At https://phys.org/news/2019-06-large-boulders-huge-canyons.html … river valleys in hilly or mountainous country have the remnants of water surges in them, discarded rocks of various shape and size. In the UK these can be seen even in small streams. A good example might be the river bed between Lynton (up high) and Lynmouth (down below), just a short distance but jammed with rocks and boulders. Some of these originate from a storm surge in 1953 when the lower town was flooded by a swollen river Lyn. However, other rocks had a much older origin, in earlier storms and torrential rain episodes. What is a small and insignificant river became a raging torrent – pushing rocks and boulders before it. Water can be very powerful. A recent paper in the journal Geology (to be published in July 2019) says boulders and rocks play a role in the geological evolution of canyons in the US – over a long geological span of time. They imply that current uniformitarian mainstream theories on canyon evolution may be overly simplistic. In other words, storm surges play an important role in canyon formation (and can obviously carve out canyons much quicker than a gradualist scenario). Canyons may have come into being much quicker than uniformitarians have until now allowed – even if such surges in water were few and far between.
It seems that a computer model was created in an attempt to understand the interplay between river boulders and the sides of hillsides. Presumably they already were questioning the uniformitarian mantra otherwise why make a model to disprove it. Rivers that flow through soft geology tend to be wide and flat (with flood plains). Rivers that flow across hard rock formations tend to favour narrow canyons with steep upper hill sides (just like the river Lyn in Devon). Hard rocks slow down the process of erosion. It is mostly soil and sediment that is leached out first of all and carried down stream but doing so also dislodges rocks, small at first but over time bigger rocks and boulders become unstable and are moved by the water flow (during storms etc). They found that the presence of boulders in the channel created a feedback loop with steeper hillsides resulting eventually in to an erosional force simply because the rocks were bouncing around and moving other rocks at the same time, even from the more solid rock faces on the sidelines. Large boulders thus help in canyon evolution. They can actually shape canyons.
Later, the model was tested in the field by going to New Mexico (and the Rio Grande) where they found evidence to support their ideas. They also found evidence that some large boulders become wedged between other boulders and had been in place for a very long period of time. It follows that the odd severe flood and storm surge can move a lot of rocks and they might not move again until the water becomes a raging torrent again – which could be in intervals of hundreds of years. The same thing applies to the river Lyn.