NIWA – Technical Report on Coastal Hydrodynamics and Sediment Processes in Lyall Bay

Home , Lyall Bay, Wind generated current

Technical Report 17

Wellington Airport Runway Extension Technical Report on Coastal Hydrodynamics and Sediment Processes in Lyall Bay

Prepared for Wellington International Airport Ltd

March 2015 (updated March 2016)

Prepared by: Mark Pritchard Glen Reeve Richard Gorman Iain MacDonald Rob Bell

For any information regarding this report please contact: Rob Bell Programme Leader: Hazards & Risk Coastal & Estuarine Processes +64-7-856 1742 [email protected]

National Institute of Water & Atmospheric Research Ltd PO Box 11115 Hamilton 3251

Phone +64 7 856 7026

NIWA CLIENT REPORT No: HAM2015-003 Report date: March 2015 (updated March 2016) NIWA Project: WIA15301

Quality Assurance Statement

Reviewed by: Dr S. Stephens

Formatting checked by: A. Bartley

Approved for release by: Dr A. Laing

Front page photo: Lyall Bay aerial photograph (2013-14 LINZ aerial photography series).

© All rights reserved. This publication may not be reproduced or copied in any form without the permission of the copyright owner(s). Such permission is only to be given in accordance with the terms of the client’s contract with NIWA. This copyright extends to all forms of copying and any storage of material in any kind of information retrieval system.

Whilst NIWA has used all reasonable endeavours to ensure that the information contained in this document is accurate, NIWA does not give any express or implied warranty as to the completeness of the information contained herein, or that it will be suitable for any purpose(s) other than those specifically contemplated during the Project or agreed by NIWA and the Client.

Contents

Executive summary ...... 11

1 Introduction and Background ...... 17

2 Hydrodynamic field programme: methods and results (Lyall Bay)...... 18 2.1 Instrumentation ...... 18 2.2 Results ...... 19

3 Hydrodynamic flow and sediment-transport modelling (Delft2D) ...... 33 3.1 The approach to hydrodynamic and sediment-transport modelling ...... 33 3.2 Delft3D modelling Suite ...... 35 3.3 Model Development and Setup ...... 36 3.4 Modelled hydrodynamic scenario results ...... 46 3.5 Sediment-transport modelling ...... 52 3.6 Discussion and summary ...... 58

4 Wave climate: Spectral wave modelling (SWAN) ...... 60 4.1 South Wellington coast wave modelling (SWAN) ...... 60 4.2 External forcing ...... 61 4.3 Model verification ...... 62 4.4 Lyall Bay wave conditions ...... 64 4.5 Wave statistics at selected output locations ...... 70 4.6 Extreme wave climate ...... 77

5 Phase-resolving wave model results (ARTEMIS) ...... 79 5.1 ARTEMIS wave model ...... 80 5.2 ARTEMIS mesh development...... 80 5.3 ARTEMIS model set-up (physical and numerical parameters)...... 83 5.4 Boundary reflection ...... 84 5.5 Validation ...... 84 5.6 Results showing wave changes due to the proposed runway extension ...... 85 5.7 Discussion/Summary ...... 120

6 Suspended-sediment plume modelling (Delft3D-SED) ...... 122 6.1 Rationale and scenario modelling approach ...... 122 6.2 Delft3D-SED (Cohesive) ...... 124

Wellington Airport Runway Extension

6.3 Suspended-sediment plume results ...... 126 6.4 Discussion/summary ...... 139

7 Acknowledgements ...... 140

8 Glossary of abbreviations and terms ...... 141

9 References ...... 143

Tables Table 2-1: Locations (WGS-84) and 2014 deployment information for the Lyall Bay instruments. 19 Table 2-2: Additional deployment information. 20 Table 2-3: Summary of principal component analysis for currents at Site 1. 26 Table 2-4: Summary of the overall current drift at Site 1 over the deployment period. 29

Table 3-1: RMSE and model Skill for observed and modelled M2, S2 and N2 semi- diurnal tidal amplitude at three sites in the model domain. 42

Table 3-2: Observed and modelled (parentheses) M2, S2 and N2 semi-diurnal tidal current major-axis amplitude at three sites in the model domain. 44 Table 4-1: Locations of wave measurements used to validate the SWAN simulations. 62 Table 5-1: Physical and numerical parameters used in the set-up of ARTEMIS. 83 Table 5-2: Model predicted changes in wave height (m) at site P1 that result from a change in model boundary conditions (ranging from 1–5 m). Note: that the Table values represent the change in significant wave height locally. 88 Table 5-3: Change in predicted wave heights at data extraction site P2. 89 Table 5-4: Change in predicted wave heights at data extraction site P3. 89 Table 5-5: Change in predicted wave heights at data extraction site P4. 90 Table 5-6: Change in predicted wave heights at data extraction site P5. 90 Table 5-7: Change in predicted wave heights at data extraction site P6. 91 Table 5-8: Change in predicted wave heights at data extraction site P7. 92 Table 5-9: Change in predicted wave heights at data extraction site P8. 92 Table 5-10: Percent change in predicted wave heights at data extraction site P1. 93 Table 5-11: Percent change in predicted wave heights at data extraction site P2. 93 Table 5-12: Percent change in predicted wave heights at data extraction site P3. 94 Table 5-13: Percent change in predicted wave heights at data extraction site P4. 94 Table 5-14: Percent change in predicted wave heights at data extraction site P5. 95 Table 5-15: Percent change in predicted wave heights at data extraction site P6. 95 Table 5-16: Percent change in predicted wave heights at data extraction site P7. 96 Table 5-17: Percent change in predicted wave heights at data extraction site P8. 96 Table 6-1: Maximum predicted SSC above background levels after reasonable mixing at ~150 m for discharges at D1–D3 and sediment-discharge rates of 1 and 2 kg/s. 128

4 Wellington Airport Runway Extension

Figures Figure 1-1: Aerial image of Wellington International Airport and runway. 17 Figure 2-1: Site map showing oceanographic instrument locations overlying the seabed bathymetry to WVD-53. 21 Figure 2-2: Wind rose for winds measured at Wellington Aero over a period of 28-years (January 1985 to December 2012). 22 Figure 2-3: Wind rose for winds over the deployment period (18-Aug to 09-Oct 2014). 23 Figure 2-4: Current and wind time series for deployment at Site 1. 24 Figure 2-5: Scatter plots of currents at three elevations above the bed at Site 1. 26 Figure 2-6: Progressive current drift at Site 1 at three elevations above the sea bed (bottom-left panel for near the surface) and the depth-average (bottom- right panel). 28 Figure 2-7: Wave parameters measured at Site 1. 30 Figure 2-8: Wave rose for measurements at Site 1. 31 Figure 2-9: Wave parameters measured at Site 2 in outer Lyall Bay. 32 Figure 3-1: Delft3D-WAVE model grid (red) of south Wellington coast superimposed on top of the Delft3D-FLOW model grid (black) that also includes Wellington Harbour. 37 Figure 3-2: Areal extent of the hydrodynamic model bathymetry. 38 Figure 3-3: Close-up of the hydrodynamic model bathymetry in Lyall Bay. 39 Figure 3-4: Winds over the deployment period (18-Aug to 09-Oct 2014). 40 Figure 3-5: Idealised 1-month simulation of a sequence of two southerly (180) and one northerly (0) wind events. 41 Figure 3-6: Comparison of observed (blue) and modelled (red) tide heights for two sites in Wellington Harbour and one site in Lyall Bay (C). 43 Figure 3-7: Cross-correlation of modelled versus observed tides, showing a zero lag (i.e., both time series are in phase). 43 Figure 3-8: Modelled net or residual circulation over the ~7-week field deployment period in 2014. 46 Figure 3-9: Residual (net) current circulation in Lyall Bay over the ~7-week field deployment period. 47 Figure 3-10: Current circulation near peak southerly wind at low water. 49 Figure 3-11: Current circulation near peak southerly wind at mid-flood tide (HW-3 hrs). 49 Figure 3-12: Current circulation near peak southerly wind at high water. 50 Figure 3-13: Current circulation near peak northerly wind at low water. 51 Figure 3-14: Current circulation near peak northerly wind at mid-flood tide (HW-3 hrs). 51 Figure 3-15: Current circulation near peak northerly wind at high water. 52 Figure 3-16: Geomorphic features in Lyall Bay incl. the 1941 shoreline. 53 Figure 3-17: Predicted changes in seabed height for the existing situation over the ~7-week field period in 2014. 56 Figure 3-18: Predicted changes in seabed height for the proposed runway situation over the ~7-week field period in 2014. 56 Figure 3-19: Net or relative change in the response of sea-bed heights with the proposed runway in place over the ~7-week simulation. 58

Wellington Airport Runway Extension

Figure 4-1: Extent of the SWAN modelling domain, with instrument sites marked. 61 Figure 4-2: Comparison of significant wave height (Hm0) measured by instruments (black) with corresponding values predicted by the SWAN simulation (red) at the three measurement locations. 63 Figure 4-3: Comparison of wave statistics measured by the Baring Head Waverider buoy (black), with corresponding values predicted by the SWAN simulation (red). 63 Figure 4-4: Wave conditions at 00:00 NZST on 22-Aug 2014 from the SWAN simulation, in Lyall Bay (top panel) and over the full model domain (bottom panel). 66 Figure 4-5: Wave conditions at 02:00 NZST on 19-Sept 2014 from the SWAN simulation, in Lyall Bay (top panel) and over the full model domain (bottom panel). 67 Figure 4-6: Wave conditions at 00:00 NZST on 23-Sept 2014 from the SWAN simulation, in Lyall Bay (top panel) and over the full model domain (bottom panel). 68 Figure 4-7: Wave conditions averaged over the 55-day SWAN simulation, in Lyall Bay (top panel) and over the full model domain (bottom panel). 69 Figure 4-8: Locations of sites selected for output of wave statistics from the ~7-week SWAN simulation. 70 Figure 4-9: Wind roses for output sites 3–Baring Head (left) and 6–inner Lyall Bay (right) derived from the SWAN simulation. 71 Figure 4-10: Probability distribution functions for significant wave height derived from the SWAN simulation at the 18 output sites. 73 Figure 4-11: Probability distribution functions (vertical axis) for peak wave period derived from the SWAN simulation at the 18 output sites. 74 Figure 4-12: Probability distribution functions (vertical axis) for second moment mean wave period derived from the SWAN simulation at the 18 output sites. 75 Figure 4-13: Wave roses, i.e., joint occurrence distributions for mean wave direction and significant wave height at the 18 output sites. 76 Figure 4-14: Probability distribution functions for root-mean-square bed orbital velocity derived from the SWAN simulation at the 18 output sites. 77 Figure 4-15: “Waitangi-Day” 2002 waves off Baring Head with NIWA’s Brodie Building at Greta Point for scale. 78 Figure 5-1: Spatially averaged terrain node data illustrating the zone between 0 - 2m where no bathymetric data exists. 81 Figure 5-2: Left image shows the Lyall Bay bathymetry with the existing runway and the finite element mesh used in the ARTEMIS wave model. Right image shows a 3D illustration of bathymetry looking to the NW. 82 Figure 5-3: Left image shows the Lyall Bay bathymetry with the proposed airport extension added and finite element grid used in the ARTEMIS wave model. Right image shows a 3D illustration of bathymetry focusing on the proposed airport extension looking to the NW. 83 Figure 5-4: ARTEMIS model predicted surface water level as a result of a 1.5 m wave at the boundary with a wave period of 12 seconds (LEFT). Google Earth image of Lyall Bay taken on 24 July 2014 for a wave height of 1–1.5m and a period ~10 seconds (RIGHT). 85

6 Wellington Airport Runway Extension

Figure 5-5: Lyall Bay bathymetry plots showing the extraction transects (T1-T8) and point extraction sites (P1-P8) for analysis. 86 Figure 5-6: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when the boundary is forced with a boundary incident wave height of 1 m (10-sec period). 98 Figure 5-7: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when |the boundary is forced with a boundary incident wave height of 1.5 m (10-sec period). 99 Figure 5-8: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when the boundary is forced with a boundary incident wave height of 2 m (10-sec period). 99 Figure 5-9: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when the boundary is forced with a boundary incident wave height of 3 m (10-sec period). 100 Figure 5-10: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when the boundary is forced with a boundary incident wave height of 4 m (10-sec period). 100 Figure 5-11: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when the boundary is forced with a boundary incident wave height of 5 m (10-sec period). 101 Figure 5-12: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 1 m (10-sec period). 101 Figure 5-13: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 1.5 m (10-sec period). 102 Figure 5-14: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 2 m (10-sec period). 102 Figure 5-15: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 3 m (10-sec period). 103 Figure 5-16: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 4 m (10-sec period). 103 Figure 5-17: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 5 m (10-sec period). 104 Figure 5-18: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 1 m (10-sec-period). 104 Figure 5-19: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 1.5 m (10-sec period). 105

Wellington Airport Runway Extension

Figure 5-20: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 2 m (10-sec period). 105 Figure 5-21: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 3 m (10-sec period). 106 Figure 5-22: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 4 m (10-sec period). 106 Figure 5-23: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 5 m (10-sec period). 107 Figure 5-24: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 1 m (10-sec period). 107 Figure 5-25: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 1.5 m (10-sec period). 108 Figure 5-26: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 2 m (10-sec period). 108 Figure 5-27: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 3 m (10-sec period). 109 Figure 5-28: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 4 m (10-sec period). 109 Figure 5-29: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 5 m (10-sec period). 110 Figure 5-30: Free surface elevation spatial plots of Lyall Bay for a 1.5 m incident wave with a period of 8 seconds. 113 Figure 5-31: Wave height spatial plots of Lyall Bay for a 1.5 m incident wave with a period of 8 seconds. 113 Figure 5-32: Spatial plots of Lyall Bay for the predicted change in wave height as consequence of the airport development, given a boundary incident wave height of 1.5 m with a wave period of 8 seconds. Positive values (grey) represent an increase in wave height, negative values (green) are a decrease in wave height. 114 Figure 5-33: Free surface elevation spatial plots of Lyall Bay for a 1.5 m incident wave with a period of 12 seconds. 115 Figure 5-34: Wave height spatial plots of Lyall Bay for a 1.5 m incident wave with a period of 12 seconds. 115 Figure 5-35: Spatial plots of Lyall Bay for the predicted change in wave height as consequence of the airport development, given a boundary incident wave height of 1.5 m with a wave period of 12 seconds. Positive values (grey) represent an increase in wave height, negative values (green) are a decrease in wave height. 116

8 Wellington Airport Runway Extension

Figure 5-36: Free surface spatial plots of Lyall Bay for a 3 m incident wave with a period of 8 seconds 117 Figure 5-37: Wave height spatial plots of Lyall Bay for a 3 m incident wave with a period of 8 seconds. 117 Figure 5-38: Spatial plots of Lyall Bay for the predicted change in wave height as consequence of the airport development, given a boundary incident wave height of 3 m and a wave period of 8 seconds. Positive values (grey) represent an increase in wave height, negative values (green) are a decrease in wave height. 118 Figure 5-39: Free surface spatial plots of Lyall Bay for a 3 m incident wave with a period of 12 seconds. 119 Figure 5-40: Wave height spatial plots of Lyall Bay for a 3 m incident wave with a period of 12 seconds. 119 Figure 5-41: Spatial plots of Lyall Bay for the predicted change in wave height as consequence of the airport development, given a boundary incident wave height of 3 m with a wave period of 12 seconds. Positive values (grey) represent an increase in wave height, negative values (green) are a decrease in wave height. 120 Figure 6-1: Generalised wind forcing used for turbid plume simulations, consisting of a sequence of southerly, northerly and another southerly individually spanning 2.5 days peaking at 22 m/s (43 knots). 123 Figure 6-2: Schematics of a NW decant discharge location (top) and a SW discharge location (bottom) in the perimeter rock dyke and options for associated fill operations. 124 Figure 6-3: Composite surface-layer maximum SSC over a 30-day simulation (top) and averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 1 kg/s sediment flux discharged from site D1 (NW corner). 129 Figure 6-4: Composite surface-layer maximum SSC over a 30-day simulation (top) and averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 2 kg/s sediment flux discharged from site D1 (NW corner). 130 Figure 6-5: Composite surface-layer maximum SSC over a 30-day simulation (top) and averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 1 kg/s sediment flux discharged from site D2 (SW corner). 131 Figure 6-6: Composite surface-layer maximum SSC over a 30-day simulation (top) and averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 2 kg/s sediment flux discharged from site D2 (SW corner). 132 Figure 6-7: Composite surface-layer maximum SSC over a 30-day simulation (top) & averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 1 kg/s sediment flux discharged from site D3 (mid-south end). 133 Figure 6-8: Composite surface-layer maximum SSC over a 30-day simulation (top) & averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 2 kg/s sediment flux discharged from site D3 (mid-south end). 134

Wellington Airport Runway Extension

Figure 6-9: Predicted instantaneous plume snapshots of SSC at 24, 48, 72 and 96 hours after a 1 kg/s decant slug release for locations D1 (left panels) and D2 (right panels). 136 Figure 6-10: Predicted instantaneous plume snapshots of SSC at 24, 48, 72 and 96 hours after a 2 kg/s decant slug release for locations D1 (left panels) and D2 (right panels). 137 Figure 6-11: Predicted instantaneous plume snapshots from location D3 of SSC at 24, 48, 72 and 96 hours after a 1 kg/s (left panels) and 2 kg/s (right panels) decant slug release. 138

10 Wellington Airport Runway Extension

Executive summary Wellington International Airport operates in a constrained area in the coastal suburb of Rongotai. The Airport has a single runway of 1945 m Take Off Runway Available (TORA).

Wellington International Airport Limited (WIAL) propose to extend the runway to 2300 m TORA to further enable long-haul flights, particularly Code E aircraft (e.g., Boeing 777 or Airbus A330) and the ensuing economic benefits for the Wellington region.

NIWA were engaged by WIAL in August 2014 to provide environmental services in relation to coastal physical processes, sediment contamination and marine ecology in the Lyall Bay area.

WIAL engaged AECOM (formerly URS Ltd) to provide engineering services in the development of engineering design options and the final specimen design to progress through the environmental assessment and consenting process. Engineering services incorporates aspects of coastal hazard risk (wave overtopping and storm-tide) and marine geology into the final specimen design. This information is covered in the Engineering Report (AECOM, 2015a).

After initial engineering and economic feasibility studies considering both Evans Bay and Lyall Bay, the runway is to be extended south into Lyall Bay by approximately 350 m.

Coastal environmental assessments are required to support applications for coastal permits and consents for the airport runway upgrade in the coastal marine area (CMA) and adjacent coastal environment.

This technical report covers the field and modelling investigations focused on coastal physical processes (hydrodynamic, wave and sediment dynamic processes), with modelling simulations covering both the present runway configuration and the predicted changes arising from the extended runway further into Lyall Bay. The modelling approach was tailored to supporting an assessment of relative changes arising from the proposed runway extension compared to the existing environment (which includes the existing runway revetment). This report also forms the detailed background technical basis to the findings of an AEE report on assessing the effects of the airport runway extension on coastal physical processes (Bell, 2016), which also considers the effects of climate change.

Companion NIWA technical reports are MacDiarmid et al. (2015) covering a description of the marine ecology and Depree et al. (2016) concerning marine sediments and contaminant levels in Lyall Bay.

Field investigations in Lyall Bay Field investigations were designed to provide information on the existing environment within Lyall Bay and for the verification of numerical hydrodynamic and wave models. A multi-beam sounding survey was undertaken in January 2014 to fill gaps in the bathymetry dataset of Lyall Bay, and a seamless digital elevation model of the seabed was produced (Mackay & Mitchell, 2014).

A deployment of current meters and wave sensors in Lyall Bay from 18 August to 9 October 2014 measured tidal water levels and currents, wind-driven currents, and waves. This included capturing the response of Lyall Bay to a strong southerly gale on 22-23 September, with winds peaking at 80 km/h that produced significant wave heights up to 4.7 m (wave period = 11 seconds). The main lunar tidal current (M2) within the inner Bay was <0.006 m/s (and not detectable above the noise threshold in the measurements), but wind-generated currents in the strong southerlies reached 0.3 m/s or more.

Wellington Airport Runway Extension 11

Update: a further deployment of surface-following wave gauges was undertaken at three sites within Lyall Bay in late 2015 to provide additional high-resolution wave data for DHI to assist with assessing mitigation options to address surfing quality. This report does not include these recent data but are covered in DHI (2016).

Hydrodynamic and sediment-transport modelling The hydrodynamic modelling investigations were designed to describe the existing tidal and wind- driven circulation and sediment transport in Lyall Bay and then establish the potential impact of the proposed Wellington Airport runway extension on hydrodynamics and sediment transport inside the Bay.

The model predicted, for pre-runway extension bathymetry and coastline boundary, that the mean tidal or residual circulation is weak through inner Lyall Bay. The model predicted only localised changes in tidal currents due to construction of a runway extension. These changes were predicted to the south of where the present runway extends and appear as a weakening of time-averaged residual flow either side of where the new runway extension would extend. However, these are near- field effects around the runway extension and have no effect on the overall time-averaged residual circulation and flushing through the rest of the wider bay. The residual currents will also be unaffected in the outer Bay by the runway extension and therefore will not affect the dispersion of treated wastewater from the marine outfall discharging from the Moa Point wastewater treatment plant.

Simulations of wind-driven circulation at wind-event timescales within Lyall Bay showed that a southerly wind drives northward current flows on both sides of the Bay. On the west side of the bay, this northward flow is forced to turn clockwise in the head of the inner bay and return on the eastern side of the inner bay. This wind-driven current then converges with the northward current flow on the eastern side of the bay into the cove off Moa Point beach, curving round to the west towards the spur groyne. This effectively forms a clockwise eddy at the head of the Bay and a counter-clockwise eddy in the middle of the Bay. The local re-steering of the northward current flow along the eastern Bay would also produce slightly faster velocities across the end of the extended runway than presently occurs at the same location 350–400 m off the present runway terminus, as would be expected by the presence of the reclamation in what is presently open water.

In contrast under northerly wind conditions, southward current flows are generated along the eastern and western periphery of the Bay, and exiting Lyall Bay, with a return northward flow through the centre of the Bay up to Lyall Bay Beach, where it bifurcates. The hydrodynamic model predicts that the proposed runway extension produces a negligible alteration of the northerly-wind driven circulation but will produce local, inconsequential, changes in wind-driven circulation during southerlies in the vicinity of the runway extension, deviating the wind-driven flow path around the perimeter of the extension and reducing wind-driven current speeds.

Sediment-transport modelling predicted only small net changes in sea-bed heights over much of Lyall Bay and particularly within the inner Lyall Bay, following the construction of the runway extension. The main changes in sea-bed height were predicted locally around the extended runway, where the existing sea-bed topography would need to adjust to the presence of the rock dyke and the associated wave interaction with the structure. Sand was predicted to mostly erode beside the dyke and to deposit further offshore, especially to the south, to form a new equilibrium profile.

12 Wellington Airport Runway Extension

The model predicts that there may be localised deposition on the southwest and southeast corners of the rock dyke, which is caused by the reduction of bed shear stress for sediment transport in the lee of the dyke during southerly winds, compared to the present situation. This is not to say that over much longer periods than those simulated, that varying seasonal weather conditions and swell frequency could cause the simulated sand deposits next to the runway extension to re-mobilise and move, given the high wave exposure of this area.

Wave modelling of general wave patterns in Lyall Bay Model simulations of waves was undertaken for the wider south Wellington coast between Baring Head (Ōrua-pouanui) and Sinclair Head (Te Rimurapa), where the spectral wave model SWAN was used to simulate spatial variations in the wave climate for the existing environment. The SWAN model was verified against the Baring Head wavebuoy dataset and the Lyall Bay measurements from the 2014 field investigations.

Waves entering Lyall Bay, predominantly from due south, largely maintain their energy through the deeper water of the outer Bay, but refract shoreward and lose energy in the shallower water on both sides of the outer Bay, and in the inner Bay west of the airport. There are areas of increased wave height produced by shoaling over two submerged reefs (the submerged southward extension of the headland at Te Raekaihau, and another reef east of the headland between Te Raekaihau and Waitaha Cove). The north-south alignment of these reefs will tend to focus waves from the south, causing wave energy to increase over the reefs.

Locations inside Lyall Bay tend to have minimal wind sea during northerly winds, so any southerly swell present will dominate the mean period of waves within the Bay. The deeper Lyall Bay sites are dominated by southerly waves, but in shallower waters on both east and west coasts of the Bay, the dominant wave direction shifts towards a locally shore-normal direction through refraction processes.

The high exposure to the long southerly fetch means wave heights along south Wellington coast including Lyall Bay can reach extremely high wave heights. The largest recorded event at the Baring Head wave buoy was on 20 June 2013, when the significant wave height reached 9.5 m and where the maximum wave height of 15 m is likely to be the largest wave height experienced in the area in the last 50 years, including Cyclone Gisele (the Wahine storm). On 22–23 September 2014 during the 2014 field-deployment period, a southerly storm event with 80 km/h winds produced significant wave heights over 6 m at the Baring Head buoy and at the Dobie wave gauge site in outer Lyall Bay, and just under 5 m at the nearshore wave gauge near the end of the existing runway. The two Lyall Bay deployments in tandem with the wavebuoy indicate that extreme wave heights off the end of the runway will be partially attenuated compared with the more exposed area off Baring Head where the wave buoy operates and in outer Lyall Bay.

The high-resolution ARTEMIS wave model, using regular1 (sinusoidal) wave trains, was observed to reproduce the spatial wave-crest patterns in Lyall Bay, when compared to satellite images. The model is therefore replicating the refraction and diffraction of the incident wave-train, within the inner Bay. Results show that the proposed runway extension will affect wave heights, predominantly on the eastern side of Lyall Bay near the west and east of the extended runway revetment and along and off the western side of the existing runway revetment i.e., just north of the spur groyne.

1 Pertains to the use of a repeating regular sinusoidal wave time series used for input on the offshore boundary of wave models such as ARTEMIS (otherwise known as monochromatic waves).

Wellington Airport Runway Extension 13

Firstly, considering the effect of the runway extension on wind waves. In the lee of the spur groyne, for example 1.5 m incident wind waves (period=8 seconds) from Cook Strait would be attenuated by a further 0.2 to 0.6 m in a zone extending approximately 250 m west of the existing runway revetment. The greatest wave-height reduction in this inner sector would be 80–100 m northwest of the spur groyne, with an increase in the height of waves immediately behind the spur groyne, caused by diffraction. In The Corner (the nearshore wave shoaling and surfing area adjacent to the existing runway revetment) the wave-height reduction would be somewhat less at 0.2 to 0.4 m.

The largest reduction in wave height would occur adjacent to the west side of the extended revetment, due to both the friction of the dyke causing wave energy dissipation, and to a subtle wave-sheltering effect as the runway extension alignment of ~183S is slightly west of the predominant wave approach direction from due south.

For swell, simulations of a 1.5 m incident wave height (period=12 seconds) showed a similar pattern of wave height change to the 8-second waves, after the runway is extended. The main differences would be a somewhat smaller reduction in wave height as swell peels along the western side of the extended revetment and a slightly wider shadow zone in eastern part of Lyall Bay with reductions in wave height of 0.3 to 0.6 m. The eastern-most swath of propagating swell in towards The Corner, parallel with and 50 m off the existing runway revetment, is predicted to be slightly narrower and around 0.2–0.4 m lower in wave height than for the existing situation, for a good-quality surfing combination of 1.5 m incident swell height and 12 seconds period.

The ARTEMIS wave model also predicts an increase in wave heights: i) adjacent to and in the immediate lee of the spur groyne and submerged rock outcrop; ii) slight increases in a narrow central part of the inner Bay of the main Beach; and iii) in a patch close to the beach towards the eastern end of Lyall Bay Beach near The Corner. This latter area arises due to the reduced wave height offshore from this part of the Beach, with waves able to shoal further inshore than is the case for the existing situation, given the same wave height and period.

Waves reaching the central part of Lyall Bay Beach are likely to be slightly increased in height by the runway extension, otherwise the changes will be negligible elsewhere in the Bay, including the western side, based on simulations of regular waves.

The cove to the east of the runway (adjacent to Moa Point Road) will also experience a reduction in wave height as a result of the proposed extension. This reduction in wave height is more appreciable inshore and in the corner immediately adjacent to the existing runway. There is also likely to be more resonant standing wave, or wave-reflection, behaviour within the more-enclosed cove after the extension has been constructed.

The ARTEMIS results indicate the predicted scale of general changes in waves throughout the Bay, relative to the existing situation, based on regular sinusoidal wave sets. More detailed high- resolution wave modelling was undertaken by DHI, using irregular measured waves propagating through the Entrance for occasions when surfing quality was good, to further investigate the effect of the runway extension on wave heights and characteristics (e.g., wave steepness, wave peakiness). These aspects are important considerations for assessing the effects on surfing quality and safety of recreational users, are described in the assessment report by DHI (2016). The DHI results also indicate that the ARTEMIS model overestimates the reduction of wave heights behind the spur breakwater as a result of the runway extension (Dr S. Mortensen, DHI, pers. com.).

14 Wellington Airport Runway Extension

Turbid-plume dispersion modelling The purpose of suspended-sediment plume modelling was to determine the spatial extent of turbid plumes that might occur during construction of the airport runway extension from the discharge of decant water or stormwater containing silts and clays. The sediment model simulates suspended- sediment concentration (SSC), which can be readily related to turbidity measures (e.g., NTU) but does not directly simulate the visual clarity of water. This linkage and thresholds for ecological effects are covered in the AEE report by James et al. (2016).

On completion of the dyke wall that would form the perimeter of the extended runway, de-watering discharges during the embankment fill operations have the potential to introduce high-turbidity plumes into the receiving waters of Lyall Bay if a fine-sediment fraction is present in the fill material. The main source of turbidity in the water column would be from any clays, muds, or silts present in the fill material, even if they are only a small percentage of material by volume.

Three alternative discharge locations in the dyke walls were assessed in the turbid-plume model simulations – two on the western side of the airport runway extension at the NW and SW corners of the dyke and one in the middle of the southern wall of the dyke. The discharge was introduced into the top layer of the 5 depth-layers in the dispersion model, to cover the case of the discharge being mainly freshwater from rainfall runoff within the fill area.

As with most reclamation construction projects in the coastal marine area, the type of fill material (especially the proportion of fine silts and clays) and the exact construction methodology and erosion and sediment control measures to be deployed are not prescribed in detail – but rather the selected contractor would design the construction plan, fill flow and sequencing generally in accordance with the construction methodology. Therefore at this stage it is not possible to quantify ahead of time, the sequencing of the sediment discharge rate of fine-grained material (e.g., silts, clays) in the decant discharge(s). This situation is usually managed by way of an adaptive management approach, setting an environmental turbidity limit not to be exceeded, after allowing for reasonable mixing2, while monitoring the receiving waters at a few locations during the construction phase. Two SSC limits of 25 and 40 mg/L were considered to safeguard different ecological effects in the Marine Ecology Assessment Report (James et al. 2016). Sediment-discharge rates were simulated to find the rates at which these two different ecological SSC limits of 25 and 40 mg/L are not exceeded, after reasonable mixing of the discharge.

Based on Delft3D-SED plume simulations using two sediment discharges of medium silt at rates of 1 kg/s and 2 kg/s, maximum suspended-sediment concentrations (SSC) predicted were in the range 11–16 mg/L and 22–36 mg/L (above background) respectively for those rates across the three discharge points modelled at a distance of ~150 m. An upper SSC environmental limit, to apply after reasonable mixing, is proposed to be set to 25 mg/L (or equivalent NTU magnitude), derived from considering the limit at which sea-bird foraging is curtailed (James et al. 2016). This implies the predicted fine-sediment discharges of medium-silt at a rate of 2 kg/s for two of the three discharge locations modelled would be just under the 25 mg/L limit at ~150 m, but not the NW discharge nearest the spur breakwater, which would reach a maximum of 36 mg/L in the shallower water and more constrained receiving waters in that area.

2 “reasonable mixing” in the RMA (e.g., s107) implies the existence of a zone in which the compliance standard need not be met – but important to distinguish between the near-field mixing zone, the point of complete mixing, and the non-compliance zone (MfE, Resource Management Ideas, No. 10, Aug 1994).

Wellington Airport Runway Extension 15

Modelling results show that due to the slow tidal currents during fair weather, the turbid plume exhibits small footprints (relative to higher winds) with a radial dispersive pattern from the discharge point. In conditions with high winds (northerly or southerly), currents are higher, which increases both the lateral spread of plumes and reduces the SSC. Consequently, the highest concentrations in the near to intermediate areas (a few 100s of metres from the discharge) will occur in calm weather, while in the far-field areas of the Bay, the highest concentrations, albeit only 1–5 mg/L above ambient conditions, will occur during high winds when the entire inner Bay is more turbid anyway from wave activity and possibly stormwater.

16 Wellington Airport Runway Extension

1 Introduction and Background Wellington International Airport operates in a constrained area in the coastal suburb of Rongotai (Figure 1-1). The Airport has a single runway of 1945 m Take Off Runway Available (TORA).

NIWA were engaged by WIAL in August 2014 to provide environmental services in relation to coastal physical processes, sediment contamination and marine ecology in the Lyall Bay area.

WIAL engaged AECOM (formerly URS Ltd.) to provide engineering services in the development of engineering design options and the final specimen design to progress through the environmental assessment and consenting process. Engineering services incorporates aspects of coastal-hazard risk (wave overtopping and storm-tide) and marine geology into the final specimen design. This information is covered in the Engineering Report (AECOM, 2015a).

After initial engineering and economic feasibility studies considering both Evans Bay and Lyall Bay, the additional 350 m runway extension is proposed to be to the south into Lyall Bay.

Coastal environmental assessments are required primarily to support applications for coastal permits and consents for the Airport runway upgrade in the coastal marine area (CMA) and adjacent coastal environment.

Companion NIWA technical reports are MacDiarmid et al. (2015) on marine ecology and Depree et al. (2016) on marine sediments and contaminant levels.

Figure 1-1: Aerial image of Wellington International Airport and runway. LINZ (2003-04).

Wellington Airport Runway Extension 17

2 Hydrodynamic field programme: methods and results (Lyall Bay) Given the importance in quantifying the changes in waves from the proposed Wellington Airport runway extension, a field deployment of instruments was undertaken in Lyall Bay over a 7-week period from August to October 2014 to provide observational data to validate both wave and tidal models.

This Section describes the instrumentation used and the analyses of the field datasets, including wind.

2.1 Instrumentation

2.1.1 Currents Current velocities in the Lyall Bay were measured using Acoustic Doppler Current Profilers (ADCP). ADCPs measure currents in “bins” throughout the water column, thus providing the user with a velocity profile over the entire water column.

ADCPs were deployed on the sea bed facing the water surface, which is known as a bottom-mounted upward-looking configuration. ADCP acoustic transducers were nominally positioned 0.4 m above the sea bed.

2.1.2 Waves 3 Significant wave height (Hs), mean spectral period (Tm) and, where possible, peak direction of wave propagation (Dp) were measured using an ADCP and a DOBIE Wave Gauge (DWG).

The ADCP calculates the wave parameters Hs, Tm and Dp from the current velocity measured in the bins closest to the sea surface. For a comprehensive overview of how the wave parameters are derived by the ADCP, the reader is directed to the Waves primer technical report4 published by the ADCP manufacturer (Teledyne RD Instruments Inc.). All wave parameters were calculated using the software supplied by the ADCP manufacturer.

In contrast to the ADCP, which calculates wave parameters from current velocities measured close to the sea surface, the DWG calculates wave parameters from pressure fluctuations measured at the height of the instrument. Pressure fluctuations beneath waves attenuate with depth through the water column at a rate that depends on the wave frequency (where wave frequency is the inverse of wave period). The rate of depth attenuation is higher for high-frequency waves than it is for low- frequency waves (e.g., swell). Therefore, not only will the total variance of the pressure spectrum at depth be less than the total variance of the pressure spectrum at the sea surface, but also the shape of the spectrum will be different, due to the relatively greater attenuation of the high-frequency

(short-period) components. Hs is calculated from the total variance of the pressure spectrum, and must be adjusted for the depth-attenuation of pressure. Tm is calculated from the shape of the pressure spectrum, and can either be adjusted for depth attenuation so that it accurately represents the period of the waves at the surface, or it can be left un-adjusted, in which case it can be interpreted as the mean spectral period of the wave-orbital motions at the level of the measurement. Tm adjusted for depth attenuation will always be smaller than Tm at the level of the measurement, because the higher-frequency components of the wave spectrum are attenuated more than the lower-frequency components. Care is required when comparing wave parameters

3 Defined as the average of the top 33% (1/3) of wave heights over the measurement cycle 4 http://www.rdinstruments.com/pdfs/waves_primer.pdf

18 Wellington Airport Runway Extension

from difference sources (e.g., field measurements and a wave model) to ensure that like terms are being compared. No information about wave direction can be derived from pressure measured at a single location.

2.1.3 Winds Winds prior to 1 January 2013 were obtained from the Automatic Weather Station at Wellington airport operated by the NZ Met Service (NIWA Climate Database; Agent No. 3455). The station at Wellington airport is located 4 m above mean sea level. Winds for Lyall Bay after 1 January 2013 were obtained from NIWA’s meteorological forecasting model (EcoConnect) and the Baring Head weather station that NIWA operates as access to NZ Met Service data is now restricted.

2.2 Results

2.2.1 Instrument locations and deployment information Figure 2-1 shows the locations occupied during the instrument deployment in Lyall Bay. The details of what instruments were deployed at each of the sites and over which period are in shown in Table 2-1.

Additional deployment information relating to the sampling scheme for each instrument is shown below in Table 2-2.

Table 2-1: Locations (WGS-84) and 2014 deployment information for the Lyall Bay instruments.

Location Latitude Longitude Instrument Deployment Depth Mean (S) (E) (parameter measured) Period below water Water depth Surface (m) (m)

Site 1 41˚ 20.197 174˚ 48.140 ADCP (currents and waves) 18-Aug 14:00 to 11.0 11.0 (inshore) 09-Oct 09:00

Site 2 41˚ 20.932 174˚ 47.933 ADCP (currents and waves)5 N/A (offshore) DWG (waves) 04-Sep 11:00 to 7.6 19.0 09-Oct 09:00

5 The ADCP at Site 2 was lost to a storm during the deployment. No current data is available at Site 2.Wave data at Site 2 comes from the DWG.

Wellington Airport Runway Extension 19

Table 2-2: Additional deployment information. List of symbols: Ibb = interval between bursts and BD = burst duration.

Location Instrument Parameter Ibb BD Sampling (mins) (mins) rate (Hz)

Site 1 ADCP Currents 2.5 2.5 0.66

Waves 30 20 2

Site 2 DWG Waves 15 3.41 5

20 Wellington Airport Runway Extension

Figure 2-1: Site map showing oceanographic instrument locations overlying the seabed bathymetry to WVD-53. The black polygon outlines the approximate area covered by the proposed runway extension.

Wellington Airport Runway Extension 21

2.2.2 Winds The long-term wind climate for the Lyall Bay area was derived from hourly observations at Wellington Aero Automatic Weather Station (operated by the NZ Met Service) covering a 28-year period from 1 January 1985 to 1 January 2013. The long-term wind rose is shown in Figure 2-2.

The mean wind speed (for 10-min averages) over the 28-year period to the end of 2012 was 7.2 m/s, with a maximum of nearly 30 m/s (108 km/hr) which was reached on the on the 15 May 1985 when the wind blew from the SSW (220˚). A more recent southerly storm recorded a similar maximum wind speed of 28 m/s (101 km/hr) on 20 June 2013.

Long-term wind directions exhibit a bi-modal distribution, dominated by winds from a northerly and southerly direction.

Figure 2-2: Wind rose for winds measured at Wellington Aero over a period of 28-years (January 1985 to December 2012). Meteorological convention is used in expressing the direction that the wind "blows from".

The wind rose for the 2014 Lyall Bay deployment period (18 September to 9 October 2014) is shown in Figure 2-3. The mean wind speed over the deployment period was 6.9 m/s, which is similar to the mean wind speed from the 28-year record. The maximum wind speed during the deployment period was 22.1 m/s (80 km/hr), which was reached on the 22 September 2014 when the wind blew from the SSW (200˚). Overall, the distribution of wind speeds and directions during the field deployment was similar to the long-term distribution (cf. Figure 2-2 and Figure 2-3), with somewhat more frequent southerlies (and hence higher wave conditions). Therefore, the range of wind conditions experienced during the 2014 field programme was representative of the long-term wind climate, plus included a strong southerly gale.

22 Wellington Airport Runway Extension

Figure 2-3: Wind rose for winds over the deployment period (18-Aug to 09-Oct 2014).Meteorological convention is used in expressing the direction that the wind "blows from".

2.2.3 Currents

Site 1 (inshore) The results the ADCP deployment at Site 1 are shown in this section in the form of:

. Time series plots of current speed and direction at various heights above the sea bed, water depth, and wind speed and direction.

. Scatter plots of current velocity components u (east–west component, with east positive) and v (north–south, with north positive).

. Tidal analysis of the u and v velocity components.

. Progressive vector plots for the cumulative current-velocity run at various elevations above the sea bed. Although net drift can be inferred from these plots, they cannot account for spatial changes in currents (speed and direction) elsewhere in the Bay away from the actual deployment site, and therefore need to be interpreted with caution.

Time series plots for Site 1 are shown in Figure 2-4.

Wellington Airport Runway Extension 23

Figure 2-4: Current and wind time series for deployment at Site 1. Panels: (A) wind speed, (B) wind direction (in meteorological convention "blowing from"), (C) water depth, (D) current speeds at three elevations above the bed, and (E) current direction (in oceanographic convention "flowing to"). List of symbols: mab = metres above bed.

24 Wellington Airport Runway Extension

Figure 2-4C shows that tide ranges at Site 1 were small, typically around 1 m. Figure 2-4D shows that the currents at Site 1 are substantially affected by wind conditions. Larger current speeds between 0.35–0.4 m/s were measured during periods of strong southerly winds. During calmer periods, Figure 2-4D shows that the currents at Site 1 were typically less than 0.1 m/s. Nearer the sea-bed, the maximum recorded current speed of 0.32 m/s was measured on 22-Sept 2014 at 16:17 hrs during the strong gale. In the 12-hour period prior, wind speeds ranged between 12.4 and 22.0 m/s (mean 18.4 m/s) with direction from the S-SSE. Higher in the water column, at 9.0 metres above bed (mab), the maximum recorded current speed of 0.37 m/s was measured at same time as the maximum near- bed current.

Scatter plots of the current-velocity data from the three elevations above the sea bed are shown in Figure 2-5. Here, the east–west velocity component (u) is plotted against the north–south velocity component (v). In addition to the velocity data, Figure 2-5 displays the results of a principal component analysis (PCA) of the current data. PCA is an orthogonal transformation in which the u and v velocity components are oriented (rotated) such that the variance (or energy) along the principal (major) axis is a maximum, and consequently the variance along the minor axis is a minimum. The usefulness of PCA is that the major axis defines the overall orientation of the flow over the deployment period, although if the currents swing around as in this case, it should be interpreted carefully . In Figure 2-5, the major and minor axes are displayed as the thick and thin lines respectively (both lines span ± 1 standard deviation). The results of the PCA are also summarised in Table 2-3. Currents measured at Site 1 had a principal orientation of SE–NW, but with considerable variability, especially nearer the surface, caused by various wind-generated responses.

Postscript: Following the wave modelling undertaken by NIWA (Sections 4 and 5 in this Report) and complementary wave modelling by DHI, it has become clear that the directions measured by the compass in the ADCP deployed at Site 1 (and after conversion using the magnetic declination of 22.4), appear to have a bias of approximately 8 to the east of south, when compared with satellite images of waves around that site propagating from due south. The modelling simulations match closely with these satellite images. Directional current and wave data presented below report the data as processed without the 8 bias included, but subsequent interpretation of the directional wave and current data should factor in this bias.

Taking into account the directional bias, the principal orientations shown in Figure 2-5 and listed in Table 2-3 would be slightly more to the SSE.

Wellington Airport Runway Extension 25

Figure 2-5: Scatter plots of currents at three elevations above the bed at Site 1. Panels: (A) data from 2.0 mab, (B) data from 5.5 mab, and (C) data from 9.0 mab. Note the u and v velocity components relate to the east-west and north-south directions, respectively. The thick line is the major axis and the thin line is the minor axis (both lines span ± standard deviation).

Table 2-3: Summary of principal component analysis for currents at Site 1. These directions exclude the addition of the apparent 8 bias.

Depth Major axis orientation Major axis std. dev. Minor axis std. dev. (mab) (˚ True north) (m/s) (m/s)

2.0 144 (SE) ±0.043 ±0.018

5.5 138 (SE) ±0.048 ±0.030

9.0 125 (SE) ±0.046 ±0.037

A tidal analysis of the ~52-day record of near-bed currents at 2.0 mab at Site 1 was conducted. This consisted of fitting a number of tidal constituents to the data by adjusting amplitudes and phases of the constituents such that the least-squares difference between fitted and measured data was minimised. The results of the tidal analysis did not yield any significant tidal constituents over and above the “noise” threshold in the data, although the main lunar M2 peak tidal current (12.4 hr

26 Wellington Airport Runway Extension

period) of 0.005 m/s was just under the noise threshold. These results show that tidal currents at Site 1 within inner Lyall Bay are very weak.

Net-current drift plots show the time sequence of currents at the measurement site. Figure 2-6 shows the PVD at Site 1 at three elevations above the bed and the depth-averaged trace. At Site 1 the three elevations have quite different net drift directions. The period of southerly winds that occurred on 21–24 Aug resulted in a SSE drift near the sea bed, while closer to the surface the drift direction was to towards the E-ESE. Another period of southerly winds that occurred on 22–23 Sept resulted in different set of drift directions, with a WNW drift near the sea bed, while closer to the surface the drift direction was to towards the N-NNE. The different drift patterns that occurred during the two southerly wind events shows that wind-driven current circulation patterns are complex and that cause and effect is not simply determined by the general wind approach direction. Currents at Site 1 will be influenced by the underwater rock outcrop and nearby spur groyne (see local bathymetry in Figure 2-1), along with circulation throughout the bay as a whole, and by processes other than wind, such as wave-induced currents particularly around the submerged rock outcrop. The apparent directional bias of 8 would also slightly alter the residual circulation direction.

The overall net drift speeds and directions at Site 1 for the ~7 week deployment are listed in Table 2-4. The net depth-averaged current drift over the nearly 52-day period was only 7 mm/s or 630 m per day. This slow drift results from the very low tidal currents or low-to-moderate wind-generated currents from opposing wind directions (north and south) punctuated by occasional strong gales, which for the 2014 field period were two southerly events (2). The PVD plots and net-residual drifts inferred from the ADCP data at Site 1 are difficult to interpret because in addition to tidal dynamics they also include the effect of wind driven circulation driven by the two gales that occurred during the instrument deployment. This complicated interaction in addition to the localised bathymetric steering of flows at Site 1 is therefore potentially not a true indication of the bay wide net residual drift direction as suggested by a PVD analysis.

Wellington Airport Runway Extension 27

Figure 2-6: Progressive current drift at Site 1 at three elevations above the sea bed (bottom-left panel for near the surface) and the depth-average (bottom-right panel). Yellow dot marks the starting position and changes in line thicknesses mark different concurrent time periods of wind events referred to in the text.

28 Wellington Airport Runway Extension

Table 2-4: Summary of the overall current drift at Site 1 over the deployment period. Mean depth at site is 11 m

Depth Net drift speed Net drift speed Mean drift direction ADCP Bin No. (mab) (m/s) (km/day) (˚ True north)

1 (near bed) 2.0 0.009 0.777 270

8 (mid-depth) 5.5 0.011 0.918 326

15 (near surface) 9.0 0.004 0.343 179

Depth-average 2–9 0.007 0.629 311 (1–15)

2.2.4 Waves Measurements of waves are shown in this section in the form of:

. Time series plots of significant wave height (Hs), mean spectral period (Tm), peak 6 direction of wave propagation (Dp) , wind speed and wind direction.

. Wave-rose diagrams, which illustrate the relationship between wave height and direction.

Site 1 Time series of the wave parameters measured by the ADCP at Site 1 are shown in Figure 2-7. The largest Hs recorded at Site 1 was 4.7 m, which had a Tm of 11.10 seconds and which arrived from the south. This maximum Hs was recorded on 22-Sept 2014 at 17:30 during the strong southerly gale, when the wind speed reached ~ 22 m/s (~80 km/hr).

6 From the ADCP deployment at Site 1 only.

Wellington Airport Runway Extension 29

Figure 2-7: Wave parameters measured at Site 1. Panels: (A) wind speed, (B) wind direction (in meteorological convention "blowing from"), (C) significant wave height (Hs), (D) mean spectral period (Tm) at the sea surface, and (E) peak wave direction (Dp) (wave direction is the direction that the waves are coming from). Note: an apparent 8 bias to the east of south for the ADCP directions has not been included.

30 Wellington Airport Runway Extension

Figure 2-8 shows that the wave directions at Site 1 are dominated by waves directly from the south. Over 90% of the waves with wave heights greater than 1 m approach from this direction. Note: an apparent 8 bias to the east of south for the ADCP directions has not been included in the compass rose, which would reduce the small SSE arm and increase the SSW arm in Figure 2-8.

Figure 2-8: Wave rose for measurements at Site 1. (Wave direction is the direction that the waves come from). Note: an apparent 8 bias to the east of south for the ADCP directions has not been included.

Site 2 Time series of the wave parameters derived from the DWG measurements at Site 2 are shown in Figure 2-9.

The mean spectral wave period (Tm) is reported here as the DWG “sees it”, that is, it is not adjusted for depth attenuation (Section 2.1.2). In this case, Tm can be interpreted as the mean spectral period 7 of the wave-orbital motions at the level of the DWG . The estimates of Hs have however been adjusted for the depth-attenuation of pressure so it is more representative of surface waves.

The largest Hs recorded at Site 2 was 6.1 m, which had a Tm of 10.9 seconds. This maximum Hs was also recorded on 22-Sept 2014 (at 17:00) during the strong southerly gale, when the wind speed reached ~ 22 m/s (~80 km/hr).

In comparison, the 22-Sept southerly gale produced significant wave heights over 6 m (6.5 m maximum – Figure 4-2) at the Baring Head wavebuoy (site 3), just slightly higher than the 6.1 m

7 Tm adjusted for depth attenuation will always be smaller than Tm at the level of the measurement, because the higher-frequency components of the wave spectrum are attenuated more than the lower-frequency components.

Wellington Airport Runway Extension 31

recorded in outer Lyall Bay (DWG site 2), and attenuating to under 5 m at the ADCP near the runway (site 1).

Figure 2-9: Wave parameters measured at Site 2 in outer Lyall Bay. Panels: (A) wind speed, (B) wind direction (in meteorological convention "blowing from"), (C) significant wave height (Hs) and (D) mean spectral period at the instrument (Tm).

32 Wellington Airport Runway Extension

3 Hydrodynamic flow and sediment-transport modelling (Delft2D) A key component of the AEE investigations, is to predict and assess the effect of the proposed runway extension into Lyall Bay on coastal physical processes, within the Bay. This includes any likely before and after changes in tidal currents and tidal circulation, wind-driven circulation cells, sediment transport due to waves and currents and the impact of any changes caused by the proposed runway extension on sediment deposition on near-shore morphology. Such before-and-after predictions for coastal processes in relation to modifications or discharges to the coastal marine area are routinely undertaken using numerical hydrodynamic models and associated modules for sediment transport and plume dispersion and assessing the degree and extent of changes.

The prime focus of the modelling investigations was to assess the changes before and after the construction of the proposed runway extension. This was accomplished by freezing the boundary and initial conditions and model parameters (e.g., bed friction) for the validated model simulations of the existing coastal environment and then changing the bathymetry to include the perimeter rock dyke of the runway extension.

The hydrodynamic model was calibrated sufficiently well to enable the relative change in wind-driven currents and residual circulation patterns to be assessed before and after the runway extension.

3.1 The approach to hydrodynamic and sediment-transport modelling We used the Deltares Delft3D modelling suite in the investigations of the changes before and after construction of the extended runway on the tide, wind and wave-driven dynamics of Lyall Bay and on the before-and-after effect of changes in current circulation and waves on sediment transport and morphology in the Bay. The modelling approach adopted was based on the following components:

• For this present application to the south Wellington coast and Wellington Harbour, the Delft3D modelling system was used in depth-averaged (2-D) mode.

This assumes no density stratification of the water column occurs (e.g., from differential warming or cooling of the surface layer or substantial freshwater sources). The same assumption was used in applying a depth-averaged model to plume dispersion from the ocean outfall in outer Lyall Bay by Bell et al. (2013). The waters in outer Lyall Bay have been shown previously (e.g., Beca Carter–Caldwell Connell, 1980; Bell, 1989) to be generally well mixed or only weakly-stratified with surface to bottom layer density differentials of only 0.02 to 0.16 kg/m3. In the outflowing ebb-tide plume from Wellington Harbour, the density stratification becomes more important, especially with high discharges from the Hutt River, but tidal modelling and observations in Bell (1989) show that the Harbour plume travels in a W to WSW direction offshore from Lyall Bay and does not directly enter the Bay. Besides mixing of the water column from stronger tidal currents off Lyall Bay, the Bay itself is exposed to cycles of strong southerly and northerly winds that causes regular episodes of strong shear and vertical mixing along with turbulent mixing from waves.

The other aspect where a 2-D model may be limiting are situations of complex three- dimensional wind-driven circulation, such as for a bowl-shaped semi-enclosed embayment, where a downwelling offshore-directed current in deeper waters can counter a down-wind onshore-directed surface current in the centre of the embayment. A previous near-bed and near-surface mooring of two current meters in outer Lyall Bay (Bell, 1989) and the recent ADCP mooring at Site 1 (Section 2.2.3), show there are some difference in wind-driven

Wellington Airport Runway Extension 33

response between the surface and bottom layers, but no evidence of counter-currents down the water column in the centre of Lyall Bay e.g., the outer current-meter deployment in Lyall Bay (Bell, 1989) responded to strong southerlies with generally southerly-directed (up-wind) currents at both top and bottom current meters, but at somewhat different current speeds.

The use of the depth-average 2-D mode, which still incorporates all the main forcing mechanisms, can therefore be justified, alongside run-time constraints on running the entire model in 3-D (which included part of the deep Cook Strait waters and Wellington Harbour) and using the model primarily as a comparative tool to determine the effects on hydrodynamics after the runway extension is constructed, relative to the present situation.

. Residual hydrodynamic circulation in Lyall Bay was investigated using forcing from tides (multiple constituents) at the open-sea boundary (in Cook Strait) and local wind speed and directions applied over the sea-surface of the model. The simulations were run for the same time frame as the field observational period of approximately 7 weeks (see Section 2).

• Combined tide and wind-driven circulation investigations in Lyall Bay focused on isolating the wind response to currents (leaving aside waves) and were forced by a sequence of 3- day wind events peaking at a speed of 22 m/s (strong gale), directed from the prevailing south and north directions. The winds were superimposed on a background of a repeating

mean (M2) tide applied as a boundary condition on the open-sea boundary (see section 3.3.3) to isolate the wind-driven response in currents.

• For assessing sediment transport and morphological change, the 2-D hydrodynamic model was forced with tide (multiple constituents), local wind speed and directions and coupled with a SWAN wave model on the same grid for the south Wellington coast. The simulations covered the same time frame as the ~7-week field-measurement period in 2014. The predicted currents, including wave-orbital motions, from the coupled hydrodynamic and wave models, were used by the sand sediment-transport model to predict sand transport. The sediment-transport modelling was then used to determine the before and after changes in sand deposition or scour only within Lyall Bay. The mobility of sediments on the seabed outside Lyall Bay was assumed of no consequence to this study focused on the proposed runway extension. Therefore, only the surficial sand layer on the seabed in Lyall Bay was made erodible in the model domain, with the seabed elsewhere made non-erosive in the model.

• Investigation of the intensity and extent of suspended-sediment plumes from de-watering discharges during construction was supported by simulations using the 2-D hydrodynamic model coupled to a non-cohesive sediment-transport and settling model. This assumed that the water in the discharge is the same density as water in the bay, which would be the case for barged fill material sourced from dredging the seabed or from discharging wave- overtopping waters. In this case, there would be no buoyancy effect in the water column associated with the sediment plume. However if freshwater from stormwater runoff was a significant component of the discharge, then further simulations would require that buoyancy effects are included, through coupling with a near-field mixing model. The plume model simulations are covered in Section 7.

34 Wellington Airport Runway Extension

3.2 Delft3D modelling Suite The south Wellington coast, between Baring Head (Ōrua-pouanui) and Sinclair Head (Te Rimurapa) together with Wellington Harbour (Te Whanganui a Tara), were modelled using the Deltares Delft3D hydrodynamic and sediment transport modelling suite.

The curvi-linear, 2-dimensional or 3-dimensional sigma coordinate (multi-layer) semi-implicit model suite finds numerical solutions for 2 and/or 3-dimensional flows (Delft3D-FLOW), wave–current interaction (Delft3D-WAVE), non-cohesive/cohesive sediment transport (Delft3D-SED) which can incorporate morphological evolution by scaling the hydrodynamic and sediment transport predictions (Delft3D-MOR).

3.2.1 Delft3D-FLOW Delft3D-FLOW solves the Navier-Stokes equations for momentum whilst conserving mass through the principle of continuity (Deltares, 2011). Physical processes in the model can be parameterised and simulated through specifying for example, eddy scales, turbulent-closure schemes, surface and bottom boundary conditions, surface winds and pressure fields, wave-current interaction, surface heating, salinity & temperature structure and the earth’s rotational effects.

The Delft3D-FLOW model can be forced at open and source input boundaries by oceanic tides, freshwater and heat sources. These forcing mechanisms produce the essential boundary physics required to simulate barotropic (surface-pressure gradients) and baroclinic (internal pressure gradients driven by horizontal and vertical water-density gradients) in the model domain which allow variation in seawater density to be included in model solutions. The superimposed effect of currents and waves on the bed shear stress is taken into account by means of the current-wave interaction model of Fredsøe (1984).

For this present application to the south Wellington coast, Delft3D-FLOW was set up as a 2- dimensional depth-averaged model and run in barotropic mode, assuming waters are generally well- mixed vertically (see section 3.1).

3.2.2 Delft3D-WAVE Within the Delft3D modelling suite, the wave module Delft3D-WAVE simulates the evolution of random, short-crested wind-generated waves in estuaries and tidal inlets and based on the third- generation Simulating WAves Nearshore or SWAN model (see Booij et al. 1999; Ris et al. 1999).

The SWAN model is a spectral wave model intended for shallow water applications in coastal and estuarine environments (Booij et al. 1999; Ris et al. 1999). It computes the evolution of the wave energy spectrum in position (x,y) and time (t), explicitly taking into account the various physical processes acting on waves in shallow water. These include the effects of refraction by currents and bottom variation, and the processes of wind generation, white-capping, bottom friction, quadruplet wave-wave interactions, triad wave-wave interactions and depth-induced breaking. The model can incorporate boundary conditions representing waves arriving from outside the model domain.

For all model simulations of Lyall Bay and Cook Strait flow and sediment transport predictions, Delft3D-FLOW was ‘online’ coupled with Delft3D-WAVE which has a two way wave-current interaction i.e., the effect of flow on the waves (via set-up, current refraction and enhanced bottom friction) and the effect of waves on current (via forcing, enhanced turbulence and enhanced bed

Wellington Airport Runway Extension 35

shear stress). Section 4 describes the stand-alone simulations of SWAN to determine spatial variability in the wave climate of Lyall Bay.

3.2.3 Delft3D-SED The non-cohesive (sand) sediment transport and morphodynamic computations are carried out using the Delft3D-SED module. The module’s default sediment-transport dynamics are based on the formulation of van Rijn (1984a; 1984b). This distinguishes between sediment transport below a reference height which is treated as bed-load transport and that above the reference height which is treated as suspended-load. Sediment is entrained in the water column by imposing a reference concentration at the reference height. The velocities used in these computations are provided by the flow fields predicted by Delft3D flow and if coupled, Delft3D-WAVE.

Mobile bed sediment was assumed to be sandy, as ratified by coring surveys in Lyall Bay (Depree et al. 2015), and therefore cohesion-less, with a mean grain size (D50) of 0.15 mm, and a sediment density equal to 2650 kg/m3. The dry bed density was set equal to 1600 kg/m3. Suspended sediment diameter at the beginning of the computation has a representative diameter equal to the D50 of the seabed sediment. A minimum water depth equal to 0.2 m was set in the model for a sediment- transport calculation to be undertaken in any model cell. The transversal and longitudinal calibration factors for bed-slope effects were respectively set equal to 1.5 and 1.

The simulation of suspended sediment plumes, generated during construction operations, used the cohesive-sediment component of the Delft3D-SED module, given the prime focus is on the finer fractions such as silts, which is discussed in Section 7.

3.2.4 Delft3D-MOR The Delft3D–MOR module applies sediment transport formulae (both suspended and bed total load) to estimate the morphological change through the model domain (Lesser et al. 2004). Throughout a simulation, elevation of the sea-bed is dynamically updated at each computational time-step by computing the change in the mass of bed material that has occurred as a result of the sediment sink and source terms and the sediment-transport gradients. The mass is then translated into a bed level change based on the dry-bed densities of the sediment fraction (sand in the case of Lyall Bay). This means that subsequent hydrodynamic calculations are always carried out using the continually- updated bathymetry as the sea-bed erodes or accretes.

3.3 Model Development and Setup

3.3.1 Model Grid The curvi-linear grid for the Delft3D-FLOW and WAVE model used in this study is a modified version of the grid used by Bell et al. (2013) for effluent dispersion modelling from the Moa Point outfall. This grid model in turn was developed from the gridded bathymetry data used by NIWA previously to establish a model of Greater Cook Strait for modelling hydrodynamics for tsunami and general oceanography.

Bathymetry data for the model grid used by Bell et al. (2013) was sourced from:

. NIWA’s general ocean bathymetric archive

. high-resolution swath-bathymetry surveys undertaken by NIWA for studies of sea-floor stability in Cook Strait, and

36 Wellington Airport Runway Extension

. shallow-water soundings from Land Information NZ Hydrographic Charts NZ4633 and NZ4634 for Wellington Harbour.

The flow and wave regime within Lyall Bay is driven by tide, wind and wave dynamics of the Cook Strait and also the Harbour entrance tidal flows. The grid used in this study required sufficiently-high resolution in Lyall Bay to resolve flow and sediment transport around the site of the proposed runway extension. Therefore, model grid resolution in the bay included grid cells of approximately 50 m, compared to grid cells in the Cook Strait of approximately 100–200 m where less detail was required.

The curvi-linear grid generator in the Deltares modelling suite was used to generate the grid shown in Figure 3-1. This shows the refinement of grid cells around Lyall Bay as compared to coarser resolution grid cells in open Cook Strait waters. Wellington Harbour was also gridded to reasonably fine-scale cells.

The Delft3D-WAVE model grid is also shown superimposed on top of the Delft3D-FLOW model grid (red) to show the extent of the domain covered by WAVE simulations. Figure 3-1 shows that Wellington Harbour was eliminated from WAVE simulations to speed up model run times, given waves within the Harbour have no influence on Lyall Bay.

Figure 3-1: Delft3D-WAVE model grid (red) of south Wellington coast superimposed on top of the Delft3D- FLOW model grid (black) that also includes Wellington Harbour.

Wellington Airport Runway Extension 37

New bathymetry data collected for this study from multi-beam surveys of Lyall Bay was then incorporated (or replaced) into the existing bathymetric data and then interpolated on to the new model grid(s). The bathymetry behind the revised hydrodynamic-model grid of the south Wellington coast and Wellington Harbour is shown in Figure 3-2 and a close-up for Lyall Bay in Figure 3-3. Depths are relative to Wellington Vertical Datum-1953.

Figure 3-2: Areal extent of the hydrodynamic model bathymetry. Sites marked are tide gauges (TG) or current meters referred to in the text. Depth colour scale capped at 50 m to show detail inshore.

38 Wellington Airport Runway Extension

Figure 3-3: Close-up of the hydrodynamic model bathymetry in Lyall Bay. Sites marked are current meter deployments referred to in the text.

3.3.2 Model tidal boundary conditions The Delft3D outer model boundaries in the Cook Strait were set between a westerly limit in the Strait beyond Sinclair Head (Te Rimurapa) and an eastern limit to the south-east of Baring Head (Ōrua- pouanui) as shown in Figure 3-2. The model was forced along the open-sea boundary in Cook Strait between these two locations with spatially and temporally varying tidal elevations generated by harmonic constituents extracted from the EEZ tidal model (Stanton et al. 2001).

3.3.3 Model boundary wind speed and direction The regional wind speed and directions used for modelling the Lyall Bay hydrodynamic circulation and sediment transport, pre and post runway construction, were extracted from the NIWA EcoConnect weather modelling system for a point in the centre of Lyall Bay (Figure 3-4). The peak wind speed was 22 m/s (80 km/hr) reached on 22 September (Year Day = 264) blowing from the SSW (200). The surface wind stress computed from the data drives both flow and also generates local wind waves. The data had a 1 km2 spatial scale and 30 minute temporal resolution and was implemented over the entire Delft3D model grid (area shown in Figure 3-2).

Wellington Airport Runway Extension 39

Figure 3-4: Winds over the deployment period (18-Aug to 09-Oct 2014). Wind direction in meteorological convention ("blowing from"). Source: NIWA EcoConnect weather model from the centre of Lyall Bay.

To investigate the response of Lyall Bay to strong northerly and southerly winds, an idealised simulation was run with winds reaching a peak wind speed of 22 m/s (strong gale), as shown in Figure 3-5.

This simulation was used to determine the wind-driven circulation within Lyall Bay due to southerly winds and northerly winds respectively – noting that no strong northerlies were observed in the 2014 field deployment, so modelling such an event is important to compare with a similar magnitude southerly. The wind events were interspersed with calm wind conditions to ensure a return to the background tidal conditions before the next event. This approach provides more clarity on the wind- response of current patterns in Lyall Bay from each wind direction, on a background of a repeating average tidal cycle. This idealised scenario was implemented for both pre and post runway-extension scenarios to determine the likely change in wind-driven circulation.

The same wind time series was also used, along with background tidal currents, to estimate the suspended-sediment plume dispersion within Lyall Bay. This relates to the potential discharges of fine sediments from one or two sites on the western side of the proposed runway extension during the embankment infill phase of construction (see Section 7).

40 Wellington Airport Runway Extension

Figure 3-5: Idealised 1-month simulation of a sequence of two southerly (180) and one northerly (0) wind events. Each event covers 3 days and peaks at 22 m/s interspersed with calm conditions.

3.3.4 Model wave boundary conditions Wave spectral boundary conditions for the Delft3D FLOW + WAVE + SED simulations were spatially and temporally varying along the model outer boundary in Cook Strait. Spatially and temporally varying winds were used to force surface wave conditions through the fetch-limited JONSWAP empirical algorithms. See Section 4.2 for more detail on the generation of wave boundary conditions on the southern boundary of the Delft3D model.

3.3.5 Sediment transport and seabed morphology initial conditions The Delft3D-SED modules initial seabed conditions were set to a uniform sand bed layer thickness of 10 cm inside Lyall Bay. In the remainder of model domain outside Lyall Bay, the seabed layer was set to zero thickness (i.e., no sediment transport), given there would be negligible influence from the proposed runway extension altering sediment exchange between Cook Strait and Lyall Bay. The sand 3 median diameter (D50) was set to 0.15 mm with a specific and dry density of 2650 kg/m and 1600 kg/m3 respectively. The Delft3D-MOR module was set to default conditions for all other parameters.

Critical bed shear stress for sedimentation and erosion through the model domain were set to 0.075 N/m2 and 0.15 N/m2 respectively (Whitehouse et al. 2000).

Wellington Airport Runway Extension 41

3.3.6 Model Calibration and Skill Tide levels 8 The Delft3D-FLOW model was calibrated for the three largest M2, S2 and N2 semi-diurnal tidal constituent elevations, at three sites in the model domain shown in Figure 3-2, corresponding to tide gauges at the Port of Wellington and Somes Island inside the Harbour and tide levels measured by a current meter moored at site C in 1989 (Bell, 1992). These sites were selected as tidal constituent amplitude and phase were known from analysis of tide gauge data that had been collected at these locations as part of previous studies in the area, particularly for the marine outfall investigations (Bell, 1992; Carter & Lewis, 1995).

The M2, S2 and N2 tidal constituents were then to drive a hydrodynamic model simulation of tidal elevations coinciding with the 2014 field deployment, which were compared to time series of semi- diurnal tidal elevations computed from constituents extracted from model time series data at the same locations (Figure 3-6). Note: because the solar S2 tide is small in Wellington area, the tide pattern doesn’t exhibit a strong fortnightly spring/neap cycle but rather is dominated by the monthly perigean/apogean cycle (represented by the M2 + N2 tidal constituents).

A measure of model performance was derived for water levels by estimating model root mean square error (RMSE), which is a measure of the difference in the variance between the observed and predicted signal; and model skill (SKILL) where values span 1 (high) to 0 (poor) skill decreases towards zero as described by Warner et al. (2005) and Haidvogel et al. (2008). SKILL is defined as:

2 N 2 SKILL = 1 − [|Xm − Xo| ]⁄[∑i=1(|Xmi − X̅̅o̅| + |Xoi − X̅̅o̅|) ] , (1) where X is a variable with N values in a time series and X̅ is a time average of the variable. Subscript m and o are for modelled and observed values respectively. Results from the RMSE and Skill analysis are shown in Table 3-1. The results shown indicate excellent agreement between the observed and modelled tides at the three sites. Skill values indicated a very good match between the observed and predicted values plus the RMSE values were small being in the order of a few centimetres difference, indicating a low mean variance between the model and tide gauges.

Results from a cross-correlation analysis between the observed and modelled semi-diurnal tides are presented in Figure 3-7. The lag analysis shown scores 1 at zero lag indicating that the observed and modelled tides are in phase – so the timing of the tides is well modelled.

Table 3-1: RMSE and model Skill for observed and modelled M2, S2 and N2 semi-diurnal tidal amplitude at three sites in the model domain.

Gauge site RMSE (cm) Skill Port of Wellington 7 0.99 Somes Island 5 0.99 Lyall Bay 7 0.99

8 twice-daily tides –with M2 the main lunar tide of 12.4 hr period (equivalent to an average tide range), S2 the main solar tide of 12 hr period (M2 + S2 generates the spring tide) and N2 the tidal constituent associated with the Moon’s elliptical orbit around the Earth (12.66 hr period), with M2 + S2 + N2 generating perigean-spring or “king” tides

42 Wellington Airport Runway Extension

Figure 3-6: Comparison of observed (blue) and modelled (red) tide heights for two sites in Wellington Harbour and one site in Lyall Bay (C). Based on the combination of the 3 main twice-daily tides (M2, S2, N2).

Figure 3-7: Cross-correlation of modelled versus observed tides, showing a zero lag (i.e., both time series are in phase). HBR= Wellington Harbour, SI= Somes Island, LB=Lyall Bay.

Wellington Airport Runway Extension 43

Tidal currents

The tidal current amplitude was computed for the same three twice-daily tidal constituents (M2, S2 and N2) at three different locations in the model domain shown in Figure 3-2 (site C and D) and Figure 3-3 (site 1 – also site C again). The calibration sites selected correspond to two sites (C and D) where there were previous deployments of current meters during earlier studies in 1989 and 1978 respectively as described by Bell (1992) and Carter & Lewis (1995) and site 1 is where the ADCP was moored for the recent 2014 field deployment (see Section 2).

Based on a threshold signal to noise ratio 2 in the least-squares analysis of tides from the current meter data at Site 1 (near the spur groyne) indicated that the semi-diurnal tidal constituents for currents were statistically inseparable from back ground noise in the current meter measurements. Hence, tidal currents at site 1 were slow enough to be below the statistical detection threshold (ND=not detected) within the accuracy performance of the ADCP. The comparisons between the observed and predicted major-axis current amplitudes at the two other locations outside Lyall Bay (site C and D) indicated good agreement between the modelled and observed tidal current magnitudes for the three main semi-diurnal tidal constituents.

Table 3-2: Observed and modelled (parentheses) M2, S2 and N2 semi-diurnal tidal current major-axis amplitude at three sites in the model domain. ND is not detected to a significant level above data noise.

Current-meter site [year] M2 (cm/s) S2 (cm/s) N2 (cm/s) 1 [2014] (near spur groyne) ND (2) ND (0) ND (0) C [1989] (Bell, 1992) 10.5 (13) 2 (3) 2 (3) D [1978] (Bell, 1992) 29.1 (26) 4 (6) 5 (6)

The Deflt2D model performed well for the depth-averaged tide levels and tidal currents in the two contrasting areas of the inner Lyall Bay (very slow currents) and offshore from Lyall Bay (sites C & D), where tidal currents dominate over winds (Bell, 1989; Bell, 1992). This indicates that the model application is well representing the tidal response of the south Wellington coast and Lyall Bay.

Wind-driven circulation The current circulation pattern for combined tide and wind forcing can be displayed as a residual (or net) circulation vector plot to summarise net current flows over a given period of several weeks for various tidal and wind forcing combinations. In this case, the residual circulation was modelled and calculated for the ~7-week deployment period in 2014 (Section 2).

The residual circulation pattern using the tidally-calibrated Delft2D model incorporating the wind forcing is summarised in Figure 3-8.

The model prediction for the residual wind and tidal current direction at Site 1 (the dot in Figure 3-8) is a fair match (WSW directed) versus NW directed for the depth-averaged net residual current from the ADCP record (bottom-right panel; Figure 2-6). The net-drift speed modelled at Site 1 is around 0.05 m/s compared to a considerably slower 0.007 m/s for the ADCP data, which on this basis indicates that the model over-predicted the net or residual flow at this site. However, it is noted that the ADCP depth-averaged residual plot shows the drift current generated at Site 1 is in opposite directions for two southerly-wind events (21-24 August and 22-23 September 2014), which would have led to the slower net residual and illustrates the local complexity of flows and waves at Site 1

44 Wellington Airport Runway Extension

near the spur breakwater. The model did not show this reversal in flow direction, but predicted a depth-averaged WSW-directed flow throughout the period.

Possible reasons why a closer match between the model and field data was not obtained are: a) Site 1 is located just off the submerged rock outcrop that extends beyond the spur groyne (see bathymetry in Figure 2-1) and therefore wind-driven currents are likely to be influenced locally by this complex bathymetry and non-linear wave effects around and over the rock outcrop. b) The bathymetry schematised in the model grid (50 m cells) will be smoother than reality and the currents averaged over the approximately 5050 m cells, so some differences between a point measurement and a cell-averaged current are expected where the local bathymetry is complex (ideally the ADCP should have been further out in the Bay, but the prime focus was on capturing waves near the spur groyne). c) Some differences in residual current will be present arising from the application of Delft3D in the depth-averaged mode. The ability to check the model performance in an area of smoother bathymetry further out in Lyall Bay at Site 2 was thwarted by the loss of the outer ADCP during a southerly gale.

Other features of the residual circulation for the present-day shoreline, match reasonably well with other confirmatory evidence. Further offshore at the southern entrance to Lyall Bay, the net west then stronger south-west residual current is driven largely by the stronger ebb-tide stream off Lyall Bay than the flood-tidal stream – a pattern that was observed by Bell (1989) including a NW-directed residual current at site C from a 1-month current-meter deployment in 1989 (triangle in Figure 3-8). Also the modelled weak clockwise residual flow in the NE corner of inner Lyall Bay also fits with the long-term sedimentation that has occurred adjacent to the existing runway revetment, and particularly behind the spur groyne, since the reclamation was undertaken in the 1950s (Figure 2; Pickrill, 1979).

The poor match between model-predicted and observed currents at Site 1 primarily results from localised bathymetric steering effects of the rocky reef where the instrument was deployed. These local detail changes in bathymetry are not able to be resolved well in the circulation and sediment transport models (unless very high-resolution model grids are used). Nevertheless, the patterns in the overall residual circulation in the Bay that results from forcing by tide and wind follows the physical laws of mass conservation where water cannot continually setup or set down in an embayment and has to flow in and out to conserve mass. In Lyall Bay where the earth’s rotational effects are negligible, a dipole type circulation occurs, as was reproduced in the tide and wind-forced model scenarios.

While a close match was not obtained at Site 1 off the spur-breakwater rocky outcrop, overall the pattern of residual circulation from wind and tide forcing within the Bay appears reasonable from our experience in coastal hydrodynamic modelling and knowledge of oceanographic processes, taking into account the planform shape of inner Lyall Bay and the hydrodynamic controls exerted by the end of the existing runway and the spur breakwater and associated rocky outcrop. A better match of the residual flow was obtained with field observations in the outer Bay from this and previous studies for the Wellington wastewater outfall project. The hydrodynamic model therefore still forms a reliable basis for comparing and isolating the relative changes before and after extension of the proposed runway.

Wellington Airport Runway Extension 45

Figure 3-8: Modelled net or residual circulation over the ~7-week field deployment period in 2014. Dot marks ADCP Site 1 and the triangle marks site C (1989 deployment). Graduated colour scale according to the magnitude of the residual current speed (see Figure 3-9 for scale).

3.4 Modelled hydrodynamic scenario results

3.4.1 Lyall Bay residual circulation Residual or net current circulation is the current pattern remaining after the to and fro tidal and wind-generated movements have been averaged out. The residual is generated by vector-averaging the velocities (taking into account both speed and direction) over a given period – in this case the ~7- week field deployment period in 2014.

The modelled magnitude and direction of residual circulation within Lyall Bay both before and after the proposed runway construction is presented in Figure 3-9. The results are based on the same time span as the observations i.e., 55 days during Aug to Oct 2014. The model was forced by time series of spatially-varying tidal boundary and surface winds for that field deployment period. The same simulation was undertaken for the existing situation and then repeated with the proposed runway extension included, all else being the same.

The circulation in the outer part of Lyall Bay beyond the end of Hue te Taka Peninsula and Te Raekaihu Point is governed by the stronger ebb-tidal currents that start to enter the Bay and exit back out into Cook Strait on the west side (Figure 3-9) and likewise the flood tide flows that enter and exit on the eastern side around West Ledge Reef at the Harbour Entrance (not shown). The residual (net) current analysis indicates the formation of two weak residual eddy-like circulation patterns within Lyall Bay with small net-current amplitudes of order <0.06 m/s.

46 Wellington Airport Runway Extension

In the mid-section (up to the end of the present runway), a residual anti-clockwise circulation occurred for the sequence of winds (mainly southerly) and tides during the field deployment period. This residual current pathway heads towards Moa Point and the eastern shoreline before turning west at the end of the present runway embankment (panel A, Figure 3-9). The introduction of the proposed runway extension (panel B, Figure 3-9) would partially interrupt the westerly net flow inshore beside the present runway, given the same set of tidal and wind conditions, and would form two shadow zones (negligible net currents) either side of the proposed runway extension. These shadow zones will in turn slow the dispersion of plumes from discharges during construction that might introduce suspended sediment (see Section 7).

Further inshore, off Lyall Bay Beach, a weak residual clockwise circulation occurred, generating a weak north-eastwards net transport in the nearshore towards The Corner and southwards along the existing runway embankment (panel A, Figure 3-9). Along with wave diffraction around the spur groyne and rock outcrop, this albeit weak net drift towards the NE corner could partially explain the overall seabed accretion that has occurred in the eastern side of the Bay adjacent to the runway embankment (Figure 2; Pickrill, 1979). Simulations predict that currents to the eastern flank of the proposed runway extension form a net flow that exits the eastern flank of the reclamation and hugs the south extension wall before being directed out into the wider bay. This localised change in current in the vicinity of the runway extension will have a negligible effect on the import/export flushing of water mass or the sediment transport within the inner Bay (panel B, Figure 3-9).

The inter-comparison of results from these net circulation simulations indicate that overall the proposed runway extension is likely to have only a minor localised effect on the residual circulation within the eastern mid- section of Lyall Bay and a negligible effect on residual circulation in the nearshore area off Lyall Bay Beach. The residual currents are also unaffected in the outer Bay by the runway extension and therefore will not affect the dispersion of treated wastewater from the marine outfall discharging from the Moa Point wastewater treatment plant.

Figure 3-9: Residual (net) current circulation in Lyall Bay over the ~7-week field deployment period. Existing environment (A) and proposed runway extension included (B).

Wellington Airport Runway Extension 47

3.4.2 Lyall Bay wind-driven circulation Both current-meter observations and model simulations demonstrate that tidal currents within inner Lyall Bay are exceptionally weak. The current-meter observations show that substantial wind-driven currents can be generated within the Bay (relative to tidal currents).

As previously mentioned, the wind fields during the 2014 field deployment were dominated by southerly winds, with one event reaching 22 m/s (80 km/hr). However, analysis of the longer term climatic records of the region (Section 2) show that strong northerly winds also occur. To investigate wind-driven circulation inside Lyall Bay, we forced the model with a time series of idealised northerly- and southerly-directed winds peaking at 22 m/s for both the existing and the proposed runway extension scenarios, superimposed on a repeating M2 tide cycle, as discussed in Section 3.3.3. The results of the simulations are shown as a series of snap-shots of current vectors at low water, mid tide and high tide. The timing of the tide and the winds were varied so that every snapshot coincides with the 22 m/s peak in wind speed (Figure 3-5).

Lyall Bay circulation in response to southerly winds Figure 3-10, Figure 3-11, Figure 3-12 show that in response to southerly wind conditions, the most intense wind-driven currents flow northwards (down-wind) in the shallow water along the periphery of both the western and eastern shores of Lyall Bay – more so on the former.

At the head of the Bay, the currents then turn clockwise and flow eastwards along nearshore zone, then south along the existing-runway embankment before being deflected WSW by the spur groyne and rock outcrop into the centre of the Bay. This clockwise eddy then converges with counter- clockwise eddy that extends from the eastern shore of the outer Bay at Moa Point and westwards across the end of the existing runway out into the central Bay. This mid-bay eddy would be partially intercepted by the proposed runway extension (right-hand panels) that forces the down-wind drift into the cove at Moa Point to turn into the wind around the extension, weakening the flow into the central Bay. The instantaneous circulation pattern shown in Figure 3-10, Figure 3-11, Figure 3-12 are similar to the residual flow pattern for the field deployment period that was also dominated by southerly winds (Figure 3-9).

The observed net residual current at Site 1 (marked by the square in Figure 3-10 to Figure 3-12) during the 22 m/s southerly on 22-23 September 2014, was directed to the NW (Figure 2-6) for the depth-averaged vector-plot as discussed above, whereas the model for the southerly-wind scenario shown below (left panels), has a more westerly direction. The model is mainly picking up the westerly component of the flow observed at the ADCP current-meter site, but differences between observed and modelled could also arise from the factors outlined in Section 3.3.6.

Based on a visual comparison of the modelled southerly-wind scenario for before and after the runway is extended, there is likely to be only relatively localised effects on the wind-generated current flows during southerlies caused by the extended runway embankment from local re-steering of the downwind flow along the eastern Bay and slightly faster velocities across the end of the extended runway than presently occur at the same location (some 350-400 m off the present runway).

48 Wellington Airport Runway Extension

Figure 3-10: Current circulation near peak southerly wind at low water. (Left) existing situation, (right) proposed runway extension.

Figure 3-11: Current circulation near peak southerly wind at mid-flood tide (HW-3 hrs). (Left) existing situation, (right) proposed runway extension.

Wellington Airport Runway Extension 49

Figure 3-12: Current circulation near peak southerly wind at high water. (Left) existing situation, (right) proposed runway extension.

Lyall Bay circulation in response to northerly winds The predicted wind-driven circulation that develops under northerly-wind conditions peaking at 22 m/s is presented in Figure 3-13 to Figure 3-15 for a sequence from low water to high water. In this scenario the most intense currents were also predicted on the periphery of the western and eastern Lyall Bay but in this instance flows were directed southwards (but similarly down-wind).

The southward outflows along the eastern and western sides of the Bay cause waters to be drawn in from Cook Strait through the central Bay to maintain water volume equilibrium. This is seen as a tongue of inflowing water directed up-wind to the north. At the head of Lyall Bay, the inflowing tongue of water then bifurcates and forms a dipole circulation. The flow on the west side of the bay turns counter-clockwise and merges with the wind-driven southerly outflow, and the flow on the eastern side forms a clockwise eddy that eventually converges with the northerly inflow in the central Bay. The eastwards flow into The Corner area (see Figure 3-16 for location) is in the same direction as for southerly winds, although considerably weaker, thus both northerly and southerly winds contribute to the easterly residual current there (Figure 3-9) and fits with the seabed accretion that has occurred in this area since the runway was extended into Lyall Bay in the late 1950s (Figure 2; Pickrill, 1979).

50 Wellington Airport Runway Extension

Figure 3-13: Current circulation near peak northerly wind at low water. (Left) existing situation, (right) proposed runway extension.

Figure 3-14: Current circulation near peak northerly wind at mid-flood tide (HW-3 hrs). (Left) existing situation, (right) proposed runway extension.

Wellington Airport Runway Extension 51

Figure 3-15: Current circulation near peak northerly wind at high water. (Left) existing situation, (right) proposed runway extension.

3.4.3 Effects of runway extension construction on wind-driven circulation The Delft2D simulations indicate that there would be little change in Lyall Bay wind-driven circulation during northerly winds, other than minor localised changes around the runway embankment area. Circulation during northerly winds would remain largely unaffected by the proposed extended runway.

During strong southerly winds, the simulations indicate that the proposed runway extension would slightly weaken the counter-clockwise eddy-driven outflow through the central area of Lyall Bay that passes off the existing-runway terminus. This mid-bay eddy would be partially intercepted by the proposed runway extension that would force the down-wind drift into the cove at Moa Point beach to double back in order to pass around the extension, thereby slightly weakening the wind-generated flow into the central Bay as it will be in deeper water (compared to the shallower water off the end of the present runway). The simulations suggest that the extension would somewhat weaken wind- driven currents over a broader convergence zone between the inner-bay clock-wise eddy and mid- bay eddy, relative to the existing situation. This slightly weaker and broader flow convergence zone would only have a minor effect on the wind-driven hydrodynamics of the central Bay and is unlikely to affect flushing or morphological change in the inner Bay. The local re-steering of the downwind flow along the eastern Bay would also produce slightly faster velocities across the end of the extended runway than presently occurs at the same location 350–400 m off the present runway terminus, as would be expected.

3.5 Sediment-transport modelling

3.5.1 Geomorphic setting of Lyall Bay Lyall Bay formed when a tombolo connected the Miramar Peninsula (to the east of the Airport) to the hills to the west of Kilbirnie. The growth of the tombolo has occurred within the last 900 years. Māori history identifies the Miramar Peninsula as a separate island (Motu-Kairangi or Miramar Island) some 900 years ago (Raukura Consultants, 2014), but the area has since been subject to

52 Wellington Airport Runway Extension

tectonic uplift by ~3 m in a 15th century earthquake event and ~2 m in the 1855 Wairarapa earthquake (Pickrill, 1979).

The construction of Wellington Airport at Rongotai (1952–1959) reclaimed 14 hectares of the eastern Lyall Bay for an 850 m runway extension, connecting with a rocky outcrop towards Moa Point, which formerly acted as a natural breakwater. The original 1941 shoreline and the former rocky outcrop are shown in Figure 3-16, along with other features mentioned in the text, including The Corner where surfing is popular.

Lyall Bay is 1.3 km wide at the southern entrance narrowing to 1.1 km in the inner Bay (post-runway extension in 1959) and is 2 km in length over a north-south orientation.

Figure 3-16: Geomorphic features in Lyall Bay incl. the 1941 shoreline. Blue line shows the approx. 1941 shoreline (from Pickrill, 1979) and green dotted line the rock outcrop in 1941 that joined the submerged rock outcrop the spur groyne currently sits on. [Background image: Google Earth, 8 October, 2014].

Wellington Airport Runway Extension 53

The bathymetry of Lyall Bay (Figure 2-1) reflects the Pleistocene geological history, with the former steep-sided valley infilled with sands and reworked by wave activity to form a flat surface gently sloping seaward (Pickrill, 1979). Inshore of the present runway terminus, the Bay is asymmetrical, with a deeper trough extending down the eastern side and a wide shallow sandy platform extending from the western shoreline (Figure 2-1), which greatly influences the wave refraction patterns from east to west off Lyall Bay Beach (Figure 3-16).

3.5.2 Model set-up and application The effects of the proposed runway extension on sand transport in Lyall Bay were investigated using the non-cohesive component of the Delft3D-SED and Delft3D-MOR modules. Sand transport in shallow coastal areas is a complex interaction between waves, which primarily re-suspend seabed sediments into the water column, which are then subsequently transported by currents until the sediment particles settle again. In nearshore waters, including the surf-zone, non-linear wave dynamics along can also transport re-suspended sediments, especially if waves propagate towards the nearshore zone at an oblique angle to the beach. The Delft3D-SED and Delft3D-MOR modules incorporate tidal and wind-driven hydrodynamics from the Delft2D simulation coupled with waves simulated by the SWAN spectral-wave model to combine these wave-current interactions with mobile sediments on the seabed. The SWAN wave model development and results are presented in detail in the following Section 4.

To provide some context in the Lyall Bay situation, the fine sand largely present on the seabed of median diameter of 0.15 mm, would require a current of ~0.16 m/s acting alone to mobilise these sediments. Based on the currents measured at Site 1 (Figure 2-4), only the two highest southerly events would have been capable of mobilising sediments in the absence of waves. On the other hand, wave orbital velocities are more efficient at mobilising these fine sands, and increases dramatically as the wave height and wave period increases. Taking the example of a 1 m wave height, sands in Lyall Bay would be mobilised in water depths up to 24, 31 and 35 m for wave periods of 8, 10 and 12 seconds. Sediment transport from one location to another only occurs when mobilised sand, usually by waves, is carried along by a current – often in a series of short steps or hops, before it re-settles on the seabed.

In this section we present results in context of the relative change between the start and end of the ~7 week instrument deployment period in August to October 2014, which included a strong southerly gale of 22 m/s (80 km/hr) that would have generated substantial sediment transport and morphological change through wave processes, besides other southerly events during the field- monitoring programme. Such an event would be at a recurrence interval of a season through to an annual time scale. Northerly winds are unlikely to generate any substantial sediment transport acting alone, leaving aside the higher currents of 0.4–0.5 m/s along the rocky west coast and outer east coast of the Bay (Figure 3-13 to Figure 3-15) and wave energy is much more subdued at short wave periods (e.g., 2-4 seconds) with a very short fetch. Consequently, under northerly-wind conditions, re-suspension of sand by wave orbital motions is limited, even though northerly winds can produce wind-generated currents in the Bay that could otherwise transport wave-suspended sands.

The Delft3D-SED model was forced by tide, wind and SWAN wave boundary conditions (Section 4) derived for the field deployment period in 2014 and applied equally to both the pre and post runway extension bathymetry in Lyall Bay to ascertain the difference in morphological response over the ~7- week period, including the southerly gale event.

54 Wellington Airport Runway Extension

Sediment-transport and associated morphological models require intensive field measurements of seabed changes and bed shear stress thresholds for erosion and deposition under a wide range of wave and current combinations, to calibrate the various model parameters to best match the time- varying sea-bed changes. The required fieldwork would have been a costly and extensive undertaking and in our opinion is unwarranted when comparing the before and after construction situation, where widely-used sediment transport formulae can suffice for determining relative change. We have therefore not undertaken an exact sediment-transport calibration, but have used an un- calibrated sand-transport model to approximate the relative morphological change both with and without the proposed runway extension.

The default values for sediment transport for the median sand size in Lyall Bay (fine sand of 0.15 mm median grain size) in the Delft3D-SED model were invoked, which are based on the widely-applied sediment-transport formulations of van Rijn (1984a; 1984b). So while the adoption of this sediment- transport formulae may not necessarily have reproduced the exact seabed erosion and accretion changes that would have occurred over the field-deployment period, applying the same formulae, parameters and all other model forcing (waves, winds, tides) to both existing and post runway extension situations, sets up a comparable pair of simulations where the relative changes between the simulations is directly related to the effects of the runway extension.

Sea-bed response comparing pre and post runway extension The results of running the coupled Delft3D-SED and Delft3D-MOR modules, including waves from the SWAN model embedded in Delft3D-SED, over the ~7-week simulation period are presented in Figure 3-17 for the existing situation and Figure 3-18 for the proposed runway extension bathymetry included (with all other aspects of the model simulation kept the same). The plots show the areas of net erosion, net deposition or negligible change over the ~7-week period, rather than numerical changes in sea-bed height. In most cases the seabed height of erosion or deposition is of the order of <1 cm down to mm, but could be several cm where cross-shore processes such as storm-cut and offshore bar formation occurs and in the periphery of the proposed rock dyke and associated rock apron. Since the sediment-transport model is un-calibrated (but instead uses default sediment- transport formulae), the absolute magnitudes of sea-bed height changes predicted by each model run may not be accurate, but we focus instead on the relative change in deposition and erosion patterns with and without runway extension using the same default sediment-transport parameters in both before and after simulations.

Both simulations shown in Figure 3-17 and Figure 3-18 predict that over the 2014 field-deployment period for both pre and post runway extension an erosional cut occurs close inshore at Lyall Bay Beach with sand migrating offshore and depositing further offshore in the form of a sand bar.

Wellington Airport Runway Extension 55

Figure 3-17: Predicted changes in seabed height for the existing situation over the ~7-week field period in 2014. Categorized as erosion (E), deposition (D) or negligible change (0).

Figure 3-18: Predicted changes in seabed height for the proposed runway situation over the ~7-week field period in 2014. Categorized as erosion (E), deposition (D) or negligible change (0).

56 Wellington Airport Runway Extension

The erosion band in shallow water adjacent to, and including, the lower beach was mainly due to near-bed wave orbital velocity and current-wave interaction increasing bed shear stress, particularly during the strong southerlies, especially the southerly gale on 22–23 September 2014, when the significant wave height reached 4.7 m, with a mean period of 11 seconds off the spur groyne (Section 2). Furthermore, the incident wave field for the ~7-week simulation was mainly from the south and propagated inshore into Lyall Bay, frequently having the energy to mobilise the seabed. Wind- generated currents during southerlies mainly play a transporting role for wave re-suspended sediments, particularly along the eastern and western perimeters of Lyall Bay where higher currents are predicted as shown in Figure 3-10 to Figure 3-12. The predicted erosion areas along the western and outer eastern perimeters by the sediment-transport model (Figure 3-17 and Figure 3-18) can be explained by the combination of wave re-suspension and transport by both wind-generated currents and non-linear wave motions in the shallow near-shore waters. The simulated effect of forcing with a northerly wind alone, coupled with local wave generation through the SWAN model (not presented), on sand transport throughout Lyall Bay was negligible. This is largely due to the very limited fetch within inner Lyall Bay and then only limited to local wind wave generation (i.e., no swell, which can mobilise seabed sediments more readily).

A shore-parallel band of deposition further offshore from Lyall Bay Beach is indicative of the development of an offshore bar 150–250 m from the shoreline, which was built by sand eroding from the nearshore zone in the sediment model and migrating further offshore during the ~7-week period of mainly southerlies. This cut and fill behaviour of the beach profiles at Lyall Bay was also observed in a previous study by Pickrill (1979), especially in winter when southerly swells are more frequent, with a bar forming between 200–250 m offshore. Therefore there is some confidence in the sediment-transport model predictions that the overall behaviour of the storm cut and fill pattern off Lyall Bay Beach is being reasonably replicated for an active southerly-wind period with the attendant energetic wave dynamics.

It should be noted that uncertainties in the sand depositional and erosion seabed heights in this near shore region are likely to be present because of inaccuracies in the local bathymetry, such as the interpolation between the beach topography and the beginning of the multi-beam bathymetry in about 2 m water depth (Section 5.3), the model grid spatial resolution (~ 50 m cells), and the application of the sediment pick-up and transport functions used in the model to the Lyall Bay situation. Nevertheless, the model presents realistic beach-profile behaviour on scales of weeks to months that is representative of the dominant near-shore processes that would occur in response to the dominant southerly wave field, with only weak tidal currents.

The relative change in erosion or deposition patterns is shown in Figure 3-19, where the sea-bed heights predicted for the runway-extension were subtracted from the sea-bed heights for the existing situation. In Figure 3-19 erosion (E) means the sea bed would have eroded more than for the existing situation, and deposition (D) means the sea bed would have accreted more than for the existing situation.

The predicted net change in sea-bed heights, following the construction of the runway extension, would be small (<1–2 cm at seasonal timescales) over much of Lyall Bay (Figure 3-19), particularly within the inner Lyall Bay, the main beach and the areas along the western and outer-eastern perimeter of the Bay. The main changes in sea-bed heights, of the order of a few cm’s at seasonal timescales, are localised around the extended-runway, where the existing sea-bed has to adjust to the presence of the rock dyke and the associated wave interaction with the structure, which it does

Wellington Airport Runway Extension 57

by eroding sand from adjacent to sections the dyke and depositing the sand further offshore, especially to the south, to form a new equilibrium sea-bed profile.

Under the time scales and environmental conditions that were simulated, the model predicts that there may be some localised deposition on the SW and SE corners of the dyke. This is not to say that over much longer periods, varying seasonal weather conditions and swell frequency that such sand deposits next to the runway-extension rock dyke will not re-mobilise and move e.g., a cut and offshore fill response or alongshore drift around the end of the revetment, given the high wave exposure of this area.

The model also predicts narrow bands of both increased erosion and deposition in the order of cm to the SW of the extended runway, which are likely to be due to changes in wind-generated current patterns during southerlies (e.g., right-hand panel of Figure 3-11) transporting wave re-suspended sediment, rather than changes in waves (which are shown to be minimal in this area as shown by the high-resolution ARTEMIS modelling in Section 5).

Figure 3-19: Net or relative change in the response of sea-bed heights with the proposed runway in place over the ~7-week simulation. Categorized as erosion (E), deposition (D) or negligible change (0) for the proposed runway extension compared to the existing situation.

3.6 Discussion and summary This Section of the modelling investigations described the existing tidal and wind-driven circulation in Lyall Bay and established the potential impact of the proposed Wellington Airport runway extension on hydrodynamics (tidal and wind-generated) and sediment transport inside the Bay.

The model, forced by observed tide and wind, indicated that the mean or residual circulation alone within Lyall Bay is weak. The maximum tide- and wind-driven residual currents predicted during the 2014 field study of the region were in order of a few cm/s. The residual circulation pattern shows the

58 Wellington Airport Runway Extension

formation of a clockwise eddy in head of the Bay and a counter-clockwise eddy in the wider mid- section of the Bay. The proposed airport extension is predicted to intercept the residual flow off the southern extent of the existing runway, forming a weaker residual current shadow zone in the vicinity of the proposed runway extension. Model simulations suggest the new runway extension would have a limited effect on the scale and flow direction of the two predicted residual eddies in proximity to the extension, and minimal effect on flushing of water masses within Lyall Bay. The residual currents will also be unaffected in the outer Bay by the runway extension and therefore will not affect the dispersion of treated wastewater from the marine outfall discharging from the Moa Point wastewater treatment plant.

To isolate the effect of the runway extension on wind-driven circulation within Lyall Bay, simulations for the pre- and post-construction configurations utilised idealised 3-day wind forcing peaking at 22 m/s for both strong southerlies and northerlies superimposed on a repeating mean tide cycle. This demonstrated the hydrodynamic effects of strong winds directed from the south and north with both directions dominating the wind climate in the Bay. The simulations showed that a southerly wind drives water northward (down-wind) on both the west and east perimeters of Lyall Bay, forming a clockwise eddy at the head of the Bay and a counter-clockwise eddy in the middle of the Bay.

In contrast under northerly wind conditions, southward flows along the eastern and western periphery of the Bay (also down-wind) exit from Lyall Bay, with a return northward inflow through the centre of the Bay up to the nearshore waters off Lyall Bay Beach, where it bifurcates. The hydrodynamic model simulations, comparing the existing situation with the extended runway in place, predicts that the proposed runway extension will have a negligible impact on the northerly- wind driven flow circulation but does cause some minor reduction to southerly-wind driven circulation in the vicinity of the extension.

Introducing waves as well as tide and wind-driven forcing into a sediment-transport model, showed the predicted net change in sea-bed heights, following the construction of the runway extension, would be negligible over much of Lyall Bay, including the nearshore area off Lyall Bay Beach. Localised changes in sea-bed heights (within 50 m) around the extended runway are likely to be minor, particularly after the existing sea-bed topography has adjusted to the presence of the rock dyke and the associated wave interaction with the structure. The local morphological adjustments seaward of the rock apron, adjacent to the rock/accropode dyke, will occur through some erosion of sand from adjacent to the dyke and depositing the sand further offshore, especially to the south, to form a new equilibrium seabed profile – although this is likely to be a dynamic “equilibrium” depending on varying seasonal weather conditions and swell frequency or persistence.

Under the seasonal time scales and environmental conditions that were simulated, the model predicts that there may be some localised deposition on the SW and SE corners of the dyke. This is not to say that over much longer periods, varying seasonal weather conditions and swell frequency that such sand deposits next to the runway-extension rock dyke will not re-mobilise and move, given the high wave exposure of this area. Note: the terminus of the existing runway revetment is positioned in an area of submerged rock outcrop (Figure 2-1) so the existing in-situ morphological analogue of the seabed response built up over decades is not easily amenable to being applied to the end of the extended runway dyke where a deeper layer of mobile sand cover will be present without rock outcrops.

Wellington Airport Runway Extension 59

4 Wave climate: Spectral wave modelling (SWAN) Model simulations of waves focused on Lyall Bay for the AEE investigations have been undertaken at two different spatial scales:

. the wider south Wellington coast between Baring Head (Ōrua-pouanui) and Sinclair Head (Te Rimurapa), where the spectral wave model SWAN has been used to simulate spatial variations in the wave climate for the existing environment, verified against the Baring Head wavebuoy dataset

. solely Lyall Bay, using a phase-resolving wave model to capture both wave refraction and diffraction processes on a high-resolution grid, and hence compare the local effect of the runway extension on wave processes.

This Section describes the wider-scale SWAN modelling, to provide information on the spatial variability in wave climate within Lyall Bay and how relates to waves outside the Bay.

The finer-scale wave modelling is described in Section 6, where the effects of the proposed runway extension are tested.

4.1 South Wellington coast wave modelling (SWAN) The SWAN model (Booij et al. 1999; Ris et al. 1999) is a spectral wave model intended for shallow water applications in coastal and estuarine environments. It computes the evolution of the wave energy spectrum in position (x, y) and time (t), explicitly taking into account the various physical processes acting on waves in shallow water. These include the effects of refraction by currents and bottom variation, and the processes of wind generation, white-capping, bottom friction, quadruplet wave-wave interactions, triad wave-wave interactions and depth-induced breaking. The model can incorporate boundary conditions representing waves arriving from outside the model domain.

Hence the SWAN model is an appropriate tool for simulating wave conditions over a study region incorporating Lyall Bay and surrounding waters of Cook Strait off the south Wellington coast. We carried out SWAN simulations both in standalone mode, to test and validate the model and to characterise spatial variability in wave conditions within the study region, and also as a fully-coupled component of the hydrodynamic/morphodynamic modelling described in the next Section. Except where noted, the same methods were used in both applications of the SWAN model.

SWAN simulations were carried out on a curvilinear grid developed for hydrodynamic modelling of Lyall Bay and approaches as described above. The extent of this grid is shown in Figure 4-1. New Zealand Map Grid (NZMG) coordinates were used.

A SWAN simulation was run from 17-Aug 2014 (12:00 NZST = 00:00 UTC) to 11-Oct 2014 (12:00 NZST = 00:00 UTC), covering the period of Lyall Bay instrument deployments described in Section 2.

60 Wellington Airport Runway Extension

Figure 4-1: Extent of the SWAN modelling domain, with instrument sites marked. See Figure 2-1 for a closer view of the Lyall Bay sites and associated sea-bed bathymetry.

4.2 External forcing In order to simulate wave development in a given region, it is necessary to specify the winds blowing over the region. Waves entering the region through any open boundaries also need to be included, while the effects of changing water levels and currents can also be accounted for if these can be provided. In general all of these inputs vary both in space and time.

Weather, tide and wave models have been developed as part of NIWA’s EcoConnect forecasting system for weather-related hazards. The presently operational version of this system now includes a New Zealand Convective Scale (NZCSM) weather model that provides 42-hour forecasts four times daily, on a rotated pole grid with a spatial resolution of approximately 1.5 km.

NIWA has recently developed a wave forecast (NZWAVE-2) on a latitude/longitude grid at approximately 2 km resolution, taking NZCSM winds as inputs, providing 36-hour forecasts four times daily. Outputs from these forecasts since July 2014 have been validated monthly against wave buoy measurements at Baring Head, and off Steep Head, Banks Peninsula. These forecasts have not yet been implemented into the operational EcoConnect forecast delivery system.

To provide inputs for the present SWAN simulations, wind fields at 10 m elevation from NZCSM forecasts were interpolated to a regular grid, specified by NZMG eastings from 1738000 to 1762000, and NZMG northings from 5407000 to 5436000, at 1000 m intervals in both directions. Wind fields were available at 30 minute intervals, and fields from the first 6 hours of each successive forecast were merged.

Wellington Airport Runway Extension 61

Swell entering the SWAN model domain was represented by applying full directional spectra obtained from NZWAVE-2 forecasts, interpolated to 19 points (every 20th grid cell) along the open southern boundary of the curvilinear model grid (Figure 4-1). As for the wind fields, these data were available at 30 minute intervals, merging records from the first 6 hours of each successive NZWAVE-2 forecast.

In the fully-coupled simulations described in the next Section, in which the SWAN wave model was incorporated into hydrodynamic/sediment transport simulations, it was found that a temporal smoothing needed to be applied to these swell boundary conditions to avoid numerical instabilities, using a 12 hour running mean (i.e., replacing each boundary directional spectrum at time t by the time-averaged of spectra at the same location from t-6 hours to t+6 hours). This averaging was not applied in the standalone SWAN simulations reported in the present Section.

In the fully-coupled simulations, wave conditions computed by SWAN were influenced by water levels and currents (both tidal and non-tidal) provided by the hydrodynamic model. For standalone simulations, purely tidal levels and currents were used, using the NIWA EEZ tidal model. This is based on an unstructured mesh that provides much finer resolution in coastal waters than the 12 km regular grid used to display operational outputs in the EcoConnect forecast system. Hence for the present study, tidal currents and sea levels were input on a regular grid covering the same extent as used for the wind fields described above, but at 100 m resolution.

The SWAN model performs interpolation of input wind, sea level and current fields to the curvilinear computational grid used for wave model computations.

4.3 Model verification NIWA operates a Waverider™ buoy on behalf of Greater Wellington Regional Council (GWRC) off Baring Head, south of the entrance to Wellington Harbour. The buoy location lies inside the SWAN model domain (Figure 4-1). Two wave records were also obtained in the field component of the present AEE investigation (Section 2): from a near-surface Dobie wave gauge, via a pressure sensor, at an outer site in the entrance of Lyall Bay, and a moored Acoustic Doppler Current Profiler (ADCP) at a near shore site, west of the end of the existing runway. The locations of these instruments are given in Table 4-1, and marked on Figure 4-1 (and for Lyall Bay in Figure 2-1). Wave statistics from the SWAN simulation (17-Aug to 11-Oct 2014) are compared with data from the three sites in the following Figures.

Table 4-1: Locations of wave measurements used to validate the SWAN simulations.

Site No. Instrument Longitude Latitude NZMG easting NZMG northing 1 ADCP (inshore) 174.7988 -41.3488 1750796.15 5422253.85 2 Dobie (offshore) 174.8022 -41.3371 1750483.38 5420964.52 3 Baring Head Waverider 174.8470 -41.4020 1754391.31 5414969.52

The variation of significant wave height (Figure 4-2) through the study period is well matched by the model at all three sites. While there is some under- and over-prediction at different times, there is no consistent bias evident. Also, the reduction of wave energy at the nearshore ADCP site relative to the more offshore Baring Head and Dobie sites is well reproduced.

62 Wellington Airport Runway Extension

Where other wave statistics are available, e.g., from the Directional Waverider™ at Baring Head (Figure 4-3), these are also represented well by the SWAN model.

Figure 4-2: Comparison of significant wave height (Hm0) measured by instruments (black) with corresponding values predicted by the SWAN simulation (red) at the three measurement locations.

Figure 4-3: Comparison of wave statistics measured by the Baring Head Waverider buoy (black), with corresponding values predicted by the SWAN simulation (red). From top to bottom: peak wave period, mean wave direction, peak wave direction and directional spreading.

Wellington Airport Runway Extension 63

4.4 Lyall Bay wave conditions The early part of the simulation, from 17 August (day = 228 on Figure 4-2 and Figure 4-3) to 10 September (day = 252) was a period of consistent southerly wave conditions in Cook Strait (apart from a brief shift in modelled, but not measured, mean wave direction around day 245). In this early part of the simulation, peak and mean wave directions were both consistently in the range 180-200°, peak period was generally in the 8-15 second range, while directional spread was in the range 20-40°, indicating a swell-dominated sea state.

An example of the model output during this period is shown in Figure 4-4, which shows the spatial distribution of significant wave height predicted by the model at 00:00 NZST on the 22 August (day = 233.0), during the first storm event of the record. The wind field (plotted with red arrows) shows strong southerly conditions over the model domain, and presumably over a more extensive fetch extending further south. This results in significant wave heights around 5 m over most of the model domain (lower panel), with mean wave direction (black arrows) aligned with the wind. Wave heights reduce in some nearshore areas, e.g., Fitzroy Bay in the lee of Baring Head and entering Wellington Harbour.

In these conditions, waves entering Lyall Bay (top panel of Figure 4-4) largely maintain their energy through the deeper water of the outer Bay, but refract shoreward and lose energy in the shallower water on both sides of the outer Bay, and in the inner Bay west of the airport. We note that at this stage of the tide the drying of the extensive rock platform at the western entrance of the outer Bay provides some sheltering, but that there are areas of increased wave height produced by shoaling over two nearby submerged reefs (the submerged southward extension of this headland at Te Raekaihau, and another reef east of the headland between Te Raekaihau and Waitaha Cove – see Figure 2-1). The north-south alignment of these reefs will tend to focus waves from the south.

From 10–21 September (day 252–263), there were several occasions in which mean and peak wave directions shifted towards the north, while directional spreading increased, indicating a mix of wind sea locally generated by northerlies combined with some continuing component of southerly swell.

An example model output from this period is shown in Figure 4-5, which shows wave conditions at 02:00 NZST on the 19th of September (day = 261.083). At this time a northerly wind blows over the model domain, producing wave heights steadily increasing in an offshore direction, with values in the 1-2 m range, and mean directions from the north, in the southern half of the domain. Within Lyall Bay, however, the limited growth of the wind sea means that the remaining southerly swell is still predominant, resulting in mean wave direction still being generally southerly.

On 22-23 September (day 264-265), a large southerly storm event occurred, producing significant wave heights over 6 m at Baring Head and the Dobie wave gauge site, and just under 5 m at the nearshore ADCP site (Figure 4-2).

An example of wave conditions during this storm is shown in Figure 4-6, which shows model outputs at 00:00 NZST on the 23 September (day = 265.0). While wave heights are scaled up somewhat, the pattern is generally consistent with that seen in the earlier southerly storm (Figure 4-4).

Finally, wave statistics output from the model were averaged over the 55-day simulation period, excluding an initial 12-hour spin-up time. While a simple time average was taken for significant wave height, a mean value of mean wave direction was obtained by vector averaging, weighted by wave height. This avoids the problem of averaging circular quantities, where a simple average of directions

64 Wellington Airport Runway Extension

that fluctuate across the jump between 360° and 0° would erroneously give an average of approximately 180°.

The results are plotted in Figure 4-7 which, though scaled down in magnitude, shows a similar pattern of spatial variability in wave height to what was seen in the two southerly events of Figure 4-4 and Figure 4-6.

Wellington Airport Runway Extension 65

Figure 4-4: Wave conditions at 00:00 NZST on 22-Aug 2014 from the SWAN simulation, in Lyall Bay (top panel) and over the full model domain (bottom panel). The colour scale indicates significant wave height. Black arrows show mean wave direction, red arrows show wind speed and direction, and blue arrows show tidal currents (all at different subsampling intervals).

66 Wellington Airport Runway Extension

Figure 4-5: Wave conditions at 02:00 NZST on 19-Sept 2014 from the SWAN simulation, in Lyall Bay (top panel) and over the full model domain (bottom panel). The colour scale indicates significant wave height. Black arrows show mean wave direction, red arrows show wind speed and direction, and blue arrows show tidal currents (all at different subsampling intervals).

Wellington Airport Runway Extension 67

Figure 4-6: Wave conditions at 00:00 NZST on 23-Sept 2014 from the SWAN simulation, in Lyall Bay (top panel) and over the full model domain (bottom panel). The colour scale indicates significant wave height. Black arrows show mean wave direction, red arrows show wind speed and direction, and blue arrows show tidal currents (all at different subsampling intervals).

68 Wellington Airport Runway Extension

Figure 4-7: Wave conditions averaged over the 55-day SWAN simulation, in Lyall Bay (top panel) and over the full model domain (bottom panel). Mean significant wave height is shown by the colour scale, and contours at 0.1 m intervals. Black arrows show mean wave direction.

Wellington Airport Runway Extension 69

4.5 Wave statistics at selected output locations The 55-day (~7-week) wave simulation undertaken is shorter than would be required to produce robust and fully-representative wave climate statistics for the region. However, summary statistics from the simulation illustrate spatial variability within Lyall Bay if not a full representation of longer- term temporal variability.

To this end, a set of 18 output locations was selected, as shown in Figure 4-8 (except for site 3 at the Baring Head Waverider location). The first three sites correspond to the instrument sites shown in Figure 4-1. Time series of wave statistics were extracted at these locations.

Figure 4-8: Locations of sites selected for output of wave statistics from the ~7-week SWAN simulation. Site 3 (Baring Head) is not shown.

To provide context, Figure 4-9 shows wind roses, i.e., the joint distribution of wind speed and direction, from sites 3 (Baring Head) and 6 (central Lyall Bay). The latter can be taken as representative of all Lyall Bay sites: the 1.5 km resolution of the NZCSM winds used allows for little meaningful variability of wind statistics within the Bay.

At both sites we see a predominance of winds from either the northerly or southerly quadrants, with both reasonably equally represented. There is a small clockwise rotation of the wind alignment in going from the Baring Head site into Lyall Bay. That is, at site 3 (Baring Head), the generally “southerly” winds have a slightly more south-easterly component than in Lyall Bay, consistent with local alignment of the coast. In northerly conditions, winds in Lyall Bay have a slightly wider spread of directions than at Baring Head.

70 Wellington Airport Runway Extension

Figure 4-9: Wind roses for output sites 3–Baring Head (left) and 6–inner Lyall Bay (right) derived from the SWAN simulation.

The following Figures show occurrence distributions of wave statistics from the 18 sites. Each Figure has plots arranged in a grid, roughly corresponding to the geographic location of the 18 sites, with most of the Lyall Bay sites in the upper 4×4 part of the grid, while site 3 (Baring Head) and site 18 are shown in the bottom row. The Figures are small, but the main aim is to show the pattern of spatial variability in these wave statistical distributions within Lyall Bay.

Figure 4-10 shows the occurrence of significant wave height (Hm0), in the form of probability distribution functions scaled to the number of occurrences. Starting with site 3 (Baring Head, bottom row), we see a range of wave heights, with two maxima, around 1 m and 2 m, tailing off to progressively lower occurrences for heights out to a maximum between 5–6 m. The bimodal distribution may be an artefact of the relatively short record containing an over-representation of larger wave events, and such a bimodal form is not being seen in distributions from multi-year buoy or model records from this location.

At sites along the outer entrance to Lyall Bay (second row from the bottom, and site 18 in the bottom row), the distribution has shifted somewhat to include a higher representation of wave heights less than 1 m than at site 3, largely due to the influence of northerly conditions, but the higher-energy tail has not been substantially modified, as these events arise in southerly conditions when there is a relatively small variation in wave height between these locations. Moving to nearshore and inner sites (top and sides of Figure 4-10), subject to sheltering, refraction, depth- limited breaking and bed friction, the height distributions become more limited in range.

Figure 4-11 shows occurrence distributions for the peak wave period (Tp). This statistic represents the period of the dominant (highest energy) component of sea state at any time, and will usually correspond to that of any swell present. These distributions show a similar range and distribution of peak periods at each site.

Figure 4-12 shows occurrence distributions for the second-moment mean wave period (Tm02). This statistic is derived from the second frequency moment of the wave spectrum, i.e., a weighted average of the mix of wave periods present in the sea state at any time, with a relatively high weighting given to short waves. This approximately corresponds to the mean period derived by zero- crossing analysis. We see that site 3 (Baring Head) has a relatively high occurrence of Tm02 below

Wellington Airport Runway Extension 71

5 seconds compared to all the locations within Lyall Bay. Site 3 will generally experience a relatively large wind sea component of sea state at most times, including northerlies, which will result in a small value of Tm02.

Locations in Lyall Bay, on the other hand, will have minimal wind sea during northerlies, so any southerly swell present will dominate the mean period calculated in this way.

Figure 4-13 present wave roses giving the joint occurrence of mean wave direction (indicated by the length of bars in directions oriented towards the direction from which waves arrive) and significant wave height (indicated by the length of bins into which each direction bar is subdivided). At site 3 there is a strong predominance of waves from the south and SSW, but with some representation of waves from other directions in the NW and SW quadrants. At all the Lyall Bay sites, the occurrence of mean wave directions from the NW quadrant vanishes. The deeper of the Lyall Bay sites are dominated by southerly waves, but in shallower water the dominant direction shifts towards a locally shore-normal direction. The simulated wave rose at the ADCP (site 1), which is the panel three in from the left and two down in Figure 4-13, showing a predominantly southerly wave climate is closely aligned with the rose generated from the ADCP data in Figure 2-8.

Finally, Figure 4-14 shows occurrence distributions for root-mean-square bed orbital velocity (Urms). This is a parameter of direct relevance to resuspension of sediment from the sea bed. It is strongly dependent on water depth, due to the rapid attenuation of wave orbital motion with distance below the water surface. Hence the deeper sites show distributions of Urms heavily weighted to lower values, while shallower sites are subject to much higher bed orbital velocities.

72 Wellington Airport Runway Extension

Figure 4-10: Probability distribution functions for significant wave height derived from the SWAN simulation at the 18 output sites. Plots are arranged to approximate the geographical distribution of the sites. Cumulative probability integrated across the distribution = 1.

Wellington Airport Runway Extension 73

Figure 4-11: Probability distribution functions (vertical axis) for peak wave period derived from the SWAN simulation at the 18 output sites. Plots are arranged to approximate the geographical distribution of the sites. Cumulative probability integrated across the distribution = 1.

74 Wellington Airport Runway Extension

Figure 4-12: Probability distribution functions (vertical axis) for second moment mean wave period derived from the SWAN simulation at the 18 output sites.Plots are arranged to approximate the geographical distribution of the sites. Cumulative probability integrated across the distribution = 1.

Wellington Airport Runway Extension 75

Figure 4-13: Wave roses, i.e., joint occurrence distributions for mean wave direction and significant wave height at the 18 output sites. Plots are arranged to approximate the geographical distribution of the sites. Bars are oriented towards directions from which waves arrive.

76 Wellington Airport Runway Extension

Figure 4-14: Probability distribution functions for root-mean-square bed orbital velocity derived from the SWAN simulation at the 18 output sites. Plots are arranged to approximate the geographical distribution of the sites. Cumulative probability integrated across the distribution =1.

4.6 Extreme wave climate No further specific work was undertaken by NIWA on extreme waves for this Report, but was instead undertaken by AECOM. as part of the coastal engineering design work on the perimeter rock dyke and the accropode protection (AECOM, 2015a). For completeness in describing the existing coastal environment, where waves play a dominant role, a brief overview of extreme wave climate is provided here, largely based on the Baring Head wavebuoy record and drawing from the desk-top review (AES Ltd. and NIWA, 2013).

Wellington Airport Runway Extension 77

Wellington’s south coast, including Lyall Bay, is very exposed to waves and swell with a long fetch from the Southern Ocean for south to south-east winds. Maximum wave heights above 10 m occur occasionally at Baring Head, mostly from deep low-pressure systems to the east of central New Zealand.

NIWA has operated a surface non-directional wave buoy in 45 m water depth off Baring Head since 21 April 1995, with the 18-year record now one of the two longest wave datasets in New Zealand (the other is the Maui platform record).

The largest waves seen in the Baring Head record were measured during the late evening of the recent storm on 20 June 2013 that battered Wellington and closed the airport. The highest (maximum) waves measured were typically 15 m for the period around midnight (NIWA web site: http://www.niwa.co.nz/news/storm-and-snow-information-update).

Anecdotal information records the wave height at the time of the Wahine disaster in April 1968 at 12 to 14 m high (Beca Carter–Caldwell Connell, 1980; Harris, 1990).

Another sizeable storm occurred on Waitangi Day (6 February) of 2002 from a deep low off Banks Peninsula, with maximum wave heights reaching 13 m (Figure 4-15) and an unnamed storm in 1989 generated up to 12 m waves (Carter et al. 2002). A more representative measure – the significant wave height (or highest one third of the waves) – exceeded 8 m during the 6 February 2002 storm.

Therefore, the 20 June 2013 wave height of 15 m is likely to be the largest wave height experienced at Baring Head in the last 50 years.

The above SWAN modelling focused on Lyall Bay over the ~7-week period covering the 2014 field deployment. The results, based on the largest 80 km/hr southerly wind event of 22 Sept 2014, indicates the extreme significant wave height off the end of the runway will be partially attenuated compared with the more exposed area off Baring Head where the wavebuoy is located.

Figure 4-15: “Waitangi-Day” 2002 waves off Baring Head with NIWA’s Brodie Building at Greta Point for scale. [From: Carter et al. 2002].

78 Wellington Airport Runway Extension

5 Phase-resolving wave model results (ARTEMIS) In order to resolve wave patterns around a structure such as the Wellington International Airport runway extension into Lyall Bay, a high-resolution wave model is required that incorporates local changes in wave processes (US Army Corps, 1984) particularly:

. Refraction – process by which the direction of a wave moving in shallow water at an angle to the seabed contours is changed (e.g., the part of the wave in shallower water moves more slowly than the part still advancing in deeper water). Refraction explains why swell and longer-period wind waves straighten up across inner Lyall Bay, even if the waves approach the outer Bay slightly to the east or west of a southerly direction.

. Diffraction – where wave energy is transmitted laterally along a wave crest when part of a train of waves is interrupted by a barrier such as a rock revetment or breakwater of a comparable physical scale to the wave length. Diffraction is the "spreading" of waves into the sheltered region within the barrier's geometric shadow. This currently occurs in the lee of the spur groyne and associated rocky outcrop on the western side of the existing runway revetment.

. Reflection– part of an incident wave that is returned seaward when a wave impinges on a steep beach, rocky outcrop, rock revetment or other reflecting structure. Porous multi-layer rock armour layers and akmons or accropodes provide substantial absorption of wave energy, thereby reducing reflection, which is controlled in models by a reflection coefficient.

. Wave breaking – depth-induced steepening and eventual breaking of waves as they shoal into shallower water, and depends on the wave period how far offshore they break e.g., swell waves feel the bottom in greater depths than wind sea.

NIWA has access to the ARTEMIS9 model (Aelbrecht, 1997), which is part of the widely-applied TELEMAC–MASCARET modelling system developed by Laboratoire National d'Hydraulique (Electricite de France).

ARTEMIS was used to compare wave patterns in Lyall Bay with the existing runway revetment to the wave patterns and heights predicted for the proposed extended runway revetment. The results will be presented as a change in significant wave height relative to the existing runway situation, which is in line with assessment of effects relative to the existing environment. ARTEMIS only simulates the propagation of regular (sinusoidal) waves applied to the boundary, but is not able to simulate irregular waves and their characteristics. More detailed high-resolution wave modelling was undertaken by DHI, using irregular measured waves propagating through the Entrance for occasions when surfing quality was good, to further investigate the effect of the runway extension on wave heights and characteristics (e.g., wave steepness, wave peakiness). These aspects are important considerations for assessing the effects on surfing quality and safety of recreational users, are described in the assessment report by DHI (2016).

In consideration of post-extension changes on the range of waves in Lyall Bay we have focused mainly on the more commonly-occurring wave heights of 1–3 m for longer period wind waves (e.g., 8 seconds) and the 10 – 15 second swell wave spectrum. Surfing is undertaken primarily at ‘The Corner’ which is located to the eastern end of Lyall Bay (Figure 3-16), but also in the central part of

9 Agitation and Refraction with Telemac on a MIldSlope http://www.opentelemac.org/index.php/presentation?id=19

Wellington Airport Runway Extension 79

the Bay. DHI have undertaken a complementary high-resolution modelling study on irregular waves, specifically addressing effects on surfing-wave quality and public safety of beach recreational users (DHI, 2015).

5.1 ARTEMIS wave model The ARTEMIS wave model is a finite element phase-resolving wave module for the TELEMAC– MASCARET software. It is used to describe the combined effects of diffraction, refraction, reflection and wave breaking for water waves propagating over bathymetry and due to interactions with lateral boundaries—like breakwaters and coastlines.

The ARTEMIS code solves the extended elliptic mild-slope equation using finite element mesh techniques. The mild slope equation which was proposed by (Berkoff 1972) is applicable for computations of refraction and diffraction of linear waves. The following elliptic form of the mild slope equation includes additional dissipative effects (Booji, 1981).

2 ∇(퐶퐶 𝑔 ∇φ ) + 퐶퐶 𝑔 (푘 + 𝑖푘휇)φ = 0 (5.1)

푊 where φ is the reduced two-dimensional velocity potential, μ = 1/2 is the dissipation (퐶퐶𝑔) 2 coefficient with W a dissipation function, k is the wavenumber, C is the wave celerity and Cg is the group velocity of the waves. Usual approximations associated with linear theory apply, with irregular waves considered as the linear superposition of regular (sinusoidal) waves (Booji, 1981).

It is an approximate model, deriving its name from being originally developed for wave propagation over mild slopes of the sea floor. The method is suitable for modelling wave resonance and seiching in harbours and wave fields due to combined refraction/diffraction/reflection in small bays. However refraction by currents is not included.

Since Release 6.1, the TELEMAC ARTEMIS wave module can take into account the effects of a rapidly varying topography in the mild-slope equation. The main results, for every node of the mesh, provide a global solution of the significant wave height Hs and phase, wave incidence, breaking rate and direction.

5.2 ARTEMIS mesh development

5.2.1 Bathymetry For Lyall Bay, an independent ARC GIS terrain model was developed that has a considerably higher resolution (5 metres) compared with the seabed terrain model developed for the full Wellington Harbour bathymetry (50 m) used in the Delft3D hydrodynamic and sediment transport modelling (Section 3).

The ARC GIS terrain model data included high resolution multi-beam sounding data collected by NIWA (Mackay & Mitchell, 2014) and LiDAR data for the intertidal areas in the Bay obtained from Greater Wellington Regional Council (Figure 5-1). The multi-beam data was collected to depths of no shallower than ~2 m below Chart Datum (CD) and the LiDAR data covers intertidal zones to down to approximately the 0 m CD contour.

There was no available seabed survey data in the intervening 0–2 m below CD nearshore zone. Consequently, a linear interpretation of seabed elevations for this zone was implemented in the

80 Wellington Airport Runway Extension

terrain model bathymetry (see zone marked in Figure 5-1), which approximately matches offshore profile slopes (excluding cut & fill bar features) for Lyall Bay demonstrated by Pickrill (1979) for 1977 measurements. This nearshore zone is generally where waves break before running up the intertidal beach, and will naturally vary especially after southerly storms, so even if this seabed bathymetry was available for one point in time, it wouldn’t necessarily match the seabed profile for different storms or post-storm situations. The assumption is that any wave refraction and diffraction processes influenced by the existing and proposed runway extensions into Lyall Bay mainly occur seaward of the nearshore breaker zone, more so for swell, so linear interpolation of the bathymetry in this area is justified. In any case, the results are presented as relative changes in waves before and after the proposed runway extension, where the seabed bathymetry, excluding the proposed runway extension area, is the same in both cases.

Figure 5-1: Spatially averaged terrain node data illustrating the zone between 0 - 2m where no bathymetric data exists. Mesh development One of the challenges with ARTEMIS is that a fine or high-resolution mesh is required to cover at least 7 points within a wave length for the peak period and no less than 3 points across a wavelength for the shortest period. For the Lyall Bay wave modelling, the wave periods of interest are between 4–20 seconds. This means that wave lengths range from approximately 25 to 600 m, so to resolve a 4-second period the grid resolution had to be refined down to a node spacing of approximately 5 m. For Lyall Bay two finite element mesh grids were developed: one of the currently available bathymetry and the existing runway revetment, and a second grid which included the proposed runway extension footprint of ~400 m (see Figure 5-2 and Figure 5-3).

Wellington Airport Runway Extension 81

Figure 5-2: Left image shows the Lyall Bay bathymetry with the existing runway and the finite element mesh used in the ARTEMIS wave model. Right image shows a 3D illustration of bathymetry looking to the NW.

82 Wellington Airport Runway Extension

Figure 5-3: Left image shows the Lyall Bay bathymetry with the proposed airport extension added and finite element grid used in the ARTEMIS wave model. Right image shows a 3D illustration of bathymetry focusing on the proposed airport extension looking to the NW.

5.3 ARTEMIS model set-up (physical and numerical parameters). ARTEMIS is implemented on the domains illustrated in section 5.2. The coastal boundary conditions are parameterised with three key variables which are used to “drive” the model: direction of approaching waves, phase -shift, and incident wave height in this case for monochromatic waves. Table 5-1 represents the physical and numerical parameters used when running the ARTEMIS wave model.

Table 5-1: Physical and numerical parameters used in the set-up of ARTEMIS. Monochromatic Waves : Yes Period Scanning : Yes /Range [5 to 20 seconds in 1 sec increments] Bathymetric Breaking : Yes /Dally formulations for regular waves GDALLY : 0.35 KDALLY : 0.100 Bottom friction : Yes /constant Rapidly varying topography : 3 /both gradient and curvature effects Dissipation coefficient : 2 / Putman and Johnson‘s formula

Wellington Airport Runway Extension 83

5.4 Boundary reflection Wave reflection at the shoreline is an important aspect of harbour wave agitation models. The reflection coefficient represents a ratio of the amplitude of the wave that approaches the coast to the amplitude that is reflected away from the coast. The reflection coefficient used in the model is based on results from Zanuttigh and van der Meer (2006). The ARTEMIS boundary file was modified to incorporate the following equations. Sensitivity testing showed that only the groyne structure and the front southern exposed end of the proposed airport were sensitive to the reflection coefficient.

Nevertheless, the ARTEMIS model coastal boundary was divided into various sections and Kr was calculated for each section inside the model. This allowed the model to be run with period scanning and Kr to change accordingly.

푏 퐾푟 = 푡푎푛ℎ(푎. 휉휊 ) (5.2) where the parameter 휉 (expanded Iribarren number) represent bimodal spectra for shallow water with flat spectra (Zanuttigh and van der Meer, 2006), defined as:

tan 훼 휉 = (5.3) 2 √(2휋퐻푚0)/(푔푇푚−1,0)

with the calibration values of the coefficients a and b obtained from Zanuttigh and van der Meer (2006) and Engineers U.A.C.O. (2002) which depend entirely on the structure surface. H is wave height, g is gravity and T wave period for a singled-peaked spectra.

5.5 Validation Initially the ARTEMIS model was to be validated against two ADCPs deployed in Lyall Bay at the sites in Figure 2-1. However, the outer boundary ADCP was lost during deployment with the only wave data from this site being sourced from the DOBIE pressure gauge moored below the surface (with no direction available). This poses some technical difficulties comparing waves at each site as the data is sourced from different instruments and processed differently. Further, an attempt was made to find periods of a few hours in the ~7-week field deployment when the period and wave height were reasonably constant to mimic a monochromatic or regular wave situation, which is what is modelled by ARTEMIS, however the measured wave conditions exhibited considerable temporal variability, which precluded such a comparison.

Consequently, a visual validation of the model was made comparing spatial wave-crest patterns and breaking zones throughout Lyall Bay from historic aerial photos archived in the Google Earth software (an example is shown in Figure 5-4). This visual comparison with the vertical view of the predicted wave fields for similar incident waves, shows a close match with a number of features, including the extent of surf zones (white areas in Figure 5-4), zones of refracting shoaling waves turning into both coasts of the outer Bay, more quiescent areas in coves and the western side of the Bay and a particularly close match with the overall refracting and diffracting wave crest patterns including in The Corner area adjacent to the existing runway revetment. There are some slight local differences in the wave-crest pattern in the predicted wave field at the tip of the spur groyne – an area of rapidly changing bathymetry associated with the rock outcrop. Furthermore, results from the ARTEMIS model suggest that for 1 and 1.5 m monochromatic waves applied at the outer-Bay boundary, the wave attenuation between the two mooring sites (Site 1–inner ADCP and Site 2–outer

84 Wellington Airport Runway Extension

Bay) is between 0.03 and 0.05 m, which is within the limits of measurable accuracy of the instruments.

Figure 5-4: ARTEMIS model predicted surface water level as a result of a 1.5 m wave at the boundary with a wave period of 12 seconds (LEFT). Google Earth image of Lyall Bay taken on 24 July 2014 for a wave height of 1–1.5m and a period ~10 seconds (RIGHT). Wave measurements for the latter were derived from the Baring Head wavebuoy.

5.6 Results showing wave changes due to the proposed runway extension

5.6.1 Changes in significant wave height at selected locations Table 5-2 to Table 5-17 show the effect of the proposed Wellington International Airport runway extension on the Lyall Bay wave patterns from modelled results extracted from 8 point locations (“P”) illustrated in Figure 5-5, to provide a limited array of locations to compare the before-and-after construction scenarios. A denser spread of locations was focused on the eastern side of the Bay, including The Corner. However, as will be shown later, extracting results from specific points only provides part of the picture – which complements the analysis of the spatial changes in wave heights in Section 5.6.3.

The first set of eight Tables show the absolute change in significant wave height locally for each extraction site P1–P8, ordered as matrix of wave period and wave height applied at the outer-Bay boundary (not the local wave height). When examining the table values, remember that they list differences in wave height at the relevant point after the construction of the runway extension, whereas the table axes give the offshore wave boundary conditions at the Bay entrance. Note, the wave height at the “P” sites will usually be less than the incident wave height at the offshore boundary as waves are attenuated as they propagate into the Bay. A negative value for change listed in the Table implies a reduction in wave height by that amount at the relevant site after the runway

Wellington Airport Runway Extension 85

extension. The second set of eight Tables show the % relative change in significant wave height locally for each extraction site P1–P8, ordered as matrix of wave period and wave height applied at the outer-Bay boundary. These latter % change results are somewhat more difficult to comprehend, particularly where the wave height is small at the “P” site for the existing situation and any small change from the proposed runway extension can register as a large % value.

The greatest effects on wave height can be seen to occur at data extraction site P1 (The Corner) and P5 (the cove east of the runway extension). The main results for each extraction site are outlined below.

Figure 5-5: Lyall Bay bathymetry plots showing the extraction transects (T1-T8) and point extraction sites (P1-P8) for analysis. Origin of transects shown by open circles.

At site P1 (‘The Corner’) the proposed airport extension produces the greatest change on the heights of regular waves (Table 5-2, Table 5-10) for incident significant wave heights of 3–5 m and periods of 5–13 seconds (i.e., mainly steep wind-sea or medium-range swell), with reductions of over 0.4 m in wave height for these combinations. The relative change at P1 for these combinations would be 22 to 40% percentage reduction of the wave height relative to the existing situation. For surfing wave conditions of 1–3 m wave heights at 10–15 second periods, the reduction is slightly lower, with a 16– 17% reduction for incident regular waves of 1–1.5 m at P1 for those wave periods. The highest reduction at P1 in significant wave height at P1 is 0.78 m or 40% for a steep 5-metre wind wave of only 5 seconds period.

At site P2 (located north behind the spur-groyne breakwater and rocky outcrop) the proposed airport runway extension has mixed effects on the wave heights. There is a 0.07–0.37 m (11%) reduction in wave height for incident wave heights from 3–5 m when the periods are greater than 7 seconds. The greatest reduction is for incident wave heights of 4–5 m and 10-11 seconds, with local wave heights reductions of 0.28–0.37 m or 10-11%. Conversely, wave heights from 1–2 m show a slight increase in

86 Wellington Airport Runway Extension

wave height of 0.02–0.06 m at P2 for incident waves of 10 seconds, and higher increases of 0.07– 0.55 m for short-period wind waves of 5–9 seconds. These results imply that higher waves of greater than 7-second periods have less energy before being diffracted around the spur groyne and submerged rock outcrop.

At site P3, due west of where the proposed airport runway extension will join on to the existing runway revetment, all simulated incident wave heights with periods 6 seconds or greater will be reduced to some degree by the extension, with slight increases for wind-waves of 5-second periods. The more substantial reductions in wave heights at P3 are in the range 0.3–0.76 m or 15–17% reduction for incident wave heights of 3–5m and periods 8–15 seconds. These results indicate that wave energy is reduced as the waves propagate down the side of the revetment of the proposed runway extension, which accounts for some of the reduction predicted for inner sites such as The Corner.

At site P4, in the middle of inner Lyall Bay due west of P2, there is no appreciable change predicted in wave heights as a result of the proposed airport runway extension (Table 5-5 and Table 5-13).

At site P5, in the cove east of the proposed runway extension, the largest reduction in % terms will occur in range 40–60%, but starting in most cases from a rather small present-day wave height anyway because of the semi-sheltered wave exposure of the cove adjacent to the Moa Point Road. The greatest absolute changes of 0.35 to 0.54 m reduction in local wave heights are predicted for incident wave heights of 4–5 m and 7–12 second periods, with % changes of 40–50% (Table 5-6 and Table 5-14). These reductions are due to the more enclosed nature of the cove once the proposed runway is extended.

At site P6, on the western side of Lyall Bay, the proposed airport extension produces a relatively small effect on local wave heights of a few cm for most wave combinations (Table 5-7), as expected. The main change occurs for a small set of wind-wave periods from 5-8 seconds and incident wave heights of 3–4 m, with local increases in wave height of 0.1–0.25 m.

At site P7, south of the proposed runway extension, the extension would have a mix of small effects on the wave heights. There is a small reduction in wave height when wave periods are greater than 7 seconds, with this reduction in wave height ranging from 0.05–0.2 m (< 7% reduction) (Table 5-8 and Table 5-16). However, 5-m incident waves show an increase in wave height when the wave period is shorter than 7 seconds (i.e., for a wind sea), with up to 0.3 m for 5 second waves (Table 5-8 and Table 5-16).

At site P8 in central Lyall Bay there is a small increase in wave height of 0.05–0.24 m (3–9%) for waves with periods less than 9 seconds (Table 5-9 and Table 5-17). For waves from 1–4 m there is no appreciable difference in wave heights when periods are larger than 9 seconds (i.e., swell).

Overall, the main changes in the height of regular waves predicted for these eight selected locations occur for sites near or “downstream” of the proposed runway extension, with smaller changes in wave heights across the wider inner Bay for wind-waves but less so for swell.

Wellington Airport Runway Extension 87

Table 5-2: Model predicted changes in wave height (m) at site P1 that result from a change in model boundary conditions (ranging from 1–5 m). Note: that the Table values represent the change in significant wave height locally.

88 Wellington Airport Runway Extension

Table 5-3: Change in predicted wave heights at data extraction site P2.

Table 5-4: Change in predicted wave heights at data extraction site P3.

Wellington Airport Runway Extension 89

Table 5-5: Change in predicted wave heights at data extraction site P4.

Table 5-6: Change in predicted wave heights at data extraction site P5.

90 Wellington Airport Runway Extension

Table 5-7: Change in predicted wave heights at data extraction site P6.

Wellington Airport Runway Extension 91

Table 5-8: Change in predicted wave heights at data extraction site P7.

Table 5-9: Change in predicted wave heights at data extraction site P8.

92 Wellington Airport Runway Extension

Table 5-10: Percent change in predicted wave heights at data extraction site P1.

Table 5-11: Percent change in predicted wave heights at data extraction site P2.

Wellington Airport Runway Extension 93

Table 5-12: Percent change in predicted wave heights at data extraction site P3.

Table 5-13: Percent change in predicted wave heights at data extraction site P4.

94 Wellington Airport Runway Extension

Table 5-14: Percent change in predicted wave heights at data extraction site P5.

Table 5-15: Percent change in predicted wave heights at data extraction site P6.

Wellington Airport Runway Extension 95

Table 5-16: Percent change in predicted wave heights at data extraction site P7.

Table 5-17: Percent change in predicted wave heights at data extraction site P8.

96 Wellington Airport Runway Extension

5.6.2 Changes in wave height along transects While the previous section focused on wave-height change at single locations for a wide range of heights and periods for regular waves, this section of the report focuses on the results for a particular incident wave combination extracted from the cross sections T1–T8 (outlined in Figure 5-5). T1 and T2 are north-south transects down the Bay (T2 is nearer the present runway), T8 is a transect off the cove east of the runway and T3–T7 are east-west transects progressively moving offshore. Transect T3 spans across the inner Bay including The Corner at the eastern end of the transect. Each Figure represents the extracted significant wave height for a series of incident waves of 1, 1.5, 2, 3 ,4 m height from the model grid cells that lie on or close to the transects for a constant wave period of 10 seconds. The black dots represent the significant wave heights associated with the simulation for the existing environment and the red crosses represent the significant wave heights for the proposed runway extension.

Figure 5-6 to Figure 5-11 shows the north-south Transects T1 – T2 (upper and lower panels, respectively). Each of the Figures shows the results of increasing incident significant wave heights of 1, 1.5, 2, 3, 4, 5 m on the southern boundary of Lyall Bay. The north-south Transect T1 on the western side of the Bay shows no appreciable change in wave height as a result of the proposed runway extension, as expected. The northern end of Transect T2 (eastern side of the Bay adjacent to the existing runway) shows the predicted reduction in wave height, as a result of the proposed extension. This reduction in wave height for T2 is predicted to occur for a distance 820 m from the beach shoreline, which is approximately level with the end of the existing runway, but the reduction in wave height diminishes considerably for offshore incident wave heights 3 m or higher.

Figure 5-12 to Figure 5-17 displays the results for the inner-Bay east-west Transects T3 – T4 (upper and lower panels, respectively), for the sequence of incident wave heights from 1 to 5 m. Predictions along the innermost Transect T3 show a reduction in the wave height as a result of the proposed runway extension. The predicted changes occur at the eastern end of the transect, starting 800 m from the western origin of the transect (or 320 m west of the existing-runway revetment), which is in the wave influence zone behind the spur groyne. The greatest reductions on transect T3 occur in two spots approximately 850 m and 920 m from the transect origin (or 270 m and 200 m off the existing runway revetment), with the second reduction spot exhibiting the highest reduction for the larger waves of 3 m or more. Transect T4, which terminates at the spur groyne, indicates a substantial increase in wave heights around 45 m off the spur groyne and associated rocky outcrop for the smaller incident wave heights, with the change decreasing to minimal differences for incident waves 3 m or more.

Figure 5-18 to Figure 5-23 show mid-Bay east-west Transects T5 – T6 (upper and lower panels, respectively) for the sequence of incident wave heights from 1 to 5 m. The eastern end of the transect T5 (near the proposed runway extension) shows that from 630 m eastwards along transect T5, there would be an increase in variability in wave height, presumably from wave reflection and refraction patterns generated at the end of the proposed runway. The minor changes predicted would be a slight decrease in wave height from 630–690 m along transect T5, and a slight increase near the extended-runway revetment at 780–860 m along T5 (or approx. 90–170 m west of the extended runway). Transect T6 across Lyall Bay further offshore, and beyond the end of the extended runway, shows no appreciable change in wave height arising from the proposed runway extension, as expected.

Wellington Airport Runway Extension 97

Figure 5-24 to Figure 5-29 show Transects T7 – T8 (upper and lower panels, respectively) for the sequence of incident wave heights from 1 to 5 m. The furthest-offshore Transect T7 shows no discernible change in wave height as a result of the proposed runway extension. Transect T8, starting in the cove east of the runway and heading SSW past the terminus of the runway extension, shows that there will be a reduction in wave height as a result of the proposed extension within the cove. This reduction in wave height for T8 tends be more appreciable closest inshore at the start of transect T8, out to around 270 m offshore. The ARTEMIS predictions also show that wave heights will oscillate more than for the existing situation, probably due to increased local resonance within a more enclosed cove formed by the extension, compared to the existing situation.

Figure 5-6: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when the boundary is forced with a boundary incident wave height of 1 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights with the proposed runway extension in place.

98 Wellington Airport Runway Extension

Figure 5-7: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when the boundary is forced with a boundary incident wave height of 1.5 m (10-sec period).

Figure 5-8: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when the boundary is forced with a boundary incident wave height of 2 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights with the proposed runway extension.

Wellington Airport Runway Extension 99

Figure 5-9: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when the boundary is forced with a boundary incident wave height of 3 m (10-sec period).

Figure 5-10: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when the boundary is forced with a boundary incident wave height of 4 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights with the proposed runway extension.

100 Wellington Airport Runway Extension

Figure 5-11: Wave heights predicted by the ARTEMIS model for Transects 1 and 2 when the boundary is forced with a boundary incident wave height of 5 m (10-sec period).

Figure 5-12: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 1 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights with the proposed runway extension.

Wellington Airport Runway Extension 101

Figure 5-13: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 1.5 m (10-sec period).

Figure 5-14: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 2 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights for the proposed runway extension.

102 Wellington Airport Runway Extension

Figure 5-15: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 3 m (10-sec period).

Figure 5-16: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 4 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights for the proposed runway extension.

Wellington Airport Runway Extension 103

Figure 5-17: Wave heights predicted by the ARTEMIS model for Transects 3 and 4 when the boundary is forced with a boundary incident wave height of 5 m (10-sec period).

Figure 5-18: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 1 m (10-sec-period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights for the proposed runway extension.

104 Wellington Airport Runway Extension

Figure 5-19: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 1.5 m (10-sec period).

Figure 5-20: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 2 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights for the proposed runway extension.

Wellington Airport Runway Extension 105

Figure 5-21: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 3 m (10-sec period).

Figure 5-22: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 4 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights for the proposed runway extension.

106 Wellington Airport Runway Extension

Figure 5-23: Wave heights predicted by the ARTEMIS model for Transects 5 and 6 when the boundary is forced with a boundary incident wave height of 5 m (10-sec period).

Figure 5-24: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 1 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights for the proposed runway extension.

Wellington Airport Runway Extension 107

Figure 5-25: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 1.5 m (10-sec period).

Figure 5-26: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 2 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights for the proposed runway extension.

108 Wellington Airport Runway Extension

Figure 5-27: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 3 m (10-sec period).

Figure 5-28: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 4 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights for the proposed runway extension.

Wellington Airport Runway Extension 109

Figure 5-29: Wave heights predicted by the ARTEMIS model for Transects 7 and 8 when the boundary is forced with a boundary incident wave height of 5 m (10-sec period). Black dots: predicted wave heights for the existing environment; red crosses: wave heights for the proposed runway extension.

5.6.3 Spatial change in wave patterns in Lyall Bay This Section focuses on the overall spatial pattern of changes in wave heights and refraction and diffraction of wave crests for regular waves. A limited set of results is presented for the more common sets of waves, covering incident wave heights of 1.5 and 3 m and periods of 8 and 12 seconds, spanning wind-waves (period 8 seconds) and the combination of 1.5 m and 12-second swell that represents waves often surfed (although these ARTEMIS results are for regular waves).

Spatial plots of the free-surface elevation (wave crests and troughs) and wave heights are presented side-by-side to compare the ARTEMIS results for the existing environment with those for the proposed runway extension in place, followed by a spatial plot of the differences in wave heights that are predicted to arise following the extended runway. The former two sets of plots are annotated with the Transects and model extraction points that have been presented in the previous sections.

Figure 5-30 to Figure 5-32 show the simulated free surface elevation, wave height and the predicted change in wave height (respectively) for a 1.5 m boundary incident wave height with a period of 8 seconds. The main features of the spatial wave pattern for this combination for the existing environment (left-hand panels) are:

. the zone of highest waves inshore occur in the central part of the Bay and the beach (between transects T1 and T2), with little attenuation from the southern entrance to the Bay

110 Wellington Airport Runway Extension

. high attenuation in wave height through refraction in the western part of inner Lyall Bay, where depths shoal quickly (blue and yellow areas of Figure 5-31)

. in the eastern sector north of P2, including The Corner, the wave field spreads out due to both diffraction and refraction, with attenuation of wave heights e.g., 0.7 to 1.2 m for an incident wave of 1.5 m

. substantial attenuation of wave heights within the cove adjacent to Moa Point Road to the east of the existing runway (P5) in the wave shadow of Hue te Taka reducing 1.5 m incident wave heights to 0.1 to 0.5 m.

These same features are also evident in the spatial pattern of the mean significant wave height simulated by the SWAN model for the 2014 field-deployment period in Figure 4-7.

Turning to the spatial wave patterns predicted after the runway has been extended, the main changes shown for a 1.5 m incident wave height and 8-second period are (mainly Figure 5-32):

. the zone of highest waves in the central part of the Bay and the beach (between transects T1 and T2) would largely be unaffected – with only small differences due to the extension, but with some patches of increases wave height for the given scenario (grey areas of Figure 5-32)

. the greatest change from the existing situation would be a reduction of 0.7–0.8 m in wave height along the western revetment of the runway extension through to the existing spur groyne, which is due to the frictional (drag) effects on the wave trains as they pass alongside the extension, and a partial shadow zone created by the slightly- west of south alignment of the runway revetment

. in the eastern sector of the Bay north of the spur groyne, the waves would be attenuated by a further 0.2 to 0.6 m in a zone extending out ~250 m from the existing runway revetment. The greatest wave-height reduction would be 80–100 m NW of the spur groyne, with a local increase in these 8-second wind waves immediately behind the spur groyne through diffraction. In The Corner, the wave-height reduction would be somewhat less at 0.2 to 0.4 m

. an increase in wave height adjacent to and in the lee of the spur groyne and submerged rock outcrop (grey areas of Figure 5-32)

. a patch of increased wave height (relative to present) near the beach towards the eastern end of Lyall Bay Beach (grey area of Figure 5-32 near The Corner). This arises due to the reduced wave height offshore from this part of the Beach, able to propagate (shoal) further inshore than is the case for the existing situation, given the same wave height and period offshore

. reduction in wave heights within the cove immediately adjacent to the east side to the extended-runway revetment

. negligible changes in the heights of regular waves in the rest Lyall Bay including the western side, outside the proximal areas near the runway (existing and extension).

Wellington Airport Runway Extension 111

Figure 5-33 to Figure 5-35 show the simulated free surface elevation, wave height and the predicted change in wave height (respectively) for a 1.5 m boundary incident swell with a longer period of 12 seconds. The differences in features of the spatial wave pattern for this swell combination for the existing environment (left-hand panels), compared with the 8-second wind waves are:

. the zone of highest waves in the central part of the Bay and the beach exhibits a wider spread than for the 8-second wave trains (Figure 5-31)

. comparatively higher waves propagate into The Corner than for wind waves, with a 60- 65 m swath of higher waves propagating into The Corner parallel and some 50 m off the existing runway revetment, east of P1 (Figure 5-34)

. swell is refracted and diffracted around the lee of the spur groyne and associated rock outcrop, leading to less of a shadow zone than for wind waves.

Figure 5-35 shows the predicted changes, after the runway is extended, would show a similar spatial pattern of changes in wave heights as for the 8-second waves. The main differences would be a somewhat smaller reduction in wave height as swell peels along the western side of the extended revetment and a wider shadow zone in eastern part of inner Lyall Bay with reductions of 0.2 to 0.6 m. The eastern most swath of propagating swell, adjacent to the existing runway revetment, is predicted to be slightly narrower and around 0.2–0.4 m lower in wave height than for the existing situation for a 1.5 m incident wave height. There would also be a wider area of lower wave heights centred on site P1 (Figure 5-34). The patches where wave heights are predicted to increase, would cover a smaller area around and in the lee of the spur groyne, but a slightly wider area in the central inner-Bay area and in the nearshore Beach area just off the breaker zone (grey areas in Figure 5-35).

Figure 5-36 to Figure 5-38 show the simulated free surface elevation, wave height and the predicted change in wave height (respectively) for a higher 3 m boundary incident swell with a period of 8 seconds. The spatial wave pattern for this higher wind-wave combination for the existing environment is similar to that predicted for the 1.5 m wave of the same wave period (Figure 5-32). The only obvious difference is the breaking zone is further offshore as expected for higher waves.

The predicted spatial changes in wave height arising from the runway extension are also similar to the smaller 8-second wave (compare Figure 5-38 with Figure 5-32), but with a smaller increase in wave heights immediately behind the spur groyne for the higher waves.

Figure 5-39 to Figure 5-41 show the simulated free surface elevation, wave height and the predicted change in wave height (respectively) for a 3 m boundary incident swell height with a period of 12 seconds. Again, the spatial pattern of changes in wave height (Figure 5-41) is similar to that predicted for the 1.5 m wave of the same wave period, apart from breaking further offshore and less of an increase in wave height immediately behind the spur groyne.

112 Wellington Airport Runway Extension

Figure 5-30: Free surface elevation spatial plots of Lyall Bay for a 1.5 m incident wave with a period of 8 seconds. Left image: predicted free surface elevation for existing environment; Right image: predicted free surface elevation for the proposed runway extension.

Figure 5-31: Wave height spatial plots of Lyall Bay for a 1.5 m incident wave with a period of 8 seconds. Left image: predicted free surface elevation for existing environment; Right image: predicted free surface elevation for the proposed runway extension.

Wellington Airport Runway Extension 113

Figure 5-32: Spatial plots of Lyall Bay for the predicted change in wave height as consequence of the airport development, given a boundary incident wave height of 1.5 m with a wave period of 8 seconds. Positive values (grey) represent an increase in wave height, negative values (green) are a decrease in wave height.

114 Wellington Airport Runway Extension

Figure 5-33: Free surface elevation spatial plots of Lyall Bay for a 1.5 m incident wave with a period of 12 seconds. Left image: predicted free surface elevation for existing environment; Right image: predicted free surface elevation for the proposed runway extension.

Figure 5-34: Wave height spatial plots of Lyall Bay for a 1.5 m incident wave with a period of 12 seconds. Left image: predicted free surface elevation for existing environment; Right image: predicted free surface elevation for the proposed runway extension.

Wellington Airport Runway Extension 115

Figure 5-35: Spatial plots of Lyall Bay for the predicted change in wave height as consequence of the airport development, given a boundary incident wave height of 1.5 m with a wave period of 12 seconds. Positive values (grey) represent an increase in wave height, negative values (green) are a decrease in wave height.

116 Wellington Airport Runway Extension

Figure 5-36: Free surface spatial plots of Lyall Bay for a 3 m incident wave with a period of 8 seconds Left image: predicted free surface elevation for existing environment; Right image: predicted free surface elevation for the proposed runway extension.

Figure 5-37: Wave height spatial plots of Lyall Bay for a 3 m incident wave with a period of 8 seconds. Left image: predicted free surface elevation for existing environment; Right image: predicted free surface elevation for the proposed runway extension.

Wellington Airport Runway Extension 117

Figure 5-38: Spatial plots of Lyall Bay for the predicted change in wave height as consequence of the airport development, given a boundary incident wave height of 3 m and a wave period of 8 seconds. Positive values (grey) represent an increase in wave height, negative values (green) are a decrease in wave height.

118 Wellington Airport Runway Extension

Figure 5-39: Free surface spatial plots of Lyall Bay for a 3 m incident wave with a period of 12 seconds. Left image: predicted free surface elevation for existing environment; Right image: predicted free surface elevation for the proposed runway extension.

Figure 5-40: Wave height spatial plots of Lyall Bay for a 3 m incident wave with a period of 12 seconds. Left image: predicted free surface elevation for existing environment; Right image: predicted free surface elevation for the proposed runway extension.

Wellington Airport Runway Extension 119

Figure 5-41: Spatial plots of Lyall Bay for the predicted change in wave height as consequence of the airport development, given a boundary incident wave height of 3 m with a wave period of 12 seconds. Positive values (grey) represent an increase in wave height, negative values (green) are a decrease in wave height.

5.7 Discussion/Summary The high-resolution ARTEMIS wave model is reasonably reproducing the spatial wave-crest patterns in Lyall Bay, compared to vertical satellite images, as incident wave-trains of regular waves are refracted and diffracted, particularly in shoaling waters of the inner Bay. Results from modelling the before and after situations, indicate that the proposed runway extension will have an effect on the heights of regular waves, mainly on the eastern side of Lyall Bay in proximity to the existing runway revetment and along the western side of the extended-runway revetment and more localised reduction on the eastern side of the extension.

In the north-eastern sector in the lee of the spur groyne (breakwater), 1.5 m incident waves of 8- second period propagating from Cook Strait would be attenuated by a further 0.2 to 0.6 m in a zone extending out ~250 m from the existing runway revetment. The greatest wave-height reduction in this sector, based on the ARTEMIS simulations, would be 80–100 m NW of the spur groyne, with a local increase in these 8-second wind waves immediately behind the spur groyne through diffraction. In The Corner, the wave-height reduction would be somewhat less at 0.2 to 0.4 m. The model also predicts an increase in wave heights: i) adjacent to and in the lee of the spur groyne and submerged rock outcrop; ii) slight increases in a narrow central part of the inner Bay of the main Beach; and iii) in a patch close to the beach towards the eastern end of Lyall Bay Beach near The Corner. This latter area arises due to the reduced wave height offshore from this part of the Beach, with waves able to propagate further inshore than is the case for the existing situation, given the same wave height and period and a fixed bathymetry.

120 Wellington Airport Runway Extension

The largest reduction in the height of regular waves overall would occur adjacent to the west side of the extended revetment, as the wave peels along the revetment subject to dissipation by friction (drag) exerted by the accropodes and also the influence of a subtle wave-sheltering effect as the runway extension alignment of ~183S is slightly west of the predominant wave-approach direction from due south.

Predicted changes for a 1.5 m incident swell of 12-second period, after the runway is extended, would show a similar spatial pattern of wave height change to the 8-second waves. The main differences would be a somewhat smaller reduction in wave height as swell peels along the western side of the extended revetment and a wider shadow zone in eastern part of Lyall Bay with reductions of 0.3 to 0.6 m. The eastern most swath of propagating swell, parallel with and 50 m off the existing runway revetment, is predicted to be slightly narrower and around 0.2–0.4 m lower in wave height than for the existing situation for a surfing combination of 1.5 m incident swell height of 12 seconds. The increase in wave height adjacent to and in the lee of the spur groyne would cover less area for the longer swells compared to an 8-second wave, given the same incident wave height.

The height of wave-trains reaching the central part of Lyall Bay Beach are likely to be only slightly affected by the runway extension, otherwise elsewhere in the Bay, including the western side, the changes will be negligible based on simulating regular waves.

The cove east of the runway (adjacent to Moa Point Road) will also exhibit a reduction in wave height as a result of the proposed extension. This reduction in wave height is more appreciable inshore and in the corner immediately adjacent to the existing runway. There is also likely to be more resonant or wave-reflection behaviour within the more-enclosed cove after the extension has been constructed.

The ARTEMIS results indicate the predicted scale of general changes in waves throughout the Bay, relative to the existing situation, based on regular sinusoidal wave sets. More detailed high- resolution wave modelling was undertaken by DHI, using irregular measured waves propagating through the Entrance for occasions when surfing quality was good, to further investigate the effect of the runway extension on wave heights and characteristics (e.g., wave steepness, wave peakiness). These aspects are important considerations for assessing the effects on surfing quality and safety of recreational users, are described in the assessment report by DHI (2016). The DHI results also indicate that the ARTEMIS model overestimates the effect on the reduction of wave heights behind the spur breakwater as a result of the runway extension (Dr S. Mortensen, DHI, pers. com.).

Wellington Airport Runway Extension 121

6 Suspended-sediment plume modelling (Delft3D-SED)

6.1 Rationale and scenario modelling approach During the construction of the extended runway, de-watering discharges during the embankment fill phase have the potential to introduce high-turbidity plumes into the receiving waters of Lyall Bay if a fine-sediment fraction is present in the fill material. The main source of turbidity in the water column would be from fine sediments (clays, muds, silts) present in the fill material, even if they are only a small percentage of material by volume.

The potential dispersal footprint of suspended-sediment discharges during the construction phase were modelled using the Delft3D-SED cohesive sediment module. Discharge locations will most likely be on the west side of the rock dyke at the northwest (NW) or southwest (SW) corners, with a further scenario in the centre of the southern terminus rock dyke, although the locations of discharges can be varied to match the contractors approach to performing the reclamation (AECOM, 2015b).

The decant water may either be saltwater, from pumped dredged or barged fill slurries, or freshwater generated by rainfall within the in-fill area (e.g., if trucked quarry fill was to be used) or a mixture of both freshwater and saline water. To cover for these different situations, the sediment- laden discharge in the dispersion model simulations was placed in the surface (top) layer in the five depth-layer dispersion model, and both a continuous and a pulsed discharge (e.g., following a rainstorm) were simulated. It was assumed that the discharge will have negligible momentum when entering the receiving waters i.e., it is not discharged as a high-velocity jet, but a more slowly- introduced inflow (to be more conservative with less dilution).

Possible locations for a weir (or decant discharge outlet) in the dyke walls on the western side of the airport runway extension are shown in Figure 6-2 labelled D1 to D3. Also shown are options for undertaking the fill operation, including truck dumps (e.g., a NW or S discharge), barge entry (probably a NW discharge only) or piped dredged-sediments as a slurry from an anchored barge (any of the discharge locations).

As with most reclamation construction projects in the coastal marine area, the type of fill material (especially the proportion of fine silts and clays) and the exact construction methodology and erosion and sediment control measures to be deployed are not prescribed in detail – but rather the selected contractor would design the construction plan, fill flow and sequencing generally in accordance with the construction methodology as outlined in AECOM (2015b). Further, the additional contribution and sequencing of rainwater from fronts and storms is not able to be predicted at this juncture. Therefore at this stage it is not possible to quantify ahead of time, the sequencing of sediment discharge rates of fine-grained material (e.g., silts, clays) in the decant discharge(s). This situation is usually managed by way of an adaptive management approach, setting an environmental turbidity limit not to be exceeded after allowing for reasonable mixing, while monitoring the receiving waters at a few locations during the construction phase. Such a limit would be achieved by the contractor putting in place sufficient sediment and erosion control measures, or altering the fill sequencing and spatial distribution to keep the turbidity levels below the agreed limit. As discussed below, sediment- discharge rates were simulated to find the rates at which some ecological SSC limits are not exceeded after reasonable mixing of the discharge.

122 Wellington Airport Runway Extension

The hydrodynamic conditions for the plume modelling were also generalised to a repeating mean M2 tide (given tidal currents are very low within the inner Bay) over a 30-day period interspersed with two strong southerly and one strong northerly wind events (Figure 6-1) as used for isolating the effect of winds versus tidal hydrodynamics in Lyall Bay (Section 3).

Figure 6-1: Generalised wind forcing used for turbid plume simulations, consisting of a sequence of southerly, northerly and another southerly individually spanning 2.5 days peaking at 22 m/s (43 knots). Calm conditions prevail for the last 11 days and the entire wind sequence is combined with a repeating mean- tidal cycle.

Wellington Airport Runway Extension 123

Figure 6-2: Schematics of a NW decant discharge location (top) and a SW discharge location (bottom) in the perimeter rock dyke and options for associated fill operations. Supplied by URS Ltd.

6.2 Delft3D-SED (Cohesive)

6.2.1 Model features Cohesive suspended sediments are transported by dispersion processes such as advection (transport) and turbulent diffusion, where suspensions are advected by the flow and diffused vertically and horizontally by small-scale turbulent mixing and current-velocity shearing processes. In addition, sediment fractions have a specified settling velocity (related to particle size via the Stokes settling equation). The process of flocculation with cohesive suspended sediments can be simulated by modifying the settling velocity of the localised sediment suspension through both suspension concentration and the effects of salinity in the water column (Deltares, 2011), but were not applied in this application as flocs tend to have a higher settling velocity than individual grains.

The more-complex subsequent wave re-suspension simulations were not undertaken, based on the assumption that relatively-frequent southerly waves or more likely swell, which is able to re- suspended fine sediment on the seabed at the range of depths adjacent to the rock dyke, would generate elevated turbidity ubiquitously throughout Lyall Bay during such wave episodes. This would apply to previously-settled fine sediments originally from a variety of sources along the coast and from stormwater discharges and the Wellington wastewater outfall sedimentation, besides the sedimentation from the airport construction.

124 Wellington Airport Runway Extension

6.2.2 Suspended-sediment plume scenarios The Delft3D-SED cohesive sediment module was used to generate generic sediment-plume simulations that reached two different environmental turbidity limits within 100–150 m of the discharge outlet. This allows the input of fine suspended particles (fine clay, silts, muds) from either of the alternative discharge points.

A representative slow settling velocity of the suspended sediment in the receiving waters was set at 0.13 mm/s for medium-silt particles (grain size = 15 µm or 0.015 mm), representing the types of fine- grained sediments that may be present in and winnowed from the placed fill material. Other model parameters used were specific and dry bed densities of 2650 kg/m3 and 500 kg/m3 respectively. Critical bed shear stress for sedimentation and erosion through the model domain were set to 0.075 N/m2 and 0.15 N/m2 respectively (Whitehouse et al. 2000).

One of the key input parameters for the sediment-plume model is the sediment discharge rate in kg/sec of the fine-grained sediment fraction of the suspended material, leaving aside the discharge of sand-sized material which will settle relatively quickly and only have a small contribution to changes in receiving-water turbidity. As the fine sediment content of the fill material is not precisely known, nor the water discharge rates and temporal variability, sediment-discharge rates are not able to be pre-determined. Therefore, plume model simulations with varying sediment discharge rates were undertaken to check the exceedance of the relevant SSC limit within 100–150 m of the discharge outlet.

Two environmental SSC limits were selected (Dr Mark James, Aquatic Environmental Sciences Ltd., pers. com., and James et al. 2016).

An upper limit on in-situ SSC could be around 40 mg/L (or equivalent NTU turbidity-sensor magnitude) derived from considering both a suggested benthic-ecological environmental limit of 35 mg/L (James et al. 2016) and also was close to the natural environmental maximum of 38 mg/L (MacDiarmid et al. 2015) measured at the bottom sensor of the outer Lyall Bay optical mooring during the ~7-week field deployment (only ~16 mg/L at the top sensor), although inshore in Lyall Bay, the SSC during the field deployment would have been similar to the bottom outer sensor with more intense wave stirring on the seabed in shallower waters)10. A sediment-discharge rate of ~2 kg/s of medium-silt material was needed in the plume model to ensure the SSC limit of 40 mg/L is not exceeded in the surface layer (assuming a background SSC of zero), after a mixing distance of around 100–150 m from the outlet.

Simulations were also undertaken for a lower-threshold SSC of 25 mg/L, which is the suggested limit for foraging seabirds (James et al. 2016), and while at the outer mooring in Lyall Bay the maximum reached of only ~16 mg/L at the top sensor (8 m depth) did not exceed this limit (only the 16-m deep sensor), it is expected that natural levels of SSC could rise to around 30–40 mg/L within inner Lyall Bay during high wave events in Lyall Bay, particularly if substantial turbid stormwater discharges coincide with a significant wave event. A sediment-discharge rate of around 1 kg/s of medium-silt material was needed in the model to ensure this SSC of 25 mg/L is not exceeded at around 100– 150 m from the outlet.

Delft3D-SED is a far-field dispersion model, so doesn’t adequately predict the near-field dilution and mixing processes within 10’s of metres of the discharge point. The maximum SSC within the initial

10 The synoptic survey during the field deployment in 2014 was necessarily carried out during calm, clear conditions and showed TSS concentrations of 0.5 mg/L at the outer-bay mooring increasing to about 1 mg/L inshore (James et al. 2015).

Wellington Airport Runway Extension 125

model cell in which the discharge is injected will be diluted and averaged over the cell (as presented below, therefore SSC will be higher than modelled in very close proximity to the discharge point. For scale reference, the cells into which the discharge is injected are around 3535 m in size.

The outputs from the suspended-sediment plume simulations for each of the three decant-discharge locations shown in Figure 6-2 (D1 in the NW corner; D2 in the SW corner, D3 in the mid-section of the southern end) were as follows:

 Composite maps of the maximum surface-layer SSC, reached in each model cell at any point during the sediment-plume simulations resulting from a continuous flux of 1 kg/s and 2 kg/s of medium-silt sediment over a 30-day period. Note: these SSC maps are composites of the maximum that could occur at different times for different locations, not instantaneous snapshots of the plume at any one time. These provide an upper envelope of SSC for each location or cell in the plot.

 Maps of the mean surface-layer SSC for periods of the sediment-plume simulation when calm, southerly and northerly winds occurred, averaging the time series from each model cell resulting from a continuous flux of 1 kg/s and 2 kg/s of medium-silt sediment over a 30-day period. Note: these SSC maps are an averaged snapshot of the plume over a period of at least 12 hours when the relevant wind conditions occurred in the simulation (and longer at several days for calm conditions). These averaged SSC plots for a subset of the simulation period show the effect of different wind conditions on the predicted SSC from the discharge alone (background = 0).

 Simulated time series of the surface-layer SSC extracted from the 2 field mooring locations in Lyall Bay (outer mooring site C and the inner site 1 to the south-west of the spur breakwater).

 Simulated instantaneous 12-hour snapshots of surface SSC in a sediment plume following a 12-hour pulse discharge of a 1 kg/s and 2 kg/s sediment flux commencing at the peak of a 22 m/s southerly gale event, and from the 2 decant-discharge locations.

The background SSC in the simulations was set to zero, so these results for the near-surface layer apply to the situations when the Lyall Bay waters are reasonably clear (e.g., calm or fair weather northerly conditions). From the field deployment in the outer Bay, the turbidity is low for the majority of the time (based on the top sensor at 8 m depth), where the measured 33-percentile SSC was 1 mg/L, the median 1.7 mg/L and for 99% of the time, SSC was <12 mg/L (data analysis; MacDiarmid et al. 2015). Higher SSC levels occurred during southerly wind events when significant wave heights in the outer Bay were generally above 2 m, especially if accompanied by longer-period swell, with turbidity generated by wave-stirred fine sediments winnowed from the seabed (Bell, 2016).

The lower threshold displayed in the SSC plots for the model simulations was set to around 3 mg/L, as unlikely to be ecologically meaningful below this level (James et al. 2016) – while the accuracy of in-situ monitoring of SSC is usually around 1 mg/L.

6.3 Suspended-sediment plume results Dispersion model results are presented initially for a month-long continuous sediment discharge, then a 12-hour pulsed discharge (e.g., following a rainstorm).

126 Wellington Airport Runway Extension

6.3.1 Continuous sediment discharge Results of the various plume-model simulations are set out on a page for each discharge location (D1–D3) for a particular sediment discharge rate (1 and 2 kg/s). The page layout Figure 6-3 to Figure 6-8 includes:

. (top-left): composite envelope for the maximum SSC in each cell at any time for the entire simulation

. (bottom panels): SSC averaged over calm, northerly and southerly wind conditions (left to right)

. (top-right): time series of predicted SSC for the outer field-deployment site C (lower trace) and the inner Lyall Bay site 1 (just to the south-west of the spur breakwater), with approximate locations shown as a star and triangle respectively in the adjacent maximum SS composite plot (and also same as site 2 in Figure 2-1).

The main findings in relation to assessment of effects of sediment-plume discharges are:

a) Discharges during calm weather show a classic radial dispersion pattern (bottom left-hand plots) arising from the very weak tidal currents within the inner Bay – so basically slower spreading by diffusion processes. At each location there is a small tidal influence on SSC, a shown in the time series plots (top-right panels).

b) The wind events in the simulations show the plume is elongated more than the calm- conditions as wind-generated currents add a transport or advection component to the dispersion of the plume (bottom middle (northerly) and bottom-right (southerly) panels).

c) Arising from the addition dispersion by the currents, the SSC arising from the discharge would decrease (relative to calm conditions) during wind events in the near-field (0–150 m) and the intermediate zone out to about 600-800 m from the discharge. Further out in the far-field, SSC would be slightly higher during windy periods than for calm conditions, but at low levels (generally < 3-4 mg/L), due to the additional plume excursion from the wind- generated currents.

d) Mostly the increase in SSC from a sediment plume is a linear function of the sediment discharge rate e.g., doubling the rate from 1 to 2 kg/s of fine sediment, generally doubles the SSC generated at most sites, although there are some differences in the near field (see Table 6-1).

e) Of the 3 discharge locations, the discharge at the NW (D1) location has the most effect on inner Lyall Bay due to its closer proximity, but SSC would only reach a maximum of <3 mg/L and 6-8 mg/L above the background SSC for sediment discharges of 1 and 2 kg/s respectively, based on the 30-day simulation with high wind events (Figure 6-3 and Figure 6-4). Conversely, a sediment discharge at the southern terminus of the rock dyke (D3) would have a negligible contribution to turbidity levels in inner Lyall Bay, but would generate higher SSC in the Moa Point Beach area and associated fringing reefs, with highest values during calm conditions and lowest during a southerly-wind event (Figure 6-7 and Figure 6-8).

f) After discharge, a mixing distance of 150 m is deemed to be sufficient to allow for reasonable mixing (and extend beyond any barge or machinery activities), before compliance through monitoring is considered. Table 6-1 shows the maximum SSC values estimated from the

Wellington Airport Runway Extension 127

plume simulations to be generated at ~150 m from the discharge for the 3 discharge locations and two sediment discharge rates for relatively calm wind/wave conditions.

Table 6-1: Maximum predicted SSC above background levels after reasonable mixing at ~150 m for discharges at D1–D3 and sediment-discharge rates of 1 and 2 kg/s. These maximum SSC contributed by the proposed discharges apply during calmer wind and wave conditions – but reduce for higher wind events.

SSC (mg/L) above background at ~150 m

Sediment discharge of Discharge site D1 (NW) Discharge site D2 (SW) Discharge site D3 (S) medium silt

1 kg/s 14–16 13–14 11–13

2 kg/s 33–36 22–24 22–23

128 Wellington Airport Runway Extension

S N S

Figure 6-3: Composite surface-layer maximum SSC over a 30-day simulation (top) and averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 1 kg/s sediment flux discharged from site D1 (NW corner). Note: different SSC scales between the top and bottom plots. Star = site 1, triangle = site C.

Wellington Airport Runway Extension: Coastal Processes 129

S N S

Figure 6-4: Composite surface-layer maximum SSC over a 30-day simulation (top) and averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 2 kg/s sediment flux discharged from site D1 (NW corner). Note: different SSC scales between the top and bottom plots. Star = site 1, triangle = site C.

130 Wellington Airport Runway Extension: Coastal Processes

S N S

Figure 6-5: Composite surface-layer maximum SSC over a 30-day simulation (top) and averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 1 kg/s sediment flux discharged from site D2 (SW corner). Note: different SSC scales between the top and bottom plots. Star = site 1, triangle = site C.

Wellington Airport Runway Extension: Coastal Processes 131

S N S

Figure 6-6: Composite surface-layer maximum SSC over a 30-day simulation (top) and averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 2 kg/s sediment flux discharged from site D2 (SW corner). Note: different SSC scales between the top and bottom plots. Star = site 1, triangle = site C.

132 Wellington Airport Runway Extension: Coastal Processes

S N S

Figure 6-7: Composite surface-layer maximum SSC over a 30-day simulation (top) & averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 1 kg/s sediment flux discharged from site D3 (mid-south end). Note: different SSC scales between the top and bottom plots. Star = site 1, triangle = site C.

Wellington Airport Runway Extension: Coastal Processes 133

S N S

Figure 6-8: Composite surface-layer maximum SSC over a 30-day simulation (top) & averaged SSC for calm, northerly & southerly winds (bottom) reached in each model grid cell for a continuous 2 kg/s sediment flux discharged from site D3 (mid-south end). Note: different SSC scales between the top and bottom plots. Star = site 1, triangle = site C.

134 Wellington Airport Runway Extension: Coastal Processes

6.3.2 Pulsed 12-hour sediment discharge If the method of infilling the reclamation behind the rock dyke is undertaken through truck dumping, the sediment discharge may be more intermittent, or pulsed with much higher flows during rainstorm events. Therefore further simulations were undertaken for a pulse discharge of 1 and 2 kg/s over a 12-hour period, starting at the peak wind speed of the southerly wind used in the 1- month continuous discharge simulation, before reverting to calm conditions.

Snapshots of the plume SSC are shown in Figure 6-9 to Figure 6-11 for 24, 48, 72 and 96 hours after a 1 and 2 kg/s decant pulse release has commenced, noting the discharge ceases after 12 hours. Note: Figure 6-11 presents both sediment discharge rates for the D3 (southern) discharge location. The plots have the same SSC scale as that for the maximum composite SSC plots in the previous section.

As expected, the plume from a pulse discharge would have a smaller areal extent than for the continuous discharge. For the pulse discharge, the greatest extent and peak concentrations in the intermediate and far field for the times shown, occur mostly at 48 hours after the discharge commences and 36 hours after it ceases. Thereafter the plume dissipates through ongoing slower mixing during calm conditions as present in the simulation.

Wellington Airport Runway Extension: Coastal Processes 135

Figure 6-9: Predicted instantaneous plume snapshots of SSC at 24, 48, 72 and 96 hours after a 1 kg/s decant slug release for locations D1 (left panels) and D2 (right panels). Minimum SSC cut-off is 3 mg/L.

136 Wellington Airport Runway Extension: Coastal Processes

Figure 6-10: Predicted instantaneous plume snapshots of SSC at 24, 48, 72 and 96 hours after a 2 kg/s decant slug release for locations D1 (left panels) and D2 (right panels). Minimum SSC cut-off is 3 mg/L.

Wellington Airport Runway Extension: Coastal Processes 137

Figure 6-11: Predicted instantaneous plume snapshots from location D3 of SSC at 24, 48, 72 and 96 hours after a 1 kg/s (left panels) and 2 kg/s (right panels) decant slug release. Minimum SSC cut-off is 2 mg/L.

138 Wellington Airport Runway Extension: Coastal Processes

6.4 Discussion/summary Modelling results show that due to the slow tidal currents during fair weather, the turbid plume exhibits small footprints (relative to higher winds) with a radial dispersive pattern from the discharge point. In conditions with high winds (northerly or southerly), currents are higher, which increases both the lateral spread of plumes and reduces the SSC. Consequently, the highest concentrations in the near to intermediate areas (a few 100s of metres from the discharge) will occur in calm weather, while in the far-field areas of the Bay, the highest concentrations, albeit only 1–5 mg/L above ambient conditions, will occur during high winds when the entire inner Bay is more turbid anyway from wave activity and possibly stormwater.

Further refinement of the construction methodology, including the key roles of infilling procedures and erosion and sediment control measures to reduce the turbidity in the discharge “at source”, will be undertaken by the appointed contractor (AECOM, 2015b). The turbidity plume simulations above, in tandem with turbidity monitoring against appropriate turbidity limits during construction, will provide a mechanism to adaptively manage the effect of suspended-sediment discharges during the construction phase. Further assessment of the effect of turbidity plumes is provided in the Coastal Processes Assessment Report (Bell, 2016).

Wellington Airport Runway Extension: Coastal Processes 139

7 Acknowledgements NIWA acknowledges AECOM Ltd. for providing several datasets or image files (e.g., the 3D rock-dyke shape and footprint, likely de-watering discharge location options, extreme wave analysis and a broad outline of the construction methodology).

Dr Mark James (Aquatic Environmental Sciences) provided advice and input throughout the proposal and project stages and liaising with WIAL.

Dr Craig Stevens (NIWA) provided project management and regular reports on progress to WIAL and Dr Scott Nodder (NIWA) provided oversight of the field-deployment programme in 2014 carried out by the NIWA Wellington team and supplied sediment data.

Miles Dunkin (NIWA) produced the image for Fig 2-1.

140 Wellington Airport Runway Extension: Coastal Processes

8 Glossary of abbreviations and terms AEE Assessment of Environmental Effects. ADCP Acoustic Doppler Current Profiler – measures currents throughout the water depth and waves. accropode A single-layer artificial heavy armour unit developed by SOGREAH in 1981, later modified by chipping away excess materials and adding friction features in the form of small pyramids to increase interlocking. advection Transport of a suspended or dissolved substance by current velocities. akmon Heavy unreinforced pre-cast concrete shapes that interlock to form a robust coastal-wave protection layer, as used at the southern end of the existing runway into Lyall Bay. ARTEMIS A high-resolution phase-resolving wave model that is part of the TELEMAC model suite that simulates regular waves. CD Chart Datum. cohesive Relates to the physical bonding of fine-grained sediments (compared to sand/gravel which is non-cohesive). diffraction Diffraction is the process by which the waves are bent around into the lee zone behind a structure or barrier (e.g., rock revetment or groyne) by energy transmitting laterally along the wave crests. diffusion Mixing of suspended or dissolved material in the water column by the physical process of turbulent eddies at a range of spatial scales from molecular up to kilometres. dispersion Mixing and spreading of suspended or dissolved material in the water column by physical processes of: turbulent diffusion; transport by current flows; and shearing/stretching processes induced by spatially non-uniform flows. DWG DOBIE wave gauge. GWRC Greater Wellington Regional Council.

Hs or Hm0 Significant wave height (the average of the highest 1/3 of wave heights). irregular waves Non-repeating wave shapes that are not regular (sinusoidal) and can comprise interacting wavetrains of different wave periods from different directions – more akin to time series of continuously measured waves. LiDAR Light Detection And Ranging – a laser scanner used to measure heights of land surfaces. mab metres above bed. mean spectral wave Average of all wave periods in a time-series representing a certain sea state. period

Wellington Airport Runway Extension: Coastal Processes 141

monochromatic waves Wave time series on the sea-boundary of a wave model comprising a repeating regular-wave form of the same height and period. MSL Mean sea level. phase-resolving wave A wave model with grid cells of high enough resolution that have several cells model resolving a single wave length, and therefore resolve refraction and diffraction processes. Usually based on the wave mild-slope algorithm and waves applied at the seaward boundary are usually monochromatic (waves of constant height and period with time). regular waves Pertains to the use of a repeating regular sinusoidal wave time series used for input on the offshore boundary of wave models such as ARTEMIS (otherwise known as monochromatic waves). refraction Process where wave crests arriving at an angle (rather than parallel) to the seabed contours in shallow water are changed, with the part of the wave in shallower water moving more slowly than the part still advancing in deeper water. A consequence of this is that the wave fronts tend to become aligned with the depth contours as they shoal. significant wave height The average of the highest 1/3 of wave heights over a measured or simulated period. Akin to the average an observer would perceive, which is usually biased towards noticing the larger waves. SSC Suspended-sediment concentration. swell Generally waves with periods greater than 8–10 seconds for medium-distance swells and above 10 seconds for longer-distance ground swells, reaching up to 20 seconds for trans-Pacific swell. WIAL Wellington International Airport Ltd. WVD-53 Wellington Vertical Datum–1953 (regional vertical survey datum).

142 Wellington Airport Runway Extension: Coastal Processes

9 References

Aelbrecht, D. (1997) ARTEMIS 3.0: a finite element model for predicting wave agitation in coastal areas and harbours including dissipation. Transactions on the Built Environment, 27: 343-352. Published by WIT Press, Ashurst, Southampton, UK.

AECOM (2015a) Wellington Airport Runway Extension: Concept Feasibility and Design Report prepared by AECOM Ltd, for Wellington International Airport Ltd.

AECOM (2015b) Construction Methodology Report, Appendix L. In: Wellington Airport Runway Extension: Concept Feasibility and Design Report, prepared by AECOM Ltd, for Wellington International Airport Ltd.

AES Ltd and NIWA (2013) Desk-top review of existing information and gaps on South Wellington Coast and Wellington Harbour: Extension of the Wellington Airport runway. NIWA Client Report HAM2013-101, prepared for Wellington International Airport Ltd: 105.

Beca Carter–Caldwell Connell (1980) Moa Point Wastewater Treatment Plant and Outfall Study. Study of wastewater treatment and disposal into Cook Strait for Wellington City Corporation, prepared by Beca Carter Hollings & Ferner Ltd and Caldwell Connell Engineers Pty Ltd, Melbourne: 177.

Bell, R.G. (1992) Circulation in South Wellington coastal waters, Cook Strait, New Zealand. DSIR Water Quality Centre Internal Report, No. 0608/1.

Bell, R.G. (1989) Fitzroy and Lyall Bay outfalls model study. Vol. 1. Report prepared for Port Nicolson Wastewater Treatment Committee (Wellington City Council & Hutt Valley Drainage Board) by the DSIR Water Quality Centre and Water Research Centre (UK). DSIR Water Quality Centre Consultancy Report, No. 8036/1. 164 p. + App.

Bell, R.G. (2016) Wellington International Airport Runway Extension: Coastal Processes Assessment. NIWA Client Report HAM2015-079, prepared for Wellington International Airport Ltd., August 2015 (revised March 2016).

Bell, R.G., Pritchard, M., Palliser, C. (2013) Plume dispersion modelling for Moa Point outfall by-pass events. Prepared for Opus International Consulting Ltd, NIWA Client Report No. HAM2013–102: 58.

Berkoff, J.C. (1972) Computation of combined refraction-diffraction. Proceedings 13th International Conference on Coastal Engineering, Chapter 24: 471-490. doi.org/10.9753/icce.v13.%25p

Booji, N. (1981) Gravity waves on water with non-uniform depth and current. The Netherlands, Technical University of Delft.

Booij, N., Ris, R.C., Holthuijsen, L.H. (1999) A third-generation wave model for coastal regions 1. Model description and validation. Journal of Geophysical Research, 104(C4): 7649-7666.

Wellington Airport Runway Extension: Coastal Processes 143

Carter, L., Laing, A., Bell, R.G. (2002) Wild southerlies of summer. Water & Atmosphere, 10(3): 22-23.

Carter, L., Lewis, K. (1995) Variability of the modern sand cover on a tide and storm driven inner shelf, South Wellington, New Zealand. NZ Journal of Geology and Geophysics, 38: 451-470.

Deltares (2011) Delft-3D-Flow: Simulation of multi-dimensional hydrodynamic flows and transport phenomena, including sediments. User Manual Version 3.15: 672.

Depree, C., Olsen, G., Nodder, S., Northcote, L. (2016) Wellington Airport Runway Extension: Marine sediments and contaminants (Lyall Bay). NIWA Client Report HAM2015-004, prepared for Wellington International Airport Ltd. February 2015, revised February 2016.

DHI (2016) Wellington Airport Runway Extension Surf Break Impact Assessment. Numerical Modelling, Preliminary Mitigation Investigations and Feasibility Study. Technical Report prepared for Wellington International Airport Ltd.

Engineers, U.A.C.O. (2002) Coastal engineering manual. Engineer Manual, 1110: 2-1100.

Fredsøe, J. (1984) Turbulent boundary layer in wave-current motion. Journal of Hydraulic Engineering (ASCE), 110(8): 1103-1120.

Haidvogel, D.B., Arango, H., Budgell, W.P., Cornuelle, B.D., Curchitser, E., Di Lorenzo, E., Fennel, K., Geyer, W.R., Hermann, A.J., Lanerolle, L., Levin, J., McWilliams, J.C., Miller, A.J., Moore, A.M., Powell, T.M., Shchepetkin, A.F., Sherwood, C.R., Signell, R.P., Warner, J.C., Wilkin, J. (2008) Ocean forecasting in terrain-following coordinates: Formulation and skill assessment of the Regional Ocean Modelling System. Journal of Computational Physics, 227: 3595-3624.

Harris, T.F.W. (1990) Greater Cook Strait: form and flow. Published by DSIR Marine and Freshwater, Wellington: 212.

James, M., MacDiarmid, A., Beaumont, J., Thompson, D. (2016) Assessment of ecological effects of the reclamation and extension to Wellington Airport. Aquatic Environmental Sciences and NIWA report, prepared for Wellington International Airport Ltd.

Lesser, G.R., Roelvink, J.A., van Kester, J.A.T.M., Stelling, G.S. (2004) Development and validation of a three-dimensional morphological model. Coastal Engineering, 51(8-9): 883-915. doi:10.1016/j.coastaleng.2004.07.014.

MacDiarmid, A., Anderson, T., Beaumont, J., Chang, H., D'Archino, R., Dunkin, M., Fenwick, M., Gall, M., Gerring, P., Nelson, W., Nodder, S., Roux, M-J., Stewart, R., Thompson, D., Watts, A. (2015) Ecological characterisation of Lyall Bay, Wellington. NIWA Client Report WLG2015-10, prepared for Wellington International Airport Ltd: 270.

Mackay, K., Mitchell, J. (2014) Evans and Lyall Bay Bathymetry Report. NIWA Client Report WLG2014-03, prepared for Wellington International Airport Ltd, January 2014.

144 Wellington Airport Runway Extension: Coastal Processes

Pickrill, R.A. (1979) Beach and nearshore morphology, Lyall Bay, Wellington. NZ. NZOI Oceanographic Field Report, 13: 23. Published in April 1979 by the NZ Oceanographic Institute, Wellington.

Raukura Consultants (2014) Cultural Values Report, Wellington Airport Limited – South Runway extension. Report in association with Port Nicholson Block Settlement Trust & Wellington Tenths Trust, November, 2014: 30.

Ris, R.C., Holthuijsen, L.H., Booij, N. (1999) A third-generation wave model for coastal regions 2. Verification. Journal of Geophysical Research, 104(C4): 7667-7681.

Stanton, B.R., Goring, D.G., Bell, R.G. (2001) Observed and modelled tidal currents in the New Zealand region. New Zealand Journal of Marine and Freshwater Research, 35(2): 397-415.

US Army Corps of Engineers (1984) Shore protection manual, 2 Vols. Prepared by the Coastal Engineering Research Center, Dept. of the Army, Waterways Experiment Station, Us Army Corps of Engineers, Washington, USA.

van Rijn, L.C. (1984a) Sediment transport, Part I: bed load transport. Journal of Hydraulic Engineering (ASCE), 110(10): 1431-1456.

van Rijn, L.C. (1984b) Sediment transport, Part II: suspended load transport. Journal of Hydraulic Engineering (ASCE), 110(11): 1613-1641.

Warner, J.C., Geyer, W.R., Lerczak, J.A. (2005) Numerical modeling of an estuary: A comprehensive skill assessment. Journal of Geophysical Research, 110, C05001, doi: 10.1029/2004JC002691.

Whitehouse, R.J., Soulsby, R.L., Roberts, W., Mitchener, H. (2000) Dynamics of estuarine muds. Thomas Telford, UK.

Zanuttigh, B., van der Meer, J.W. (2006) Wave reflection from coastal structures. Proceedings of the 30th International Conference on Coastal Engineering, San Diego, Vol. 5: 4337-4349, American Society of Civil Engineers (ASCE).

Wellington Airport Runway Extension: Coastal Processes 145