1987 Volume 60 Issue 2 Pages 195-202
Flume experiments suggest that, in a braided stream, terraces are formed from bars under erosional conditions. The terraces are often soon eroded out or replaced by newly-forming bars, while the experimental parameters change very slowly. This paper examines these processes of terrace formation with the aid of field data on the Holocene terraces of the Nedori River, in the Nikko volcanic area. Five terrace surfaces, T1, T2, T3, T4 and T5 ranging in order from the older to the younger, stretch along a steep gravel-bed channel of the Nedori; here T1 and T3 are fill terrace surfaces formed in the Late Glacial time and the hypsithermal period, respectively. The other surfaces were formed by the downcutting of the T1 and T3 surfaces. The downcutting which formed the T2 surface was rapid and intensive, reflecting a highly erosional hydraulic condition of the Late Glacial-Holocene transitional period. The formation of the T4 and T5 surfaces is well elucidated by means of the processes observed in the course of the flume experiments, firstly because the former channels and bars on the T4 and T5 surfaces have a distribution pattern quite similar to that observed on the channel bed of the flume; and, secondly, becuuse the profiles of these surfaces and the present channel bed show a remarkably undulated form which is also characteristic of the channel bed of the flume. As the level of the terrace surface is included in the amplitude of channel bed undulation, the terrace surface is easily eroded or overlapped by newly-forming bars, as was observed in the flume experiments. These facts suggest a relatively small change in hydraulic conditions after the hypsithermal period, a change which was insufficient to provoke an intensive downcutting.