Scour and fill in ephemeral streams

The classical concept that mean bed elevation over an entire stream reach is lowered by scour during flood-wave passage and is restored by deposition in the waning flood phase (mean-bed scour and fill) can be challenged. The alternative that both scour and fill occur concurrently at different migrating loci within a reach (local scour and fill) is more consistent with published field data. The field and laboratory investigations reported herein suggest that mean-bed scour and fill in a uniform channel is minor compared to local scour and fill caused by bedform migration, and that maximum local scour and fill may occur on the waning flood in some instances. The field experiment, utilizing a rectilinear array of buried maximum-scour indicators (scour-cords), produced data for contouring of maximum scour and fill in an ephemeral streambed during two floods. In the first flood, 24 em of scour and fill was measured for a bankfull flow depth of 23 cm. In the second, maximum scour and fill was at least 66 cm for a bankfull flow depth of 34 cm. Estimates of antidune amplitudes for the two floods, based on theoretical models and laboratory and field observations, are 28 to 64 cm and 48 to 97 cm, respectively. This indicates that all scour and fill measured by the scour-cord array could have been caused by antidune migration. Laboratory experiments were conducted in an 18 m-long open-circuit flume with automated sediment and water input-rate controls. A series of experiments in a 26.7 cm-wide sand-bed channel with rigid walls, at grade for a simulated flood patterned after those typical of ephemeral streams, showed that mean-bed scour and fill was less than 3 percent of local scour and fill. For these experiments, mean sand size was 0.3 mm, channel slope was .009, maximum water depth was 40 mm, maximum local scour and fill was 22 mm, and maximum mean-bed scour and fill was 0.6 mm. Maximum mean bed elevation variation was thus only two sand-grain diameters. Fill occurred at peak flow followed by scour to the pre-flood mean bed elevation on the waning flood. Maximum local scour and fill took place near the end of the simulated floods, when bedform amplitudes were the greatest. A series of simulated-flood experiments in a sand-bed channel with erodible sand banks showed scour and fill behavior qualitatively similar to that of the rigid-wall channel. Bank erosion, channel meandering, and braiding prevented quantitative scour and fill measurements in these alluvial-bank experiments. Measured flow and bedform parameters and scour and fill data derived from small laboratory scour-chains were compatible with those estimated from the theoretical model used in the field experiment