8
K. Black· M. Green· T. Healy· R. Bell· J. Oldman· T. Hume
4
Model Comparisons
4.1
Water Levels and Currents
No water level measurements were made at the model boundary in the entrance
to the Manukau Harbour during the RALPH deployment. To overcome this deficiency, model boundary levels were generated using water levels recorded
within the estuary at Onehunga (Fig. 1) as follows. First, we modelled the estuary
using Onehunga tides at the open model boundary. The phase lag and amplitude
adjustment was then obtained using a lagged regression analysis of model-predicted tides at Onehunga and the original measurements. The lag was found to
be 0.58 hours and amplitude factor was 0.784, plus a mean sea level (MSL) shift
of 0.48 m was identified due to superelevation of MSL and a datum shift. These
factors were then used to generate sea level boundary conditions from the Onehunga records. Winds were taken from the nearby Auckland airport anemometer and were assumed to be spatially-uniform throughout the model grid. To allow for land boundary layer effects, speeds were multiplied by 1.1 (a 10% augmentation), although the effect of this factor could not be separated from the effect of bed frictional resistance, later set by calibration.
Water levels measured at RALPH were found to be in agreement with the
model predictions after a small additional correction was applied; the amplitude
of the sea level oscillation at the boundary was increased by 6% and the mean
vertical shift was lowered by 8 cm from 0.48 m to 0.40 m. The 0.4 m vertical shift
was subsequently found to be a result of a datum shift between the eastern and
western segments of the harbour, as applied by the Navy hydrographer. The spatial variation in the datum shift remains unknown and is later shown to be highly important in the very shallow conditions being simulated. Accentuation of the
overtides (evident in Fig. 2a) is caused by the simple linear regression adopted
to find the relationship between Onehunga and open boundary water levels. In
essence, the overtides are represented twice; once in reality as the tides travel to
Onehunga and then again in the model during the simulation.
To identify any boundary layer velocity gradients, current measurements at elevations of 48 and 97 cm are presented. The currents from the model are mostly in
good agreement (Fig. 2b). The speeds are least well predicted during the windy ebb
phase when the model over-estimates the magnitudes. This may be caused by unrepresented wave-driven currents on the intertidal flat or wind-induced shear
through the water column which is not simulated by a 2-dimensional model. Notably, the measurements and the model are in close agreement during the previous
two half cycles when waves were not present and the measurements at the two instrument elevations are also mostly similar except during the windy ebb phase.
The measured current directions rotate more smoothly than the model but
are generally in good agreement (Fig. 2c). Small deviations in direction are undoubtedly caused by poor bathymetry resolution at the 200 m grid scale adopted
K. Black· M. Green· T. Healy· R. Bell· J. Oldman· T. Hume
4
Model Comparisons
4.1
Water Levels and Currents
No water level measurements were made at the model boundary in the entrance
to the Manukau Harbour during the RALPH deployment. To overcome this deficiency, model boundary levels were generated using water levels recorded
within the estuary at Onehunga (Fig. 1) as follows. First, we modelled the estuary
using Onehunga tides at the open model boundary. The phase lag and amplitude
adjustment was then obtained using a lagged regression analysis of model-predicted tides at Onehunga and the original measurements. The lag was found to
be 0.58 hours and amplitude factor was 0.784, plus a mean sea level (MSL) shift
of 0.48 m was identified due to superelevation of MSL and a datum shift. These
factors were then used to generate sea level boundary conditions from the Onehunga records. Winds were taken from the nearby Auckland airport anemometer and were assumed to be spatially-uniform throughout the model grid. To allow for land boundary layer effects, speeds were multiplied by 1.1 (a 10% augmentation), although the effect of this factor could not be separated from the effect of bed frictional resistance, later set by calibration.
Water levels measured at RALPH were found to be in agreement with the
model predictions after a small additional correction was applied; the amplitude
of the sea level oscillation at the boundary was increased by 6% and the mean
vertical shift was lowered by 8 cm from 0.48 m to 0.40 m. The 0.4 m vertical shift
was subsequently found to be a result of a datum shift between the eastern and
western segments of the harbour, as applied by the Navy hydrographer. The spatial variation in the datum shift remains unknown and is later shown to be highly important in the very shallow conditions being simulated. Accentuation of the
overtides (evident in Fig. 2a) is caused by the simple linear regression adopted
to find the relationship between Onehunga and open boundary water levels. In
essence, the overtides are represented twice; once in reality as the tides travel to
Onehunga and then again in the model during the simulation.
To identify any boundary layer velocity gradients, current measurements at elevations of 48 and 97 cm are presented. The currents from the model are mostly in
good agreement (Fig. 2b). The speeds are least well predicted during the windy ebb
phase when the model over-estimates the magnitudes. This may be caused by unrepresented wave-driven currents on the intertidal flat or wind-induced shear
through the water column which is not simulated by a 2-dimensional model. Notably, the measurements and the model are in close agreement during the previous
two half cycles when waves were not present and the measurements at the two instrument elevations are also mostly similar except during the windy ebb phase.
The measured current directions rotate more smoothly than the model but
are generally in good agreement (Fig. 2c). Small deviations in direction are undoubtedly caused by poor bathymetry resolution at the 200 m grid scale adopted