NSW Coastal Wave Model State Wide Nearshore Wave Transformation Tool
25 January 2017 12359.101.R2.Rev0
NSW Coastal Wave Model
State Wide Nearshore Wave Transformation Tool
Prepared For Prepared By
Office of Environment & Heritage Baird Australia Pty Ltd Project Manager: Dr Michael Kinsela As Trustee for the Baird Australia Unit Trust 59-61 Goulburn Street ACN: 161 683 889 Sydney NSW 2000 ABN: 92 798 128 010
For further information please contact Sean Garber at +61 2 9994 8977 [email protected] www.baird.com
12359.101.R2.Rev0 T:\2017\Report.2017\12359.101.R2.Rev0.NSWWave_NearshoreTransferFunction_Report.docx
Revision Date Status Comments Prepared Reviewed Approved A 2 July 2015 Draft Initial Draft for Client Review JMB DRT DRT B 15 September 2015 Draft Final Draft awaiting Toolbox SJG DRT SJG Implementation C 4 November 2016 Draft For Final Review and Approval SJG DRT SJG 0 25 January 2017 Final Issued to Client SJG DRT SJG
Disclaimer / Copyright
"© 2017 Baird Australia Pty Limited All Rights Reserved. Copyright in the whole and every part of this document, including any data sets or outputs that accompany this report, belongs to Baird Australia and may not be used, sold, transferred, copied or reproduced in whole or in part in any manner or form or in or on any media to any person without the prior written consent of Baird Australia. The outputs from this report are designated only for application to the intended purpose which is documented in the report. The outputs from the report should not be used for any other site or project. Model outputs and results have been obtained from suitable models which have used the available data sets to provide input data and to calibrate the model. It should be noted that there are uncertainties associated with results presented in this report, and when details of the uncertainty or accuracy of the model is specified, these may need to be considered in the application of the outputs from this report.”
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page i Executive Summary
A detailed understanding of spatial and temporal variability in oceanic and nearshore wave climates is essential to better assess the present and future risks of coastal hazards such as erosion and inundation. To this end, the New South Wales Office of Environment and Heritage (OEH) has commissioned the next phase of the NSW Coastal Wave Model project: the State Wide Nearshore Wave Transformation Tool. The toolbox enables users to investigate and visualise over 35 years of historical nearshore wave conditions along the entire NSW coast, generate nearshore wave hindcast datasets at any location, and view nearshore wave conditions for offshore design wave scenarios.
Previously, the OEH developed a NSW Coastal Ocean Wave Model System, comprising coupled WAVEWATCH-III (WW-III) and SWAN spectral wave models. The system has been used to develop a deep-water 35-year wave hindcast based on Climate Forecast System (CFS) data, and explore a 60-year wave record based on down-scaled reanalysis data developed by the NSW/ACT Regional Climate Modelling (NARCliM) project. In addition to wave model hindcast data sets, NSW has a comprehensive long-term measured wave data set collected through the deep-water WaveRider Buoy (WRB) network.
Due to the computational effort required to transfer deep-water wave data sets to the shoreline at a state-wide scale, the process of dynamically simulating long-term nearshore wave climates is only feasible for selected locations at present, even when utilising high-performance computing.
This report describes the development and validation of shelf-scale state-wide Wave Transfer Functions (WTFs), which are applied to transfer measured or modelled deep-water directional wave data to any nearshore location along the NSW coast, in a computationally efficient manner. To date, validation of the WTFs has focused on the Sydney northern beaches and NSW Central Coast where concurrent data from two inshore directional WRBs (Narrabeen and Wamberal) are available. The validation region also includes the long-term Sydney (Long Reef) directional WRB directly offshore, allowing the direct comparison of the WTF to the measured data.
Three different methods for the WTF have been developed and evaluated over a 3.5 month period when concurrent nearshore wave data was available at Narrabeen and Wamberal. The three methods were: parametric WTF (WTF-P), quasi-spectral WTF (WTF-QS) and a full spectral WTF (WTF-S). While the methods vary in complexity and input data requirements, each demonstrated good validation to the hindcast WW-III model and measured WRB data at the two inshore validation locations. The WTF-S method was found to offer superior wave transformation predictions at relatively low computational cost.
For all WTF methods, the computational requirement to develop a long-term wave hindcast along the NSW coastline is less than 2% of that required for the WW-III state-wide model, demonstrating
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page ii Executive Summary
that the WTF approach provides a computationally efficient method to develop nearshore wave hindcast data sets for the NSW coastline.
Wave Transformation Functions have been developed at 14,510 nearshore locations for the entire NSW coast, at 100 m spacing along the 10 m depth contour (13,313 locations) and at 2 km spacing along the 30 m depth contour (1,197 locations). The WTFs can be used to generate and visualise nearshore wave data via the web-based Nearshore Wave Transformation Toolbox.
The Nearshore Wave Transformation Toolbox is managed by the NSW Office of Environment and Heritage. Further information is available online:
http://www.environment.nsw.gov.au/research/ocean-and-coastal-waves.htm
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page iii Table of Contents
1.0 Introduction 1
1.1 Study Motivation and Scope 1
1.2 Previous Studies 3 1.3 Metocean Environment of NSW 4 1.3.1 Coastal Geomorphology 4 1.3.2 Climatology 5 1.3.3 Wave Climate 6 1.3.4 Oceanography 8
2.0 Project Methodology 10
2.1 Datasets 10 2.1.1 Bathymetry 11 2.1.2 Measured Wave Data 13
3.0 The Nearshore Wave Transformation Function 17
3.1 Wave Transfer Function Methods 17 3.1.1 Parametric Transfer 18 3.1.2 Quasi-Spectral 18 3.1.3 Full Spectral 19 3.2 Numerical Wave Model 20 3.2.1 WaveWatch-III Model 20 3.2.2 Wave Transformation Matrix Simulations 21 3.3 Nearshore Validation of Wave Transfer Functions 23 3.3.1 Transferred Modelled (WW-III) Deepwater Data Compared to Modelled (WW-III) Nearshore Data 24 3.3.2 Transferred Modelled (WW-III) Deepwater Data Compared to Measured (WRB) Nearshore Data 25 3.3.3 Transferred Measured (WRB) Deepwater Data Compared to Measured (WRB) Nearshore Data 26 3.3.4 Verification of Directional Wave Climate 27
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page iv Table of Contents
3.3.5 Summary 28 3.4 Development of WTF for Entire NSW Coastline 29 3.4.1 Quality Rating Assessment 30 3.4.2 Assessment of Offshore Forcing Location 32 3.5 NSW Extreme Wave Climate Simulations 33 3.6 Application of the WTF 37
4.0 The Nearshore Wave Transformation Toolbox 38
4.1 Description 38 4.1.1 Quality Coding 39 4.2 Toolbox Code 39 4.3 Toolbox Updating Potential 39 4.4 Compatibility with Existing OEH Tools 40
5.0 Summary 41
6.0 References 43
Appendix A
Project Scope and Deliverables
Appendix B
Wave Model System
Appendix C
Bathymetry Datasets
Appendix D
Measured Wave Data Spectral Analysis Methods
Appendix E
Quantitative Validation Metrics Description
Appendix F
User Manual for the NSW Nearshore Wave Transformation Toolbox
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page v Table of Contents
Tables
Table 1.1: Tidal levels above LAT (Camp Cove Datum, Sydney) (in metres) for the NSW sites used in this study, from the National Tide Tables (ANTT 2011) 8 Table 2.1 Summary of the bathymetric data sets used to create the NSW wave model bathymetry 12 Table 2.2: Summary of wave measurements utilised in this study 14 Table 3.1: Summary of WW-III Hindcast Comparison to WTF Methods at Narrabeen 24 Table 3.2: Summary of WW-III Hindcast Comparison to WTF Methods at Wamberal 24 Table 3.3: Summary of Measured WRB data compared to WTF Methods forced with WW-III boundary conditions at Narrabeen 25 Table 3.4: Summary of Measured WRB data compared to WTF Methods forced with WW-III boundary conditions at Wamberal 25 Table 3.5: Summary of Measured WRB data compared to WTF Methods forced with offshore WRB boundary conditions at Narrabeen 26 Table 3.6: Summary of Measured WRB data compared to WTF Methods forced with offshore WRB boundary conditions at Wamberal 26 Table 3.7: Summary of Statistics for all 14,510 Wave Toolbox Output Locations 30 Table 3.8: Quality flagging definitions for the WTF-P and WTF-S methods 31 Table 3.9: Influence of Offshore Boundary Forcing Location at Narrabeen (WTF-S method) 33 Table 3.10: Influence of Offshore Boundary Forcing Location at Wamberal (WTF-S method) 33 Table 3.11: Extreme Wave Scenarios developed for application in the Nearshore Wave Transformation Toolbox based on Shand et al. (2010) 34 Table 3.12: Percentage Exceedance Wave Heights (m) at Narrabeen 37 Table 3.13: Percentage Exceedance Wave Heights (m) at Wamberal 37
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page vi Table of Contents
Figures
Figure 1.1: Locality plan showing the location of the NSW Waverider buoys and the Sydney wave model domain (Image Source: Google Earth) 2 Figure 2.1: Sample of multiple bathymetry datasets around Batemans Bay (left) and Byron Bay (right) areas 13 Figure 2.2: Study area for the development and calibration of the WTF on the mid- NSW coastline showing the nearshore calibration locations of Narrabeen and Wamberal and the Sydney deepwater WRB on the model boundary 15 Figure 3.1: Comparison of WTF methods against the WW-III spectral model for
offshore wave heights up to 9m (Hs) at Narrabeen (left) and Wamberal (right) 23 Figure 3.2: Cumulative directional energy for the measured Sydney Offshore spectral data (WRB-R) for the 2011 calibration period 27 Figure 3.3: Comparison of cumulative directional energy for the WTF-QS and WTF-S methods (red) against the measured spectral data (blue) at Narrabeen for the 2011 calibration period 28 Figure 3.4: Comparison of cumulative directional energy for the WTF-QS and WTF-S methods (red) against the measured spectral data (blue) at Wamberal for the 2011 calibration period 28 Figure 3.5: Distribution of quality control flags along the NSW coastline: WTF-P method (left) and WTF-S method (right) 32
Figure 3.6: Estimated 100 year ARI extreme nearshore wave heights (Hm0) for three directional sectors from the WW-III shelf-scale model using deepwater extreme scenario boundary conditions 36 Figure 4.1: Nearshore Wave Transformation Toolbox Interface (Spatial Transform) 38
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page vii Acronyms and Abbreviations
AHD Australian Height Datum AHO Australian Hydrographic Office ARI Average Recurrence Interval Baird Baird Australia CFSR Climate Forecast System Reanalysis CORDEX COordinated Regional climate Downscaling EXperiment DEM Digital Elevation Model EAC East Australian Current EMEP Extended Maximum Entropy Method ENC Electronic Navy Charts GA Geoscience Australia HAT Highest Astronomical Tide
Hm0 Significant wave height (spectral moment) Hs Significant wave height IMLM Iterated Maximum Likelihood Method LADS Laser Airborne Depth Sounder LAT Lowest Astronomical Tide MHL Manly Hydraulics Laboratory MHWN Mean High Water Neap MHWS Mean High Water Spring MLWN Mean Low Water Neap MLWS Mean Low Water Spring MSL Mean Sea Level MS Model Skill NARCliM NSW/ACT Regional Climate Modelling NOAA US National Oceanic and Atmospheric Administration NSW New South Wales OEH NSW Office of Environment and Heritage R Pearson’s correlation coefficient RMSE Root Mean Square Error PWD Public Works Department SI Scatter Index ss Symmetric Slope SWAN Simulating WAves Nearshore Spectral Wave Model
Tm01 Spectral Mean Wave Period
Tm02 Spectral Mean Wave Period
Tp Peak Wave Period of the Energy Spectrum
Tz Mean Zero-Crossing Wave Period WRB WaveRider Buoy WRB-P Parameteric WaveRider Buoy Data WRB-R Spectral Re-analysis WaveRider Buoy Data WRL Water Research Laboratory WTF Wave Transfer Function WTF-P Parametric Wave Transfer Function WTF-QS Quasi-Spectral Wave Transfer Function WTF-S Spectral Wave Transfer Function WW-III WAVEWATCH-III Spectral Wave Model
θm Mean Wave Direction
θp Peak Wave Direction
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page viii
1.0 Introduction
The New South Wales Office of Environment and Heritage (OEH) has commissioned the next phase of the NSW Coastal Wave Model project, which has aimed to develop a state wide nearshore wave transformation tool capable of providing long-term nearshore wave hindcasts and design nearshore wave conditions. This project has been jointly undertaken by Baird Australia (Baird) and the NSW Public Works’ Manly Hydraulics Laboratory (MHL), who have partnered to deliver the Nearshore Wave Transformation Toolbox.
The framework developed during this project, comprising a web-based tool containing the transformation algorithms, quality control, model skill, disclaimers and access release conditions will provide NSW with a longstanding process for continuous improvement towards representation of nearshore wave climate for both near real-time and long term design statistics.
This report documents the methods, datasets and numerical modelling behind the development of the nearshore wave transformation tool and the working of the web-based toolbox.
This report contains:
• Section 1: Background to the study, previous studies and the NSW metocean setting; • Section 2: Overview of the project methodology and datasets; • Section 3: Derivation, calibration and validation of the wave transfer functions; • Section 4: Overview of the Nearshore Wave Transformation Toolbox; • Section 5: Project summary.
1.1 Study Motivation and Scope
A detailed understanding of the spatial and temporal variability in oceanic and nearshore wave climates is essential to better assess the present and future risks of coastal hazards such as erosion and inundation. Previously, OEH developed a NSW Coastal Ocean Wave Model System, comprising coupled WAVEWATCH-III (WW-III) and SWAN spectral wave models. The system has been used to develop a 35-year wave hindcast using Climate Forecast System Reanalysis (CFSR) wind forcing (Cardno 2012; Kinsela et al. 2014), and a 60-year wave hindcast (1950-2009) using down-scaled reanalysis wind fields generated by the NSW/ACT Regional Climate Modelling (NARCliM) project (Dent et al. 2014).
The wave datasets developed previously have been analysed and compared to measurements from the NSW deepwater WaveRider Buoy (WRB) network, which is one of the most comprehensive long-term wave data sets by global standards. Figure 1.1 provides a locality plan indicating the locations of the seven deepwater WRB data collection sites and the location of the wave transfer calibration area (Sydney wave model domain) on the mid-NSW coast.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 1
Figure 1.1: Locality plan showing the location of the NSW Waverider buoys and the Sydney wave model domain (Image Source: Google Earth)
The long-term measured and modelled data sets provide non-directional and directional wave data in water depths of 80 m to 100 m which are typically 10 to 20 km from the NSW coastline. Full spectral numerical modelling of the propagation of wave conditions is the most comprehensive approach to generating spatial definitions of wave climates in coastal areas. This is particularly the case when multi-modal sea states (wind-sea and swell/s and/or concurrent source conditions) are encountered at a site. However, the computational effort required to numerically downscale the long term deepwater wave data sets to the nearshore along the whole 2,137 km of
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 2
the NSW coastline has been prohibitive and to date deep water wave data has been only been transferred to the nearshore for selected locations and time periods.
The purpose of this project is to develop an accurate and efficient Wave Transfer Function (WTF) to enable near shore wave data sets to be developed from measured and modelled deepwater wave data along the whole coast without the need for numerical modelling of each timestep of historical or hindcast data. The transformation matrix is derived from the relationship between measured offshore and nearshore wave conditions and further developed from selected numerical modelling of a range of sea states that can be encountered along the entire NSW coastline.
This state-wide Nearshore Wave Transformation Toolbox is to be delivered as a functional web- based tool for use with multiple input data sources. The toolbox consists of an architecture to enable the generation of nearshore wave hindcast datasets at depths of 10-15 m along all NSW nearshore coastal waters, including major bays, from deepwater conditions off the NSW coast. The transformation toolbox has the flexibility to take input from a range of available offshore wave data sources, including the NSW Offshore Buoy network, the existing NSW Coastal Wave Model, AUSWAVE and NOAA WW-III. It has the flexibility for various components to be updated and optimised in future projects for improved numerical modelling methods, updated bathymetric datasets and calibration datasets as they become available.
This study has been completed in accordance with the methodology and requested deliverables in the project technical specification (OEH, 2014). A catalogue of how this study has addressed each of the scope items requested by OEH is presented in Appendix A. 1.2 Previous Studies
OEH have been undertaking a series of projects aimed at developing the NSW Coastal Wave Model since 2011 with the aim of developing the capability to simulate nearshore coastal wave conditions along the NSW Coast. Previous phases of the project have developed models and toolboxes that enable nearshore wave models to be generated and driven by an archived deepwater hindcast to derive site specific shallow water hindcasts, however a comprehensive and efficient method of deriving similar output along the entire NSW coast in shallow water has not yet been achieved.
The projects commissioned by OEH in recent years include:
• WRL study: NSW Coastal Storms and Extreme Waves (Shand et al. 2011). • Cardno (2012): Development and calibration of NSW Coastal Wave Model that included a global WW-III model, and then a whole of NSW shelf scale wave model. That study included a 10-year continuous hindcast of wave conditions using the WW-III model, and simulation of a further 30 storms back to 1980. A toolbox was also developed in that project
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 3
to assist with running the coastal SWAN model and to prepare statistical and graphical outputs from model results. Since the 2011 study, OEH has extended the wave hindcast back to 1979 which is the start date of the CFSR hindcast wind data set. • Baird Australia (2013): Baird Australia were commissioned by the Office of Environment and Heritage (OEH) under contract OEH-600-2013 to develop a 60-year (1950-2009) continuous deep-water wave hindcast for New South Wales using a recently derived downscaled climate model data set. The intention of the project was to provide state wide deep-water wave time series for the duration of the NCEP-NCAR climate reanalysis period, using downscaled wind outputs from the Australian domain of the COordinated Regional climate Downscaling EXperiment (CORDEX) and the NSW/ACT Regional Climate Model (NARCliM) as forcing to a global scale Wavewatch-III model system. From that study, time series of deepwater directional spectra at the seven deepwater WRB’s and 42 offshore locations scattered within the NSW domain and gridded wave parameters from the 0.05 degree NSW grid in netCDF format were archived.
1.3 Metocean Environment of NSW
The coast of New South Wales, Australia, extends 2,137 km from Tweed Heads in the north to Cape Howe in the south (Figure 1.1). The coastline is oriented NNE to SSW, extending from the subtropical latitude of 28°S to almost 38°S and forms the eastern margin of the Australian continent. The NSW coastal environment is tectonically stable with a narrow and steep continental shelf, and sediment deficient. It is microtidal and wave dominated, with a highly energetic and persistent background ocean swell mainly from the south to southeast quarter (Short & Trenaman 1992).
1.3.1 Coastal Geomorphology
The NSW coastline has been shaped by the post-glacial marine transgression and the Holocene highstand, which in places is superimposed on earlier highstand deposits. Its present form is a result of the underlying geology, the sediments overlying that geology and the processes contributing to the particular geomorphic form, including sea-level history, currents, wave regime, tidal processes, climate, supply of sediments over time and tectonic setting. It is characterised by rocky headlands that separate drowned river valleys which have become partially infilled by Holocene sand to form barriers, tidal flats, lagoons and deltaic plains (Roy et al. 1980, Roy et al. 2001). NSW has 974 km of sandy coast, covering 61.2 % of the coastline (Short 2006). On the central to south coast of NSW, there is a tendency towards strong embayment, headland and deep estuaries (Roy & Thom 1981). On the north coast, there is a greater abundance of sediment and headland bypassing can occur, resulting in a stronger northerly littoral drift (Roy & Crawford 1977).
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 4
The sediments on the NSW continental shelf lie in three shore-parallel bands (Roy et al. 1980, Matthai and Birch 2001), with the inner shelf (<60 m depth) having medium to coarse-grained quartoze sand and less than 2 % mud, the middle shelf (60 to 120 m depth) having carbonate-rich mud to muddy sand of Holocene age and the outer shelf (>120 m depth) having coarse-grained, calcareous, gravely sands that accumulated in the lowstand of sea level during the last glacial maximum (Ferland et al. 1995).
1.3.2 Climatology
The climate of coastal NSW varies from subtropical in the north to temperate in the south, and is moderated by the Pacific Ocean. Average daily maximum and minimum temperatures and annual rainfall generally decreases from north to south. Summer is the rainiest season for the north of the state, whilst rainfall is more even across the seasons from Sydney southwards, reflecting the impact of a range of meteorological sources. The dominant winds along the coast are the northeast sea breeze in summer interrupted by intermittent southerly changes, and south to southwest winds in winter associated with mid-latitude lows.
The New South Wales coast is impacted upon by several different meteorological systems capable of generating storm wave events, including decaying ex-tropical cyclones, east coast lows, mid- latitude cyclones and anticyclonic highs (Shand et al. 2010). Each has a characteristic seasonality, latitudinal range and directional spectrum, described further below.
Tropical cyclones originating in the Coral Sea occasionally track down the NSW coast before degenerating into more extensive low pressure systems. They are below maximum intensity when influencing the NSW coast, affecting mostly the northern part of the state, but can influence the whole coast. They generally affect the coast between December and April, peaking in March. The highest recorded tropical cyclone waves impacting on the NSW coast were 10 m with a period of 10 s from the 4th to 5th March, 1976 due to TC Colin tracking 500 km east of Sydney (PWD 1985). Tropical cyclones in northern NSW are slightly less frequent than in south Queensland, with one occurring on average every 2.3 years (Oliver 1979).
East coast lows are cyclonic depressions that develop off the New South Wales coast, sometimes in association with inter-anticyclonic front and trough formations in the upper troposphere (Holland et al. 1987). They can develop into complex coastal depressions with strong south to southeastery winds producing heavy swell and often accompanied by rainfall. East coast lows usually form within a synoptic scale blocking pattern where a low pressure system is cut off from the mid-latitude low by a ridge of high pressure (Hopkins & Holland 1997). They tend to track parallel to the coast. East coast cyclones can be divided into several categories including easterly trough lows and inland trough lows (Shand et al. 2010). Easterly trough lows generally occur between 25 and 40°S and are usually associated with a blocking high in the Tasman Sea, creating a steep pressure gradient and
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 5
persistent onshore winds (Holland et al. 1987). Inland trough lows originate from the quasi- permanent low pressure trough over inland Queensland. These can develop into deep depressions when they cross the coast. They gradually weaken as they move offshore, although they may reintensify, and move in an irregular south-easterly track across the Tasman Sea. They can occur in any month, but most frequently in May, June and July (PWD 1985).
Mid-latitude cyclones tend to track below the Australian continent in an easterly direction south of 35°S in winter, and well south of the continent over the Southern Ocean in summer (Gentilli 1971). These systems pass every three to four days and are responsible for the persistent low to moderate ocean swell as they travel between Australia and New Zealand (Short & Trenaman 1992). Extreme wave events can be generated when these systems persist for a number of days in the Tasman Sea, usually as a result of a blocking high further east. Complex secondary low pressure systems can form in association with these lows, creating strong onshore winds close to the NSW coast and extreme wave events. Also included in this category are continental lows, which originate as depressions in the Indian Ocean or Great Australian Bight. These can intensify rapidly when they cross the east coast, and occur mainly in the second half of the year (PWD 1985). Mid-latitude lows tend to cause extreme wave events mostly in winter, although they are the most common influence on the NSW wave climate year round (PWD 1985).
High-pressure cells migrate across the Australian continent in an easterly direction approximately every ten days. The latitudinal track of these systems changes with the season, migrating between 36ºS in summer and 30ºS in winter. These high-pressure systems are dominated by light winds, clear skies, warmer temperatures and low nearshore seas, allowing background swell to dominate the wave regime. Although the most common systems, these rarely contribute towards extreme wave events. They can develop into anticyclonic intensifications which result from the strengthening of the equatorial low to the north of the subtropical high off NSW and a subsequent northeast flow of air, especially in late summer and autumn (PWD 1985).
Storms in NSW have a tendency to cluster in time. In the central region (encompassing Sydney), 40 % of storms occur within 30 days of the previous storm, and 60 % occur within 60 days of the previous storm. By region, the central region is the stormiest (PWD 1986). The south coast has the most uniform storm distribution throughout the year, although January is the quietest month. On the north coast, August to December is significantly less stormy.
1.3.3 Wave Climate
NSW has an energetic and highly variable wave climate, due to the different forcing meteorological systems as described in the section above. The general wave climate of Sydney and the NSW coastline have been documented by Lawson and Abernethy (1975), Webb (1983), PWD (1985,1986), Trenaman and Short (1987), Short and Trenaman (1992), Lord and Kulmar
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 6
(2000) and Shand et al. (2010) using data from Waverider buoys currently operated by the Manly Hydraulics Laboratory (Section 2.1.2).
There are three general types of waves that affect NSW coast. The first is the dominant, persistent long period swell waves generated outside the region (Thom et al. 1973). Whilst this swell can come from all directions, the majority of wave energy comes from the south to southeast. These waves are long-crested, and sinusoidal with periods of 6 to 14 seconds. The longest period swell also tends to be the highest, but least frequent. Locally-generated wind waves have shorter periods, variable direction and energy and are directly associated with wind systems. Storm waves are the highest waves, between 5 and 10 m, generated by periodic storm events within or close to the state. In general, 18 % of waves impacting NSW arrive from the northeast, 41 % from the east and 40 % from the southeast (Short & Wright 1981). Shand et al. (2010) report that the mean significant wave height along the NSW coastline is relatively consistent, ranging from 1.43 m at Batemans Bay to 1.63 m at Sydney. They also report the design wave return levels for the NSW coastline, ranging from 9.1 m at Botany Bay to 7.6 m at Byron Bay for the 100 year ARI
Short and Wright (1981) found that 95 % of waves over 2.5 m in height are the result of low pressure systems, with mid-latitude cyclones producing 200 days of southeast waves per year, averaging 2.3 m high and with ten to twelve second periods. East coast lows generating extreme waves occur three to four times a year and last for four and five days, producing waves averaging 2.8 m in height and having periods of 10 to 12 s. Tropical cyclones were found to affect the wave climate of Sydney on average 16.5 days per year, particularly in February and March. Waves averaging 2.7 m in height with 9 to 10 s periods are generated, primarily from the east and northeast. High pressure systems by comparison generate waves that average less than 1.5 m in height with periods of nine to ten seconds, which can approach from any direction. Lord and Kulmar (2000) found no significant trends over time in the NSW wave climate from analysis of long term Waverider buoy data.
Some extreme wave events do not correlate to the appearance of low pressure systems on the synoptic charts. These may be related to Southern Ocean depressions beyond chart limits. Some other meteorological systems impacting NSW do not have sufficient intensity, extent or duration to generate extreme waves. These include southerlies, which are too short in duration, thunderstorm squalls, major travelling depressions, which are too wide to generate extreme waves, and local winds sources such as small anticyclones, sea breezes and katabatic winds.
Historical storm wave, ocean inundation and erosion events along the NSW coast have been documented in Thom (1974), Chapman et al. (1982), PWD (1985, 1986, 1988, 1989d) and Eagle and Malone (1986). Major events threatening property have occurred every 10 to 20 years since the mid-nineteenth century. The most severe erosional events have been generally associated with a series of storms rather than a single storm; for example, 1912, 1950, 1967 and 1974 (Hanslow &
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 7
Howard, 2005). The most erosional recorded events in NSW, in terms of property damage, were the 1944 to 1946 and 1974 events (Hanslow & Howard, 2005).
The narrow and steep nature of the NSW continental shelf means that the loss of wave energy due to friction is negligible and the development of storm surge is also limited. Wright (1976) estimated that the dissipation of wave power due to bed friction over the continental shelf averages 3.5 % for the NSW coast between Sydney and Jervis Bay. While this value may be slightly higher for Northern NSW where shelf depths are generally shallow, the small loss of wave power due to friction along the NSW coast is not typical in comparison with other locations worldwide. For instance, the average power loss for Georgia, Atlantic coast, USA is 84 %, Santa Rosa Island, Florida is 58 % and Cape Henry to Cape Hatteras, Atlantic Coast, USA is 48 %. However, the degree of frictional attenuation of wave power is dependent on wave height. During low energy conditions, the frictional power loss is less than average. Under extreme (top 1 %) of wave conditions, the nearshore zone is widest and more than 16 % of power can be dissipated by bed friction.
1.3.4 Oceanography
The New South Wales tidal regime is microtidal and semidiurnal with a distinct diurnal inequality (Easton 1970, Hamon 1984) (Table 1.1). The mean spring tidal range ranges from 1.1 to 1.3 metres, with a maximum tidal range of approximately 2.0 metres. The tides are mostly uniform in timing and magnitude along the coast reflecting the uniformity of the continental shelf. The four most important harmonic constituents of the NSW tidal regime are M2, S2, K1 and O1. The greatest tidal ranges occur near the solstices.
Table 1.1: Tidal levels above LAT (Camp Cove Datum, Sydney) (in metres) for the NSW sites used in this study, from the National Tide Tables (ANTT 2011) Site HAT MHWS MHWN MSL MLWN MLWS Ballina 1.9 1.4 1.1 0.8 0.5 0.2 Coffs Harbour 2.1 1.6 1.3 0.9 0.6 0.3 Sydney 2.1 1.6 1.4 1.0 0.6 0.4 Batemans Bay 1.9 1.5 1.3 0.9 0.5 0.3 Eden 2.1 1.8 1.2 1.0 0.8 0.2 HAT: Highest Astronomical Tide, MHWS: Mean High Water Spring, MHWN: Mean High Water Neap, MSL: Mean Sea Level, MLWN: Mean Low Water Neap, MLWS: Mean Low Water Spring, LAT: Lowest Astronomical Tide.
Several types of currents occur off the NSW coast including the East Australian Current, wind- and wave-induced currents and tidal currents. The offshore East Australian Current (EAC) operates on a regional scale as the western boundary current of the Pacific Ocean. It flows near the edge of the continental shelf, generally in a north to south direction at speeds between 1 and 6 ms-1 (Creswell 1994, Oke & Middleton 2001), before separating from the coast and heading towards
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 8 the north tip of New Zealand. It separates from the coast as far north as Cape Byron and as far south as Ulladulla, but typically near Sugarloaf Point (Godfrey et al. 1980). The EAC is generally stronger in summer and weaker in winter (Ridgeway & Godfrey 1997). It can come close to the shore, producing nearshore currents that head south. It consists of a series of large eddies which flow both clockwise and anticlockwise from north to south close to shore and south to north further from shore.
Wave-induced currents associated with the predominately south to southeasterly swell include the south to north longshore current and littoral drift (Davies 1979, Bryant 1991). This drift is more pronounced in the north of the state where the greater amount of sediment allows headland bypassing. Tidal currents are weak over the steep profile of the NSW shelf. Wind induced currents flow on the surface in the direction of the wind and return flow at the bottom in the opposite direction. The predominant westerlies generate landward flows near the bottom, leading to beach accretion (Boyd 1980). Coastal-trapped waves affect the NSW coast, and have an amplitude of up to 0.2 to 0.3 m and a period of several days (Griffin & Middleton 1991, 1992). They have been linked to southerly winds (Reid & Leslie 1999) and generally progress anticlockwise around the Australian continent up the NSW coastline (Middleton 1995, Middleton et al. 1996).
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 9
2.0 Project Methodology
The project can be grouped into two major tasks; development of the nearshore wave transformation algorithms and development of the nearshore wave toolbox, however the overall construction of the NSW Nearshore Wave Transformation Toolbox has comprised numerous steps, including:
1. Measured data analysis: a) Bathymetry b) WRB wave data sets
2. Numerical wave model: a) Model development b) Implementation on a High Performance Computing facility c) Calibration and sensitivity analysis d) Validation e) Transfer matrix simulations
3. Wave transfer function: a) Method derivation b) Calibration against modelled data c) Calibration against measured data d) Validation: hindcast dataset
4. Web-based toolbox a) Toolbox Construction b) Quality assurance
Validation of the WTF has focused on the Sydney northern beaches (Narrabeen) and NSW Central Coast (Wamberal), where directional WRB data was available in shallow water. These primary validation sites are also proximate to the long-term Sydney (Long Reef) directional WRB located directly offshore, allowing direct comparison of modelled and measured transformations to measured data. 2.1 Datasets
The following datasets have been compiled and utilised in this study:
• Bathymetry (Section 2.1.1 and Appendix C); • Measured Waves: Deepwater and nearshore (Section 2.1.2); • Modelled Winds: CFSR winds for hindcast validation (Appendix B); • Modelled Waves: boundary conditions from global/regional WW-III model (Appendix B).
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 10
2.1.1 Bathymetry
The accuracy of a spectral wave model’s output is in part dependent on the quality of bathymetry applied to the model, regardless of the spatial resolution of the model grid. Multiple bathymetry datasets were sourced for this project from a range of federal, state and local government agencies. A summary of the bathymetric data applied in the model along with the resolution and area that this data covers is provided in Table 2.1. For the large-scale global and regional wave model grids, ETOPO1 and Geoscience Australia 9-second DEM data sets were applied. For the high resolution nearshore grid (an unstructured mesh – see Section 3.2), a combination of bathymetry data was compiled in priority order of accuracy, resolution and date.
The higher quality data, being the OEH supplied LADS (Laser Airborne Depth Sounder), OEH multibeam surveys and OEH single beam surveys, was applied in priority order. In the Sydney region, there was sufficient high quality data to cover most of the nearshore area except for the Illawarra. For the northern and southern regions of NSW, contours extracted from Electronic Navy Charts (ENC) provided the base dataset in the nearshore, supplemented with the higher quality survey data where available.
All datasets were converted to the same vertical datum, Australian Height Datum (AHD), prior to inclusion in the model. The data was processed to reduce the data size to enable interpolation onto the wave model nearshore grid, while maintaining spatial resolution close to shore. Figure 2.1 shows the multiple datasets that are used to represent the bathymetry in the wave model. The various datasets showed noticeable offsets and inconsistencies. These inconsistencies were addressed by creating a blank buffer around the higher quality datasets to allow a distance for blending and interpolating thereby producing a smooth bathymetry suitable for wave modelling.
Note that limitations in the bathymetry may limit the applicability of the numerical wave model in some locations. With limited availability of nearshore measured wave data, it is difficult to assess the impact of this on the resulting wave parameters. The coastline for the model was sourced in high resolution vector format from NSW Waterways.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 11
Table 2.1 Summary of the bathymetric data sets used to create the NSW wave model bathymetry Data Source Priority Resolution Area used in Processing Notes of Base Model Data E-TOPO1 Global 7 Interpolated Grid 1: Global 1- Bathymetry Supplied with Dataset 1 arc- degree resolution derived from the WaveWatchIII by minute grid GEBCO 1 arc- NOAA (Amante minute global and Eakins, 2009) bathymetry Geoscience 6 0.0025° Grid 2: Australian Used in Grid 4 > Whiteway (2009) Australia (GA) region 65 m depth in 250 m (9 second) 0.25°resolution; areas with no 2009 Grid 3: NSW other bathymetry coastline and data Tasman Sea region 0.05- degree Grid 4: only where no other data is available Royal Australian 5 4XXXX Contours along Conversion from Navy Electronic series charts the NSW LAT to AHD Navy Charts coastline from (constant 0.925 (Navy ENC) 1 Tweed Heads to m) Cape Howe to 50 m LAT Geosience 4 50 m Grid 4: Sydney Larger coverage Mostly covers Australia region only areas retained areas beyond Multibeam and narrow nearshore model Dataset swaths removed extent (Spinoccia 2013) OEH Multibeam 3 Point data Individual Gridded to 20 m Small, patchy and surveys datasets shown in resolution long swath Appendix B surveys neglected 2 Small, patchy and Individual OEH Single beam Gridded to 20 m long swath Point data datasets shown in surveys resolution surveys Appendix B neglected 1 Individual OEH LADS Gridded to 20 m Point data datasets shown in surveys resolution Appendix B 1. “Under licence by permission of The Australian Hydrographic Service. © Commonwealth of Australia 2015. All rights reserved. This information may not be copied, reproduced, translated, or reduced to any electronic medium or machine readable form, in whole or part, without the prior written consent of the Australian Hydrographic Service.”
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 12
Figure 2.1: Sample of multiple bathymetry datasets around Batemans Bay (left) and Byron Bay (right) areas
Buffer
Single Beam Survey Single Beam Survey
Multibeam Survey
Buffer
Navy Chart Gridded Contours
Navy Chart Gridded Contours
2.1.2 Measured Wave Data
NSW has a long-term measured offshore wave data set that extends back to 1974 (Kulmar et al. 2013) and real-time wave data continues to be collected by directional WRBs at seven locations. The offshore data coverage is summarised in Table 2.2. Furthermore, there exist a number of short-duration nearshore wave data along the NSW Coast as presented by MHL (2014), however no continuous long-term nearshore deployments.
2.1.2.1 Site Selection
A review of the available wave data, with particular focus on directional measurements was completed in consultation with OEH to identify sites where suitable concurrent high quality offshore and nearshore wave data exist. From the available data, two representative sites were selected, nearshore Narrabeen and Wamberal, to be used as calibration and testing of the nearshore transformation algorithms.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 13
Table 2.2: Summary of wave measurements utilised in this study
Lon. Lat. Depth First Record End Record Length Directional Site Name (°E) (°N) (m) Date Date (years) Buoy (Since)
Offshore Byron Bay 153.73 -28.82 71 14/10/1976 31/12/2014 38.2 26/10/1999 Coffs Harbour 153.27 -30.36 72 26/05/1976 31/12/2014 38.6 12/02/2012 Crowdy Head 152.86 -31.83 79 10/10/1985 28/12/2014 29.2 19/08/2011 Sydney 151.42 -33.78 85 3/03/1992 31/12/2014 22.8 Whole Record Port Kembla 151.03 -34.48 78 7/02/1974 25/12/2014 40.9 20/06/2012 Batemans Bay 150.34 -35.71 73 27/05/1986 31/12/2014 28.6 23/02/2011 Eden 150.19 -37.3 100 8/02/1978 31/12/2014 36.9 16/12/2011 Nearshore Wamberal 151.452 -33.432 11 5/08/2011 16/04/2012 0.61 Whole Record Narrabeen 151.304 -33.721 10 27/07/2011 14/11/2011 0.30 Whole Record
The development and calibration of the WTF was focused on Narrabeen on Sydney’s northern beaches, and Wamberal on the NSW central coast (Figure 2.2). Those two sites were selected as they had the most comprehensive nearshore wave data (≈ 10 m depth) for a concurrent time period in late 2011, as well as high resolution bathymetry data sets. The other key consideration was the availability of suitable offshore directional for the calibration period, which was available from the continuous Sydney offshore directional WRB (Long Reef). The offshore directional WRB is located at a similar depth to the archived directional wave data from the 35-year CFSR and NARCliM hindcast projects (available at 0.05-degree resolution) as shown on Figure 2.2.
The archived WRB data consists of spectral and time-domain parametric wave parameters, for example significant wave height (Hs and Hm0), mean crossing wave period (Tz, Tm02) and peak wave period (Tp). Directional wave data includes mean wave direction (θm) and peak wave direction (θp).
Narrabeen and Wamberal are relatively close together (Wamberal is 33 km NE of Narrabeen) and would experience similar offshore wave climates. The nearshore measurement sites at both locations were mid-compartment. Narrabeen Beach is 3.4 km long with an easterly exposure, bounded by the 1.7 km long, Long Reef Peninsula to the south and the entrance to the Narrabeen lakes system in the north. The sheltering afforded by the Long Reef Peninsula might limit energy from southerly directions. Wamberal Beach is approximately 4.3 km long with east-south-east exposure, bounded by Terrigal headland to the south, Forresters Point to the north and the entrance to intermittently open Terrigal Lagoon lies approximately one-third the way north along the beach.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 14
The validation of the wave model and WTF at these two sites would provide some confidence in the model’s performance at open ocean beaches with similar exposures. While more distinctive validation sites might be desirable, no further nearshore wave measurements were available for this study.
Figure 2.2: Study area for the development and calibration of the WTF on the mid-NSW coastline showing the nearshore calibration locations of Narrabeen and Wamberal and the Sydney deepwater WRB on the model boundary
2.1.2.2 Wave Data Processing
An accurate representation of the measured directional wave spectra is required as input to the wave transformation process. Of the three wave transformation functions considered in this study (Section 3.0), two require 2D direction-frequency spectra data at the deepwater input location.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 15
A continuous record of full resolution 2D directional wave spectra has not been stored for the directional wave data sets. Instead integrated directional parameters (e.g. Hs, Tp, θp) and a summarised directional spectrum are archived by the Directional WRB. The Datawell (Fourier decomposition) method for directional spectra, which identifies the peak energy of the seastate, tends to show generic directional spreading due to the condensed nature of the archived information. The directional spread of the Datawell results are generally much broader than those of the other analysis methods and provide negative energy at 90° either side of the peak and secondary spurious small peaks at 180° from the primary peak. While the integrated parameters are considered accurate descriptions of the bulk energy, the condensed spectral data is not considered an accurate description of the full directional seastate and therefore not appropriate for the wave transformation project.
Manly Hydraulics Laboratory (MHL) maintains an archive of the buoy displacement time series and a full re-analysis was performed using the archived XYZ displacement data from the Sydney deepwater, Narrabeen and Wamberal directional WRBs using functions from the DIWASP toolbox (Johnson 2012). Note that while the buoy instrumentation records acceleration, it is immediately double integrated to provide displacement in three axis and the raw acceleration data is not available for analysis. The DIWASP software requires horizontal inputs as velocity so the heave data is differentiated prior to reconstruction. Two alternative directional wave analysis methods were examined; the Iterated Maximum Likelihood Method (IMLM) (Pawka 1983) and the Extended Maximum Entropy Method (EMEP) (Hashimoto et al, 1993). The three data processing methods identified herein (Fourier, IMLM and EMEP), are described further in Appendix D.
The re-analysis of the deepwater Sydney WRB data using the IMLM and EMEP methods produced good agreement with the buoy derived integrated parameters. The IMLM method indicated a slightly more southerly dominant wave direction compared to the EMEP method; however, both methods produced a similar cumulative directional energy spectra for the calibration period. The IMLM data set was slightly more robust with fewer spurious data values. For the Narrabeen and Wamberal nearshore data sets, the IMLM was significantly more robust and showed good agreement with the buoy derived integrated parameters. The EMEP method was somewhat more unstable than the IMLM, with some records failing to analyse correctly. Direct comparison between analysis of single records consistently shows good agreement between the IMLM and
EMEP. The IMLM and EMEP have almost identical Hs and Tp values, since the frequency spectrum is developed identically.
In summary, following the directional re-analysis investigations, two data sets were used for the spectral wave model and WTF calibration:
• WRB-P: parametric directional WRB data derived from the on-buoy analysis; • WRB-R: full re-analysis directional wave spectra data derived from the IMLM method.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 16
3.0 The Nearshore Wave Transformation Function
Wave Transformation Functions (WTFs) have been adopted for a long period of time to provide a computationally efficient method of transferring wave data from a reference location to a point of interest. Commonly, WTFs have been applied for shallow water wave transformation from a deepwater reference location, or for propagation of waves into embayments or ports. WTFs normally require a calibrated physics-based wave model to be developed which is used to simulate a base set of wave conditions whose results are used to provide the interpolation matrix that underpins the transformation. A sufficient number of model simulations are required to develop an interpolation matrix that will provide an accurate transfer function of the physics based model.
For this study, three methods to undertake the wave transformation from offshore to nearshore locations are formulated and comparatively evaluated. While each method has perceived advantages over other methods, the performance and validity of the available methods remains largely untested along the NSW coast and hence the most appropriate method could not be selected a priori. The three methods assessed for this study were:
• Parametric Transformation • Quasi-Spectral Transformation • Full Spectral Transfer Function
A detailed summary of the performance of each transformation method on the NSW coast is presented in Section 3.3 below or in Taylor et.al. (2015).
The resulting transformation method/s then form the basis of the Nearshore Wave Transformation Toolbox (the Toolbox), described in Section 4.0, that provides WTFs for over 14,000 nearshore locations along the NSW coast at the 10 and 30m depth contours. Further, each nearshore output location also contains metadata, such as matrix skill based on bathymetry quality, level of validation, and error/uncertainty based on the available validation data. The basis for these measures is summarised in this Section. 3.1 Wave Transfer Function Methods
The theoretical basis behind each of the three Wave Transfer Function (WTF) approaches considered in this study (parametric, quasi-spectral and full spectral), are described in the following sections.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 17
3.1.1 Parametric Transfer
The parametric WTF (WTF-P) is the simplest of the three methods and has been widely adopted in projects over the last 30 years to efficiently transfer deepwater wave time series data to nearshore locations. The WTF-P utilises the parametric WRB data sets (WRB-P) that are readily available for the whole duration of the NSW measured and hindcast wave data sets.
For the WTF-P, offshore data is characterised as integrated parameters of the total spectrum, that
is, a single wave height (Hs or Hm0), wave period (Tp) and direction (θp or θm) for each record in the dataset.
An underlying transformation matrix of spectral wave simulations (for example from the WW-III model) covering the full wave period and direction range is then used to transform (scale) the offshore wave energy to nearshore locations and adjust the wave period and direction. At time steps where the offshore wave period and direction does not exactly match a modelled matrix element, a linear interpolation of the wave transformation coefficients is applied. The spectral shape parameters adopted for the transformation matrix simulations, including spectral peakedness and directional spreading, is characterised from the measured offshore data for each wave period and direction element. A range of directional spreading and frequency peakedness values were tested and optimised for this project.
Advantages: This method is simple, calculations are fast and require minimal storage of input data.
Disadvantages: This simplistic method treats every offshore wave condition in a generic way based on total wave parameters. It does not allow for variations in spectral shape or multi-modal seastates. The scaling of wave energy is also carried out based on a unit wave height (adopted for each spectral wave simulations in the transformation matrix).
3.1.2 Quasi-Spectral
The Quasi-Spectral WTF (WTF-QS) is a hybrid method that combines the simplicity and computational efficiency of the WTF-P algorithm, with enhanced capability to estimate the directional spectra of the transformed wave condition. The WTF-QS has been demonstrated to provide an accurate and highly efficient method for large open coast regions, for example the entire Chilean coast of South America (Monardez et al. 2008), and this particular algorithm applied in this study. A comparison of this method to the parametric transformation approach along the Chilean coast (Dominguez et. al, 2014) showed an improved description of the nearshore wave climate.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 18
The WTF-QS requires a time series of full 2D wave spectra at the offshore reference location to derive the nearshore wave data set.
In this method, the transformation matrix is populated by a set of numerical model runs that cover the range of wave period and direction condition at the deep water reference location. Nearshore
conditions from each matrix simulation are characterised by the wave parameters, Hm0, Tp and θp. The WTF consists of parametric wave transfer variables for each discrete offshore condition and is derived from the model results as the change in wave energy (or wave height) and the change in wave direction for each offshore frequency and directional ordinate. WTF-QS assumes no frequency change between offshore and the target nearshore location for each modelled ordinate. The WTF then remaps the offshore spectra (ordinate-by-ordinate) based on the parametric transfer variables for the relevant ordinate in the matrix.
Advantages: Whilst more complex than the WTF-P method, this method is still computationally efficient and requires less storage of input data than the full spectral method.
Disadvantages: This method may not adequately capture shifts in wave frequency/period that may occur along complex coastlines due to processes such as diffraction around islands.
3.1.3 Full Spectral
The full Spectral WTF (WTF-S) is the most comprehensive and computationally intensive WTF considered in this study. Spectral WTFs have been demonstrated to be robust and accurate tools to transfer detailed directional offshore wave to nearshore sites but have only been routinely applied for specific sites and projects owing to their complexity and computational requirements, for example Branson et al. (2013).
The offshore data requirements and transformation matrix for this method are the same as for the WTF-QS. The WTF-S then uses the nearshore 2D directional energy spectra to develop the WTF. With this method, the wave transformation matrix is populated for each offshore direction frequency and bin by the directional spectra modelled at the nearshore location (response spectrum). The WTF-S is then applied to an offshore directional spectra data set, and for each ordinate of the WTF-S an nearshore 2D spectra is calculated by scaling the response spectrum of a particular ordinate based on the actual energy in that ordinate of the offshore spectrum. For a particular time step, each of the transferred spectra from all ordinates of the WTF-S are summed to estimate the combined nearshore directional spectra. Spectral wave parameters such as wave height, period and direction are calculated from the combined spectra.
Advantages: This method robustly captures the directional and frequency as experienced in a full spectral numerical wave model.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 19
Disadvantages: This method has the highest computational and data storage overheads. 3.2 Numerical Wave Model
Each of the WTF methods described in Section 3.1 requires a numerical wave model to develop the underlying transformation matrices. As part of Baird Australia’s in-kind support for this project, a NSW-wide shelf scale WaveWatch-III (WW-III) has been developed for this purpose. A full description of the model setup, calibration and sensitivity assessment is presented in Appendix B.
The numerical wave model was applied in this study in two ways. Firstly the WW-III model was run for a set of discrete boundary wave periods (frequencies) and directional ordinates covering the frequency and directional range observed in the deepwater wave data during the calibration period. The wave transformation from the model boundary to the selected nearshore locations from these ‘steady-state’ wave transformation simulations provided the inputs for the WTF. Secondly, the WW-III shelf scale model was run for the duration of the calibration period to hindcast the nearshore wave conditions at the Narrabeen and Wamberal WRB sites. The results from that simulation were used to verify the accuracy of the WTF and calculate calibration statistics.
3.2.1 WaveWatch-III Model
The latest WW-III version 4.18 (Tolman 2014) was adopted for this study, which includes shallow- water physics to allow wave transformation to be modelled from deepwater up to the zone of wave breaking. An irregular grid mesh was developed to allow efficient distribution of model resolution; with similar gradation in element area from offshore to nearshore as the WW-III demonstration models (NOAA 2013) and featured 100 m element resolution at the 10 m depth contour which defines the output sites for the transfer function. A detailed description of the WW- III model developed as in-kind support for this project is provided in Appendix B.
For the calibration phase, a smaller Sydney-region model consisting of a shelf scale, unstructured mesh was developed for the inner shelf (up to a depth of ≈ 100m) for the area from Port Kembla in the south to Lake Macquarie in the north. This spectral wave model was applied to calculate the wave transformation across the inner shelf from the deepwater Sydney WRB site to the two nearshore WRB sites (Narrabeen and Wamberal). The WW-III shelf scale model can be forced with boundary spectra from the OEH global WW-III model or the deepwater WRB-R data set which is located at an equivalent depth offshore. Model calibration for the July to November 2011 nearshore measurement period at Narrabeen and Wamberal showed that the WW-III model performs well using a typical model WW-III model setup and is suitable as the basis for the WTF transformation simulations. Further description of the model set-up and calibration is provided in Appendix B.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 20
Previous numerical wave modelling of nearshore wave conditions at sites along the NSW coastline have typically adopted the SWAN wave model. Whilst SWAN is an accurate and proven nearshore wave model, other studies including Cardno (2012) and Mortlock and Goodwin (2013) have identified limitations with SWAN when applied to shelf scale wave simulations along the NSW coastline. One of the key issues with SWAN identified in Cardno (2012) was a tendency for the modelled wave climate at the deepwater buoys and at nearshore locations along the NSW coast to be biased to the high frequency end of the wave energy spectrum compared to the measured WaveRider Buoy data sets. Compared to the results of hindcast modelling using the SWAN model for the Wamberal data set presented in Mortlock and Goodwin (2013), the results from the WW-III model developed in this study (Appendix B) show improved validation of the hindcast nearshore wave climate when the model is forced with global WW-III modelled spectra in deepwater. The nearshore modelled wave data (WW-III) in this study demonstrate negligible bias for the mean
wave period (Tm02). As a comparison, the nearshore wave hindcast from SWAN models forced with global WW-III deepwater spectra for Wamberal and Newcastle presented in Cardno (2012)
and Mortlock and Goodwin (2013) respectively, have a consistent mean period (Tm02) under-bias of 1 to 2 s at the nearshore WRB locations.
3.2.2 Wave Transformation Matrix Simulations
To derive the WTFs along the NSW coast, the full suite of transformation matrices were developed for all nearshore locations along the 10 and 30 m depth contour. The transformation matrices are populated through a series of model runs appropriate to the transformation algorithm, where each nearshore location is coupled to a single deepwater point.
The wave transformation matrices used for the various WTF methods were compiled from a set of WW-III model simulations for discrete offshore wave directions and frequency boundary conditions defined as follows:
• Wave period (Tp): 3.6 s – 24.3 s @ log-scale spacing (25 ordinates);
• Wave direction (θp): 0 to 225 °TN @ 15 ° increments.
Note, the WW-III model was found to poorly represent for very short period wave conditions (<3.6 s) in such ‘steady-state’ runs, limiting the performance of the WTF methods for low wave period conditions (which are unusual for the open NSW coast).
The adopted spectral shape for the matrix of simulations was dependant on the WTF method.
For the parametric WTF (WTF-P) an analysis of the offshore wave record was completed to characterise the variation in spectral shape for the range of wave periods and directions, with the
most likely values adopted. For Tp less than 7 s, a Gamma parameter (JONSWAP peakedness) of
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 21
1.1 and a directional spreading parameter of 3 was adopted. For wave periods (Tp) equal to or greater than 7 s, a Gamma parameter of 2.3 and a directional spreading parameter of 5 were adopted.
For the Quasi-spectral and Spectral methods (WTF-QS & WTF-S) a very narrow frequency and direction spectrum was applied, with a Gamma parameter of 9 for the frequency shape and a directional spreading parameter of 80.
The wave height was maintained at 1 m for all simulations to provide ease of scaling and energy conservation. Sensitivity testing showed limited frictional losses until wave heights approached the breaking limit, as would be expected for open coast beaches and a narrow, steep continental shelf environment and therefore changes in the input wave height were insensitive for this application (in terms of transfer coefficients) where the results are typically extracted beyond the depth of wave breaking for the majority of conditions. For the WTF-QS & WTF-S methods, the wave energy corresponding to a wave height of 1 m is modelled per frequency/direction bin, which would cover the range of wave heights expected for the NSW wave climate.
However, caution should be used when applying the WTF for extreme offshore waves that would be approaching the breaking wave limit at the 10 m depth contour. Figure 3.1 demonstrates that for large offshore wave heights, frictional losses start to influence the nearshore wave height a process that is described by the WW-III spectral wave model (black lines). The WTFs do not adequately describe this frictional loss for large offshore wave heights and diverge from the WW-III result (approximate thresholds of Hs> 7 m at Narrabeen and Hs > 5 m at Wamberal). The different thresholds between the calibration sites are due to the differing exposures and nearshore slopes that would influence wave energy dissipation resulting from frictional losses. These results are typical of sites along the entire coast and it is recommended that for offshore wave heights greater than 5 m (Hs) the pre-computed extreme wave simulations (Section 3.5) be preferentially adopted.
Extending the range of wave heights up to and included breaking conditions would significantly increase the computational requirements to generate the transfer matrices and any improvement cannot be quantified due to lack of extreme wave heights in the calibration data.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 22
Figure 3.1: Comparison of WTF methods against the WW-III spectral model for offshore
wave heights up to 9m (Hs) at Narrabeen (left) and Wamberal (right)
3.3 Nearshore Validation of Wave Transfer Functions
Validation of the WTFs was undertaken over the same period for which the WW-III model was calibrated, where high quality nearshore wave measurements were available at two locations during 2011; Narrabeen, July 2011 to November 2011 and Wamberal, July 2011 to November 2011 (as described in Section 2.1.2 - Table 2.2). Calibration of the shelf scale WW-III model is described in Appendix B.
Validation of the WTF methods was undertaken according to the following steps:
• The wave transformation matrices used for the various WTF methods were compiled from a set of WW-III model simulations for discrete offshore wave directions and frequencies at the boundary of the model (Section 3.2.2); • Transfer of the deepwater WW-III hindcast to the nearshore calibration sites; • Transfer of the measured deepwater WRB directional data to the nearshore calibration sites; • Each WTF method was then assessed by completing the following comparisons at the Narrabeen and Wamberal nearshore WRB locations: • Transferred deepwater WW-III hindcast against the nearshore WW-III hindcast; • Transferred deepwater WW-III hindcast against the measured nearshore WRB data sets; • Transferred measured deepwater WRB directional data against the measured nearshore WRB data sets.
Results from the various wave model, WTF and WRB data sets were quantitatively compared with a range of statistical parameters including:
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 23
• Model Skill (MS); • Root Mean Square Error (RMSE); • Bias; • Scatter Index (SI); • Pearson’s correlation coefficient (R); and • Symmetric Slope (ss).
These metrics are described in detail in Appendix E.
3.3.1 Transferred Modelled (WW-III) Deepwater Data Compared to Modelled (WW-III) Nearshore Data
Comparison of the hindcast nearshore wave conditions from the shelf scale grid (Section 3.2.1) of the global WW-III model with the results from the three WTF methods are made in Table 3.1 and
Table 3.2. The results indicate that the WTF-P provides a good estimate of the nearshore Hm0, Tp
and θp. Both the WTF-QS and WTF-S methods demonstrate improvement over the WTF-P results.
The WTF-QS and WTF-S methods produce identical estimates of Hm0, which were better across all
statistical metrics than the Hm0 estimated by the WTF-P method. All metrics for Tp were higher for the spectral methods compared with the WTF-P method, with the WTF-QS method marginally
improved, which was supported by slightly higher metrics for the Tm02 parameter. The peak and
mean wave direction (θp and θm), generally had improved metrics for the spectral methods, with the WTF-S method slightly higher compared to the WTF-QS method.
Table 3.1: Summary of WW-III Hindcast Comparison to WTF Methods at Narrabeen WTF-P WTF-QS WTF-S
Hm0 Tp θp Hm0 Tp θp Tm02 θm Hm0 Tp θp Tm02 θm MS 0.96 0.90 0.85 1.00 0.96 0.90 0.99 0.99 1.00 0.95 0.96 0.95 0.99 Bias -0.10 -0.37 7.05 -0.05 0.00 -0.95 0.08 -0.39 -0.05 -0.20 1.25 -0.54 0.57 RMSE 0.26 1.42 16.7 0.09 0.93 11.4 0.34 3.65 0.09 1.10 8.34 0.70 3.46 SI 0.22 0.15 0.16 0.07 0.10 0.12 0.05 0.04 0.07 0.12 0.09 0.06 0.04 R 0.95 0.83 0.81 0.99 0.93 0.83 0.98 0.97 0.99 0.90 0.93 0.98 0.98 ss 0.95 0.95 1.07 0.97 1.00 0.98 1.01 1.00 0.97 0.97 1.02 0.92 1.01
Table 3.2: Summary of WW-III Hindcast Comparison to WTF Methods at Wamberal WTF-P WTF-QS WTF-S
Hm0 Tp θp Hm0 Tp θp Tm02 θm Hm0 Tp θp Tm02 θm MS 0.98 0.9 0.88 1.00 0.95 0.93 0.98 0.99 1.00 0.94 0.96 0.91 0.99 Bias -0.06 -0.42 2.64 -0.02 -0.12 1.3 0.16 0.87 -0.02 -0.37 0.83 -0.64 0.24
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 24
RMSE 0.22 1.5 17.7 0.09 1.06 12.5 0.46 4.65 0.09 1.22 8.73 0.86 4.25 SI 0.18 0.16 0.14 0.07 0.12 0.10 0.06 0.04 0.07 0.13 0.07 0.08 0.04 R 0.97 0.83 0.84 0.99 0.91 0.87 0.97 0.98 0.99 0.89 0.94 0.97 0.98 ss 0.98 0.94 1.03 0.99 0.98 1.01 1.02 1.01 0.99 0.95 1.01 0.90 1.00
Overall the validation metrics for the two spectral methods are similar and provide better statistical metrics compared to the WTF-P method for the modelled data inputs. These outcomes demonstrate that the spectral WTFs are able to derive a comparable nearshore wave climate to a full spectral wave model such as WW-III.
3.3.2 Transferred Modelled (WW-III) Deepwater Data Compared to Measured (WRB) Nearshore Data
Comparisons of the deepwater global WW-III hindcast transferred to nearshore via the WTF with the measured nearshore WRB data set are made in Table 3.3 and Table 3.4. The results indicate that the WTF-QS and WTF-S have similar validation metrics for wave height, period and direction as compared to the nearshore WW-III hindcast simulation results presented in Appendix B. This result provides assurance that the WTF has sufficient frequency and directional ordinates to represent the deepwater wave climate of the Sydney and Central Coast region.
Table 3.3: Summary of Measured WRB data compared to WTF Methods forced with WW-III boundary conditions at Narrabeen WTF-P WTF-QS WTF-S
Hm0 Tp θp Hm0 Tp θp Tm02 θm Hm0 Tp θp Tm02 θm MS 0.89 0.61 0.74 0.93 0.68 0.77 0.86 0.84 0.93 0.67 0.79 0.85 0.85 Bias -0.12 -0.75 7.77 -0.07 -0.38 0.34 0.19 0.37 -0.07 -0.58 2.30 -0.36 -0.26 RMSE 0.40 2.69 19.30 0.29 2.57 14.7 1.06 1.08 0.29 2.58 15.9 1.04 0.95 SI 0.35 0.27 0.19 0.25 0.27 0.15 0.14 0.15 0.25 0.26 0.17 0.13 0.14 R 0.86 0.46 0.72 0.90 0.52 0.65 0.78 0.77 0.90 0.52 0.70 0.77 0.75 ss 0.95 0.88 1.09 0.97 0.92 1.00 1.02 1.05 0.97 0.90 1.02 0.95 0.95
Table 3.4: Summary of Measured WRB data compared to WTF Methods forced with WW-III boundary conditions at Wamberal WTF-P WTF-QS WTF-S
Hm0 Tp θp Hm0 Tp θp Tm02 θm Hm0 Tp θp Tm02 θm MS 0.87 0.60 0.76 0.92 0.59 0.73 0.83 0.81 0.92 0.59 0.75 0.77 0.78 Bias -0.23 -1.08 0.11 -0.18 -0.71 -1.58 0.01 0.21 -0.17 -1.02 -1.27 -0.63 -0.51 RMSE 0.40 2.75 24.80 0.31 2.79 22.8 1.03 0.99 0.31 2.84 21.8 1.12 0.97
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 25
SI 0.27 0.27 0.20 0.21 0.29 0.19 0.14 0.15 0.21 0.28 0.18 0.13 0.12 R 0.84 0.48 0.66 0.89 0.44 0.58 0.71 0.70 0.89 0.45 0.62 0.72 0.71 ss 0.83 0.85 1.00 0.87 0.88 0.98 1.00 1.03 0.87 0.86 0.98 0.91 0.92
3.3.3 Transferred Measured (WRB) Deepwater Data Compared to Measured (WRB) Nearshore Data
A final validation was completed by comparing transferred offshore measured WRB data for the three WTF methods to the nearshore measured data at the Narrabeen and Wamberal WRB locations, as presented in Table 3.5 and Table 3.6. The validation metrics for the WTF-QS and WTF-S methods are similar at the calibration sites, with the WTF-S providing marginally improved validation metrics.
Table 3.5: Summary of Measured WRB data compared to WTF Methods forced with offshore WRB boundary conditions at Narrabeen WTF-P WTF-QS WTF-S
Hm0 Tp θp Hm0 Tp θp Tm02 θm Hm0 Tp θp Tm02 θm MS 0.95 0.77 0.76 0.92 0.91 0.95 0.80 0.79 0.90 0.91 0.95 0.77 0.76 Bias -0.05 -0.30 6.48 0.22 1.71 -0.04 -0.38 5.67 -0.30 2.06 -0.05 -0.30 6.48 RMSE 0.20 2.19 14.83 0.68 6.85 0.20 2.05 13.0 0.68 6.95 0.20 2.19 14.8 SI 0.18 0.23 0.14 0.10 0.07 0.18 0.22 0.12 0.09 0.07 0.18 0.23 0.14 R 0.91 0.64 0.67 0.86 0.88 0.91 0.68 0.70 0.86 0.88 0.91 0.64 0.67 ss 0.95 0.94 1.06 1.03 1.01 0.95 0.93 1.05 0.95 1.01 0.95 0.94 1.06
Table 3.6: Summary of Measured WRB data compared to WTF Methods forced with offshore WRB boundary conditions at Wamberal WTF-P WTF-QS WTF-S
Hm0 Tp θp Hm0 Tp θp Tm02 θm Hm0 Tp θp Tm02 θm MS 0.94 0.75 0.77 0.88 0.92 0.95 0.76 0.79 0.91 0.92 0.94 0.75 0.77 Bias -0.11 -0.04 1.07 0.46 -0.50 -0.10 -0.24 2.80 -0.19 -0.12 -0.11 -0.04 1.07 RMSE 0.22 1.99 18.17 0.70 7.81 0.22 1.94 17.0 0.55 7.77 0.22 1.99 18.2 SI 0.15 0.23 0.15 0.08 0.07 0.15 0.22 0.14 0.08 0.07 0.15 0.23 0.15 R 0.92 0.61 0.63 0.87 0.88 0.92 0.63 0.66 0.88 0.88 0.92 0.61 0.63 ss 0.90 0.96 1.00 1.07 0.99 0.91 0.94 1.01 0.96 0.99 0.90 0.96 1.00
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 26
3.3.4 Verification of Directional Wave Climate
The WTF-QS and WTF-S methods provide directional spectra at nearshore locations that can be utilised for applications that require directional wave data, for example sediment transport. For the model calibration period, further analysis of the directional wave climate from the spectral WTFs has been undertaken.
Figure 3.2 presents the cumulative directional energy by directional bin for the Sydney offshore wave buoy (WRB-R).
Figure 3.2: Cumulative directional energy for the measured Sydney Offshore spectral data (WRB-R) for the 2011 calibration period
Figure 3.3 presents a comparison of the cumulative directional energy by directional bin for the WTF-QS and WTF-S methods against the measured spectra at Narrabeen. Similarly, Figure 3.4 presents the cumulative directional energy by directional bin for Wamberal. The results indicate that for Narrabeen, the WTF-QS results have a more easterly bias than the WTF-S results, and that the WTF-S results have better directional fit with the measured spectra (WRB-R - IMLM method). A consistent result is shown for Wamberal, with a more noticeable bimodality in the wave direction for the two transferred methods. The percentage difference between the measured and modelled
total energy is similar to the correlation between the measured and modelled Hs (as converted to wave energy, m0).
The nearshore directional wave climates at Wamberal and Narrabeen differ due to differences in the offshore bathymetry alignment and relative protection from the predominant southerly wave direction, with Long Reef Peninsula providing sheltering from southerly wave directions at Narrabeen and Wamberal receiving both the easterly and southerly wave energy. The comparisons provided in Figure 3.3 and Figure 3.4 show that the spectral WTF methods accurately
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 27
describe nearshore transformation of waves from offshore (as shown in Figure 3.2) and the differences in wave climates at the two sites.
Figure 3.3: Comparison of cumulative directional energy for the WTF-QS and WTF-S methods (red) against the measured spectral data (blue) at Narrabeen for the 2011 calibration period WTF-QS WTF-S
Figure 3.4: Comparison of cumulative directional energy for the WTF-QS and WTF-S methods (red) against the measured spectral data (blue) at Wamberal for the 2011 calibration period WTF-QS WTF-S
3.3.5 Summary
The evaluation of the WTF methods for the NSW coastline has produced high validation metrics and at the Narrabeen and Wamberal calibration sites the spectral WTF methods produced a similar set of validation metrics compared to the full spectral WW-III model results. The transfer functions were not changed based on the validation results. This is a promising outcome and highlights the potential to apply such methods to accurately and efficiently calculate long-duration
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 28
time series of nearshore wave data using long-term hindcast or measured deepwater wave data. The two spectral WTFs functions considered in this study produced very similar calibration results using the measured and hindcast data sets at the Narrabeen and Wamberal nearshore sites demonstrated improvement compared with the parametric WTF (WTF-P) method. The WTF-S has potential for application at a larger range of locations compared to the WTF-QS method due to the ability for frequency transformation to be represented if it is present in the numerical wave model results that are used to develop the WTF inputs. At the nearshore calibration locations considered in this study, being open coastline mid-compartment settings, the spectral WTFs (WTF- QS and WTF-S) produced very similar results to the full spectral wave model hindcast data as the measured and modelled frequency changes between offshore and the nearshore WRBs is small. Further validations at more protected and complex nearshore locations are recommended.
The transfer functions require a fraction of the computational requirements compared to a full spectral wave model and the spectral functions are capable of estimating the 2D direction- frequency spectra with reasonable accuracy. The WTF-P and WTF-S methods have been selected for implementation in the wave toolbox, the more basic WTF-P providing near instant transformation of long duration datasets while the WTF-S provides the most accurate nearshore solution of the WTF methods investigated. 3.4 Development of WTF for Entire NSW Coastline
Following the evaluation of the WTF options, wave transformation matrices for both the WTF-P and WTF-S methods were developed for entire NSW coastline along the 10m and 30m depth contours for incorporation into the Nearshore Wave Transformation Toolbox. A total of 14,510 nearshore sites were developed; 13,313 sites at 100m spacing along the 10m depth contour and 1,197 sites at 2km spacing along the 30m contour.
The WTFs were developed from a matrix of ‘steady-state’ simulations, as described in Section 3.2.2, using the NSW-wide shelf-scale WW-III model, as described in Appendix B. Note that different spectral shapes were adopted for the alternate WTF methods.
While the WTFs were shown to reproduce equivalent outcomes to a full spectral wave model and accurately replicate the nearshore wave climate at two calibration sites, the validations were completed at open coastline mid-compartment locations and may not be representative of all 14,510 sites established for the wave toolbox. Since the availability of nearshore measured wave data is limited (as discussed in Section 2.1.2), a 1-year hindcast of nearshore wave conditions at all wave toolbox locations was developed from the shelf scale WW-III model as a proxy for measured nearshore wave data.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 29
The hindcast year selected for this exercise was 2008, which is most representative of the average long-term wave climate along the entire length of the NSW coast. The WW-III model set up as previously applied was maintained and there was no factoring of the CFSR wind data set for the simulations. The shelf scale model simulation omits wind forcing over the shelf which was demonstrated in the sensitivity testing to be a minor factor when calibrating the wave model (see Appendix B).
State-wide validation metrics for the WTFs were developed from comparison of the 1-year nearshore hindcast (WW-III) against the transferred deepwater hindcast at all nearshore toolbox locations for both the WTF-P and WTF-S methods. The standard validation metrics of bias and symmetrical slope (ss), as presented in Appendix E, were evaluated comparing modelled and transferred wave height, peak period and peak direction parameters.
Summary statistics (mean and standard deviation) for the validation metrics across all sites are presented in Table 3.7 for the WTF-P and WTF-S methods. Good agreement is generally observed between the modelled and transferred wave parameters for the year-long validation dataset with the WTF-S method performing better than the WTF-P method for all parameters. However, there are outlier sites amongst the statistics. Full tables of validation metrics for each of the 13,313 10m nearshore points and 1,197 30m nearshore points are provided in a data file accompanying this report.
Table 3.7: Summary of Statistics for all 14,510 Wave Toolbox Output Locations WTF-P WTF-S
Hm0 Tp θp Hm0 Tp θp Mean StD Mean StD Mean StD Mean StD Mean StD Mean StD MS 0.97 0.02 0.83 0.08 0.84 0.11 1.00 0.01 0.92 0.02 0.95 0.04 Bias 0.05 0.04 0.43 0.23 -2.47 4.04 0.01 0.03 0.28 0.14 0.74 2.28 RMSE 0.19 0.06 1.65 0.37 18.64 4.43 0.08 0.03 1.16 0.19 10.41 3.06 SI 0.17 0.07 0.19 0.04 0.16 0.05 0.06 0.02 0.13 0.02 0.09 0.03 R 0.96 0.02 0.73 0.10 0.75 0.13 0.99 0.01 0.87 0.03 0.92 0.04 ss 1.01 0.04 1.04 0.02 0.96 0.04 1.01 0.03 1.03 0.02 1.01 0.03
3.4.1 Quality Rating Assessment
Utilising the validation metrics at each site, a quality rating system of nearshore sites was then developed to identify the sites within the Toolbox where the WTFs do not provide a comparable result to the calibrated WW-III spectral wave model. The algorithm to determine the quality classification of a site was defined as satisfying an acceptable bias and symmetrical slope. An acceptable symmetrical slope was defined as ±10 % of unity. An acceptable bias was defined as
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 30
±0.15 m for Hs, ±0.5 s for Tp and ±10 °N for θp. The same thresholds were used for the each transfer method, and based on discrete value range of acceptable metrics, rather than identifying outliers in the distribution of validation metrics (e.g. within a confidence limit of ±3 standard deviations from the mean). This ensures that sites with metrics within the acceptable range provide a physically accurate representation of the full spectral wave model. A detailed summary of the quality rating system is provided in Table 3.8. Note that the highest quality rating is reserved for locations where calibration to measured data has been completed.
Table 3.8: Quality flagging definitions for the WTF-P and WTF-S methods Percentage Percentage of nearshore of nearshore Colour Code Definition Explanation points points WTF-P WTF-S Bias and symmetrical slope outside acceptable thresholds# Red 3 Poor validation 45.1% 7.0 % for modelled data comparisons (see Table 3.7) Bias and symmetrical slope within Good validation acceptable thresholds# for Orange 2 to Modelled 54.9 % 93.0 % modelled data comparisons Data (see Table 3.7) Bias and symmetrical slope within Good validation acceptable thresholds# for Green 1 to Measured measured data comparisons 0.1% 0.1% Data (see Section 3.3) Narrabeen and Wamberal Only # Acceptable thresholds:
• Bias: ±0.15 m for Hs, ±0.5 s for Tp and ±10 °N for θp • Symmetrical Slope: ±10 % of unity
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 31
Figure 3.5: Distribution of quality control flags along the NSW coastline: WTF-P method (left) and WTF-S method (right)
3.4.2 Assessment of Offshore Forcing Location
The wave transformation function has been calibrated for both measured and modelled data input. For spatially sparse offshore data, such as the offshore buoy network, the coastline and nearshore locations is broken up into regions corresponding to each offshore data location (approximately centred on each offshore buoy location). For more spatially rich offshore model data, such as the 35-years NSW Storm Wave hindcast with boundary conditions at 5km spacing along the 80-100m contour, each nearshore output point of the Toolbox is coupled to a location directly offshore. The appropriateness of this selection was tested by forcing the transformation with offshore boundary points to the north and south of the coupled offshore point to quantify the influence on the resulting nearshore result.
Offshore boundary points to the North and South of the calibration sites were used to force the WTF and compared to the results from the Toolbox forcing (i.e. directly offshore). These alternate boundary points were approximately 50km to the North and South of each site. Both the WTF-P and WTF-S were forced with the 2008 global WW-III hindcast period for the three offshore boundary points and comparisons for the WTF-S method presented in Table 3.9 (Narrabeen) and
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 32
Table 3.10 (Wamberal). Note that any differences are a result of differences in the offshore wave conditions as the same WTF is applied in each case.
The results show that there is generally little difference between the offshore forcing cases, which is consistent with the expectation that the offshore wave climate would be very similar over the 100km offshore distance. Both the offshore and south locations show an improved result (over the north location) with negligible difference between the two. This confirms the suitability of selecting the closest offshore wave hindcast location for each Toolbox nearshore location.
Table 3.9: Influence of Offshore Boundary Forcing Location at Narrabeen (WTF-S method) North Offshore South
Hm0 Tp θp Hm0 Tp θp Hm0 Tp θp MS 0.94 0.70 0.76 0.98 0.88 0.94 0.98 0.90 0.96 Bias -0.08 -0.29 8.91 0.00 -0.22 -0.07 -0.04 -0.25 1.34 RMSE 0.28 2.18 19.96 0.17 1.39 10.01 0.14 1.31 8.37 SI 0.24 0.25 0.18 0.15 0.16 0.10 0.12 0.15 0.08 R 0.90 0.54 0.67 0.95 0.81 0.89 0.97 0.83 0.92 ss 0.95 0.94 1.07 0.99 0.96 0.99 0.96 0.95 1.01
Table 3.10: Influence of Offshore Boundary Forcing Location at Wamberal (WTF-S method) North Offshore South
Hm0 Tp θp Hm0 Tp θp Hm0 Tp θp MS 0.96 0.75 0.84 0.97 0.80 0.89 0.99 0.84 0.91 Bias 0.01 -0.10 4.17 0.05 0.21 0.69 0.02 0.05 1.43 RMSE 0.25 2.01 19.16 0.20 1.85 14.48 0.14 1.67 12.85 SI 0.21 0.23 0.15 0.16 0.21 0.12 0.12 0.19 0.10 R 0.93 0.60 0.74 0.95 0.67 0.81 0.98 0.73 0.85 ss 1.02 0.96 1.03 1.03 1.00 0.99 1.02 0.99 1.00
3.5 NSW Extreme Wave Climate Simulations
The characterisation of extreme wave conditions are an important part of coastal management and planning. However there is very little extreme wave information at most shallow water locations and the accuracy of the WTF methods to generate such data has not been tested. As a result, additional analyses to estimate first-pass extreme value wave conditions for each Nearshore Wave Transformation Toolbox location were performed using the shelf-scale WW-III model. A data set of design nearshore wave conditions representing 5, 20, and 100 years ARI wave
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 33 conditions has been computed for all 10 m depth and 30 m locations and uploaded into the Nearshore Wave Transfer Toolbox.
Table 3.11: Extreme Wave Scenarios developed for application in the Nearshore Wave Transformation Toolbox based on Shand et al. (2010) Byron Bay NE ESE SSE
ARI Hm0 Tp Hm0 Tp Hm0 Tp 5 yr 4.10 11.2 6.10 12.4 5.81 12.2 20 yr 4.57 11.5 6.80 12.8 6.48 12.6 100 yr 5.11 11.8 7.60 13.3 7.24 13.0 Sydney NE ESE SSE
ARI Hm0 Tp Hm0 Tp Hm0 Tp 5 yr 4.20 11.0 5.79 12.0 7.00 12.8 20 yr 4.80 11.4 6.61 12.6 8.00 13.5 100 yr 5.40 11.8 7.44 13.1 9.00 14.1 Batemans Bay NE ESE SSE
ARI Hm0 Tp Hm0 Tp Hm0 Tp 5 yr 4.21 11.1 5.24 11.7 5.71 12.0 20 yr 4.86 11.5 6.04 12.2 6.58 12.6 100 yr 5.50 11.9 6.84 12.7 7.46 13.1 Eden NE ESE SSE
ARI Hm0 Tp Hm0 Tp Hm0 Tp 5 yr 4.64 11.4 5.78 11.6 6.29 11.8 20 yr 5.36 11.5 6.67 11.8 7.26 12.0 100 yr 6.07 11.7 7.56 12.0 8.23 12.2
The deep water extreme wave scenarios were derived from that documented by Shand et al. (2010) based on extreme value analysis of long term measured deep water wave data sets at four sites: Byron Bay, Sydney, Batemans Bay and Eden (Table 3.11). Directional extreme wave statistics in three directional sectors (NE 0-90°, ESE 90-135°, SSE 135-225°) were available for Byron Bay,
Sydney and Batemans Bay for the 10 years ARI (Hs) scenario (Shand et al., 2010 - Table 4.10). The directional analysis was extended to the 5, 20 and 100 years ARI scenarios using proportional relationships based on the 10 years ARI scenario and the ‘all directional’ Hs ARI values reported in Appendix F of Shand et al. (2010). The analysis was extended to Eden (as the extreme wave
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 34 climate becomes more energetic south of Batemans Bay) using the same directional proportion as
Batemans Bay. The peak wave period was related to the return level Hs using a linear relationship developed with reference to the storm wave dataset documented in Cardno (2012). Note that there are uncertainties associated with each reported ARI value and these have not been accounted for in this modelling.
These scenarios were applied as steady-state point offshore boundary conditions to the shelf-scale WW-III model. The WW-III model then interpolates the sparse offshore boundary points along the offshore boundary using nearest-neighbour interpolation. The resulting nearshore wave parameters are approximate only and contain uncertainty for several reasons:
• The offshore extreme scenarios have been approximated from the best available data, which themselves contain uncertainty, at sparse intervals; • The nearshore extreme wave parameters are not necessarily (and unlikely to be) linearly related to the offshore extreme return levels, due to the non-linear processes of wave refraction, diffraction, shoaling etc.; • The use of three representative wave directions may not capture the most extreme case at a given nearshore location which may be highly sensitive to wave direction depending on the local bathymetric setting.
A more robust method to compute extreme nearshore wave conditions would be to apply a suitably long offshore wave dataset to the WTF (or model) and then compute the extreme value statistics on the resulting nearshore parameters.
Figure 3.6 shows the estimated nearshore extreme wave height (Hm0) for the three directional 100 years ARI scenarios; northeast, east-southeast and south-southeast. The figure highlights the difference extreme wave height exposure from the NE compared to the more southerly directions and the variation along the NSW coastline. These and the other extreme wave scenarios can be investigated in more detail using the visualisation provided by the Nearshore Wave Transformation Toolbox which enables zooming and individual site selection.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 35
Figure 3.6: Estimated 100 year ARI extreme nearshore wave heights (Hm0) for three directional sectors from the WW-III shelf-scale model using deepwater extreme scenario boundary conditions
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 36
3.6 Application of the WTF
An initial scope item requested from OEH was to demonstrate the functionality of the Nearshore Wave Transformation Toolbox by completing a continuous 35-year coastal wave hindcast at the two primary nearshore calibration locations: Narrabeen and Wamberal. This was to be based on the 35-year deep-water NSW storm wave hindcast being developed by OEH. This item is to be be specifically addressed once the toolbox is operational on the OEH network to demonstrate the final toolbox functionality.
As an interim deliverable, the WTF was utilised to develop a long duration hindcast at the calibration sites using the full duration of the available Sydney WRB record, from 1992 to 2013 (provided as an electronic data file accompanying this report). The measured data set was transferred using the parametric WTF (WTF-P) as a directional re-analysis of the whole data set has not been completed and therefore appropriate offshore data is not available to apply the spectral (WTF-S); that is, only the year 2008 has been re-analysed for this study. It is recommended that following this project OEH undertake a re-analysis of all available buoy displacement data to derive a reliable offshore spectral wave dataset for application in the final wave toolbox.
The 22 years of transferred Directional WRB nearshore data is compared to the limited duration nearshore Directional WRB measurements (calibration data sets) in terms of wave height exceedances within Table 3.12 and Table 3.13. The differences are in wave height exceedance values between the datasets are in keeping with the validation statistics for the WTF-P method, described in Section 3.3.
Table 3.12: Percentage Exceedance Wave Heights (m) at Narrabeen Percentage Exceedance Nearshore Data Source 99% 90% 50% 10% 1% Measured Nearshore Data (2011) 0.47 0.66 1.02 1.94 3.09 Transferred Nearshore Data (2011) – WTF-P 0.28 0.54 0.93 1.64 3.10 Transferred Nearshore Data (1992 – 2013) – WTF-P 0.26 0.53 0.94 1.63 2.74
Table 3.13: Percentage Exceedance Wave Heights (m) at Wamberal Percentage Exceedance Nearshore Data Source 99% 90% 50% 10% 1% Measured Nearshore Data (2011) 0.57 0.78 1.24 2.09 3.01 Transferred Nearshore Data (2011) – WTF-P 0.48 0.72 1.11 1.81 2.69 Transferred Nearshore Data (1992 – 2013) – WTF-P 0.39 0.65 1.09 1.90 3.19
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 37
4.0 The Nearshore Wave Transformation Toolbox
A web-based toolbox, referred to as the Nearshore Wave Transformation Toolbox (the Toolbox), has been developed to allow the full suite of nearshore wave parameters to be presented along the entire NSW coastline based on the adopted transformation method and selected input offshore wave conditions. The Toolbox enables the user to rapidly transform offshore wave conditions to the 10m and 30m depth contours along the entire NSW coastline, and visualise or extract the transformed nearshore wave conditions.
A key aspect of the Toolbox is that it can be developed and improved over time, and hence has been developed in an open source, web-based architecture. The Toolbox has the potential to be updated for future improvements in the transformation matrices, be it better bathymetric data at a site or upgraded model application.
A comprehensive description of the toolbox functionality, along with user access details, is presented in the toolbox user manual included in Appendix F. The following provides a brief summary of key features of the toolbox. 4.1 Description
The Nearshore Wave Transformation Toolbox’s web-based interface consists of a user control panel and a Google Maps-based display for map based referencing of the data.
Figure 4.1: Nearshore Wave Transformation Toolbox Interface (Spatial Transform)
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 38
Nearshore wave conditions can be derived from either deep-water historical Waverider buoy measurements or modelled hindcast data using one of two available transformation methods; Parametric and Spectral.
The toolbox consists of a number of pages each with unique functionality, as follows:
• Spatial Transform: provides a spatial map of transformed nearshore wave conditions for a selected day and time. • Timeseries Transform: provides a timeseries plot of transformed wave conditions for a selected time period at a selected nearshore location. • Spectral Transform: provides the directional spectrum of a transformed wave condition for a selected day and time at a selected nearshore location. • Data Extraction: allows the user to utilise the full capability of the toolbox and extract continuous transformed wave data across multiple nearshore locations in a range of common data formats. • Extreme Waves: provides a spatial map of pre-computed simulated nearshore extreme wave conditions derived from deep-water return period wave estimates presented by Shand et. al. (2011).
4.1.1 Quality Coding
The Toolbox presents a quality coding scheme that indicates the quality of the transformation at each location. To assist in identifying the quality of the transformation a system of quality codes has been developed each output location, using traffic light coding, as described in Section 3.4.1. 4.2 Toolbox Code
The toolbox has been developed using PHP and Perl, which are the scripting languages commonly used by MHL to develop the operational webpage and perform other functions. These are commonly used languages and employ open source compilers and development environments so the toolbox is able to be readily modified into the future. As required, other software may be used to interrogate the transformation matrix, and perform other analyses such as extreme value analysis or statistical functions. This may include Matlab or Python, both of which are widely used programming languages. 4.3 Toolbox Updating Potential
Updating of the transformation matrix is facilitated by a suite of Matlab code. Updating of the transformation matrix may be warranted as the NSW WW-III model is improved and/or other
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 39
detailed studies are completed across NSW including additional nearshore wave data collection and/or detailed site specific modelling.
It is outside the scope of this project to validate the matrix across the entire NSW coast, but it is anticipated that future coastal studies in NSW will provide refinements at each location. 4.4 Compatibility with Existing OEH Tools
A key requirement of the Toolbox was to ensure compatibility with the existing OEH Wave Model Tool. This was achieved by ensuring output formats from the Toolbox (such SWAN tab and sp1 files) are able to be read in by the OEH Wave Model Tool to allow the full functionality of that tool to be utilised.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 40
5.0 Summary
Baird Australia and MHL have successfully completed the calibration of alternate WTF methods for two nearshore sites on the Sydney (Narrabeen) and Central Coast (Wamberal) regions. In order to develop the WTF’s, Baird Australia contributed outputs from a Wavewatch-III v4.18 shelf scale model that can be coupled to OEH’s existing global scale WW-III model. The shelf scale WW-III model achieved higher validation metrics, particularly for wave period, compared to earlier studies where SWAN nearshore wave models were coupled to the WW-III model.
Three separate WTF algorithms were investigated during the calibration process. The first is a simple parametric transfer function (WTF-P) that is capable of being applied to estimate nearshore wave conditions based on integrated parameters of the total offshore sea state; that is, wave height, period and direction. The WTF-P has very low computation and data requirements, and can easily be deployed in real-time and on the web to develop nearshore long duration wave data sets. The WTF-P can also be rapidly applied to the entire record of the archived parametric wave data from the NSW directional WRB’s.
Two spectral WTF methods, the WTF-QS and WTF-S have also been developed. The WTF-QS is a hybrid method that combines the simplicity of the parametric WTF with the additional capability to approximate the transferred directional spectra at nearshore sites. The WTF-S method is a fully spectral WTF and requires the most detailed data (2D spectra) as input to the WTF. The three WTF achieved good calibration when compared to the WW-III model at the nearshore calibration sites, with the calibration of the two spectral methods being excellent. Applying deepwater WRB-R (re- analysis directional data) to the WTF-QS and WTF-S, demonstrated better validation metrics compared to applying the WW-III hindcast model data set at the deepwater reference locations; an artefact of the calibration of the WW-III hindcast model at the offshore boundary location.
Comparison of the nearshore cumulative directional energy spectra from the two spectral WTF’s over the calibration period with measured wave data indicates that the WTF-S provides a closer match to the WRB-R derived directional energy distribution nearshore.
Following extensive validations, two methods were deployed in the Nearshore Wave Transformation Toolbox, the WTF-P and WTF-S. The WTF-P provides a rapid solution for transferring long duration records to nearshore locations, and does not require full spectral offshore data as input. This makes it applicable to the majority of offshore measured data, where directional spectra are either not available or not of sufficient quality for deriving nearshore conditions. The WTF-S provides for a more accurate transformation to nearshore locations, however requires 2D spectral data as input. This study has investigated the available offshore spectral data archived by MHL and found that the directional description of the archived buoy derived analysis is inadequate. Detailed directional spectral analysis was subsequently completed
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 41 for the 2008 period to inform validation tasks and the resulting nearshore transformations produced excellent outcomes. It is recommended that OEH embark on a process to re-analyse the available archived offshore buoy displacement data using the techniques described in Section 2.1.2.2 and Appendix D. This will ensure a high quality spectral wave dataset for future use in the toolbox and other coastal studies.
WTF’s have been developed at 14,510 nearshore locations for the entire NSW coast, at 100m spacing along the 10m depth contour (13,313 locations) and at 2km spacing along the 30 m depth contour (1,197 locations). The WTF can be interrogated and applied via the web-based Nearshore Wave Transformation Toolbox. A toolbox user manual that provides access details and a comprehensive description of the toolbox functionality is presented in Appendix F.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 42
6.0 References
Amante, C. and B.W. Eakins, 2009. ETOPO1 1 Arc-Minute Global Relief Model: Procedures, Data Sources and Analysis. NOAA Technical Memorandum NESDIS NGDC-24. National Geophysical Data Center, NOAA. doi:10.7289/V5C8276M.
Ardhuin, F., W. E. Rogers, A. V. Babanin, J. Filipot, R. Magne, A. Roland, A. van der Westhuysen, P. Queffeulou, J. Lefevre, L. Aouf and F. Collard, 2010: Semiempirical dissipation source functions for ocean waves. Part I: Definition, calibration, and validation. J. Phys. Oceanogr., 40, 1,917–1941.
Baird Australia (2014a). NSW Coastal Wave Hindcast – (CORDEX/NARCliM) Wind Data and Wave Model Validation Report, prepared for Office of Environment and Heritage (NSW), March, 52 pp.
Baird Australia (2014b). NSW Coastal Wave Hindcast – 60-year CORDEX Wave Hindcast Dataset, prepared for Office of Environment and Heritage (NSW), August, 27 pp.
Benoit, M. (1993) Practical comparative performance survey of methods used for estimating directional wave spectra from heave-pitch-roll data. Proc.23rd ICCE Vol 1. ASCE pp.62-75.
Benoit, M., Frigaard, P., Schaffer, H. (1997) Analyzing multidirectional wave spectra: a tentative classification of available methods. In Proceedings of the 1997 IAHR conference, San Francisco (1997), pp. 131-158
Branson, P., McConochie, J., Churchill, J. and Williams, B. (2013). Estimation of long-term harbour wave climates using spectral transfer functions. Proceedings of Coasts and Ports Conference, (11- 13 September, Sydney).
Bryant, E. A. (1991) Natural Hazards. Cambridge University Press, Melbourne.
Bryant, E. A. & Kidd, R. W. (1975) Beach erosion, May-June, 1974, Central and South Coast, NSW. Search 6: 510-3.
Cardno (2012). “NSW Coastal Waves: Numerical Modelling – Final Report”. Prepared for the NSW Office of Environment and Heritage. Ref LJ2949/R2745. Version 3. September 2012.
Chapman, D. M., Geary, M., Roy, P.S. & Thom, B.G. (1982), Coastal Evolution and Coastal Erosion in New South Wales: A report prepared for the Coastal Council of New South Wales. Coastal Council of New South Wales, Sydney. 341 pg.
Cresswell, G.R. (1994) Nutrient enrichment of the Sydney continental shelf. Australian Journal of Marine and Freshwater Research 45: 677–691.
Datawell BV (Feb 14, 2013) Datawell Waverider Reference Manual. Service &Sales 1704 RP Heerhugowaard, The Netherlands.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 43
Davies, P. J. (1979) Marine geology of the continental shelf off Southeastern Australia. Bureau of Mineral Resources Bulletin 195.
Dent J., Kinsela M., Taylor D., Garber S, Treloar D. (2014). “Evaluation of Downscaled Climate Model Outputs for Generating Long-Term Wave Climate Statistics for the NSW Coastline.” 2014 NSW Coastal Conference (Ulladulla, November 2014).
Dominguez J., Cienfuengos R., Catalan P., Lucero F., Zamorano L. (2014). Assessment of Fast Spectral Wave Transfer Methodologies from Deep to Shallow Water. The 34th International Conference of Coastal Engineering. Seoul, Korea. 15-20 June 2014.
Eagle, R.J. & Malone, J.D. (1986) Summary of Storm Damage and Associated Sea and Weather Conditions on the NSW Coast – 1876 to 1981. PWD Report No 86014, NSW Public Works Department Coastal Branch, Sydney, NSW.
Easton, A. K. (1970) Tides of the Continent of Australia. Horace Lamb Centre, Adelaide.
Gentilli, J. (1971) Climates of Australia and New Zealand. World Survey of Climatology, 13.
Godfrey, J. S., Creswell, G. R., Golding, T. J., Pearce, A. F. & Boyd, R. (1980) The separation of the East Australian Current. Journal of Physical Oceanography 10: 430-440.
Griffin, D.A. & Middleton, J.H. (1991) Local and Remote Wind Forcing of New South Wales Inner Shelf Currents and Sea Level. Journal of Physical Oceanography 21(2): 304-322.
Griffin, D.A. & Middleton, J.H. (1992) Upwelling and internal tides over the inner New South Wales continental shelf. Journal of Geophysical Research 97: 14389–14405.
Hamon, B.V. (1987) A Century of Tide Records: Sydney (Fort Denison) 1886-1986. Flinders Institute for Atmospheric and Marine Sciences.
Hanslow, D.J. & Howard, M. (2005) Emergency Management of Coastal Erosion at NSW. In: Planning for Natural Hazards – How Can We Mitigate the Impacts? R. John Morrison, S. Quin and E.A. Bryant (Eds) Proceedings of a Symposium, University of Wollongong, 2-5 February 2005.
Hashimoto, N., Nagai, T., Asai, T. (1993). Modification of the extended maximum entropy principle for estimating directional spectrum in incident and reflected wave field. Rept. Of P.H.R.I. 32(4) 25- 47.
Holland G. J., Lynch, A. H. & Leslie, L. M. (1987) Australian east coast cyclones. Part I: Synoptic overview and case study. Monthly Weather Review 115: 3024–3036.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 44
Hopkins, L. C. & Holland, G. J. (1997) Australian heavy-rain days and associated east coast cyclones: 1958–92. Journal of Climate 10:621–635.
Johnson D., (2012). “DIWASP, a directional wave spectra toolbox for MATLAB®: User Manual.” Research Report WP-1601-DJ (V1.4), Centre for Water Research, University of Western Australia.
Kinsela M., Taylor D., Treloar D., Dent J., Garber S., Mortlock T., and Goodwin I. (2014). “NSW Coastal Ocean Wave Model: Investigating Spatial and Temporal Variability in Coastal Wave Climates”. Proceedings of 23rd NSW Coastal Conference (11-14 November, Ulladulla).
Kulmar M., Modra B., Fitzhenry M. (2013). The New South Wales Wave Climate - Improved Understanding through the Introduction of Directional Wave Monitoring Buoys. Proceedings of Coasts and Ports Conference, (11-13 September, Sydney).
Lawson, N. V. & Abernethy, C. L. (1975) Long-term wave statistics off Botany Bay. 2nd Aust. Conf. on Coastal and Ocean Engineering. Gold Coast, Queensland. Institute of Engineers, Australia. Pp. 167-176.Short, A.D. & Wright, L.D. (1981) Beach systems of the Sydney region. Australian Geographer 15: 8–16.
Longuet-Higgins M.,Cartwright D., Smith N. (1963) Observation of the directional spectrum of sea waves using the motions of a floating buoy, in Ocean wave spectra, Prentice-Hall, pp111-136
Lord, D. & Kulmar, M. (2000) The 1974 storms revisited: 25 years experience in ocean wave measurement along the south-east Australian coast. 27th International Conference on Coastal Engineering Conference Proceedings: July 16-21, 2000, Sydney, Australia.
Matthai, C. & Birch, G. F. (2001) Detection of anthropogenic Cu, Pb and Zn in continental shelf sediments off Sydney, Australia – A new approach using normalization with cobalt. Marine Pollution Bulletin, 42(11): 1055-1063.Ferland et al 1995
Middleton, J.H. (1995) The oceanography of Australian seas. In: Zann, L.P. and Kailola, P. (eds.) The Sate of the Marine Environment Report for Australia, Technical Annex 1 The Marine Environment, Ocean Rescue 2000. Great Barrier Reef Marine Park Authority, Townsville.
Middleton, J.H., Cox, D. & Tate, P. (1996) The oceanography of the Sydney region. Marine Pollution Bulletin 33: 124-131.
Monardez, P.M., Acuna, H., Scott, D. 2008. Evaluation of the potential of wave energy in Chile. OMAE2008-57887. Proceedings of the ASME 27th International Conference on Offshore Mechanics and Arctic Engineering OMAE2008. June 15-20, 2008, Estoril, Portugal.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 45
Mortlock T., and Goodwin I. (2013). Calibration and sensitivity of nearshore SWAN model performance for measured and modelled wave forcing scenarios, Wamberal, Australia. Prepared for Office of Environment and Heritage by Macquarie University.
NOAA (2013) WaveWatch III Tutorial. Version 1.1 Feb 2013. http://polar.ncep.noaa.gov/waves/index2.shtml
Oke, P. R. & Middleton, J.H. (2001) Nutrient enrichment of Port Stephens: the role of the East Australian Current. Continental Shelf Research 21: 587-606.
Oliver, J. (1979) Wind and storm hazards in Australia. Natural Hazards in Australia. Canberra.
Pawka,S.S. (1983). Island shadows in wave directional spectra. J.Geophys.Res. 88(C4) 2579-2591.
PWD (1985) Elevated Ocean Levels. Storms affecting the NSW Coast 1880-1980. Report prepared by Blain, Bremner & Williams Pty Ltd and Weatherex Meteorological Services Pty Ltd, Public Works Department of NSW Report No. 85040, May 1985.
PWD (1986) Elevated Ocean Levels. Storms affecting the NSW Coast 1980-1985. Report prepared by Lawson and Treloar Pty Ltd and Weatherex Meteorological Services Pty Ltd, Public Works Department of NSW Report No. 86026, August 1985.
PWD (1988) Coastal Storm in NSW August and November 1986 and their effect on the coast. Public Works Coast and Rivers Branch. Prepared by University of NSW Water Research Laboratory, May 1988.
PWD (1989) Batemans Bay Oceanic Inundation Study. Public Works Department Coast and River Branch, New South Wales, Sydney, Australia.
Reid, H. J. & Leslie, L.M. (1999) Modeling coastally trapped wind surges over Southeastern Australia. Part I: Timing and speed of propogation. Weather and Forecasting 14: 53-66.
Ridgway, K.R. & Godfrey, J.S. (1997) Seasonal cycle of the East Australian Current, Journal of Geophysical Research 102 (C10): 22,921–22,936.
Roy, P. S., Thom, B. G. & Wright, L. D. (1980) Holocene sequences on an embayed high-energy coast: an evolutionary model. Sedimentary Geology 28: 1-19.
Roy, P.S., Williams, R.J., Jones, A.R., Yassini, I., Gibbs, P.J., Coates, B., West, R.J., Scanes, P.R., Hudson, J.P. & Nichol, S.L. (2001) Structure and function of south-east Australian estuaries. Estuarine, Coastal and Shelf Science 53: 351–384.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 46
Shand, T. D., Goodwin, I. D., Mole, M. A., Carley, J. T., Browning, S., Coghlan, I. R., Harley, M.D., and Peirson, W. L. (2011). NSW Coastal Inundation Hazard Studies: Coastal Storms and Extreme Waves. Water Research Laboratory, Technical Report 2010/16.
Short, A.D. (2006) Australian Beach Systems – Nature and Distribution. Journal of Coastal Research 22(1):11-27.
Short, A.D. (1993) Beaches of the New South Wales Coast. Australian Beach Safety and Management Program, Sydney, 356 pg.
Spinoccia, M., 2013. NSW Continental Slopes Survey (GA-2413/SS10/2006) - Bathymetry grids. Geoscience Australia, Canberra.
Taylor, D., Garber, S., Burston, J., Couriel, E., Modra, B. and Kinsela, M. (2015). Verification of a Coastal Wave Transfer Function for the New South Wales Coastline. Proceedings of Coasts and Ports Conference, (15-18 September, Auckland).
Thom, B. G., McLean, R. F., Langford-Smith, T. & Elliot, I. (1973) Seasonal beach change, central and south coast, N.S.W. In: Engineering Dynamics of the Coastal Zone, Institute of Engineers Australia. 73/1, 35-42. Short & Trenaman 1992
Tolman, H. L. and D. V. Chalikov, 1996: Source terms in a third-generation wind-wave model. J. Phys. Oceanogr., 26, 2,497–2,518.
Tolman, H. L., 2002: Alleviating the garden sprinkler effect in wind wave models. Ocean Mod., 4, 269–289.
Tolman H. (2014). User manual and system documentation of WAVEWATCH III R° version 4.18. NOAA / NWS / NCEP / MMAB Technical Note: 316, March 2014.
Trenaman, N.L. & Short, A.D. (1987) Deepwater and breaker wave climate of the Sydney region. Coastal Studies Unit Technical Report 87/1, Coastal Studies Unit, University of Sydney.
Whiteway, T., 2009. Australian Bathymetry and Topography Grid, June 2009. Scale 1:5000000. Geoscience Australia, Canberra. http://dx.doi.org/10.4225/25/53D99B6581B9A
Wright, L. D. (1976) Nearshore wave-power dissipation and the coastal energy regime of the Sydney-Jervis Bay region, New South Wales: A comparison. Australian Journal of Marine and Freshwater Research 27: 633-640.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Page 47
Appendix A Project Scope and Deliverables
Table A.1: Scope and Deliverable Checklist Methodology / Scope Item Report Section Notes 1. Develop an nearshore wave transformation matrix for all NSW coastal waters using a shallow-water spectral wave model and a discretised array of deep-water wave conditions (e.g. wave height, period and direction) that encompasses the feasible range of NSW deep-water wave climates - Establish a general shallow-water spectral wave model (e.g. SWAN, WW-III…) for NSW waters Appendix B WW-III model developed
- Develop a series of overlapping nearshore wave model grids/meshes covering the entire NSW coastline (including major Single shelf to coastline bays), for both bathymetry, and where data permits, bottom friction scale irregular mesh was Section 3.2 & (e.g. based on seabed type) – model grids are to be of sufficient developed, removing the Appendix B offshore extent and spatial resolution to provide reliable wave need for multiple predictions in to 10-15 water depth, for measured and modelled overlapping grids. deep-water wave inputs - To the extent possible, calibrate the shallow-water spectral wave Calibration of the WW-III model and series of nearshore grids using available nearshore wave model completed at two measurement datasets Appendix B nearshore locations where measurements were available - Evaluate model skill where possible using available nearshore wave Full range of validation measurement datasets Appendix B metrics was determined, including model skill. - Develop a discretised array of offshore wave conditions that Matrix of WW-III simulations combines key parameters such as wave height, period and direction across wave period and (spectral or parametric) to cover the feasible range of deep-water Section 3 (specifically direction range, adopting wave climates for NSW – parameter bin resolutions should Section 3.2.2) different spectral shape adequately capture spatial and temporal variability in the key wave depending on WTF method. parameters - Run the wave model for the full array of discretised deep-water Completed for a total of conditions, for each nearshore grid/mesh, to derive the 14,510 nearshore locations corresponding shallow-water wave transfer functions at each grid Section 3.4 (13,313 at the 10m depth cell or mesh element for the feasible range of deep-water forcing contour and 1,197 at the conditions 30m depth contour). - Build an nearshore wave transformation matrix database with the transfer functions derived from the simulations, in a portable and Wave transformation Section 3.4 & efficient format (e.g. NetCDF) that can be readily queried by the functions are queried via the Section 4 tools developed to derive design nearshore wave conditions and to web-based Toolbox. develop coastal wave hindcast datasets
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix A
Methodology / Scope Item Report Section Notes - The design of the nearshore wave transformation matrix and model Background toolkit grids/meshes must allow for updating as new datasets (e.g. (MATLAB code) allows bathymetry, bottom type, deep-water forcing) become available, Section 4.4 updates to bathy and and to accommodate alternative wave modelling approaches (e.g. transfer matrices. Boussinesq) for complex sites 2. Develop a user-friendly nearshore wave toolbox that allows users to query the transformation matrix, to derive nearshore wave conditions at any location in NSW based on the input deep-water wave conditions from measured or modelled datasets. The toolbox should feature the following functions: - An nearshore wave processing tool that reads deep-water wave data from various sources (e.g. waverider buoys, NSW Coastal Wave Model, AUSWAVE), queries the nearshore wave transformation Draft web-based Toolbox matrix, and outputs coastal wave conditions across the nearshore Section 4 http://104.197.61.80/ wave model grids/meshes in a portable and efficient format (e.g. NetCDF), and as time series and spectra at user-selected locations (e.g. .tab) - A tool that allows the user to update the nearshore wave transformation matrix with revised transfer functions based on new Background toolkit (MATLAB code) to allow input datasets (e.g. bathymetry, bottom type, deep-water forcing) or Section 4.4 alternative wave modelling approaches (e.g. Boussinesq) updates to bathy and transfer matrices.
- An error/uncertainty qualification tool that allows the user to apply Quality flags developed for spatially defined corrections to the calculated nearshore wave all nearshore locations conditions based on error estimates derived from evaluation against based on WTF comparisons Section 3.4 nearshore wave measurement datasets (e.g. waverider buoys, to nearshore measurements satellite altimeter, HF radar) or full spectral WW-III model. - A tool that processes gridded and point output data from the Nearshore Wave Toolbox nearshore waves processing tool, in formats that support integrated provides *.tab and *.sp1 application with the existing OEH Wave Model Toolbox for output Section 4 outputs, compatible with the data analysis and visualisation (or alternatively provides similar existing OEH Wave Model output processing functionality) Toolbox - Outputs from the nearshore wave toolbox should be compatible with the existing OEH Wave Model Toolbox (or the nearshore wave toolbox should provide for similar output processing functionality), Outputs as tab and sp1 files which allows users to: plot time series and time-averaged spectra; Section 4 for interrogation by existing create spatial maps of key wave statistics; identify and analyse storm OEH Wave Model Toolbox. events using the peaks over threshold method; and, calculate extreme value statistics at locations of interest.
3. Derive design nearshore wave conditions for NSW based on extreme value analysis of measured (waverider buoy) and modelled (WW-III) historical deep-water wave records - Using extreme value analysis techniques, calculate probabilistic Analysis of Shand et. al. design statistics for key wave parameters based on measured and (2010) informed boundary simulated historical deep-water wave climate datasets (the design conditions for a series of Section 3.5 statistics should be regionally significant – i.e. guided by measured WW-III EVA runs. Spatial and simulated wave records and considering latitudinal variability in and nearshore results to be NSW deep-water wave climates) accessed via the Toolbox.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix A
Methodology / Scope Item Report Section Notes - Using the nearshore wave transformation matrix, calculate the Design (EVA) conditions coastal wave conditions across each nearshore grid/mesh modelled in the full spectral corresponding to the design deep-water wave conditions WW-III model. Section 3.4 Full-year hindcast developed for all nearshore locations. - Generate a series of spatial maps displaying key wave parameters for Map output of EVA WW-III the design nearshore wave conditions, for each of the nearshore Section 3.5 & simulations to be provided model grids/meshes Section 4.1.1 in mat format and displayed in toolbox - The design coastal wave conditions should be stored in a portable and efficient database format (e.g. NetCDF), which allows for the Toolbox to allow for extraction of wave parameters at user-defined locations and along Section 4.1.1 interrogation of the WW-III transects between deep-water and nearshore locations EVA simulations.
4. Develop coastal wave hindcast datasets for NSW using the nearshore wave transformation matrix and OEH’s continuous 35- year (1979-2013) deep-water NSW Storm Wave Hindcast - Demonstrate the functionality of the nearshore wave toolbox by 35-year deep-water NSW completing a continuous 35-year coastal wave hindcast (based on Storm Wave Hindcast not the 35-year deep-water NSW Storm Wave Hindcast), for model available in required format grids/meshes that are identified as featuring high priority sites of for application to the WTFs. interest Processing script provided to OEH to convert hindcast output to compatible WTF Section 3.6 boundary conditions. Functionality demonstrated through 2008 hindcast to all Nearshore Wave Toolbox locations and full Sydney record (1992 to 2013) to calibration sites (WTF-P). - Demonstrate integrated application between the nearshore wave Nearshore Wave Toolbox toolbox and the existing OEH Wave Model Toolbox by: plotting time provides *.tab and *.sp1 series and time-averaged spectra; generating spatial maps of key Section 4.4 outputs, compatible with the wave statistics; analysing storm events using the peaks over existing OEH Wave Model threshold method; and, calculating extreme value statistics at Toolbox. locations of interest. - The locations of high priority coastal wave hindcasts will be The toolbox provides OEH determined in collaboration with OEH once the nearshore wave the ability to generate - transformation matrix and toolbox have been developed. hindcast timeseries at any NSW nearshore location.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix A
Appendix B Wave Model System
Baird Australia have developed and implemented a new global to shelf scale spectral wave model for this study; focussing with high resolution on the NSW coastline. The model setup incorporates the most up-to-date modelling framework and provides a consistent set of wave physics from deep water to the point of wave breaking near the shoreline. B.1 Model Description: WaveWatch III 4.18
WaveWatch-III (WW-III) is a full-spectral third-generation wind-wave model developed by the USA’s National Center for Environmental Prediction (NCEP) (Tolman 2014). It solves the random phase spectral action density balance equation for wave-number-direction spectra and accounts for a wide range of wave physics. Physical processes (source terms) that can be included are wave growth and decay due to the actions of wind, nonlinear resonant interactions, dissipation (`white- capping'), bottom friction, surf-breaking (that is, depth-induced breaking) and scattering due to wave-bottom interactions. For this study, the third-order accurate numerical scheme is used describe wave propagation (Tolman 1995). The source terms are integrated in time using a dynamically adjusted time-stepping algorithm, which concentrates computational efforts in conditions with rapid spectral changes (Tolman 2014).
The WW-III version 4.18 third-generation spectral wind-wave model (Tolman, 2014) was used in this study to develop the NSW nearshore wave model for this study. Released in March 2014, it is the latest version of the WaveWatch-III model and an update to the model version (3.14) used for the NSW Coastal Wave Model (Cardno, 2012). The updates in version 4.18 include a number of developments to the modelling framework of particular interest to this project, including the addition of unstructured grid functionality and the inclusion of shallow water physics such as triad interactions. The development of shallow water wave transformation in WW-III provide the opportunity to model ocean scale wave generation and nearshore transformation in a single modelling scheme. This offers advantages over the previous NSW Coastal Wave Model approach that coupled WW-III with the Simulating WAves Nearshore (SWAN) model. As noted in Cardno (2012), previous phases of the NSW Coastal Wave Modelling project identified that there were discrepancies between the WW-III and SWAN models that appeared related to the wind growth
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix B
physics and the transformation of the frequency spectrum with wave propagation in space and time. This most notably manifested as a low bias in the modelled wave period.
Table B.1 summaries the model physics and parameters adopted in the calibrated WW-III model following the model calibration and sensitivity tests.
Table B.1: Summary of WW-III shelf scale model parameterisation. Parameter Description Mesh Resolution Variable from 2.5 km at ≈ 100 m depth to 100 m at 10 m depth. Time Step 2 min Higher order schemes with Tolman (2002) averaging technique. Propagation Scheme Third order propagation scheme. Flux Computation Friction velocity from Tolman and Chalikov input. Linear Input Cavaleri and Malanotte-Rizzoli with filter. Input and Dissipation Tolman and Chalikov (1996) source term package. Stabilisation Stability correction enabled. Non-linear Interactions Discrete interation approximation (DIA). Bottom Friction JONSWAP bottom friction formulation. Sea Ice No damping by sea ice. Reflection No reflection. Depth Induced Breaking Battjes-Janssen. Triad Interactions Lumped Triad Interaction method. Bottom Scattering Magne and Ardhuin. Shallow Water Limiter Miche-style.
B.2 Model Set-Up
The wave modelling system adopted in this project is based on the multi-domain Wavewatch- III/SWAN modelling system developed for the OEH NSW Wave Model Project as documented in Cardno (2012). The model is set up for one-way nesting, where each model grid is run consecutively in cascading order from lowest to highest spatial resolution. The three model grids were as follows:
• Grid 1: Global 1-degree resolution; • Grid 2: Australian region 0.25-degree resolution; • Grid 3: NSW coastline and Tasman Sea region 0.05-degree resolution.
Figure B.1 shows the layout of the Australian and NSW region grids in the Wavewatch-III model.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix B
Figure B.1: Global Scale NSW WW-III Model (Cardno , 2012)
A new fourth nested grid created as an unstructured mesh covering the NSW coast has been developed for this project (replacing the previous SWAN transition model). The use of an unstructured mesh negates any need to create a series of overlapping nearshore grids for site specific modelling thereby removing potential issues associated with the lateral (cross-shore) wave boundary conditions with the overlapping grids.
The model domain covers the continental shelf margin of New South Wales including portions of the Tasman Sea and Coral Sea in the South Pacific Ocean, extending from deep water (80-200 m) to the coast with variable resolution. The extent of the mesh and the interpolated model bathymetry is shown in Figure B.2. The model domain extends from Moreton Island in Queensland in the north to Gippsland Lakes in Victoria in the south so that the closed lateral boundaries do not affect the NSW coastline.
The flexible mesh consisting of non-overlapping triangular elements permits spatially varying resolution, allowing areas with complex bathymetry and coastlines to be sufficiently resolved. The NSW model consists of approximately 169,300 elements with arc lengths that range from 100 m in the nearshore to 5 km at the open ocean eastern boundary. A smoothed 30 m depth contour is enforced in the model mesh with output points specified at 1 km spacing. The 10 m depth contour was also represented in the model mesh with output points (as discrete mesh nodes) specified at 100 m spacing. A NSW coastline vector from NSW Waterways, supplied by OEH, was used as the basis of the coastline in the model.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix B
The Sydney region model is a subsection of the NSW-wide nearshore model, extending from Lake Macquarie in the north to Port Kembla in the south. It consists of approximately 27,700 elements. This Sydney regional model was used for the initial sensitivity testing and calibration using the Narrabeen and Wamberal measured data sets.
Minimisation of the model run-time was constrained due to the size of small mesh elements required to resolve the model bathymetry between the 10 m contour and the shoreline. An optimisation exercise was undertaken to simplify the mesh and hence reduce run-time while maintaining calibration.
Figure B.2: Extents of the (left) NSW-wide and (right) Sydney regional nearshore wave unstructured meshes
B.3 Computational Performance
For this project, Baird Australia’s in-house computing infrastructure as well as the new OEH high performance computing facility, IRS, were utilised. The IRS provides high-performance compute nodes with up to 20 cores per node. Simulations can be run in Message Passing Interface (MPI) and be able to utilise up to 180 cores across the nine compute nodes.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix B
Version 4.18 of WW-III was implemented in parallel on the NSW OEH’s high performance computing facility IRS utilising an OMP and MPI configurations. The computational demands to develop the WTF ordinates were not excessive, simulations at the IRS were undertaken using Open Multi-Processing (OMP) parallel processing with 40 parallel threads per run. For the four grid setup using the Sydney calibration nearshore grid, a runtime of approximately 0.25 hours per day of simulation was achieved on IRS in OMP mode. In comparison, on Baird Australia’s in-house workstations, a run time of approximately 2.2 hours per day was achieved. When running the four grid WW-III setup with the NSW shelf to coastline grid, a runtime of approximately 3.5 hours per day of simulation was achieved.
The project required the hindcast of the entire 2008 year on the NSW shelf to coastline grid. This was undertaken on the IRS in an MPI configuration utilising 5 nodes, and a total of 200 parallel threads per run. Benchmark testing indicated that in MPI mode across multiple nodes, the MPI efficiency was 53%. As a result, the run time for simulations in MPI mode with 200 threads was approximately one-third of the time compared undertaking the simulation with up to 40 threads in OMP mode. The total run time for the 1-year of simulation on the IRS was approximately 23 days. B.4 Model Forcing B.4.1 Winds
The global WW-III model is forced with the CFSR wind data product (NOAA, 2010) which was found to be the most suitable wind database for the modelling of storms along the NSW coast (Cardno, 2012) due to its high temporal resolution (hourly) that are able define the shorter duration storm events. For this study, the simulations of the global scale model adopted unadjusted CFSR winds and did not include the nearshore wind scaling, as was adopted with Cardno (2012).
B.4.2 Waves
The offshore extent of the NSW-wide and Sydney unstructured coastal WW-III meshes aligns with the archived spectral time series outputs from OEH’s 35-year CFSR hindcast dataset. The NSW shelf scale WW-III model is fully compatible with OEH’s existing 35-year CFSR hindcast data set that has archived directional wave spectra in 80 m to 200 m water depth at 0.05-degree resolution along the coastline. Additionally, its offshore boundary aligns with other available deepwater wave data (e.g., NSW Offshore Buoy network, AUSWAVE and NOAA WW-III).
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix B
B.5 Model Calibration
Table B.2 presents a comparison of the hindcast nearshore wave conditions from the shelf scale grid (Section A.2) of global WW-III model with the measured WRB-R data sets from the Sydney deepwater, Narrabeen and Wamberal WRB locations. The model physics for that initial calibration simulation are consistent with the larger global scale model.
Table B.2: Summary of WW-III model calibration compared to Sydney, Narrabeen and Wamberal WRB-R Data Sets: July to November 2011 Sydney Deepwater WRB
Hm0 Tp θp Tm02 θm
MS 0.90 0.54 0.69 0.75 0.47 Bias -0.16 0.24 -9.25 0.07 -3.22 RMSE 0.44 2.57 44.39 1.04 58.24 SI 0.24 0.29 0.32 0.16 0.46 R 0.85 0.37 0.54 0.61 0.32 ss 0.89 0.98 0.88 1.00 0.85 Narrabeen WRB
Hm0 Tp θp Tm02 θm MS 0.93 0.63 0.79 0.86 0.90 Bias -0.02 0.29 1.22 0.26 -0.15 RMSE 0.29 2.64 12.31 1.05 8.78 SI 0.25 0.28 0.13 0.15 0.09 R 0.89 0.47 0.66 0.79 0.83 ss 1.00 0.98 1.00 1.04 1.00 Wamberal WRB
Hm0 Tp θp Tm02 θm MS 0.93 0.57 0.70 0.81 0.87 Bias -0.15 0.13 -0.86 -0.01 -2.78 RMSE 0.30 2.90 19.38 1.03 11.96 SI 0.21 0.31 0.16 0.16 0.10 R 0.89 0.40 0.55 0.72 0.79 ss 0.89 0.97 0.98 1.00 0.98
B.6 Model Sensitivity Testing
Although the WW-III model performed well based on the model results for a three month calibration period from August to November 2011 at the three WRB sites, a suite of sensitivity
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix B scenarios were modelled to understand the sensitivity of the model, and confirm that the initial model setup was suitable. The sensitivity simulations involved simulating the calibration period using various model physics and propagation inputs and comparing key model result metrics to the base run (that was described in Section B.2) to determine the most reliable model setup. The calibration process was performed on both a global scale, which focussed on wave model physics, triad interactions, shallow water limiter and the directional bin widths used in the wave propagation calculations as well as a shelf model scale, which focussed on mesh setup and time steps. A summary of the model sensitivity simulations are presented in Table B.3.
Results from the various sensitivity runs were quantitatively compared with range of statistical parameters including: Model Skill (MS), Root Mean Square Error (RMSE), Bias, Scatter Index (SI), Pearson’s correlation coefficient (R) and Symmetric Slope (ss). All sensitivity cases were compared to the base model. The statistics have been previously supplied and discussed with OEH in a separate spreadsheet.
Table B.3: Summary of sensitivity simulations undertaken for the WW-III model Scale of Run WaveWatchIII Sensitivity Parameter Description ID Switch Test
C1 Global Base Run as described in Section A.2 WAM4 and variants source term C3 Global Wave Physics ST3 package Ardhuin et al. (2010) source term C4 Global Wave Physics ST4 with STAB3 package C5 Global Wave Physics ST6 BYDRZ source term package Lumped Triad Interactions (LTA) C9 Global Triad Interactions TR1 method Shallow Water C11 Global MLIM (OFF) Miche-style (OFF) Limiter C12 Global Directional Bins 10 Degrees Shelf scale model with no wind C2 Sydney Mesh Wind forcing C6 Sydney Mesh Time step 0.5 minutes C7 Sydney Mesh Time step 2 minutes Derefined Grid – 250 m nearshore C8 Sydney Mesh Spatial Resolution mesh elements
The BYDRZ source term package (ST6) showed the worst results, consistently showing lower model skill at Sydney, Wamberal and Narrabeen for significant wave height, period and peak directions.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix B
Meanwhile, the WAM4 and variants source term package (ST3), and the Ardhuin et al. (2010) source term package (ST4) showed comparable model skill for significant wave height to the Tolman and Chalikov (1996) source term package (ST2), but showed lower model skill and larger degree of bias for peak and mean periods. Including triad interactions showed very little impact on model results from the base model which did not include triad interactions. Similarly, turning off the shallow water limiter had very little impact on results from the base model which included a shallow water limiter. Refining the spectra directional bin widths to 10 ° from 15 ° also had little impact on model results.
On the Sydney shelf to coastline scale model, the grid was then tested excluding no wind forcing on the mesh grid (that is, with boundary wave conditions only). This was important to ensure that wind wave growth over the shelf scale model from the CFSR winds was minimal and hence allow the transfer matrix set of simulations to focus on boundary condition wave propagation in the absence of local wind forcing. The sensitivity case with no wind applied to the shelf scale model showed that applying wind forcing on the shelf scale model was very minor (Table B.4).
The Sydney calibration model was also tested for sensitivity to time step duration and spatial grid resolution. The model time step was tested with a halving of time step from 1 minute to 0.5 minute and then doubling to 2 minutes. Both model results showed that the model results were insensitive to these time step changes. A de-refined mesh model with a resolution of approximately 250 m along the 10 m depth contour was also tested and showed that model results at Sydney, Wamberal and Narrabeen were very comparable to results using the adopted mesh with 100 m nearshore resolution. B.7 Benchmarking against previous WW-III studies
Benchmarking between the previous WW-III v3.14 model and the current WW-III v4.18 model was undertaken for a two month period between June and July 2007. Model outputs from the NSW grid showed very comparable results at the deep water buoy locations. Figure A.3 shows a comparison of resulting time series for the Sydney deepwater WRB buoy. There are only minor differences in the two modelled wave data sets, which is to be expected as the wave physics scheme for both simulations is consistent. This outcome provides confidence that adopting the ST2 (Tolman) physics scheme with v4.18 will provide consistent wave data sets as OEH’s existing WW-III v 3.14 model at the deepwater WRBs.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix B
Figure B.3: Comparison of measured WRB data at Long Reef, and WW-III hindcast simulations with v3.14 (red line) and v4.18 (blue line)
B.8 Discussion and Summary
Baird Australia’s global to shelf scale WW-III v4.18 spectral wave model for the NSW coastline includes shallow-water physics to allow wave transformation to be modelled from deepwater up to the zone of wave breaking. Previous numerical wave modelling of nearshore wave conditions at sites along the NSW coastline have typically adopted the SWAN wave model. Whilst SWAN is an accurate and proven nearshore wave model, other studies including Cardno (2012) and Mortlock and Goodwin (2013) have identified limitations with SWAN when applied to shelf scale wave simulations along the NSW coastline. One of the key issues with SWAN identified in Cardno (2012) was a tendency for the modelled wave climate at the deepwater buoys and at nearshore locations along the NSW coast to be biased to the high frequency end of the wave energy spectrum compared to the measured WaveRider buoy data sets. Compared to the results of hindcast modelling using the SWAN model for the Wamberal data set presented in Mortlock and Goodwin (2013), the results from the WW-III model developed in this study show improved validation of the hindcast nearshore wave climate when the model is forced with global WW-III modelled spectra in deepwater. The WW-III modelled nearshore wave data in this study demonstrate negligible bias
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix B
for mean wave period (Tm02). As a comparison, the nearshore wave hindcast from SWAN models forced with global WW-III deepwater spectra for Wamberal and Newcastle presented in Cardno
(2012) and Mortlock and Goodwin (2013) respectively, have a consistent mean period (Tm02) under-bias of 1 to 2 s at the near shore WRB locations.
In summary, the results show that the WW-III model adopting the standard (ST2) physics applied in earlier OEH projects (Cardno, 2012 and Baird Australia, 2014a) performs well for the calibration period and is suitable as the basis for the WTF transformation simulations to be conducted for this project. This approach will also ensure consistency with the earlier hindcast modelling undertaken with a consistent physics description. Table B.1 shows the calibrated model setup determined at the completion of the sensitivity analysis.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix B
Appendix C Bathymetry Datasets
Table C.1: Summary of nearshore bathymetry datasets used in the construction of the NSW wave model nearshore unstructured mesh Year of Location Data Type Notes Survey Culburra Beach OEH single beam 1993 Mollymook Offshore OEH single beam 1991 Batemans Bay OEH single beam 1986 Sydney Offshore 2008 Base dataset underlying OEH single beam other sets Thirroul OEH single beam 2011 Narrabeen 2012 Much greater extent OEH single beam than 2014 dataset Manly OEH single beam 2012 Wamberal OEH single beam 2013 Mona Vale OEH single beam 2014 Bigola OEH single beam 2014 Bogin OEH single beam 2014 Dee Why OEH single beam 2014 Bate Bay OEH single beam 2011 Broken Bay OEH single beam 1978 Palm Beach OEH single beam 1999 Port Hacking OEH single beam 2012 Sawtell OEH single beam 2011 Angourie Point OEH single beam 1995 Byron Offshore OEH single beam 2002 Dixon Park OEH single beam 2013 Nobbys Beach OEH single beam 2007 Old Bar Offshore OEH single beam 2009 Stockdon Beach OEH single beam 2007 Overlaps LADS at south end Port Macquarie Offshore OEH single beam 1991 Wooli Offshore OEH single beam 2007 Woody Head OEH single beam 1994 Coffs Harbour OEH single beam 2003
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix C
Year of Location Data Type Notes Survey Nelson Bay Boat Harbour OEH single beam 8/08/20 02 Port Stephens OEH single beam 29/06/1 Large extent 994 Port Stephens OEH single beam 2012 Overlies 2007 PS datasets Tanilba Bay OEH single beam 2011 Port Stephens OEH single beam 2007 More extensive than 2012 survey Seven Mile Beach OEH multi beam 2013 SEVENMILE Seven Mile Beach OEH multi beam 2014 SEVENMILE Potato Point OEH multi beam 2011 POTATOP Murramurang OEH multi beam 2010 MURRAMB Mullumb OEH multi beam 2011 MULLIMB Montague Island OEH multi beam 2011 MONTGUE Durras OEH multi beam 2014 DURRAS Eden OEH multi beam 2009 EDEN Culburra OEH multi beam 2013 CULBBCH Kingshead point OEH multi beam 2013 KINGHPT Jervis Bay Marine Park Area OEH multi beam 2007 JBMPAREA Brush Island OEH multi beam 2011 BRSHISL Kurnell OEH multi beam 2013 KURNELL Kurnell OEH multi beam 2013 DESAL Avoca OEH multi beam 2008 AVOCBCH Bundeena OEH multi beam 2013 JIBBONB Northern Beaches - North OEH multi beam 2014 NNBEACH Port Hacking OEH multi beam 2012 OFFSHPT Port Hacking OEH multi beam 2013 PHACRONU Potters Point OEH multi beam 2013 POTTERP Northern Beaches - South OEH multi beam 2014 SNBEACH Port Hacking - Entrance OEH multi beam 2012 PTHACKENT Bate Bay OEH multi beam 2015 PTHACKBB South Head off Sydney OEH multi beam 2014 STHHEAD Harbour Crescent Head OEH multi beam 2008 CRSNTHD Hawkes Nest OEH multi beam 2012 HAWKES Port Stephens Great Lakes OEH multi beam 2012 PSGLMPN Marine Park North Port Stephens Great Lakes OEH multi beam 2012 PSGLMPS Marine Park South
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix C
Year of Location Data Type Notes Survey Nature and Environment OEH multi beam 2013 NERP (Off Coffs Restoration Project Harbour) Laurieton OEH multi beam 2008 LAURTON (overlaps with single beam (north) and LADS (south)) Nambucca OEH multi beam 2008 NAMBUC Swansea OEH multi beam 2008 NEWCSTLOAR Sawtell OEH multi beam 2011 SAWTELL ((overlaps with single beam) Seal Rocks OEH multi beam 2012 SEALRKS (overlaps with PSGLMPN) Solitary Islands Marine Park - OEH multi beam 2006 SIMPALL All Central Coast 2011 Prioritised over 2008 OEH LADS where overlapping Central Coast OEH LADS 2008 Port Stephens OEH LADS 2011 Old Bar OEH LADS 2013 Lake Cathie OEH LADS 2013 Tweed OEH LADS 2011
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix C
Appendix D Measured Wave Data Spectral Analysis Methods
All estimation methods below aim to determine the directional spectra that best matched the co- spectra and quad-spectra which are determined from a Fourier decomposition of each of the three heave records. For each frequency (or wave period), six Fourier components are determined (real and imaginary for each heave):
= + north heave
𝑛𝑛𝑛𝑛 𝑛𝑛𝑛𝑛 𝑛𝑛𝑛𝑛 𝐴𝐴 = 𝛼𝛼 +𝑖𝑖𝛽𝛽 west heave
𝑤𝑤𝑛𝑛 𝑤𝑤𝑛𝑛 𝑤𝑤𝑛𝑛 𝐴𝐴 = 𝛼𝛼 + 𝑖𝑖𝛽𝛽 vertical heave
𝑣𝑣𝑛𝑛 𝑣𝑣𝑛𝑛 𝑣𝑣𝑛𝑛 The co- and quad-spectra are𝐴𝐴 then formed,𝛼𝛼 𝑖𝑖𝛽𝛽 for example:
= . = +
𝑛𝑛𝑤𝑤 𝑛𝑛𝑛𝑛 𝑤𝑤𝑛𝑛 𝑛𝑛𝑛𝑛 𝑤𝑤𝑛𝑛 𝑛𝑛𝑛𝑛 𝑤𝑤𝑛𝑛 𝐶𝐶 = 𝐴𝐴�����×�𝐴𝐴���� = 𝛼𝛼 𝛼𝛼 𝛽𝛽 𝛽𝛽
𝑄𝑄𝑣𝑣𝑛𝑛 �𝐴𝐴��𝑣𝑣𝑛𝑛�� 𝐴𝐴���𝑛𝑛𝑛𝑛�� 𝛼𝛼𝑛𝑛𝑛𝑛𝛼𝛼𝑤𝑤𝑛𝑛 − 𝛽𝛽𝑣𝑣𝑛𝑛𝛽𝛽𝑛𝑛𝑛𝑛 D.1 Fourier Method
The on-buoy analysis calculates the first four Fourier components of the directional distribution D(θ,f):
1 1 ( , ) = + cos + + cos 2 + cos 2 2 𝐷𝐷 𝜃𝜃 𝑓𝑓 � 𝑎𝑎1 𝜃𝜃 𝑏𝑏1𝑐𝑐𝑐𝑐𝑐𝑐𝜃𝜃 𝑎𝑎2 𝜃𝜃 𝑏𝑏2 𝜃𝜃� Where a1, b1, a2 and b2 are calculated𝜋𝜋 directly from the co- and quad-spectra.
A complete description of the on-buoy analysis is provided in the Datawell Waverider Reference Manual (2013) and Longuet-Higgins (1963).
This method is limited by the on-buoy analysis to only the first four Fourier coefficients. This gives rise to the following problems with the resulting reconstruction:
a) Narrow spreading is not well represented, as the few Fourier coefficients have a theoretical limit to the spreading that they are capable of representing. This is generally problematic
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix D
for swells that have been generated from a single source far away, which are generally long period waves. This affects the development of the matrix by reducing energy at the central peak, and distributing it to directions either side of the peak; b) negative energy is present at some directions due to the limited ability of the Fourier coefficients to fit to the directional spectrum. The negative energy is clearly not a physical phenomenon, but must remain in the reconstruction to ensure the total energy is maintained; and c) bimodal spectra with the same frequency and different directions cannot be represented in the reconstruction, and results a direction somewhere between the two swell trains. This is generally a rare occurrence. D.2 Iterative Maximum Likelihood Method (IMLM)
The Maximum Likelihood Method method assumes that an estimate of the Directional Spreading function may be expressed as a linear combination of the cross-spectra (which are determined by the co- and quad-spectra):
1 ( , ) = ( , ). ( ) ( ) , 𝐷𝐷𝑀𝑀𝑀𝑀𝑀𝑀 𝑓𝑓 𝜃𝜃 � 𝛼𝛼𝑚𝑚𝑛𝑛 𝑓𝑓 𝜃𝜃 𝐺𝐺𝑚𝑚𝑛𝑛 𝑓𝑓 𝐸𝐸 𝑓𝑓 𝑚𝑚 𝑛𝑛 Where Gmn is the cross-spectra, and E(f) is the single sided energy spectrum.
The best estimate for the directional spreading function was determined to be:
( , ) = ( , ). . ( , ) 𝐾𝐾 𝐷𝐷𝑀𝑀𝑀𝑀𝑀𝑀 𝑓𝑓 𝜃𝜃 −1 ∗ Where H(f,θ) is the transfer function between∑𝑚𝑚𝑛𝑛 the𝐻𝐻𝑚𝑚 surface𝑓𝑓 𝜃𝜃 𝐺𝐺elevation𝑚𝑚𝑛𝑛 𝐻𝐻𝑛𝑛 𝑓𝑓signal𝜃𝜃 and another wave signal (heave, pressure, slope, etc)
The IMLM extends this method iteratively reduces the difference between the MLM solution and the cross-spectra computed from the wave signals.
For a detailed description of the IMLM see Pawka (1983) and Benoit (1997)
Benoit (1997) argued that the IMLM method is generally accurate in estimation of the direction peak, but can overestimate the peak when trying to match the measured co- and cross- spectra. This leads to a narrower directional spreading than is physically present.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix D
D.3 Extended Maximum Entropy Method (EMEP)
The EMEP aims to maximise the entropy H defined as
= ( ) ln ( ) 2𝜋𝜋
𝐻𝐻 �0 𝐷𝐷 𝜃𝜃 �𝐷𝐷 𝜃𝜃 � 𝑑𝑑𝜃𝜃 Where D is the directional spreading function:
1 ( , ) = 𝐾𝐾 [ . cos( . ) + sin ( . )]
𝐷𝐷 𝑓𝑓 𝜃𝜃 𝑒𝑒𝑒𝑒𝑒𝑒 �� 𝐴𝐴𝑘𝑘 𝑘𝑘 𝜃𝜃 𝐵𝐵𝑘𝑘 𝑘𝑘 𝜃𝜃 � ∆ 𝑘𝑘=1 Where Δ normalises the energy at each frequency. Ak and Bk are unknown parameters which are to be determined iteratively. For further description of the EMEP method see Benoit (1997) and Hashimoto (1993).
Figures D.1 to D.3 compare the three methods for the nearshore site of Wamberal for the 2011 calibration period.
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix D
Figure D.1: Comparison of Hm0 for Wamberal for the three methods and the single Hm0 value reported by the buoy. IMLM and EMEP values are almost perfectly overplotted
Wamberal Hm0 4 Datawell Parametric Datawell Reconst. IMLM EMEP 3.5
3
2.5
2 Hm0 (m) Hm0
1.5
1
0.5
0 Nov Dec
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix D
Figure D.2: Comparison of Tp for Wamberal for the three methods and the single Tp value reported by the buoy. IMLM and EMEP values are almost perfectly overplotted.
Wamberal Tp 20 Datawell Parametric Datawell Reconst. IMLM 18 EMEP
16
14
12
10 Tp (s)
8
6
4
2
0 Nov Dec
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix D
Figure D.3: Comparison of Direction for Wamberal for the three methods. Datawell Parametric is based on the directional analysis reported by the receiver. Wamberal Direction
Datawell Parametric Datawell Reconst. 180 IMLM EMEP
160
140
120
100 Direction (deg)
80
60
40
20 Nov Dec
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix D
Appendix E Quantitative Validation Metrics Description
The model skill score (equation E.1) provides a scaled representation of the predictive accuracy of the model against the measurements. This metric gives a value of 0 where there is no agreement and 1 for perfect agreement between the modelled and measured data (Willmott et al, 1985).
[ ] = 1 (equation E.1) ([ 𝑁𝑁 ] [ 2 ]) ∑𝑖𝑖=1 𝑀𝑀𝑖𝑖−𝑂𝑂𝑖𝑖 𝑁𝑁 2 𝑀𝑀𝑐𝑐𝑑𝑑𝑒𝑒𝑑𝑑𝑑𝑑𝑘𝑘𝑖𝑖𝑑𝑑𝑑𝑑 − ∑𝑖𝑖=1 𝑀𝑀𝑖𝑖−𝑂𝑂�𝑖𝑖 + 𝑂𝑂𝑖𝑖−𝑂𝑂�𝑖𝑖 where Oi = observed, or measured data (m for wave height), Mi = modelled data (m for wave height), and = mean of observed data (m for wave height).
𝑖𝑖 The Root Mean Squared (RMS) error𝑂𝑂� (equation E.2) provides a measure of the magnitude of the difference between the modelled and measured values.
= [ ] (equation E.2) 1 𝑁𝑁 2 𝑅𝑅𝑀𝑀𝑑𝑑 �𝑁𝑁 ∑𝑖𝑖=1 𝑀𝑀𝑖𝑖 − 𝑂𝑂𝑖𝑖 where N is the number of samples in the dataset.
Bias measures the tendency of the model to over-/under- estimate the observed value (equation E.3). A bias value of zero indicates an unbiased model. A positive bias indicates that the model is persistently overestimating the observed, and conversely a negative bias indicates that the model persistently underestimates the observed.
= ( ) (equation E.3) 1 𝑁𝑁 𝑁𝑁 ∑𝑖𝑖=1 𝑖𝑖 𝑖𝑖 The scatter index𝐵𝐵𝑖𝑖𝑎𝑎𝑐𝑐 provides an𝑀𝑀 indication− 𝑂𝑂 of the spread of the data around its mean. It is calculated as the RMW error normalised by the mean of the observations (equation E.4), giving the proportion of RMW difference with respect to the mean observation.
([ ] [ ]) = 1 𝑁𝑁 2 (equation E.4) �𝑁𝑁 ∑𝑖𝑖=1 𝑀𝑀𝑖𝑖−𝑀𝑀� − 𝑂𝑂𝑖𝑖−𝑂𝑂� 𝑂𝑂� w𝑑𝑑𝑆𝑆here = mean of modelled data (m for wave height).
𝑀𝑀�𝑖𝑖
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix E
Pearson’s correlation coefficient (R) is a measure of the strength of the linear relationship between two variables, in this case modelled and observed data (equation E.5). The value of R range from - 1 to 1, where 0 indicates no linear relationship, 1 indicates a perfect positive linear relationship, and -1 indicates a perfect negative linear relationship.
[ ][ ] = 𝑁𝑁 (equation E.5) ∑𝑖𝑖=1[ 𝑀𝑀𝑖𝑖−𝑀𝑀] � 𝑂𝑂𝑖𝑖−𝑂𝑂[ � ] 𝑁𝑁 2 𝑁𝑁 2 𝑟𝑟 �∑𝑖𝑖=1 𝑀𝑀𝑖𝑖−𝑀𝑀� ∑𝑖𝑖=1 𝑂𝑂𝑖𝑖−𝑂𝑂� Symmetrical slope (ss) calculates the linear best fit between the measured and the observed through the origin (equation E.6) and gives an indication of the strength of the relationship between the measured and observed.
[ ] = (equation E.6) 𝑁𝑁 [ ] ∑𝑖𝑖=1 𝑀𝑀𝑖𝑖𝑂𝑂𝑖𝑖 𝑁𝑁 2 𝑐𝑐𝑐𝑐 ∑𝑖𝑖=1 𝑂𝑂𝑖𝑖
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix E
Appendix F User Manual for the NSW Nearshore Wave Transformation Toolbox
NSW Coastal Wave Model 12359.101.R2.Rev0 State Wide Nearshore Wave Transformation Tool Appendix F
User Manual
Issue Date: January 2017
Toolbox Version: 1.2
Contents Introduction ...... 3 Transformation Methods ...... 4 Offshore Wave Forcing ...... 4 Quality Flags ...... 5 Toolbox Access ...... 5 Spatial Transform ...... 6 Spatial Transformation Example ...... 7 Timeseries Transform ...... 9 Timeseries Transform Example ...... 10 Spectral Transform ...... 12 Spectral Transform Example ...... 13 Data Extraction ...... 15 Data Extraction Example ...... 16 Extreme Waves ...... 18 Extreme Waves Example ...... 20 Data Availability ...... 21 Waverider buoy data (measured) ...... 21 NSW Wave Hindcast data (modelled) ...... 21 Troubleshooting ...... 22 References ...... 22
Document Version 1.2 January 2017 Page 2 of 22
Introduction
This user manual describes the functionality of a web-based tool, referred to as the New South Wales Wave Transformation Toolbox (the Toolbox), developed by Baird Australia and the Manly Hydraulics Laboratory as part of the NSW Coastal Wave Model project, funded by the Office of Environment & Heritage.
The Toolbox enables the user to rapidly transform offshore wave conditions to the 10m and 30m depth contours along the entire NSW coastline, and visualise or extract the transformed nearshore wave conditions.
Nearshore wave conditions can be derived from either deep-water historical Waverider buoy measurements or modelled hindcast data using one of two available transformation methods; Parametric and Spectral.
The toolbox consists of a number of pages each with unique functionality, as follows:
• Spatial Transform: provides a spatial map of transformed nearshore wave conditions for a selected day and time. • Timeseries Transform: provides a timeseries plot of transformed wave conditions for a selected time period at a selected nearshore location. • Spectral Transform: provides the directional spectrum of a transformed wave condition for a selected day and time at a selected nearshore location. • Data Extraction: allows the user to utilise the full capability of the toolbox and extract continuous transformed wave data across multiple nearshore locations in a range of common data formats. • Extreme Waves: provides a spatial map of pre-computed simulated nearshore extreme wave conditions derived from deep-water return period wave estimates.
Document Version 1.2 January 2017 Page 3 of 22
Transformation Methods
Underpinning the Toolbox are two Wave Transformation algorithms; a Parametric Wave Transfer Function and a Spectral Wave Transfer Function. Full details of the development and validation of the wave transformation algorithms are presented in Baird Australia (2016) and Taylor et. al. (2015).
Parametric Transformation
The Parametric Wave Transfer Function utilises offshore data, characterised as integrated parameters of the total spectrum, that is, a single wave height (Hs), wave period (Tp) and direction
(θp). An underlying transformation matrix of spectral wave simulations covering the full wave period and direction range is then used to transform the offshore wave energy to inshore locations and adjust the wave period and direction.
The advantage of the parametric transformation is that it is readily applied to the entire available offshore wave data record and provides rapid transformation for long duration datasets.
Spectral Transformation
The Spectral Wave Transfer Function is more comprehensive and computationally intensive, requiring the full offshore directional spectrum. Each energy ordinate of the offshore spectrum is transferred based on a matrix of spectral wave simulations and then combined for a full reconstruction of the inshore 2D directional energy spectrum.
The Spectral Transformation was found to provide better validation metrics when compared to inshore wave measurements.
Offshore Wave Forcing
The Toolbox provides access to measured and modelled historical offshore wave conditions along the NSW coast to drive wave transformations, as well as options to manually enter user-specified conditions or select from pre-computed extreme value offshore wave conditions.
Further details on the availability of each data type can be found at the end of this document.
Waverider buoy data
Measured wave data from the NSW deep-water Waverider buoy network operated by Manly Hydraulics Laboratory is available for the duration of directional instrument records.
NSW Wave Hindcast data
The Office of Environment and Heritage has also developed a wave hindcast dataset for NSW using the WAVEWATCH III® spectral wave model and Climate Forecast System (CFS) data, which provides continuous directional offshore wave data for all of NSW from 1979 onwards.
Document Version 1.2 January 2017 Page 4 of 22
Quality Flags
To provide an indication of the accuracy of wave transformations at each nearshore output location, two quality flags have been applied.
Bathymetry Flags
Multiple bathymetry datasets were sourced to develop the transformation algorithms, from a range of federal, state and local government agencies. A summary of the bathymetric data applied in the model along with the resolution and area that this data covers is provided in Baird Australia (2016). The accuracy of the wave transformation algorithms is in part dependent on the quality of bathymetry applied to the underlying spectral wave model. Hence, a bathymetry flag was applied each inshore location based on the quality of the bathymetry data covering that site. Higher quality datasets (such as multibeam survey) result in a higher ranked bathymetry flag.
Validation Flags
Due to the limited availability of nearshore wave measurements, a quality rating system of inshore sites was developed to identify the sites within the Toolbox where the transformation algorithms do not provide a close approximation of the calibrated WW-III spectral wave model. A detailed summary of the quality rating system is provided in Baird Australia (2016).
Note that the highest quality rating is reserved for locations where calibration to actual nearshore measured data has been completed. The Office of Environment and Heritage has a collaborative program of nearshore wave data collection at NSW beaches to support the validation and further development of the Nearshore Wave Transformation Toolbox.
Toolbox Access
The NSW Nearshore Wave Transformation Toolbox is managed by the Office of Environment and Heritage. Registered access to the toolbox is available: http://www.environment.nsw.gov.au/research/ocean-and-coastal-waves.htm
The NSW Nearshore Wave Forecast is an operational version of the tools that provides a rolling 9- day window of real-time and forecast nearshore wave conditions along the NSW coast using the parametric transform method. Public access to the forecast tool is available: http://forecast.waves.nsw.gov.au
Document Version 1.2 January 2017 Page 5 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
Spatial Transform
The Spatial Transformation allows the user to visualise the alongshore variation in nearshore wave conditions for a given date and time within a nearshore Area of Interest. The user is able to select the following:
(1) Timestep (Date and Time); (2) Transformation Method (Method); and (3) Offshore Wave Data Source (Source).
Nearshore wave conditions will be presented as coloured vectors at all nearshore locations within the Google Maps-based display on the right of the page.
Before querying the toolbox, the user should adjust the map view to the site/area of interest, as the visible map extent determines the extent and density of nearshore locations to calculate.
To move the map, the user can click and hold the mouse and then drag the map. The keyboard arrow keys can also be used to move the map north, south, east, and west. To zoom in or out, go to the bottom right and use the zoom in and zoom out buttons. The user can also double click to zoom in, or use your mouse scroll or trackpad to zoom in and out.
The Timestep is selected by using the calendar selection tool. This tool appears when the user clicks on the calendar icon next to the Date input box (1).
The Transformation Method (Parametric or Spectral) is selected by clicking on the drop-down menu in the Method input box (2).
Three Offshore Wave Data Source options are available on the Spatial Transform page:
• offshore measured Waverider buoy data (Waverider) • offshore modelled Hindcast wave data (WW3) • user defined offshore wave conditions (User Input) – parametric transform only.
The Data Source is selected by clicking on the drop-down menu in the Source input box (3).
If User Input is selected, the parametric transform method must be selected, and the user is
required to enter values for the offshore wave condition – wave height (Hs), wave period (Tp) and wave direction (Dir) – in the User Input boxes (4, 5 and 6).
Further information can be found by clicking the icons next to the input boxes.
Once the required inputs are entered, the toolbox is queried by clicking the “Submit” button (7).
Document Version 1.2 January 2017 Page 6 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
(1)
(2)
(3)
(4)
(5)
(6)
(7)
Spatial Transformation Example
As an example, a Parametric transformation of User Defined conditions is queried from the toolbox with output provided at nearshore sites along the Central Coast.
The user inputs are as follows:
• Method: Parametric • Source: User Input
• User Hs: 2.6 (metres)
• User Tp: 9.5 (seconds) • User Dir: 140 (degrees TN coming from)
Note: The Date/Time inputs are not used and are ignored for User Defined Conditions.
Document Version 1.2 January 2017 Page 7 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
The toolbox then returns coloured vectors at all nearshore locations within the map window at the time of the query.
Wave parameters can be viewed along with the site details (including bathymetry and validation flags) by clicking on a nearshore vector.
Document Version 1.2 January 2017 Page 8 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
Timeseries Transform
The Timeseries Transformation allows the user to visualise a timeseries of transformed wave parameters at a nearshore Site of Interest. The user is able to select the following:
(1) Site of interest (Location ID), (2) Start Date, (3) End Date, (4) Transformation Method (Method); and (5) Offshore Wave Data Source (Source).
The Site of Interest can be located using the Google Maps-based display on the right of the page. To move the map, the user can click and hold the mouse and then drag the map. The keyboard arrow keys can also be used to move the map north, south, east, and west. To zoom in or out, go to the bottom right and use the zoom in and zoom out buttons. The user can also double click to zoom in, or use your mouse scroll or trackpad to zoom in and out.
Once the output node of interest is located, the Location ID is entered into the input box (1).
Start and End Dates are selected using the calendar selection tool. This tool appears when the user clicks on the calendar icon next to the Start and End Date input boxes (2 and 3).
The Transformation Method and Data Source is selected by clicking on the drop-down menus in the Start and End Date Method and Source input boxes (4 and 5).
Further information can be found by clicking the icons next to the input boxes.
(1)
(2)
(3)
(4)
(5)
(6)
Once the required inputs are entered, the toolbox is queried by clicking the “Submit” button (6).
Document Version 1.2 January 2017 Page 9 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
Timeseries Transform Example
As an example, a Parametric transformation of 1 month of offshore Waverider buoy data (January 2014) is queried from the toolbox with timeseries output at a nearshore site along Stockton Beach (north of Newcastle).
The user inputs are as follows:
• Location ID: 1006377 • Start Date: 01-Jan-2014 • End Date: 01-Feb-2014 • Method: Parametric • Source: Wave Rider
Tip: the Location ID for the Site of Interest can be found by navigating to the location on the map display and clicking the nearshore location. The Location ID can then be selected by clicking the link in the Site ID balloon which will update the Location ID input box.
Document Version 1.2 January 2017 Page 10 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
The toolbox then returns timeseries plots of Significant Wave Height (Hs), Peak Wave Period (Tp) and Wave Direction (Dir). Exact values can be viewed by scrolling over the plots with the mouse.
Tip: To save a copy of the timeseries plot use the “Print Screen” button and paste image into a document format of choice.
Document Version 1.2 January 2017 Page 11 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
Spectral Transform
The Spectral Transformation page allows the user to visualise a transformed directional spectrum (2D wave spectrum) for a selected timestep at a nearshore Site of Interest. The user is able to select the following:
(1) Site of interest (Location ID); and (2) Transformation Timestep.
The Site of Interest can be located using the Google Maps-based display on the right of the page. To move the map, the user can click and hold the mouse and then drag the map. The keyboard arrow keys can also be used to move the map north, south, east, and west. To zoom in or out, go to the bottom right and use the zoom in and zoom out buttons. The user can also double click to zoom in, or use your mouse scroll or trackpad to zoom in and out.
Once located the Location ID is entered into the input box (1).
The transformation timestep is selected using the calendar selection tool. This tool appears when the user clicks on the calendar icon next to the Date input box (2).
The current version of the toolbox completes a Spectral Transformation of offshore Hindcast Wave Data (WW3) only, hence transformation method and offshore data source are not input options.
Further information can be found by clicking the icons next to the input boxes.
(1)
(2)
(3)
Once the required inputs are entered, the toolbox is queried by clicking the “Submit” button (3).
Document Version 1.2 January 2017 Page 12 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
Spectral Transform Example
As an example, a Spectral transformation of deep-water wave hindcast data at 3 pm on the 25th May 2008 is queried from the toolbox with a directional spectrum output at a nearshore site near Mossy Point (south of Batemans Bay).
The user inputs are as follows:
• Location ID: 1011046 • Date: 25-May-2008
Tip: the Location ID for the Site of Interest can be found by navigating to the location on the map display and clicking the nearshore location. The Location ID can then be selected by clicking the link in the Site ID balloon which will update the Location ID input box.
Document Version 1.2 January 2017 Page 13 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
The toolbox then returns a plot of the directional spectrum and associated integrated wave
parameters (Hs, Tp, Dir) for the selected timestep.
Tip: The spectral plot can be saved as an image by right clicking on the plot and selecting “Save image as….” from the drop down list.
Document Version 1.2 January 2017 Page 14 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
Data Extraction
The Data Extraction page allows the user to query the toolbox and download transformed wave conditions, in a variety of data formats, at nearshore Sites of Interest. The user is able to select the following:
(1) First Site of Interest (Start Site); (2) Last Site of Interest (Final Site); (3) Start Date; (4) End Date; (5) Transformation Method (Method); (6) Offshore Wave Data Source (Source); (7) The Extracted Data filename (Filename); and (8) The Extracted Data Format (Format).
The Site/s of Interest can be located using the Google Maps-based display on the right of the page. To move the map, the user can click and hold the mouse and then drag the map. The keyboard arrow keys can also be used to move the map north, south, east, and west. To zoom in or out, go to the bottom right and use the zoom in and zoom out buttons. The user can also double click to zoom in, or use your mouse scroll or trackpad to zoom in and out.
Once located either a single or multiple sites can be requested by entering the Start Location ID (1) and Final Location ID (2). All sites between the Start and Final site will be extracted from the toolbox. To request a single site, the Start and Final sites should both be entered as the same.
Start and End Dates are selected using the calendar selection tool. This tool appears when the user clicks on the calendar icon next to the Start and End Date input boxes (3 and 4).
The Transformation Method and Data Source is selected by clicking on the drop-down menus in the Method and Source input boxes (5 and 6).
The user can enter a unique identifier that will be appended to the extract Data Filename as a prefix (separated with an underscore) using the Filename input box (7). If left blank the toolbox will assign an appropriate name to the extracted data file.
A number of data formats are available for data extraction. The required format is selected by clicking on the drop-down menu in the Format Input Box (8). The following formats can be output from the toolbox:
• csv – comma separated value file. Able to be viewed in a text editor or Microsoft Excel. • mat – Matlab binary file. For loading data directly into Matlab. • tab – SWAN table output format. Able to be viewed in a text editor or loaded in to the OEH Wave Model Toolbox.
Document Version 1.2 January 2017 Page 15 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
• sp1 – SWAN 1D spectral output format. Able to be viewed in a text editor or loaded in to the OEH Wave Model Toolbox. Note this is only available for the spectral transformation method. • netCDF – WW3 2D Spectral output format, an industry standard data archiving format. Note this is only available for the spectral transformation method.
Further information can be found by clicking the icons next to the input boxes.
Once the required inputs are entered, the toolbox is queried by clicking the “Submit” button (9).
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8) (9)
Data Extraction Example
As an example, a Data Extraction from the toolbox is made for a one month period (November 2012) at two nearshore sites (30m depth) near Ballina. A spectral transformation of deep-water offshore hindcast data is requested with output saved to a comma separated value file with the prefix “ExampleExtract”.
The user inputs are as follows:
• Starting Site: 3000081 • Starting Site: 3000082 • Start Date: 01-Jan-2008 • Start Date: 01-Feb-2008 • Method: Parametric • Source: Wave Rider • Filename: “ExampleExtract”
Document Version 1.2 January 2017 Page 16 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
• Format: csv
Tip: the Location ID for the Site of Interest can be found by navigating to the location on the map display and clicking the nearshore location. The Location ID can then be copied and pasted into the Location ID input box.
The toolbox then returns a file named “ExampleExtract_3000081-3000082_2008010104- 2008020100.csv” for download.
Document Version 1.2 January 2017 Page 17 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
Extreme Waves
The Extreme Waves page allows the user to visualise the alongshore variation of transformed extreme waves at nearshore locations within an Area of Interest.
The deep-water extreme wave scenarios have been derived from Shand et al. (2010) based on extreme value analysis of long term measured deep water wave data sets at four sites: Byron Bay, Sydney, Batemans Bay and Eden. Directional extreme wave statistics in three directional sectors (NE 0-90°, ESE 90-135°, SSE 135-225°) were available for Byron Bay, Sydney and Batemans Bay for the 10 years ARI (Hs) scenario, and have been extended to the 5, 20 and 100 years ARI scenarios using proportional relationships.
The ARI conditions available in the toolbox were modelled as discrete scenarios for three offshore wave height return periods (5, 20 & 100 years ARI) and three offshore wave directions (NE, ESE and SSE) with boundary forcings derived from the analysis presented by Shand et al. (2010).
For further details, refer to Baird Australia (2016).
Note: The nearshore results from the Extreme Waves Toolbox page represent the transformation of Offshore Extreme Wave conditions, and therefore only provide a proxy for the equivalent nearshore extreme condition.
The user is able to select the following:
(1) Offshore Average Recurrence Interval Scenario (ARI); and (2) Offshore Direction of ARI Scenario (Dir).
Nearshore extreme wave conditions will be presented as coloured vectors at all nearshore locations within the Google Maps-based display on the right of the page. Before querying the toolbox, the user should adjust the map view to the area of interest, as the visible map extent determines the extent and density of nearshore locations to calculate.
To move the map, the user can click and hold the mouse and then drag the map. The keyboard arrow keys can also be used to move the map north, south, east, and west. To zoom in or out, go to the bottom right and use the zoom in and zoom out buttons. The user can also double click to zoom in, or use your mouse scroll or trackpad to zoom in and out.
Document Version 1.2 January 2017 Page 18 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
The Offshore Average Recurrence Interval (ARI) scenario and Offshore Direction of the ARI Scenario are selected by clicking on the drop-down menus in the ARI (1) and Dir (2) input boxes.
Further information can be found by clicking the icons next to the input boxes.
Once the required inputs are entered, the toolbox is queried by clicking the “Submit” button (3).
(1)
(2)
(3)
Document Version 1.2 January 2017 Page 19 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
Extreme Waves Example
As an example, the nearshore conditions resulting from a 100 year ARI offshore scenario from the SSE sector are queried from the toolbox in the Port Macquarie region.
The user inputs are as follows:
• ARI: 100 years • Dir: SSE
The toolbox then returns coloured vectors at all nearshore locations within the map window at the time of the query.
The wave parameters will be displayed in the Location Values box when a vector is scrolled over with the mouse. Alternatively, the wave parameters can be viewed along with the site details (including bathymetry flags) by clicking the nearshore vector.
Document Version 1.2 January 2017 Page 20 of 22
Spatial Transform Timeseries Transform Spectral Transform Data Extraction Extreme Waves Data Availability
Data Availability
The Data Availability page of the Toolbox provides a summary of the offshore wave data available to complete nearshore wave transformations using the various tools.
Two primary Data Sources are available: the measured offshore Waverider buoy data, and the NSW Wave Hindcast data generated using the WAVEWATCH III® spectral wave model.
Waverider buoy data (measured)
NSW has a long-term measured offshore wave data set that extends back to 1974 (Kulmar et al. 2013) and real-time wave data continues to be collected by directional WRBs at seven locations. However, the directional wave data record, as required by the toolbox, has a reduced temporal coverage and is summarised in the Table below. Note that the measured directional data record
summarised in the table below consists of wave parameters (Hs, Tp, Dir) but not directional spectra. Directional spectra are available for a more limited period that will be progressively updated as spectral analyses are completed.
Length of Available First Directional Site Name Lon. (°E) Lat. (°N) Depth (m) Directional Data Record Date (years) Byron Bay 153.73 -28.82 71 26/10/1999 16.9 Coffs Harbour 153.27 -30.36 72 12/02/2012 4.6 Crowdy Head 152.86 -31.83 79 19/08/2011 5.1 Sydney 151.42 -33.78 85 3/03/1992 24.6 Port Kembla 151.03 -34.48 78 20/06/2012 4.3 Batemans Bay 150.34 -35.71 73 23/02/2011 5.6 Eden 150.19 -37.3 100 16/12/2011 4.8
NSW Wave Hindcast data (modelled)
An offshore wave hindcast has been completed by the Office of Environment and Heritage using the multi-domain WAVEWATCH III® modelling system developed by Cardno (2012). The hindcast dataset was developed using Climate Forecast System (CFS) ocean-atmosphere forcing.
The wave hindcast provides continuous hourly deep-water wave conditions at 5 km resolution along the entire NSW coast from 1979 to present.
With full directional spectral conditions available, both the Parametric and Spectral Transformation methods can be applied using the NSW Wave Hindcast dataset.
Document Version 1.2 January 2017 Page 21 of 22
Troubleshooting
At the completion of each toolbox request, a message will appear to inform the user as to the status of the request, located below the page title. The user should always check the toolbox message before interrogating the toolbox output.
Following a successful request, the message “Transformation Successful” will appear below the page title. Note that if the transformation is only partially successful a warning messages may also appear that described the limitations of the request. Typically warning messages relate to data availability. If the user enters an invalid request an error message will be displayed.
References
Baird Australia (2016). NSW Coastal Wave Model - State Wide Inshore Wave Transformation Tool. Report Prepared for Office of Environment and Heritage. 12359.101.R2.Rev0. April 2016.
Taylor D., Garber S., Burston J., Couriel E., Modra B., Kinsela M. Verification of a Coastal Wave Transfer Function for the New South Wales Coastline. Coasts & Ports Conference, 15 - 18 September 2015, Auckland.
Cardno (2012). “NSW Coastal Waves: Numerical Modelling – Final Report”. Prepared for the NSW Office of Environment and Heritage. Ref LJ2949/R2745. Version 3. September 2012.
Document Version 1.2 January 2017 Page 22 of 22