral erosion, coupled with vertical erosion, makes a set of terraces. The terraces are identified automatically during the simulation, and in the top section are colored according to their height above the channel. And just like that,107 years after Davis’ original hypothesis, we have new data to start testing it.

Based on these simulations, we built a predictive framework for what “accidental” terraces formed under steady climate should look like and how often they should form. In the second step of the analysis, I compared the model predictions to a selection of carefully studied valleys in the central and western US, including the Wind River valley from my last post. Here’s the summary figure:


Model predictions for terrace age and geometry and estimate erosion rates for natural river valleys from our new paper in JGR-Earth Surface.

For the geomorphologists: we found that key terrace attributes commonly attributed to climate change–such as the occurrence of pairs of terraces on either side of the valley, their length along the valley, and their separation in time by thousands of years–can form without climate change. These predictions hold for particular conditions: rivers must erode laterally relatively quickly (i.e., > 10 mm/yr) and cut down relatively slowly (i.e., < 1 mm/yr). By comparing to the natural river valleys, we see that the meandering river model with steady vertical incision explains key terrace attributes for the Colorado River valley (Texas) better than the Wind River valley.

Long story short: with an idea of how meandering rivers might make terraces under steady conditions, we have a better shot at identifying the terraces that really tell us something about climate change.

Particularly in the mountains, many of the flat surfaces you see are flat because a river used to run there. So give a glance around on your next trip through a river valley: you might be walking in the footsteps of the river itself. Terraces are a hot topic with lots of active research. For more new work on decoding the terrace record, see papers on the roles of landslides [Scherler et al., 2016], bedrock [Schanz and Montgomery, 2016], and climate change [Langston et al., 2015].