Along an Old Trough for a Stretch

After some floods spill out of the main channel, they enter an old channel trough. As long as the low points along the way remain connected to one another, discharge that was originally scattered across the plain will gradually converge; water advances along the old trough and gathers again into a more concentrated flow. What matters is not the complete outline of the old river course, but the continuous drop that still remains between low places. Once discharge begins to gather, this preexisting hollow will again carry more water.

As the flood recedes, the fine mud suspended in the water also settles toward low places, so old channel troughs and other depressions receive more sediment than the slightly higher surrounding surfaces. It does not have to fill the trough. One flood lays down one layer; the next covers it with another. After several rounds of deposition, the original relief may be erased only in part. Deposition does not raise the whole plain evenly. It first takes up the deeper parts of the old trough, so that gentler segments appear along what had been a continuously descending bed. Each place is still lower than its surroundings, but the fall connecting them is no longer as continuous as before. The trough is still visible, still holds water in the rainy season, and the next flood may still enter it. Water may even travel along the old trough for a stretch, and discharge that had been dispersed begins to converge; but when it reaches several low areas that are no longer connected, that convergence becomes difficult to sustain. The water has not run into a levee. It has merely lost the conditions needed for discharge along the way to keep joining the same flow, and afterward it may spread out across the plain, waiting for another set of low places to gradually connect into a new channel.

A comparison of sixty-three modern avulsions recorded this unevenness: braided rivers more often reoccupy old channels, while avulsions in meandering rivers are more often accompanied by floodplain deposition and the formation of new channels. This is not a command that the two kinds of rivers each obey, nor can it prescribe a route for the next flood. It makes the role of fine mud concrete: deposition does not always alter the water’s path by building a conspicuous new levee. Sometimes, after each recession of the water, it slowly changes the topographic relations required for the next convergence of flow. A clear break may not appear on the river surface, yet the change has already been distributed through segments of trough bed that have become slightly shallower.

Yet on some plains, low places remain connected over long distances. In 2008, a breach in the embankment near Kusaha, Nepal, caused the Kosi River to leave its existing channel, shifting the main flow more than a hundred kilometers eastward. Researchers later used pre-avulsion topographic and channel data to run simulations, and the resulting paths broadly matched the actual avulsion. The Kosi River cannot be treated as a hundred-kilometer enlargement of the fine-mud mechanism described above. The breach triggered water to leave a constrained channel; after it left, the low places and old channels already present on the plain still influenced where discharge gathered again. The terrain had not written a route for the flood in advance, but the water after the breach was not facing an undifferentiated plane either. Old channel troughs are sometimes gradually weakened by deposition, and sometimes they still connect low places far apart from one another; the two states do not automatically collapse into a single rule.

When the flood arrives again, the water surface spreads into that old trough that has not been filled. The trough is still visible, and the water can still move along it for a stretch. At some point, the low places along the way are no longer connected, and the discharge no longer gathers into one flow.