Turbulence structure in straight gravel-bed channels.

It is clear when looking at a gravel-bed river that the flow turbulence is not random:
eddies and boils can be seen at the surface. Understanding the behaviour of gravel-bed rivers
is limited until the flow structures are described and their origins explained. This work uses a
combination of point monitoring of velocity and visualisation of flows over gravel beds in both
the field and flume to describe and explain the structure of the boundary layer and the
generation mechanisms of flow structures within it.
The boundary layer over fluvial gravel beds can be split into two main flow zones. The
inner zone consists of eddies developing and shedding off relatively large obstacles clasts and
more confused eddying in the inter-obstacle wake zone. There is some evidence to suggest
that flow directly above the inner zone is characterised by the domination of negative Reynolds
stress associated with the upwards ejection of eddies at higher than average streamwise
velocity. The visualisations show that the eddy shedding becomes more confused as the flow
rate is increased, and is best described as intermittent.
Above the obstacle crests is the outer flow zone in which there are larger flow
structures. Visualisations of the vertical streamwise plane and simultaneous logging of velocity
at three points suggest that the larger structures consist of inverted wedges of high-speed fluid
which move at more than the mean streamwise velocity as "sweep-like" structures. Low-speed
fluid is ejected away from the bed after a high-speed front has passed. These structures
become increasingly dominant as the flow rate is increased. The high-speed zones appear to
be associated with the passage of the front slopes of water surface waves, although this
cannot explain some of those monitored in the field.
The flow structures are observed to exhibit a crudely intermittent behaviour in the
visualisations. Attempts to quantify the intermittency both from the vistlalisations and by time
series analysis of both velocity and stress does not produce the clear quasi-periodic velocity
signatures that some other workers have found. There is some agreement between the periods
of dominant peaks in the spectra of Reynolds stress and calculated burst periods. This is
interpreted as meaning that a similar bursting process occurs over mixed gravel beds as that
reported over the smooth and uniformly rough surfaces. There is no evidence that eddy
shedding produces a periodic signature in either the velocity or stress series.
A mechanistic model is proposed which involves the coupling of eddy shedding with the
outer flow wedges. In conditions of high relative roughness (or low flow). eddy shedding is
predominantly responsible for the generation of turbulent structures in the boundary layer.
whereas at low relative roughness (or high flow conditions). the outer zone wedges dominate.
This allows the interpretation of flow structure in gravel bed rivers to be made with respect to
that observed in sand bed rivers and in tidal flows over gravels.
Observations of gravel entrainment in the visualisation experiments suggest that there
are two modes of entrainment. At marginal transport conditions. fines are entrained from the
wake zones of the larger obstacle clasts by vertical velocity fluctuations. At higher flow
conditions, the high speed "sweep-like" structures are responsible for entraining the larger
clasts, although the timescale of entrainment is greater that that of an individual flow structure.