12
K. Black· M. Green· T. Healy· R. Bell· J. Oldman· T. Hume
current-generated friction velocity remains relatively important beyond the
wave boundary layer. While this current-induced turbulence is important in the
water column, we are dealing with a small elevation above the bed of 7 cm only.
Diffusion is modelled as a random walk and a horizontal eddy diffusivity of
0.015 m 2 .s- 1 has been chosen (both along and across the flow streamlines). This
value is ten times larger than the vertical eddy diffusivity in order to incorporate
advective and turbulent processes associated with short-duration current oscillations and wave breaking. The selected value remains uncalibrated but its influence is relatively small on the model grid of 200 m cells. In the absence of bedload transport data for model confirmation, we do not treat bedload in this
chapter.
Model output is compared to the measurements in Fig. 2g which shows predictions from 2 adjacent cells and a prior run with a slightly altered depth. The
magnitude of the predictions are comparable with the measurements, i.e. 2
peaks in sse with magnitudes of about 300-400 mg.l- 1 and a period of low concentration of less than 50 mg.l- 1 around the high tide. The second peak is quite
well modelled, even though adjacent model cells give different results. The first
peak is less well predicted, but this is primarily because of the low wind speeds
around burst 2156, which resulted in a low orbital motion prediction. The results
are highly sensitive to depth, primarily because of the strong dependence of bottom orbital motion on water column attenuation and the interaction between
fetch, bed friction and bank emergence. Green et al. (1998) also found mostly
good agreement between measured sse and predictions by adopting the same
methods as those applied here. However, they were unable to explain the unexpectedly high concentrations in the turbid fringe, which were larger than expected for the measured orbital motion. With the full spatial model and its capacity to treat the process of grain size winnowing, changes in roughness length
and horizontal advection, we are able to predict the magnitude and the phase of
peak sse, even though it does not coincide with peak orbital motion. Secondly,
Green et al. (1998) were unable to explain concentrations recorded around peak
flood tide which were lower than expected for the measured orbital currents
(bursts 2162-2166). The model, on the contrary, predicts low concentrations at
this time which match the field measurements (Fig. 2g). The modelling shows
advection by the flood currents of lower sse from the deep channel as the cause
of the anomaly noted by Green et al. (1998).
5
Discussion
One must conclude that if the inputs can be specified accurately, effective numerical simulations of currents, water levels, surface waves, wave orbital currents and sse are possible in an estuary, even with complex morphology, strong
tidal currents and unsteady wave conditions.
While Green et al. (1998) were able to correctly predict sse for some stages of
the tidal cycle using point measurements, the numerical models were needed to
K. Black· M. Green· T. Healy· R. Bell· J. Oldman· T. Hume
current-generated friction velocity remains relatively important beyond the
wave boundary layer. While this current-induced turbulence is important in the
water column, we are dealing with a small elevation above the bed of 7 cm only.
Diffusion is modelled as a random walk and a horizontal eddy diffusivity of
0.015 m 2 .s- 1 has been chosen (both along and across the flow streamlines). This
value is ten times larger than the vertical eddy diffusivity in order to incorporate
advective and turbulent processes associated with short-duration current oscillations and wave breaking. The selected value remains uncalibrated but its influence is relatively small on the model grid of 200 m cells. In the absence of bedload transport data for model confirmation, we do not treat bedload in this
chapter.
Model output is compared to the measurements in Fig. 2g which shows predictions from 2 adjacent cells and a prior run with a slightly altered depth. The
magnitude of the predictions are comparable with the measurements, i.e. 2
peaks in sse with magnitudes of about 300-400 mg.l- 1 and a period of low concentration of less than 50 mg.l- 1 around the high tide. The second peak is quite
well modelled, even though adjacent model cells give different results. The first
peak is less well predicted, but this is primarily because of the low wind speeds
around burst 2156, which resulted in a low orbital motion prediction. The results
are highly sensitive to depth, primarily because of the strong dependence of bottom orbital motion on water column attenuation and the interaction between
fetch, bed friction and bank emergence. Green et al. (1998) also found mostly
good agreement between measured sse and predictions by adopting the same
methods as those applied here. However, they were unable to explain the unexpectedly high concentrations in the turbid fringe, which were larger than expected for the measured orbital motion. With the full spatial model and its capacity to treat the process of grain size winnowing, changes in roughness length
and horizontal advection, we are able to predict the magnitude and the phase of
peak sse, even though it does not coincide with peak orbital motion. Secondly,
Green et al. (1998) were unable to explain concentrations recorded around peak
flood tide which were lower than expected for the measured orbital currents
(bursts 2162-2166). The model, on the contrary, predicts low concentrations at
this time which match the field measurements (Fig. 2g). The modelling shows
advection by the flood currents of lower sse from the deep channel as the cause
of the anomaly noted by Green et al. (1998).
5
Discussion
One must conclude that if the inputs can be specified accurately, effective numerical simulations of currents, water levels, surface waves, wave orbital currents and sse are possible in an estuary, even with complex morphology, strong
tidal currents and unsteady wave conditions.
While Green et al. (1998) were able to correctly predict sse for some stages of
the tidal cycle using point measurements, the numerical models were needed to