8256

• UCREFRP Catalog Number: 8256
• Author: Holly, F. M. and R. Ettema
• Title: Sediment Imbalance in Rivers
• USFW Year: n.d.
• USFW - Doc Type: Simulation Possibilities and Problems
• Copyright Material: YES

section, I-D models must be considered only as
models of idealized rivers having certain gross
properties in common with the natural rivers they
are supposed to emulate. Obviously, this weakness
begins to disappear with 2- and 3-D models, al-
though they too cannot resolve spatial heterogene-
ityat a scale finer than that of their computational
grids.

Bedload and Suspended-load Source
Predictors

Prediction of bedload transport capacity and the
flux of sediment from the bed into the water col-
umn rely on a body of competing, inconsistent,
empirical relations derived from laboratory and
field experiments under supposed equilibrium con-
ditions. The range of differences in various load
predictors is well illustrated by Table 2.113 in
Vanoni (1975). Mobile-bed models are not only
subservient victims of the enOlmOUS uncertainties
in the validity of such predictors for a particular
situation, but also compound the potential elTOr in
assuming quasi-instantaneous satisfaction of
equilibrium conditions (though this can be relaxed
somewhat; see for example Holly and Rahuel
1989). mtimately, one can only remove this uncer-
tainty by incorporating a complete description of
the dynamic forces acting on individual sediment
particles. But this would require a complete de-
scription of the turbulent velocity fluctuations
near the bed, and would have to account for the
infinite variety of particle shapes, sizes, and inter-
actions, an obviously impossible task. At present,
and for the foreseeable future, the modeler's only
recourse for the study of a particular river reach is
to adopt the transport and source predictors that
are known to be best adapted to that particular
location, if possible, abandoning hope of adopting
global predictors applicable to all cases.

Stochasticity

A basic premise of the modelling descriptions
given here is that the processes involved are deter-
ministic and rational; that is, that there is no
randomness or uncertainty in the laws them-
selves. There is, however, a strong element of
stochasticity in the water-sediment transport
problem, and this has been explicitly recognized by
some investigators, starting with the work of
Gessler (1975). Sediment behavior at any particu-
lar moment at any particular location of a river

FORREST M. HOLLY. JR. AND RoBERT E'rrEMA 423

probably should be considered as one realization of
a random. process.

Cohesion

Even if one can consider bedload and sus-
pended-load predictors for noncohesive (loose)
sediments as reasonably reliable within certain
limits, all bets are off when electrochemical inter-
action of fine (clay) particles gives them a cohesive
strength. Considerable work in this area has been
done by Krone (1962) and his colleagues, but there
is as yet no consensus as to the appropriate pa-
rameters, or indeed the very fOlm of the entrain-
ment laws, that should be used when cohesion
occurs. Surface erosion of cohesive particles, nota-
bly their progressive detachment from the bed, is
quite different from mass erosion, when an entire
layer of cohesive material fails structurally and
injects a large mass of bed material into the water
column in a single event.

Bedforms

The assimilation of a river bed to a single sur-
face of UnllOlm properties, especially in I-D mod-
els, necessarily ignores the complex interplay of
bedforms and their intimate and strongly non-
linear relation with the flow hydrodynamics and
sediment properties. Attempts are made in most
models to take these interactions into account glob-
ally, for example through transport and flow-de-
pendent bed roughness and relations between
dune height and bed mixing layers, but a model of
any dimension is destined to be unable to resolve
the complex geometric features of bedforms and
their interactions with the flow and sediment
transport.

Armoring

Although sediment sorting (differential trans-
port depending on particle size) can be repre-
sented in a rather straightforward manner insofar
as bed-sediment accounting is concerned, accu-
rate representation of bed-degradation protection
through interlocking of a relatively small coverage
of immobile particles is much more complex and
challenging. This is especially true when part of a
river bed may be armored while another is . in
active bed movement; such a distinction simply
cannot be made with a 1- D model. Yet armoring
can be the primary mechanism for atTesting bed