On the East Australian Current: U pwelling and Separation

Moninya Roughan

151 E 152 E 153 E 154 E 151 E 152 E 153 E 154 E

A t hesis s u bmitted for the degree of D octor of P hilosophy at The University of New South Wales Sydney, Australia December 2001 Do not tempt me to beat my chest and say this is what I have done.

Nelson Mandela

Declaration

I hereby declare that this submission is my own work and to the best of my knowl­ edge it contains no material previously published or written by another person, nor material which to a substantial extent has been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis.

I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.

Monmya Roughan There's a song in the wind and a kiss in the rain, There's a thrill in the sound of the sea, There is joy in the grey of a stormy today, There's a song in the heart of me.

Molly Hall circa 1930 Dedication

For my mum, for her ceaseless love, support and encouragement.

11 Friends help you move. Real friends help you move bodies. The very best friends help you bury them, then forget where.

Anonymous

Acknowledgements

I would like to thank my supervisor Jason Middleton for his encouragement, support and advice. Throughout my PhD Jason has been an inspiration in how to combine work and play.

Thanks go to Greg Nippard, Iain Suthers, Richard Piola, Jocelyn Dela Cruz, David Ghisolfi, Ann Marie Wong, Augy Syahailatua, the Master and crew of the RV Franklin, as well as the shore crew for their assistance, dedication and genuine hard work in the field component of this research.

Thanks to George Cresswell, Stuart Godfrey, and Scott Condie for the insightful discussions during my visits to the CSIRO division of Marine Research in Hobart. Also from the Labs, Kim Badcock for his tireless efforts in providing me with satellite data on many occasions, in many different formats, and °Jeff Dunn for supply of the CARS data. Lixin Qi provided the GRADS graphics in Figure 3.16.

Nathan Bindof and Steve Rintoul must be thanked for helping make my Antarctic dream come true, (even after I chose sub-tropical oceanography as the subject of my research). Both trips were unforgettable for completely different reasons, and anyone who has ever been stuck in the ice on a burning ship, or has flown by helicopter over

iii ACKNOWLEDGEMENTS iv icebergs and glaciers will know what I am talking about!

My colleagues at UNSW Pedro and Mahk have been full of good ideas, and even subjected themselves to the proof reading. And thanks of course to my predecessors in the lab who made an impression in the early days of my career and have now moved on to bigger and better things: Mark, the Witchdoctor, Brad, and the Professor.

My friends and family have been a constant source of encouragement and support, their unfailing love and understanding despite months of neglect, both astounded me and inspired me. Finally Rob 'BC' Brandez must be thanked for his tireless efforts proof reading what to a non-oceanographer, are essentially boring manuscripts! Not to mention the R and R at Bronte and Coogee and the unlimited supply of banana smoothies. Abstract

A multidisciplinary experimental program was undertaken during the 1998 - 1999 Austral summer to investigate upwelling processes on the continental shelf of New South Wales, Australia. Oceanographic time series data were obtained from two arrays of current meters and thermistors moored across the continental shelf at Smoky Cape (30°55'8) and Diamond Head (31 °44'8) for a 2 month period. These shore-normal arrays spanned the point where the East Australian Current (EAC) normally separates from the coast. Two intensive hydrographic surveys were also conducted aboard the RV Franklin during mooring deployment and retrieval. The combination of the time series measurements with the hydrographic data comprise the basis of the observational data set examined in this study.

The observations show that the EAC dominates the physical processes across the narrow continental shelf at Smoky Cape. Analysis of the dynamical balances up­ stream and downstream of the EAC separation point reveals inherent variability in the system. Upstream of the separation point the along-shore pressure gradient drives the flow, and the processes of advection and bottom stress are important, whereas the variability in the across-shore currents and local acceleration tend to dominate the flow downstream. Generally the influence of wind stress on the flow is found to be minimal in the separation region.

V ABSTRACT vi

The EAC is responsible for driving colder nutrient rich water from the continental slope at Smoky Cape into the near surface waters in the coastal region south of Smoky Cape by means of Ekman transport through the bottom boundary layer. Current driven upwelling occurs concurrently with the encroachment of the EAC across the continental shelf. The acceleration of the current increases the bottom friction which allows water to be transported shoreward through the bottom bound­ ary layer for extended periods, resulting in upwelling downstream of Smoky Cape. Current driven upwelling occurs on a more massive scale than that driven by local wind forcing by an order of magnitude.

A configuration of the Princeton Ocean Model is initialised using climatological measurements of temperature and salinity, which is used to investigate the mean current field off the NSW coast. The modelled velocity field closely resembles the observed currents and is used as a basis for a series of Lagrangian particle track­ ing experiments designed to investigate the source and sink of upwelled water and nutrients. Advection of particles both forward and backward in time reveals that upwelled waters originate offshore at depth, well to the north of Smoky Cape. In­ shore, coastal recirculations form and surface waters flowing northward toward the separation point are more often than not entrained into the southward flowing EAC jet. This implies that phytoplankton blooms, which have historically been observed in this region, could have been advected northward inshore towards Smoky Cape or southward along the EAC front. Supporting Publications

Journal Articles: Roughan, M. and Middleton, J.H. 2002: A comparison of current driven upwelling mechanisms: Observations off the coast of NSW Australia. Accepted: Continental Shelf Research

Roughan, M. Oke, P.R. and Middleton, J .H. 2002: A modelling study of the climato­ logical current field and the trajectories of upwelled particles in the East Australian Current. Submitted: J. Physical Oceanography

Conference Proceedings: Roughan, M. and Middleton, J.H. 2002: Current driven upwelling about the separa­ tion point of the East Australian Current. Eos. Trans. AGU, 83(4), Ocean Sciences Meet. Suppl., Abstract OS51E-13, 2002.

Roughan, M. and Middleton, J.H. 2001: Observations and comparisons of upwelling mechanisms on the NSW shelf. In Proceedings, Eighth National Australian Meteo­ rological and Oceanographic Society Conference, Hobart, Australia, 2001.

Roughan, M. and Middleton, J.H. 2000: Nutrient enrichment on the NSW Shelf. In Proceedings, Australian Marine Sciences Association Annual Conference, Sydney,

Vll SUPPORTING PUBLICATIONS viii

Australia, 2000.

Roughan, M. and Middleton, J.H. 1999: Three dimensional structure of the East Australian Current on the NSW shelf: Upwelling and nutrient enrichment. In Pro­ ceedings, Sixth National Australian Meteorological and Oceanographic Society Con­ ference, Canberra, Australia, 1999. Contents

Declaration ii

Dedication iii

Acknowledgements iv

Abstract vi

Supporting Publications viii

1 Introduction 1

1.1 Motivation . 2

1.2 Overview .. 6

1.3 Scope of Study 9

1.4 Objectives ...... 10

IX CONTENTS X

1.5 Outline ...... 11

1.5.1 Chapter 2: The Field Project ...... 11

1.5.2 Chapter 3: Experimental Observations ...... 12

1.5.3 Chapter 4: Variability about the EAC separation ...... 12

1.5.4 Chapter 5: Current Driven Upwelling ...... 12

1.5.5 Chapter 6: A Modelling Study about the EAC Separation 13

1.5.6 Chapter 7: Recapitulation ...... 13

2 The Field Project 14

2.1 The Experimental Program ...... 15

2.1.1 The Hydrographic Survey ...... 15

2.1.2 The Current Meter Array ...... 18

2.1.3 Mooring Construction ...... 21

2.2 Other Data Sources ...... 23

2.2.1 Wind. 23

2.2.2 Sea Level Elevation 25

2.2.3 Atmospheric Pressure at the Sea Surface 26

2.2.4 Satellite Imagery 26

2.2.5 Climatology of Australian Regional Seas 27 CONTENTS xi

2.3 Summary ...... 27

3 Experimental Observations 28

3.1 The Velocity Field ...... 29

3.2 The Temperature Field ...... 33

3.3 The Wind Field ...... 37

3.4 Sea Level Elevation ...... 38

3.4.1 The Inverse Barometer Effect ...... 40

3.5 Variability of the Coastal Currents ...... 42

3.5.1 EAC Events ...... 44

3.5.2 Wind Event ...... 48

3.5.3 Reversal Event ...... 48

3.5.4 Current Event ...... 53

3.5.5 Mean Current Field...... 54

3.5.6 Mean Temperature Field ...... 59

3.6 Summary ...... 59

4 Variability about the EAC separation 62

4.1 Temporal Variability ...... 64

4.1.1 Frequency Analysis ...... 64 CONTENTS XU

4.1.2 Time Series Correlations ...... 69

4.2 Spatial Variability ...... 71

4.2.1 Vertical Structure ...... 73

4.2.2 Horizontal Structure ...... 76

4.3 Depth Averaged Velocity Field ...... 80

4.4 Momentum Balance ...... 81

4.4.1 Depth Averaged Momentum Equations ...... 83

4.4.2 Estimation of terms ...... 83

4.4.3 Uncertainties ...... 86

4.4.4 Balance of Terms 1: Temporal Analysis . 88

4.4.5 Balance of Terms 2: Statistical Analysis . 100

4.4.6 Discussion . 112

4.5 Summary .... . 115

5 Current Driven U pwelling 117

5.1 Nutrient Enrichment .. . 119

5.2 Wind Driven Upwelling . 120

5.2.1 Observations of wind driven upwelling . 122

5.3 Current Driven Upwelling ...... 127 CONTENTS xiii

5.3.1 EAC Encroachment, Pre-conditioning and Upwelling . 130

5.4 Comparison of the Wind and Current events . 138

5.4.1 Vertical Uplift ... . 140

5.5 Topographic Acceleration . 144

5.5.1 Hydrographic evidence of encroachment driven upwelling . 152

5.6 Western Boundary Current Separation and Enrichment . 157

5.6.1 Separation of the EAC . . . 158

5.7 The Magnitude of the Upwelling. . 160

5.8 Discussion . 163

5.9 Summary . 166

6 A Modelling Study about the EAC Separation. 168

6.1 Model Configuration .. . 170

6.1.1 Case 1: Ramping . 174

6.1.2 Case 2: Nudging • • • • o o o o O O I O o o , 0 0 o o o o o o o . 175

6.1.3 Case 3: Forcing . 175

6.1.4 Comparison .. . 177

6.2 Analysis of the Climatological Velocity Field . 183

6.3 Density Surfaces and Chlorophyll-a Concentrations . 190 CONTENTS XIV

6.4 Flow Dynamics ...... 193

6.5 Lagrangian Particle Tracking Experiments . 202

6.5.1 Particle Advection: Forward in time . 202

6.5.2 Particle Advection: Backward in time . . 206

6.6 Summary ...... 211

7 Recapitulation 213

7.1 Objectives Revisited . 219

7.2 Concluding Remarks . 220

Appendix A: The Princeton Ocean Model 222

Bibliography 225 List of Figures

1.1 SeaWiFS satellite images of ocean colour on the east coast of Australia...... 3

1. 2 Mean potential density and nitrate concentrations from CARS in the Smoky Cape

region, NSW, Australia. 5

2.1 The experiment region, in the vicinity of Smoky Cape, NSW, Australia...... 16

2.2 Map of experiment region showing location of the CTD sections...... 19

2.3 Schematic diagram of the across-shore location of the mooring arrays...... 21

2.4 Schematic diagram of the design of the mid-shelf mooring at Smoky Cape. 22

3.1 Current vector time series from each mooring at Smoky Cape. 30

3.2 Current vector time series from each mooring at Diamond Head...... 31

3.3 Filtered temperature time series: Smoky Cape. 34

3.4 Filtered temperature time series at Diamond Head ...... 35

xv LIST OF FIGURES xvi

3.5 Filtered wind stress at South West Rocks, Port Macquarie and Taree...... 38

3.6 Sea level anomalies at Coffs Harbour, Port Macquarie, Crowdy Head, Port Stephens

and Sydney on the NSW coast...... 39

3.7 Sea level pressure anomalies at Coffs Harbour, South West Rocks, Taree, and

Sydney...... 41

3.8 Sea level anomalies and sea level pressure anomalies at Port Macquarie...... 42

3.9 Correlations of adjusted sea level elevation...... 43

3.10 Time series of alongshore wind stress and sub-inertial alongshore currents at

Smoky Cape...... 44

3.11 Time series of alongshore wind stress and sub-inertial alongshore currents at Di-

amond Head...... 45

3.12 Time series of alongshore wind stress and sub-inertial across-shore currents at

Smoky Cape...... 46

3.13 Time series of alongshore wind stress and sub-inertial across-shore currents at

Diamond Head...... 4 7

3.14 Alongshore surface currents at Smoky Cape and sea level elevation at Port Mac­

quarie. 49

3.15 Sea level anomalies on the NSW coast, 17 December 1998 - 9 January 1999. 50

3.16 Mean Sea level Pressure and surface wind field for the 23 and 27 December 1998 . 51

3.17 Satellite images of sea surface temperature on 27 December 1998 and 7 January

1999...... 52 LIST OF FIGURES xvn

3.18 Temporal mean of the current velocity at each mooring for each event...... 55

3.19 Temporal mean of the temperature time series at each mooring for each event. . 60

4.1 Power spectrum of the currents at Smoky Cape and Diamond Head...... 65

4.2 Coherence, Phase and Gain between the depth averaged currents at Smoky Cape

and Diamond Head...... 67

4.3 Cross correlations between the across-shore currents at SCB...... 70

4.4 Cross correlations between the along and across-shore currents at SCB and DHB. 70

4.5 Scalar EOFs at Smoky Cape and Diamond Head...... 74

4.6 Vertical EOFs at Smoky Cape...... 75

4.7 Vertical EOFs at Diamond Head...... 75

4.8 Horizontal structure of the largest eigenfunction of the current meter time series

at each mooring...... 77

4.9 Horizontal structure of the second largest eigenfunction of the current meter time

series at each mooring...... 78

4.10 Time series of the depth averaged currents at Smoky Cape and Diamond Head. . 82

4.11 Time series of the terms and residual in the alongshore momentum equation. . . 89

4.12 Temporal mean of the terms and residual in the alongshore momentum equation. 90

4.13 Time series of Tb and ii.Vii at SCA and SCB...... 91

4.14 Time series of iit + Tb and Py at DHA and DHB...... 92 LIST OF FIGURES xviii

4.15 Time series of iit and Ju showing periods of ageostrophy...... 99

4.16 EOFs of the terms in the alongshore momentum equation at Smoky Cape. . .. 102

4.17 EOFs of the terms in the alongshore momentum equation at Diamond Head. . . 103

4.18 Reconstructed time series from the amplitudes of the EOFs of terms in the along-

shore momentum equation at Smoky Cape...... 105

4.19 Reconstructed time series from the amplitudes of the EOFs of terms in the along-

shore momentum equation at Diamond Head ...... 106

4.20 EOFs of the terms in the alongshore momentum equation at Smoky Cape for four

events. 109

4.21 EOFs of the terms in the alongshore momentum equation at Diamond Head for

four events...... 110

5.1 Schematic representation of wind driven upwelling ...... 122

5.2 Lowpass filtered wind stress at Smoky Cape, 2 - 26 November 1998...... 123

5.3 Cross sections of alongshore velocity, nitrate and chlorophyll-a, November 1998 .. 124

5.4 Cross sections of alongshore velocity, nitrate and chlorophyll-a, January 1999 .. 125

5.5 Schematic representation of current driven upwelling...... 129

5.6 Satellite images of sea surface temperature on 15 November 1998 and 21 November

1998...... 130

5. 7 Vector plot of the surface currents, November 1998...... 131

5.8 Schematic representation of before and after encroachment ...... 134 LIST OF FIGURES XlX

5.9 Time series of the alongshore wind stress, alongshore depth averaged currents and

temperature anomaly at Smoky Cape...... 136

5 .10 Time series of the alongshore wind stress, alongshore depth averaged currents and

temperature anomaly at Diamond Head. 137

5 .11 Cross correlations between surface currents rY at Smoky Cape and Diamond Head.139

5.12 Time series of wind driven uplift and temperature anomalies ...... 141

5.13 Observed, wind driven and calculated temperature anomalies...... 143

5.14 Time series of Burger number at each mooring...... 146

5.15 Buoyancy Frequency Nat Smoky Cape and Diamond Head...... 147

5.16 Time series of alongshore velocity, temperature in the BBL, and Burger number

at Smoky Cape...... 148

5.17 Time series of alongshore velocity, temperature in the BBL, and Burger number

at Diamond Head...... 149

5.18 Time series of alongshore wind stress, velocity, vorticity, Rossby number and

temperature at Smoky Cape ...... 151

5.19 Satellite images of sea surface temperature 18 and 26 January 1999 ...... 153

5.20 Vector plot of the ADCP surface currents, January 1999...... 154

5.21 Observed and mean nitrate concentrations at Urunga, Diamond Head and Point

Stephens...... 162

6.1 Model domain and grid...... 172 LIST OF FIGURES XX

6.2 Comparison of the CARS and POM grids...... 173

6.3 Evolution of the kinetic energy field during model spin up for the ramped case. . 179

6.4 Surface and depth averaged geostrophic velocity field...... 180

6.5 Forcing at the northern boundary...... 181

6.6 Evolution of the kinetic energy field for each of the three cases...... 181

6. 7 RMS differences in 'f/ and V for three model initialisation techniques...... 182

6.8 Depth averaged velocity field and the sea surface elevation ...... 184

6.9 Comparison of observed and modelled surface velocities...... 186

6.10 Comparisons of the observed and modelled alongshore currents upstream and

downstream of the EAC separation point ...... 187

6.11 Cross sections of the climatological currents (v,u,w) at Urunga, Smoky Cape,

Point Plamer and Diamond Head ...... 189

6.12 Observations of chlorophyll-a concentrations...... 190

6.13 Velocity and depth of the a= 25.25 kgm-3 isopycnal slab ...... 192

6.14 The local Rossby number and the depth averaged transport_streamfunction .... 195

6.15 The transport streamfunction along a surface slab from the modelled velocity

field and the geostrophic velocity field from CARS...... 197

6.16 Richardson number and the Burger number for the Smoky Cape region ...... 198

6.17 Divergence and Rossby number at the surface and along the a = 25.25 kgm-3 isopycnal surface. . 200 LIST OF FIGURES XXI

6.18 Trajectories advected forward in time by surface velocities. 204

6.19 Trajectories advected backward in time along the surface slab and along the

isopycnal surface. 207

6.20 Shore normal trajectories advected backward in time along the surface slab .... 209 List of Tables

2.1 Details of the CTD sections conducted in November 1998 and January 1999. 20

2.2 Table of current meters across the two mooring arrays...... 24

3.1 Basic current meter statistics...... 33

3.2 Mean and standard deviation from the temperature records...... 36

3.3 Basic wind record statistics. 37

3.4 Basic current meter statistics during each event...... 58

4.1 Temporal mean of the depth averaged velocities. . . . . 80

4.2 Cross correlation coefficients between terms in the alongshore momentum equa-

tion at Smoky Cape...... 94

4.3 Cross correlation coefficients between terms in the alongshore momentum equa-

tion at Diamond Head...... 95

xxii LIST OF TABLES xxiii

4.4 Cross correlation coefficients between terms in the alongshore momentum equa-

tion at Smoky Cape and Diamond Head during EA C 1 and EA C 2...... 96

6.1 RMS differences between three initialisation methods...... 179 The ocean, whose tides respond ... to the pull of the moon, the ocean which corresponds to the amniotic Buid in which human life begins, the ocean on whose surface vessels.. can ride but in whose depth sailors meet their death and monsters conceal themselves... it is un­ stable and threatening as the earth is not; it spawns new life daily yet swallows up lives; it is changeable like the moon, unregulated, yet indestructible and eternal.

Adrienne Rich, 1976

Introduction

Upwelling in the coastal waters of eastern Australia has been observed to be a per­ sistent feature during the Austral spring and summer and may be either widespread or localised in extent (Rochford, 1984). Associated with this upwelling are sporadic blooms of phytoplankton and occasional swarms of zooplankton in continental shelf waters (Hallegraeff and Jeffrey, 1993). In oligotrophic waters such as those on the east coast of Australia, any increase in nutrient concentrations is significant as this can result in an increase in phytoplankton productivity. The consequences of such an increase in primary production are evident through the food chain. Furthermore, observations of blooms of toxic dinoflagellates (commonly known as red tides) have

1 CHAPTER 1. INTRODUCTION 2 increased recently (eg Hallegraeff (1988), Anderson (1989)). Toxic algal blooms and their implications are of paramount interest in the public domain and raise ques­ tions concerning management of the coastal environment. It is therefore important that the mechanisms which drive these nutrient enrichment events are fully under­ stood. Furthermore there currently exists a disparity in the literature in exactly what mechanisms drive upwelling along the east coast.

Wind-driven upwelling is a widespread phenomenon and is commonly accepted as the dominant nutrient enrichment mechanism in many coastal regions around the world. Coastal waters on the east coast of Australia are generally poor in nutrients, and although wind-driven upwelling does occur in response to seasonal north and northeasterly breezes, it is not a persistent feature and is by no means massive. Wind-driven upwelling events along the coast of New South Wales have generally been observed to be both localised and short-lived (9 - 15 days). Despite the lack of local wind forcing upwelling has commonly been observed in distinct locations along the east coast.

1.1 Motivation

Upwelling is a common feature in the Smoky Cape region on the mid-north coast of New South Wales (NSW) Australia (Rochford, 1975), moreoyer, often this upwelling appears to be unrelated to local wind forcing. Concentrations of phytoplankton can be estimated from satellite observations of ocean colour. For example, Figure 1.1 shows two images of ocean colour for the NSW coast obtained from the NASA Sea-viewing Wide Field-of-view Sensor (SeaWiFS). Figure l.la shows widespread biological activity extending along much of the coast whilst Figure l.lb shows lo­ calised productivity extending southward from Smoky Cape (30°55'8). Smoky Cape is the approximate region where the East Australian Current (EAC) most corn- CHAPTER 1. I NTROD UCTION 3

monly separates from t he coast and moves southeastward. These regions of high productivity arc indicative of the occurrence of nutrient upwclling events. However the difference in the biological response. which is highlighted by the coastal (Fig­ ure 1. l a) and non-coastal (Figure 1.1 b) blooms. suggests that different enrichment mechanisms arc involved.

0.7

0.5 ~; 0.3 _[ m .c>­ 0.2 ~ .2 .c ()

0.1

0.07

0.05

0.03

151 E 152 E 153 E 154 E 151 E 152 E 153 E 154 E

Figure 1. 1: SeaWiFS satellite images of ocf'an colour on the cast coast of Australia. High chlorophyll- a concrntrations (2 mg m- 3 ) are represented by t he warm colours (red). low concen­ trations (0.0.'3 mg m- 3 ) arc represented by the cool colours (blue) .

T he persistent upwclling south of Smoky Cape is reflected in the mean density and nitrate concentrations in the region. Horizontal sections of potential density and mean nitrate concentrations along the cast coast of Australia arc shown in Figure 1.2. T he data were obtained from the Climatology of Australian Regional Seas (CARS) (Ridgway et al. . 2001 ). In the surface waters the mean density ranges from 24 - 2G kg m- a and mean nutrient concentrations arc < 1 µmo11- t (Figure 1.2a.d) . At CHAPTER 1. INTRODUCTION 4 a depth of 250 m the potential density of the water adjacent to the coast is ,..., 27.5 kg m-3 and the mean nutrient concentrations are as high as 15 µmo11- 1 and extend the length of the coast (Figure l.2c,f). Higher in the water column over the continental shelf at a depth of 125 m mean densities range from 26.5 - 27.5 kg m-3 and nutrient concentrations along the coast are rv 5 - 15 µmo11- 1 (Figure l.2b,e). However the maximum values of both density and nutrient concentrations are found adjacent to the coast, immediately south of Smoky Cape. As well as highlighting the oligotrophic nature of the surface waters this shows that a nutrient pool exists at depth. Most importantly however, the increase in mid-shelf, mid-depth density and nutrient concentrations south of Smoky Cape indicates a persistent pattern of uplift and advection at, and southward of, Smoky Cape. Following Rochford (1991), throughout this thesis the term uplift is used to refer to the raising of cold water and nutrients toward, but not reaching the surface, and the term upwelling refers to a raising of nutrients to the surface proper.

Over the last two decades various explanations have been proposed to account for the upwelling of dense, nutrient rich water into the coastal region (Rochford, 1975; Godfrey et al., 1980b; Cresswell, 1994; Gibbs et al., 1998; Oke and Middleton, 2000, inter alia). During each of these studies different oceanographic conditions have prevailed. As well as wind driven upwelling, significant decreases in bottom tem­ perature have been observed in association with: i) a rapidly flowing EAC (Boland and Church, 1981; Blackburn and Cresswell, 1993); ii) the. separation of the EAC from the coast (Tranter et al., 1986); and iii) eddies associated with the EAC system (Boland and Church, 1981; Cresswell, 1994; Gibbs et al., 1997). These mechanisms however tended to have been inferred from incomplete data sets which were obtained during the investigation of other processes.

Idealised modelling studies by Chapman and Lentz (1997) and Gibbs et al. (1998) designed to investigate the role of current driven upwelling, concluded that Ekman CHAPTER 1. I NTRODUCTION 5

29S 28

30S 27 31S ('") I 26 E 0) 32S -"'-

33S 25

34S 24 29S 15

30S

31S 10 -' '0 32S E :::l. 5 33S

34S 0 151E 153E 155E 151E 153E 155E 151E 153E 155E

Figure 1.2: Mean p otential density (a-c) and nitrate concentrations (d- f) in the Smoky Cape region. NSW. Australia. M ean fields a rc obtained from CARS, at three depth layers: 10 m. 125 m. 250 111 .

pumping through the bottom b oundary layer is not an effective way of transporting bottom water into the c oastal region. as the bot tom boundary layer ·shuts down' and ceases to pump water after only a few days. Hence the importance of upwelli11g driven b y the EAC has been d own-played . However as wind driven upwclling is not common in the Smoky Cape region. and regular upwclling has been observed. the most obvious forcing mechanism is the vert ical circulation associated with the swift ly flowing Ea t Australian Current a nd its separation from the coast. CHAPTER 1. INTRODUCTION 6 1.2 Overview

The East Australian Current forms the western boundary to the south Pacific sub­ tropical gyre. The EAC originates in the Coral Sea, where the westward flow bi­ furcates at around 18 °S (Church, 1987). It then flows southward along the east coast of Australia, carrying warm water from the equatorial region poleward to the Tasman Sea, forming a typical western boundary current (Gill, 1982; Church and Freeland, 1987; Church and Craig, 1998, inter alia). The EAC dominates shelf flows between 25 °S and 30 °S, where it is at its most intense. As a result the EAC has a significant impact on conditions of the shelf region.

In many respects the EAC is analogous to the Gulf Stream, which at comparable lat­ itudes interacts with coastal waters at the western boundary of the Atlantic Ocean. However Tranter et al. (1986) cite topographic differences, stating that the conti­ nental shelf off NSW is much narrower than its Atlantic equivalent. Poleward of 32 °S the coast of NSW tends westward away from the axis of the EAC, however in the case of the Gulf Stream it is the actual axis of the current that bends away from a relatively straight coastline. This perhaps contributes towards the behavioural differences between the Gulf Stream and the EAC.

Surface velocities in the core of the EAC are of the order of 1 - 2 ms-1 southward, with the velocity maximum usually found at the shelf break (Boland and Hamon, 1970; Godfrey et al., 1980a). The EAC system is highly v~riable both temporally and spatially (Hamon and Kerr, 1968). It is thought that the current has a well­ defined annual cycle with the strongest flow during the Austral summer (Godfrey, 1973; Hamon, 1965; Hamilton, 1996).

Two decades of oceanographic data reveal that the EAC tends to separate from the coast along a sharp front, somewhere between 30 - 34 °S, and most often in the CHAPTER 1. INTRODUCTION 7 vicinity of Sugar Loaf Point (32 °27'8) (Godfrey et al., 1980b). Consensus has not been reached on the specific mechanisms causing separation. Godfrey et al. (1980b) suggested that variations in alongshore topography and continental shelf alignment were significant in causing the current to move off the continental slope. More recently Mata et al. (2000) proposed that baroclinic Rossby waves may contribute to separation of the EAC as eddies are shed from the main flow. As the current leaves the coast it forms a sharp front between the warm southward flowing EAC and cooler often northward flowing shelf waters of the Tasman Sea. Temperature changes of 2 - 5 °C have been observed across the front (Godfrey et al., 1980b; Cresswell et al., 1983; Tranter et al., 1986).

It has long been recognised that upwelling on the east coast of Australia can occur when the East Australian Current is within close proximity to the coast. Rochford (1975) explained upwelling events that seemed to be unrelated to local winds as a result of EAC intrusions. Furthermore, these slope water intrusions were identified by Rochford (1975) and Pearce (1980) as the primary source of nutrient enrichment along the NSW coast.

Boland and Church (1981) found that the bottom Ekman flux associated with the poleward flowing EAC is the main mechanism which pumps nutrient rich waters up onto the continental shelf. Further to the south off the Sydney coast, Gibbs et al. (1998) found that bottom Ekman forcing alone is not a sufficient means of driving upwelling, as certain conditions can cause the bottom boundary layer (BBL) to shut down (MacCready and Rhines, 1993). They concluded that the proximity of the EAC to the coast preconditions the region by uplifting the isopleths, however upwelling favourable winds are needed to actually drive the uplifted water into the coastal region.

Hallegraeff and Jeffrey (1993) observed blooms of phytoplankton extending along CHAPTER 1. INTRODUCTION 8 the east coast, from Cape Hawke (32 °S) southward to Maria Island (43 °S). Typ­ ically these phytoplankton blooms occurred in spring and summer which Hamon (1968) attributed to the seasonal nature of the EAC. Godfrey et al. (1980b) later proposed that south of the separation point, surface waters tended to be entrained into a coastal current which flowed northward towards the point of separation. They argued that this northward counter current over the continental shelf may also con­ tribute to the vertical enrichment process.

Based on the work of Godfrey, Tranter et al. (1986) proposed that food chains would develop to the north of the point of nutrient enrichment, as they were carried with the nutrient rich counter current. This concept was contrary to previous beliefs that waters enriched by slope water intrusions were carried southward with the EAC. This northward flowing enrichment effect has also been observed further south off Sydney, associated with the intrusion of an EAC eddy (Cresswell, 1994).

Another physical mechanism that is possibly associated with uplift and upwelling is related to the actual separation of the EAC. The EAC generally separates from the coast along a line extending south east from Sugar Loaf Point (32 °30 'S) (Godfrey et al., 1980b). From the examination of accumulated bottom sediments it was concluded that this point of separation had been pronounced for a considerable period of time. Southward of the point where the EAC separates from the coast, it is common to find upwelled waters in the region bounded by the coast and the warmer current offshore. Often a temperature change of·"' 5 °C occurs over a very short distance, accompanied by a visible change in colour, from the tropical blue oligotrophic waters of the EAC to green, phytoplankton-rich upwelled waters (Church and Cresswell, 1986). Associated with the front is a rapid change in current speed which creates a disturbance in the surface wave pattern, and which is often accompanied by the accumulation of surface debris. CHAPTER 1. INTRODUCTION 9

Numerical studies by Oke and Middleton (2000) indicate that alongshore topo­ graphic variations in the vicinity of Cape Byron (northern NSW) can have a signifi­ cant impact on the near-shore and shelf circulation. They suggest that a narrowing of the continental shelf accelerates the alongshore flow, resulting in an area of high bottom stress which then drives an upwelling BBL flow. The simulations showed a high level of vertical mixing between upwelled BBL water and interior shelf wa­ ters, as well as 'along-isopycnal mixing'. This mixing provides a mechanism for the entrainment of nutrient rich slope water into the EAC mean flow.

With the exception of the well known effects of local wind forcing, these diverse theories all suggest that the highly variable influence of the EAC is possibly respon­ sible for uplift, and perhaps also upwelling and nutrient enrichment. Therefore to understand the mechanisms driving upwelling on the NSW coast, the questions that remain to be addressed are as follows:

• Why is upwelling more prolific at some geographical locations such as south of Smoky Cape;

• Exactly what role does the EAC play in driving upwelling in this region; and

• What is the role of local wind forcing in driving upwelling in this region?

1.3 Scope of Study

The purpose of this study is to increase knowledge of the physical oceanography of the inner shelf immediately upstream and downstream of the separation point of the EAC, with the subsequent goal of understanding the distribution of upwelled nutrient rich water over the inner continental shelf. CHAPTER 1. INTRODUCTION 10

This study complements the work of Oke and Middleton (2000, 2001) who found, from numerical simulations of the Cape Byron and Smoky Cape region, that changes in alongshore topography can result in upwelling driven by the EAC. Previously hydrographic surveys have identified this region as an area of preferential nutrient uplift. However, there have been no extensive physical oceanographic studies to examine why upwelling is prolific in the area.

This observational study is designed in part to test the theories of current driven upwelling where the continental shelf narrows at Smoky Cape and in part to exam­ ine the dynamics about the separation point of the EAC. A comprehensive study consisting of both hydrographic and in-situ measurements conducted in the vicinity of the separation point of the EAC is described. The study site was chosen for its complex dynamics in terms of EAC separation, upwelling, coastal processes, and frontal dynamics. This study is significant in that it encompasses the first high density near-shore hydrographic survey to span the separation point of the EAC concurrent with in-situ moored current and temperature measurements. The study enables a critical examination of the effects of the EAC on the coastal waters. It is the first study that presents in-situ observations of current driven upwelling in this region.

1.4 Objectives

The main objectives of this study are:

1. To investigate the dynamics upstream and downstream of the EAC separation point;

2. To examine upwelling south of the Smoky Cape and suggest possible mecha­ nisms driving that upwelling; and CHAPTER 1. INTRODUCTION 11

3. To understand the advection processes near the EAC separation point.

These goals are achieved through

• The combined analysis of in-situ measurements and hydrographic observations from an extensive field program, and

• A numerical modelling study of the NSW shelf region.

1.5 Outline

This thesis is organised as follows: Chapters 2 and 3 deal with the design and im­ plementation of the field project and a description of the observations. In Chapters 4 and 5, analyses and interpretations of the observations are presented. A mod­ elling study is described in Chapter 6, followed by a discussion and recapitulation in Chapter 7. The results presented in Chapter 5 of this work have been submitted for publication as listed in the Supporting Publications.

1.5.1 Chapter 2: The Field Project

In this chapter the design and implementation of the Smoky Cape experiment is documented. The methodology and techniques employed in acquiring the data, including the deployment of two mooring arrays over a three month period and two research cruises undertaken aboard RV Franklin, are described in detail. The methods used in the processing and subsequent analysis of the data are presented. CHAPTER 1. INTRODUCTION 12

1.5.2 Chapter 3: Experimental Observations

Time series of current velocities, temperature, local winds and sea level elevation obtained during the experiment are presented. The time series are described in terms of five main events that occurred during the study period, these being a Wind event, a current Reversal event, a Current event and two main EAC events EA C 1 and EA C 2. These events form the basis for the interpretation presented in the following chapters.

1.5.3 Chapter 4: Variability about the EAC separation

Time series of current and temperature are examined in terms of spatial and tem­ poral variability upstream and downstream of the EAC separation point. Empirical orthogonal functions are utilised in an analysis of the vertical and horizontal struc­ ture of the current at each of the moorings. The terms in the alongshore momentum equation are evaluated using the observed data and a study of the balance of these terms is undertaken using both time series and statistical techniques. The study focuses on the balances upstream and downstream of the separation point as well as the balances that dominate each of the events in turn.

1.5.4 Chapter 5: Current Driven Upwelling

The hydrographic data obtained during two oceanographic voyages are combined with the current and temperature time series to examine the role of the EAC in coastal upwelling. Wind driven upwelling and current driven upwelling events are examined in detail. Specifically three current induced upwelling mechanisms are investigated, these being uplift resulting from the encroachment of the EAC across CHAPTER 1. INTRODUCTION 13 the continental shelf, uplift induced by the topographic acceleration of the EAC and upwelling associated with the separation of the EAC.

1.5.5 Chapter 6: A Modelling Study about the EAC Sepa­ ration

Finally a modelling study of the NSW shelf region is presented. For this study a configuration of the Princeton Ocean Model is implemented to examine the clima­ tological current field across the mid-NSW continental shelf. The mean current field obtained from the model is then used in a series of Lagrangian particle tracking experiments to investigate the origin and fate of upwelled water.

1.5.6 Chapter 7: Recapitulation

The main results of the thesis are summarised in this chapter. The implications of the work are discussed and potential directions for further work are canvassed. If we knew what it was we were doing, it would not be called research, would it?

Albert Einstein (1879-1955)

The Field Project

In order to answer the questions posed in Chapter 1 a detailed high resolution oceanographic data set is needed. Data sets acquired in the past have not been sufficient to address the questions posed in the vicinity of the EAC separation point. Hence the Smoky Cape experiment was devised.

The design and implementation of an extensive oceanographic field project was a significant component of the candidature. Aside from the obvious scientific benefits, many new skills were learnt during the 6 months dedicated to the planning phase of the experiment: from the design and construction of oceanographic moorings where

14 CHAPTER 2. THE FIELD PROJECT 15 consideration must be given to strength and composition of materials, to mooring location, where one must position the moorings in such a way as to minimise mooring loss whilst maximising data retrieval. Many technical skills were learnt such as swaging, splicing, and the benefits of antifouling, (not just instruments, but also the limbs of various willing workers!) On a personal level, the experiment necessitated the development of many qualities, not least of which was the patience needed to coordinate and work with a menagerie of highly specialised people in a very close working environment.

2.1 The Experimental Program

The study site extends alongshore from Urunga (30°S) to Port Stephens (33°S), New South Wales, Australia, and offshore from the coast to the 500 m isobath (Figure 2.1). This region spans the widely recognised location of EAC separation, as well as Smoky Cape, the narrowest point (16 km) of the NSW continental shelf (Figure 2.1). The both the cross sectional area and width of the shelf are shown as a function of latitude in Oke et al. (2000, Fig. 8). North of the separation point the EAC tends to hug the coast and flows in excess of 2 ms-1 are observed at the shelf break. Immediately downstream of the separation point, the interaction of the current and the shelf waters results in a strong thermal front and high velocity shear.

2.1.1 The Hydrographic Survey

The experimental program was carried out during the 1998 -1999 Austral summer. Shipboard hydrographic measurements were obtained during two cruises aboard the RV Franklin (FR1498 and FR0199). CTD (Conductivity, Temperature and Depth) CHAPTER 2. THE FIELD PROJECT 16

Experiment Region 29S------..------

<)Mooring Anemometer 30S *

31S

32S

33S

34S

152E 154E

Figure 2.1: The experiment region, in the vicinity of Smoky Cape, NSW, Australia. The mooring locations are denoted by the O, and anemometer locations (referred to in the text from north to south as CH, SWR, PMQ and TRE) are denoted by the*· The 100, 200, 1000 and 2000 m isobaths are shown. The 200 m isobath is marked in bold. Distance scales are defined as follows: 1 ° of latitude= 60 nautical miles = 111 km. CHAPTER 2. THE FIELD PROJECT 17 and ADCP (Acoustic Doppler Current Profiler) measurements were taken along transects across the continental shelf during both these cruises, in conjunction with an intensive biological sampling regime which included EZ and Neuston net tows, minibat tows and water samples.

The CTD acquires measurements of conductivity (C), temperature (T), fluorescence, light, dissolved oxygen, and pressure from which from which salinity (S), and depth (D) are calculated. The fluorometer was calibrated with in-vivo fluorescence from water samples obtained underway to obtain an estimate of chlorophyll-a concentra­ tions. The calibration regression equation calculated from data obtained during the November cruise is

chlorophylla = 0.0719 x fluorescence+ 0.191 with R 2 = 0.76 (Dela Cruz pers. comm.). Standard underway sampling was also conducted. The underway dataset, which was either recorded at, or averaged over 5 minute intervals, consists of GPS measurements of latitude, longitude, ship's direc­ tion, speed and depth. Surface temperature and salinity were measured continuously using a Thermosalinograph connected to the engine cooling intake system, which takes water from a depth of 4 m. Meteorological data including atmospheric tem­ perature, humidity and pressure, wind speed and direction were also recorded at 5 minute intervals, from instruments located on the ship's mast at a height of 10 m.

During the first cruise (14 - 27 November 1998) 116 CTD casts were completed along 4 main shore normal transects, starting inshore at a depth of 25 m moving offshore to a depth of up to 2000 m. ADCP transects were also run along the CTD lines immediately after the completion of each CTD transect. These four transects captured the alongshore advection of the EAC as it moved southward.

During the second cruise (20 January -4 February 1999), 154 CTD casts were com­ pleted along a total of 9 transects. These transects included repeats of the original CHAPTER 2. THE FIELD PROJECT 18

4 transects, as well as an additional 5 transects further downstream, across the strong thermal front associated with the separation of the EAC. This overall period (November - February ) was chosen to capture the full strength and variability of the EAC during the summer season when biological productivity is at a maximum. The locations of the CTD transects are shown in Figure 2.2. The four main tran­ sects, Urunga, Smoky Cape, Point Plomer and Diamond Head, were named because of their proximity to geographical features and were repeated twice on each cruise. The 5 southern transects undertaken on the second cruise are named Crowdy Head, South Crowdy Head, Cape Hawke, Broughton Island and Point Stephens. Ad­ ditional CTD casts were also taken along the 1000 m isobath in the alongshore direction, between each of the 4 main transects. A summary of all the CTD sections completed during both cruises is presented in Table 2.1.

2 .1. 2 The Current Meter Array

The mooring program involved the deployment of two across-shore mooring arrays over a three month period from November 1998 - January 1999. One array was deployed at Smoky Cape, where the shelf is 16 km wide, the other array was deployed 90 km south of this point, directly offshore from Diamond Head where the shelf widens to 30 km. For the purpose of this thesis, the continental shelf is defined as the region extending seaward of the coast to the 200 m isobath. The mooring arrays were designed to examine the three dimensional nature of the East Australian Current.

A schematic diagram of the mooring arrays is shown in Figure 2.3. Fourteen cur­ rent meters (a mixture of 2D and 3D Falmouth Scientific Instruments (FSis), Inter Ocean S4s and Aanderra RCM4s) were deployed through the two mooring arrays, positioned shore-normal across the continental shelf. Optic Stowaway and Titbit CHAPTER 2. THE FIELD PROJECT 19

November 1998

152E 153E

Figure 2.2: Map of the experiment region showing CTD sections completed during two cruises aboard the RV Franklin, November 1998 (left) and February 1999 (right). CTD casts are denoted by•, mooring locations are denoted by 0, anemometer locations are denoted by*· The 100, 200, 1000 and 2000 m isobaths are shown. The 200 m isobath is marked in bold. Distance scales are defined as follows: 1 ° of latitude = 60 nautical miles = 111 km. CHAPTER 2. THE FIELD PROJECT 20

Transect Latitude Date Time (UTC) max depth Cast no. 0 s start stop (m) Urunga 30 30.28 16/11/98 19:27 05:18 2000 16-25 Smoky Cape 30 55.38 16/11/98 22:48 07:57 1000 30-39 Pt Plomer 3118.99 17/11/98 20:00 04:37 500 43-50 Diamond Head 31 44.36 19/11/98 19:02 01:47 300 51-57 Smoky Cape 2 30 55.34 21/11/98 21:34 05:45 1000 68-76 Pt Plomer 2 3119.01 23/11/98 01:07 08:21 300 77-83 Diamond Head 2 31 43.37 23/11/98 22:26 05:22 500 84-91 Port Hacking 34 05.84 25/11/98 19:53 03:43 500 101-107 Urunga 30 30.71 21/1/99 19:41 04:44 1000 6-15 Urunga N/S 30 35.28 22/1/99 03:57 09:03 1000 15-17 Smoky Cape 30 55.20 22/1/99 19:06 02:03 1000 22-30 Smoky Cape N /S 30 58.91 23/1/99 01:21 08:28 1000 30-32 Pt Plomer 31 19.30 23/1/99 19:07 04:17 1000 37-45 Diamond Head 31 43.63 24/1/99 20:03 02:58 200 51-59 Smoky Cape 2 30 55.16 27/1/99 00:50 05:53 300 64-70 Crowdy Head 31 50.92 27/1/99 19:42 00:50 150 74-80 Sth Crowdy Head 32 00.08 28/1/99 03:31 08:09 140 81-84 Cape Hawke 32 09.98 28/1/99 21:01 06:00 500 91-101 Broughton Is. 32 32.99 29/1/99 19:40 06:32 1000 107-119 Point Stephens 32 44.99 30/1/99 19:28 02:07 200 124-130

Table 2.1: Details of the CTD sections completed during two cruises aboard the RV Franklin in November 1998 and January 1999. temperature sensors were interspersed amongst the current meters throughout both arrays. The sampling interval was set to 3 minute averages over a 15 minute period for the FSis, S4s and Titbits, and over a 30 minute period f~r the Aanderras and Optic Stowaways. The Smoky Cape array (30°55'8), consisted of 3 moorings, Smoky Cape A (SCA), Smoky Cape B (SCB) and Smoky Cape C (SCC), spaced evenly across the continental shelf, deployed in 50 m, 100 m, 150 m of water respectively. The Diamond Head array (31 °44'8) consisted of four moorings, Diamond Head AA (DRAA), Diamond Head A (DHA), Diamond Head B (DHB), Diamond Head C (DHC), deployed in 30 m, 50 m, 95 m, 150 m water respectively. CHAPTER 2. THE FIELD PROJECT 21

The Smoky Cape Array: 30° 55

• S4 • 3D FSI * RCM4 - 50 .A. 2D FSI X Titbit ..c fil" - 100 0

- 150

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30

The Diamond Head Array: 31 ° 44

-50

..c fil" - 100 0

- 150 •

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 Distance offshore (km)

Figure 2.3: Schematic diagram of the across-shore location of the mooring arrays, showing horizontal and vertical position of the current meters and temperature sensors. (Dotted line indicates snapped mooring. )

2.1.3 Mooring Construction

The schematic diagram of the mid-shelf mooring at Smoky Cape (SCB) is shown in Figure 2.4. The Mooring Design and Dynamics program for Matlab (Dewey, 2000) was used to design the current meter moorings and estimate the amount of buoyancy and weight needed to keep the mooring securely anchored to the seabed with minimal mooring layover. CHAPTER 2. THE FIELD PROJECT 22

Smoky Cape B, 100m

Mooring Element Length[m) Height[m) 12B-3 0. 34 92. 63

unsw wire 12.00 92.28

12x10B3 pack 0 .87 80 . 20

unsw wire 1.00 79.33 30 - ACM 0 .40 78 . 18

unsw wire 10.00 77.62 unsw wire 0 . 50 67.54 13B- 2 0 . 39 67.04 13B- 2 0. 39 66.65 unsw wire 0 . 50 66 . 26

unsw wire 15 . 00 65.68

unsw wire 0 . 50 50.60 l0B-3 0 . 29 50.10 12B- 3 0.34 49.81 unsw wire 0.50 49 . 46

InterOcn S4 CM 0 .80 48 . 81

unsw wire 25 . 00 47.67

unsw wire 0 .50 22.59 nok 0. 28 22.09 12B-3 0. 34 21.81 unsw wire 0 .50 21. 46

InterOcn S4 CM 0 .80 20 . 81

unsw wire 15 .00 19. 85

unsw wire 0 .50 4.77 n ok 0 . 28 4 . 27 nok 0 . 28 3.99 nok 0.28 3. 71 unsw wire 0. 50 3 . 43

unsw sea star 0.60 2 . 61 YBffiil( Y".lr~hain 2:as ±:B 2 Railway Wh eels 0 . 35 0 . 35

Figure 2.4: Schematic diagram of the mid-shelf mooring at Smoky Cape (100 m) showing the design of the mooring, including the positioning of the current meters and acoustic release and flotation. Temperature sensors were not included in the mooring design process as because of their small size and weight their effects on the buoyancy are negligible. CHAPTER 2. THE FIELD PROJECT 23

Instrument Failings

The two offshore moorings in 150 m depth snapped at exactly the same point (50 m below the surface) due to a structural weakness in the mooring rod of the 3D FSI current meters. The top half of each of these moorings was lost, one of which was recovered some 4 weeks later, 600 miles to the south. The S4 at the bottom of the mid-shelf mooring at Diamond Head (DHBsom) showed signs of leakage, which distorted the current time series at this depth.

The instrumentation for each mooring is listed in Table 2.2 along with its depth (indicated by the numeral subscript) and position across the array (AA, A, B or C).

The current vector time series was rotated so as to align the components with the principal alongshore and across-shore directions. These angles which were obtained from the local bathymetry at each mooring site were found to range between 2-15 ° and are given for each mooring in Table 2.2.

2.2 Other Data Sources

2.2.1 Wind

Wind velocity measurements were obtained from four land-b~sed weather stations near the study site: Coffs Harbour (CH), South West Rocks (SWR), Port Macquarie (PMQ) and Taree (TRE). The locations of these stations, CH, (30.31 °S 153.12 °E), SWR (31.89 °S 152.51 °E), PMQ (30.92 °s 153.09 °E), TRE (31.44 °s 152.86 °E) are indicated in Figure 2.1. With the exception of SWR, all wind measurements were 10 minute averages logged hourly by automatic weather stations located at the respective airports. The instrument height was 10 m above the ground whilst @ ::i:,.: 'i:l ~ ~ Mooring Water Distance Instrument Instrument Instrument Record Principal ~

Depth (m) Offshore (km) Type Depth(m) Label Length (days) Axis (0 ) ~ SCA 50 2.8 S4 15 SCA1s 66 2 ttj S4 35 66 SCA3s ~ 8.9 3D FSI 20 SCB20 66 SCB 100 10 ~ S4 50 SCBso 66 t::::l 3D FSI 80 SCBso 66 ;g sec 150 14.8 3D FSI 50 SCCso 14t 10 0 RCM4 80 SCCso 67 trJ 0 DHAA 30 3.1 2D FSI 15 DHAA 1s 64 15 t--3 DHA 50 9.6 2D FSI 20 DHA20 64 15 S4 35 DHA3s 64 DHB 95 20.4 2D FSI 20 DHB20 63 9 S4 50 DHBso 63 S4 80 DHBso * DHC 150 26.4 3D FSI 50 DHCso + 9 RCM4 80 DHCso 66

Table 2.2: Table of current meters across the two mooring arrays. Showing water depth, distance offshore, type of instrument, instrument depth, given label, record length and the principal axis of each mooring. t denotes short record length, mooring failed after 7 days. * denotes spurious data returned. :j: denotes meter lost due to mooring failure.

1:...:) ,l::a,.. CHAPTER 2. THE FIELD PROJECT 25 ground height above sea level ranged from 5-10 m. The wind readings at South West Rocks were measured manually every three hours, and recorded by the light house keeper at the Smoky Cape Light house, which is 117 m above sea level. To facilitate comparison, all wind velocities were adjusted to a standard height of 10 m above sea level using a neutral stability wind profile. Furthermore, to enable comparison between wind records at different locations, the wind data was not rotated for along or across-shore alignment. Good correlation was found between the wind records at Port Macquarie and those from Taree, whilst the station at Coffs Harbour showed little correlation with any of the other three stations, and was deemed to be too far from the study site to accurately represent the wind field.

Wind speed and direction were also recorded aboard ship throughout both Franklin cruises at 5 minute intervals, in the standard underway file. Readings from each of the land based weather stations were compared with shipboard wind records in order to assess their representation of the true marine winds, and to justify the use of these winds for the three month period. Comparisons showed that whilst the wind direction was represented accurately, coastal wind records tended to slightly underestimate the magnitude of those experienced at sea.

2.2.2 Sea Level Elevation

Sea level records were obtained for 5 locations on the east coast ·of Australia, courtesy of the NSW Government, and Manly Hydraulics Laboratory. These locations are Coffs Harbour (CH, 30°17'S, 153°08'E), Port Macquarie (PMQ, 31 °25'S, 152°54'E), Crowdy Head (CRO, 31 °50'S, 152°45'E), Port Stephens (PST, 32°44'S, 152°12'E) and Sydney (SYD, 33°50'S, 151 °l5'E). This information enabled the detection of Coastally Trapped Waves (CTWs) as they propagate up the NSW coast. The sea level elevations were lowpass filtered using a Lanczos-cosine filter with a cut off CHAPTER 2. THE FIELD PROJECT 26 of 36 hours. Sea level data were adjusted for the inverse barometer effect using atmospheric pressure observations (described below) from nearby meteorological stations. The filtered atmospheric pressure anomalies were added to the filtered sea level anomalies to give the total sea surface pressure anomaly (Church and Freeland, 1987).

2.2.3 Atmospheric Pressure at the Sea Surface

Sea level pressures were obtained from 4 locations along the NSW coast, Sydney, Taree, South West Rocks (light house at Smoky Cape) and Coffs Harbour. Records are 3 hourly averages. Gaps in the data (most often at night when the light house keeper did not want to crawl out of bed) were linearly interpolated manually. The sea level pressure anomalies were calculated by subtracting the mean values which were 1014, 1012.8, 1012.6 and 1012.6 hPa from south to north. These values are fairly consistent with the recognised mean SLP of 1013 hPa, hence the actual means were us_ed in preference. The records were then lowpass filtered using a Lanczos-cosine filter with a cut off period of 36 hours.

2.2.4 Satellite Imagery

Oceanographers are now able to reap the benefits of recent ·advances in satellite technology. Satellite images of both sea surface temperature and ocean colour have been obtained for the study period. Satellite derived sea surface temperature mea­ surements were obtained from the NOAA AVHRR/HIRS programme, in the form of colour images, and as raw data. Four-hourly data were obtained for the duration of the experiment. Satellite images of Sea Surface Temperature (SST) are partic­ ularly useful for obtaining a synoptic view of the EAC system. Concentrations of CHAPTER 2. THE FIELD PROJECT 27 phytoplankton can be estimated from satellite observations of ocean colour. Im­ ages of ocean colour for the NSW coast were obtained from the NASA Sea-viewing Wide Field-of-view Sensor (SeaWiFS). Although algorithms relating light intensity to chlorophyll-a concentration can be contaminated by river runoff and sediment resuspension, in this region river runoff and sediment resuspension are minimal. These images show widespread regions of high biological productivity which are often indicative of upwelling events.

2.2.5 Climatology of Australian Regional Seas

Annual mean nutrient fields (nitrate, phosphate, silicate and dissolved oxygen) for the EAC region were obtained from the CSIRO High resolution (1/8°) Climatology of Australian Regional Seas (CARS) (Ridgway et al., 2001). Coarser resolution (1/2°) mean temperature and salinity fields were also obtained. Vertically the atlas has 56 (unequal) depth layers from O - 5500 m, with maximum resolution being from O - 300 m.

2.3 Summary

A comprehensive data set was obtained from November 1998 to February 1999 consisting of time series measurements of velocity and temperature, concurrent with two high resolution hydrographic surveys. Time series of wind, sea level elevation and atmospheric pressure were also obtained, as well as satellite imagery of sea surface temperature and ocean colour. For the duration of the experiment the mooring arrays spanned the separation point of the EAC which enables the use of this data set in subsequent chapters to answer the questions which were posed in Chapter 1. People only see what they are prepared to see.

Ralph Waldo Emerson

Experimental Observations

In this chapter the results of the mooring programme are introduced. The in-situ measurements of velocity and temperature are presented an~ discussed. Specific events which are evident in the time series are highlighted, with attention devoted to how they occurred and why. The events of interest are as follows: i) a Wind event where persistent northerly (upwelling favourable) winds blew for a three day period, concomitantly with a significant decrease in bottom temperature; ii) a cur­ rent Reversal event where the predominantly southward flowing currents oscillated in strength and direction; iii) a Current event where the current increases noticeably in strength, again, concomitantly with a decrease in bottom temperature; and iv)

28 CHAPTER 3. EXPERIMENTAL OBSERVATIONS 29 two EAC events where the EAC is flowing swiftly southward adjacent to the coast. Two events ( Wind event and Current event) coincided with significant decreases in bottom temperature across the mooring arrays, indicating that upwelling of colder water from depth has occurred. These events will be discussed in more detail in Chapters 4 and 5, in particular the mechanisms behind the decrease in bottom tem­ perature and the upwelling implications of the events. For consistency all times were either measured in or converted to UTC.

3.1 The Velocity Field

Time series of current vectors for the period of the experiment are shown in Fig­ ures 3.1 (Smoky Cape) and 3.2 (Diamond Head). The time series are labelled according to their position and depth in the mooring array which corresponds to the current meters shown in the mooring diagram in Figure 2.3. The subscript in the label refers to the depth of each current meter. For example, SCA15 is located on the Smoky Cape array, mooring A (i.e. above the 50 m isobath), at a depth of 15 m. Using a right hand coordinate system, northward currents are orientated positive, up the page.

There are several features of note in the time series records. The structure of the Smoky Cape current records are similar across the array. Qurrents are predomi­ nantly southward, with speeds up to 150 cms-1 in the surface waters (Figure 3.1). The magnitude of the current decreases with depth through the water column, and increases eastward towards the open ocean, however the overall structure remains the same at each location across the shelf. At low frequencies the current is strongly polarised in the alongshore direction, and the alongshore component is highly cor­ related vertically throughout the water column. For example from 16 - 23 January 1999 the currents at each mooring in the Smoky Cape array are flowing strongly CHAPTER 3. EXPERIMENTAL OBSERVATIONS 30

SCA35m

scc80m

SOcms-1

14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 3.1: Current vector time series from each mooring at Smoky Cape. The first two records are from the inshore mooring (SCA). The next three records are from the mid-shelf mooring (SCB) and the last two records from the offshore mooring (SCC). The depth of each instrument is indicated in the subscript adjacent to each time series. The direction convention is such that a positive current corresponds to an equatorward flow. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 31

Diamond Head current meter records

DHM1sm

DHA35m

DHBaom

50 cms-1

14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 3.2: Current vector time series from each mooring at Diamond Head. The first record is from the coastal mooring DHAA, the next two are from the inshore mooring (DHA). The next three records are from the mid-shelf mooring (DHB) and the last record is from the offshore mooring (DHC). The depth of each instrument is indicated in the subscript adjacent to each time series. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 32 southward, with little current shear.

The currents measured along the Diamond Head transect are far more variable than those 90 km to the north at Smoky Cape. These currents are also considerably weaker in strength, and less consistent in direction (Figure 3.2). There are only two short periods at the beginning and end of the time series when the currents are strongly southward and even then, only at the mid-shelf and offshore moorings (DHB, DHC). At these times the currents at DHB and DHC resemble those at SCB and SCC, which implies the EAC is having a strong influence across the shelf in both regions.

At all other times the EAC, whilst dominating the flow at Smoky Cape, does not directly influence the coastal waters at Diamond Head. This suggests that the EAC separates from the coast somewhere between Smoky Cape and Diamond Head for a significant length of time during the experimental period. At the start and end of the period however, the EAC is flowing parallel and adjacent to the coast, and driving the flow both at Smoky Cape and Diamond Head. One period of note occurs from 25 December 1998- 2 January 1999, where a complete current reversal occurs across the northern array and periodic oscillations persist across the southern array. This event will be discussed in more detail in Section 3.5.3.

The mean current strengths and the standard deviations are presented for each current meter record in Table 3.1. These results reveal the inherent variability of the currents. Inshore at Diamond Head (DHAA) the standard deviation of the across-shore flow (3.3 cms-1) is more than an order of magnitude larger than the mean current (-0.1 cms-1 ). In the alongshore direction the standard deviation of the flow (7.9 cms-1) inshore is large compared to the mean current (0.9 cms-1 ). The variability decreases at the offshore mooring (DHC) where the standard deviation of the flow (29 cms-1) is the same order of magnitude as the mean current (23.3 cms-1 ). CHAPTER 3. EXPERIMENTAL OBSERVATIONS 33

To the north at Smoky Cape, the mean and standard deviation in both the along and across-shore directions are comparable in magnitude. The maximum mean flow of 77 cms-1 southward occurs at SCB20 which decreases with depth to 47 cms-1 southward at SCB50 . At SCA and SCB the mean current in the middle of the water column is 60% of that measured at the surface. At Diamond Head the maximum mean flow (25 cms-1 southward) is again in the surface waters mid-shelf, but it is 60% weaker than the current measured along the same isobath 90 km to the north.

Smoky Cape Diamond Head u V u V SCA15 x -3.2 -54.1 DHAA1s x -0.1 0.9 (J 4.3 29.1 (J 3.3 7.9 SCA3s x -1.1 -33.6 DHA20 x -0.1 2.0 (J 3.3 25.2 (J 6.0 16.3 SCB20 x -20.6 -77.2 DHA3s x 2.2 2.3 (J 9.2 32.4 (J 5.1 10.9 SCBso x 0.5 -48.0 DHB20 x 3.5 -25.5 (J 10.8 30.1 (J 11.7 28.7 SCBso x -1.0 -28.4 DHBso x 1.1 -9.3 (J 3.9 30.4 (J 5.4 22.8 SCCso x -13.9 -56.0 DHCso x 1.6 -23.3 (J 7.1 35.6 (J 3.6 29.1

Table 3.1: Mean (x) and standard deviation (a) of the across-shore (u) and alongshore (v) components of the current time series (cms- 1) at the Smoky Cape and Diamond Head moorings. The numeral subscript indicates the depth of each instrument.

3.2 The Temperature Field

Temperature was recorded at 31 depths across the two mooring arrays for the extent of the moored experiment. Six of these records were recorded by the FSI current CHAPTER J. EXPERIJ\IENTAL OBSERVATIONS 34

meters. shown in Figure 2.3. Tweuty-, ix HDL temperature loggers were distributed throughout the water column on the moorings as indicated by the x in Figure 2.3.

The raw temperature time series were filtered using a Lanczos-cosine filter. with a cut off period of 36 hours. The lowpass filtered temperature data for the Smoky Cape array arc shown in Figure 3.3 and for the Diamond Head array in Figure 3.4. There is an obvious warmiug trencl which occurs throughout the period which spans the Austral summer and the water is noticeably stratified. There is a vertical change of 4 - 5 °C at DHA ancl SCA, and above the 100 m isobath the vertical change in temperature is 6 - 7 °C at DHB and SCB respectively.

25 8m 35m J-) 20 48m

I- 15 SCA 50m 10 ~-~--~---'---"---'-,.__~-----'----"---,.__----'-_----'------'------'--~--'

25 - 8m - 20m* J-) 20 - 55m 80m' I- 15 - 88m SCB 100m 98m 10 L__ _ __L.__ _1_ __ j__L___l_L____L _ _j__L___ L_ _ __L_ _j__j___ ...1...... c====::'..J

25 - 30m - 50m* J-) 20 - 80m 125m I- 15 - 148m sec 150m 10 ~----'---~---'---"---'-,.__~-----'----"---,.__----'------'------'------'--~--' 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 3.3: Temperature time series records at Smoky Cape. filtered with a cut off period of -16 hours. All records measmed with an HDL temperature sensor. except records measured concomitantly with current by FSI current meters (denoted by*). As in subsequent figures. each event is marked EAC 1 (El). Wind PtlPnt (v\i). Revffsal event (R). Cn11 -ent euent (C) and EA C 2 (E2). CHAPTER 3. EXPERIMENTAL OBSERVATIONS 35

25 E1 w R C E2 - Bm - 15m' J-) 20 - 16m 28m I- 15 DHA 30m 10 '------'------L--__,__--L.._~__,__-~-~~-----'---~-__,____,____ ~--~~

25 - Bm - 20m' J-) 20 - 25m 35m I- 15 - 48m DHA 50m 10

25 - Bm - 2om· 0° 20 - 55m 80m I- 15 - 93m DHB 95m 10

25 - 80m - 125m J-) 20 - 148m

I- 15

10 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 3.4: T('lnperatlll'C' timC' sNies rC'co rdC'd a t Diamond Head , filterC'd with a cut off period of 36 hums. All rC'cords nwaslll'ed with an HDL temperature sensor. except records measured concomitantly with current by FSI Ctl!TC'nt meters ( denoted by *).

Table 3.2 shows the mean and stanrl.arrl. deviation of the te11'1peraturr throughout the water column as measured across the mooring a rrays. The temperature in the surface waters remains constiu1t moving offshore at Smoky Cape. At Diamond Hearl. however. the mean temperature incrca. cs 2 °C offshore. with proximity to the EAC. As expected. temperature decreases with depth at both arrays. Inshore. surface temperatures at Diamond Hearl. arc consistently colder than those 90 km to the north at Smoky Cape. l\Iid-shrlf. apart from the surface waters which arr a similar CHAPTER 3. EXPERIMENTAL OBSERVATIONS 36 temperature, the waters at SCB are on average 3 °C warmer at depths of 20 and 50 m than at DHB in the south (Table 3.2). At a depth of 88 m mid-shelf and at the shelf break mooring the mean temperature in the north (SCC) is consistently 1 °C warmer than that at DHC. This suggests that the EAC does not directly influence the coastal waters at Diamond Head for a large part of the experiment period.

The temperature time series reveal certain events which will be discussed in further detail throughout this thesis. Of particular note are two periods where there is a significant decrease in temperature (up to 5 °C) across both mooring arrays, the first occurs from 7 - 15 December 1998 and the second occurs from 2 - 11 January 1999 (Figure 3.3 and 3.4).

depth T (J depth T (J DHAA 8 21.0 1.15 DHA 8 21.7 1.01 15 19.7 1.11 20 19.7 1.05 16 20.2 1.10 25 19.5 1.05 28 19.2 1.12 35 18.6 1.17 48 17.9 1.09 DHB 8 23.5 1.79 DHC 80 18.4 1.52 20 21.1 1.57 125 15.8 1.29 55 17.9 0.87 148 14.6 1.31 80 16.5 1.02 93 16.2 1.03

depth T (J depth T (J SCA 8 24.0 1.5 SCB 8 23.9 0.57 35 20.6 1.58 20 24:1 1.41 48 19.3 1.42 55 20.7 1.44 80 18.1 1.39 sec 80 19.6 1.38 88 17.4 1.35 125 16.5 1.20 98 17.0 1.35 148 15.3 1.25

Table 3.2: Mean and standard deviation (a) of temperature (T °C) as measured across the mooring arrays. The depth (m) of each thermistor is indicated. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 37 3.3 The Wind Field

Wind stress was calculated using the formulation of Large and Pond (1981) from wind records at three locations: South West Rocks (SWR, 31.89 °S 152.51 °E), Port Macquarie (PMQ, 30.92 °S 153.09 °E) and Taree (TRE, 31.44 °S 152.86 °E), as shown in Figure 2.1 and documented in Chapter 2. Typically, local winds along the NSW coast are highly variable in strength and direction, which is reflected in the observations presented here. Table 3.3 gives the mean and standard deviation of the wind speed and direction at the three locations, as well as the mean and standard deviation of the along and across shore components of wind stress. It is evident that the standard deviations are large when compared to the mean values. The maximum mean wind speeds were recorded at PMQ, however the mean filtered wind stress in the alongshore direction (Ty) was a maximum at SWR.

SWR PMQ TRE Speed (ms-1) x 3.0 3.7 2.9 (7 2.0 2.1 1.8

Direction (0 ) x 156.3 138.6 84.6 (7 136.5 108.9 106 Stress (Pa) Tx Ty Tx Ty Tx Ty x -0.001 -0.002 0.006 0.001 -0.008 -0.002 (7 0.007 0.023 0.008 0.015 0.010 0.013

Table 3.3: Mean (x), and standard deviation (a) of raw wind speed and direction measurements, at three land based weather stations, South West Rocks (SWR), Port Macquarie (PMQ), and Taree (TRE) at 15 min intervals. Mean and standard deviations are also shown for the lowpass filtered wind stress (Pa) in the across-shore (rz) and alongshore (ry) directions.

The filtered time series of wind stress from South West Rocks (SWR), Port Mac­ quarie (PMQ) and Taree (TRE) are shown in Figure 3.5. The maximum wind stress occurs around 13 December in a southward direction, which is driven by northerly/upwelling favourable winds. The maximum northward wind stress (down­ welling favourable) occurs from 19- 20 November. The only other time when winds CHAPTER 3. EXPERIMENTAL OBSERVATIONS 38 were from an upwelling favourable direction was from 8 - 9 January 1999, however the maximum stress was 0.07 Pa which is not considered to be very strong.

0.15 ....-----r---....-----r----.--....--..----r---.--r------r---..----.---.---..-----,----,

0.075 E1 w R C E2

-0.075 SWR -0.15 ,______._ __..__ _ __.___,__..__..____._ _ __,__..__ _ __,_ __.._____.___,_ __..,______._ _ _,

0.15 ....------,---,------,----,--.---..-----.-----,-..------,---..------.-----.----.------.---,

0.075 <'t- E ~.,. -0.075 PMQ -0.15------~---~---~

0.15 ....-----r---....------,----,--....--..-----,----,-,------,------.----,----.------.---,

0.075

-0.075 TRE -0.15 ,______._ __..__ _ __.___,__..__~__.__ __,__..__ _ __._ __~___.____._ __..,______._ _ _, 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 3.5: Filtered wind stress (r) in Pascals as calculated from wind velocities measured at South West Rocks (SWR), Port Macquarie (PMQ) and Taree (TRE). The direction convention is such that a positive wind stress corresponds to equatorward wind forcing. The location of each anemometer is shown in Figure 2.1.

3.4 Sea Level Elevation

The temporal evolution of sea level elevation anomalies are shown for five locations along the NSW coast in Figure 3.6, from 14 November 1998 - 30 January 1999. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 39

The locations are Coffs Harbour (CH, 30°17'8, 153°08'E), Port Macquarie (PMQ, 31 °25'8, 152°54'E), Crowdy Head (CRO, 31 °50'8, 152°45'E), Port Stephens (PST, 32°44'8, 152°12'E) and Sydney (SYD, 33°50'8, 151°15'E).

Adjusted sea level anomalies

C::J ~ V '\T <>v v -v

Nov Dec Jan 1998 1999

F igure 3.6: Sea level anomalies (cm) at Coffs Harbour (CH), Port Macquarie (PMQ), Crowdy Head (CRO), Port Stephens (PST) and Sydney (SYD), on the NSW coast from 14 November 1998- 30 January 1999. Sea levels are adjusted for atmospheric pressure variations, and are lowpass filtered with a 44 hour cut off. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 40

3.4.1 The Inverse Barometer Effect

The oceanic response to fluctuations in atmospheric pressure is essentially a static one and is referred to as the Inverse Barometer Effect (Wunsch and Stammer, 1997). That is, if p:(t) is an atmospheric pressure change measured in millibars, then the oceanic sea level change r,' is

r,'(t) = p:(t) (3.1) Po9 where g is the local gravity in and p0 is the water density. The units are such that a 1 mbar atmospheric pressure fluctuation generates nearly exactly 1 cm of water level change.

As atmospheric pressure can affect the height of the sea level, these variations are used to adjust the sea level anomalies. Variations in atmospheric pressure at each location are lowpass filtered with a 44 hour cut off. and are shown in Figure 3.7. As an example Figure 3.8 shows the time series of sea level anomalies (SL), the atmospheric pressure anomalies (SLP) and the adjusted sea level elevation (SLa) at Port Macquarie. It shows that variations in atmospheric pressure can affect the sea level anomalies by up to 50% (eg 19 December 1998 - 2 January 1999).

Previous studies in the EAC region have focussed on coastal trapped waves (CTWs) and the barotropic response of coastal waters to remote wind forcing (Church et al., 1986a, 1986b; Freeland et al., 1986; Church and Freeland, 1987; Huyer et al., 1988, inter alia). This is supported by the modelling study of Griffin and Middleton (1991) which demonstrated that a significant portion of the barotropic current field on the inner Sydney shelf could be explained by remotely forced CTWs generated to the south of Sydney. In particular, wind events in Bass Strait were shown to produce significant variance in the alongshore currents. Furthermore Griffin and Middleton (1991) showed that the amplitude of the sea level fluctuations that are attributable to the baroclinic mode are small compared to those attributable to the barotropic CHAPTER 3. EXPERIMENTAL OBSERVATIONS 41

Sea level pressure anomalies 10 -..; 5 cu a.. 0 :S a.. ...J en -5 ,_, TRE -10 -SYD

-15 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 3. 7: Sea level pressure anomalies {hPa) at Coffs Harbour {CH), South West Rocks (SWR), Taree {TRE), and Sydney (SYD) during the experiment period. mode. Hence in this analysis it is only possible to examine the barotropic response to CTWs.

The phase speed (c) of a mode 1 CTW scales as c rv f L where f is the Coriolis parameter and L is a length scale which in this case is the width of the continental shelf. In this region f = -7.5 x 10-5s-1 and L rv 20 - 60 km, hence the phase speed lies between 1.5 - 4.5 ms-1. The results of the Australian Coastal Experiment (ACE) showed that CTWs propagate northward through this region with a speed of 3- 5 ms-1 (Church et al., 1986a). The phase speed increases from south to north from 3.2 ms-1 at Cape Howe, to 5.2 ms-1 at Newcastle, as a result of an increase in both stratification and shelf width.

The time series of adjusted sea level elevation from each location are correlated against the adjusted sea level elevation at Sydney. Figure 3.9 shows the contours of equal correlation plotted in a plane of time lag in days and distance north of Sydney. The least squares fit through the maximum correlation coefficients indicates a prob­ able northward propagation rate of 5.51 ms-1 . These observations are slightly faster CHAPTER 3. EXPERIMENTAL OBSERVATIONS 42

40,------.--.------.----.----.-----,---,------,----,----,-----,-----,

20 PMQ

-20 -SL - - SLP D Sla -40L__ _j___ L__ _ _j___ L__ _ _j___ L__ _ _j___ L_ _ _j___ .L....'.::==:::::r:::::::::'.J 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 3.8: Sea level (SL) anomalies (cm), sea level pressure (SLP) anomalies (hPa) and adjusted sea level anomaly (SL0 ) at Port Macquarie for the duration of the Smoky Cape experiment. 3 hourly records have been lowpass filtered with a 44 hour cut off.

than those of Church et al. (1986a), however the ACE experiment was conducted to the south of Smoky Cape, and over a longer time frame, which possibly allowed for a more accurate representation of the phase speed of the CTWs.

3.5 Variability of the Coastal Currents

The variability of the coastal currents is examined for the duration of the exper­ imental period, with particular attention paid to five specific events. The events are summarised as follows: EAC 1: 21 November 1998 - 7 December 1999, Wind event: 7 December 1998-15 December 1998, Reversal event: 25 December 1998- 2 January 1999, Current event: 2 January 1999 - 13 January 1999, EAC 2: 13 Jan­ uary 1999 - 24 January 1999. Each of these events are marked on various figures, specifically Figures 3.3-3.5, and 3.10- 3.13. The uneventful period from 15 - 25 December 1998 is not examined specifically. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 43

SYD CAO

-----o.B 0. 7 ____,.. 0,7 _:__---- ___ 0 .6

-2...____ ...J______J~--...L..------1-U.n=-....L,_ ~__ ___,L ___ ....L,__..-:::::;_----1. .5______. 0 50 100 150 200 250 300 350 400 Distance (km)

Figure 3.9: Contours of correlation in adjusted sea level elevation for various time lags (days) versus distance north of Sydney. Time series from Coffs Harbour (CH), Port Macquarie (PMQ), Crowdy Head (CRO) and Sydney (SYD) are correlated against adjusted sea level elevations at Port Stephens (PST). The * indicates the positions for specific displacements of the local maxima of the cross correlation functions. The solid line is a least squares fit through those points indicating the probable northward propagation rate of 5.51 ms-1 .

The events were chosen for their salient features which will be discussed in more detail throughout this thesis. During the Wind event, strong northerly (upwelling favourable) winds are present, and are followed by a substantial decrease in the bottom temperature across the mooring arrays. The Reversal event is noted for the sudden and strong reversals in the current meter records. Prior to this event the EAC was flowing strongly southward with speeds of up to 1 ms-1 recorded at the Smoky Cape moorings. During the event the current oscillated between rapid northward and weaker southward flow. The third period of .interest, the Current event occurred immediately after the final reversal. It is distinguished by a rapid drop in bottom temperature across both mooring arrays, that is seemingly unrelated to local winds. The bottom temperature begins to rise again at Smoky Cape around 9 January 1999, indicating the end of the Current event.

Finally, at the start and end of the current meter records there are periods where the EAC is flowing swiftly southward adjacent to the coast between Smoky Cape and CHAPTER 3. EXPERIMENTAL OBSERVATIONS 44

~E -0.05·-~f z -0.1 ~: ~fil :tt:t~- 1-85:L ~~ : ~

'i _]~.___ -SC_._A:-~....__---"_· /__,_-~____.-,_6·.,_._.,.__._·-____.--..__- __,_-·__._-_·tt_1~~-/·-__,_-J_·,.·__._~-~_._--r_.~ .._L·,..__- 1_~__._-~~:~

14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 3.10: Time series of the alongshore wind stress (negative is poleward) and the sub-inertial alongshore current velocities at each mooring at Smoky Cape.

Diamond Head. These events are referred to as EA C 1 and EA C 2 and are included as a comparison. Throughout this thesis various analysis is performed in light of each of these events, which are described in more detail in the following sections.

3.5.1 EAC Events

For two periods during the experiment strong southward currents are evident at nearly all the current meters at both mooring arrays. It is during these times that the EAC is flowing adjacent to the coast past Smoky Cape, southward to Diamond Head. Figures 3.10 and 3.11 show the time series of the sub-inertial alongshore current at each mooring. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 45

0-~~ ~E -0-~ z -0.1 t' : ~fil : tr: I~~- 1-E±vr V ~ : ~ 50

DH~ ~Ar <: ,~ -100A :t:t: ·tAvt :t~ j-:1~~

DHA: ~t~ 1~E~: 1~~ ~ 50 DHB: 1~~~~ ,~ -100A ~-~sr13?1½'.

50

DH~ I-~~ ,~ -100A S:cf:tj :j ·:ti 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 3.11: Time series of the alongshore wind stress (negative is poleward) and the sub-inertial alongshore current velocities at each mooring at Diamond Head. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 46

The sub-inertial depth averaged across-shore currents from these moorings are shown in Figures 3.12 and 3.13. The alongshore wind stress Ty is included for completeness.

0.05~ ~E -0.~ z -0.1 ~: ~ J1r:t~- 1-E±-r ~~ : ~

1 SCA: ~=rt:Et::t,: ~~~

21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 3.12: Time series of the alongshore wind stress (negative is poleward) and the sub-inertial across-shore current velocities at each mooring at Smoky Cape.

During the two events marked on the figures as El and E2 the.currents are strongly southward, with velocities of more than 1 ms-1 at SCB and SCC. During EAC 2 these strong southward velocities are also seen at DHB and DHC. That the strong velocities are seen at Diamond Head, where previously they were not, implies that the EAC has encroached upon the shelf, and is now flowing adjacent to the coast.

During EA C 1 the currents at Diamond Head are weaker than those at Smoky Cape which implies that the core of the EAC is further offshore during this event than it CHAPTER 3. EXPERIMENTAL OBSERVATIONS 47

, 0-~~ ~ E -0.05 z -0.1 : ~g : f:tvAt~- 1_£±· L'J ~ : ~

DH~ ~=ti ~ [v t-~~-Er: l-:,~1

DHA: ~'ttttl:ef·r~: 1~~~ 20 , 0~ ~~:r DHB: ~~'t?~1~~: 1~~~ 20 1 0~ ~:r DH~ p~ ~ 1-: f~I: ~1v·=~At~~: l-~1 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 3.13: Time series of the alongshore wind stress (negative is poleward) and the sub-inertial across-shore current velocities at each mooring at Diamond Head. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 48 is during EAC 2.

3.5.2 Wind Event

The Wind event, (7-15 December 1998) is the only time throughout the experimen­ tal period when upwelling favourable winds were persistent (Figure 3.5). From the temperature records (Figure 3.3 and Figure 3.4) we see that these strong northerly winds coincided with a 5 °C drop in sea temperature over a 5 day period at both Smoky Cape and Diamond Head. This significant decrease in bottom tempera­ ture indicates that upwelling has occurred and will be discussed in more detail in Chapters 4 and 5.

3. 5 .3 Reversal Event

The time series of current vectors show that from the 25 December 1998 - 3 Jan­ uary 1999, the flow field across the shelf at both Smoky Cape and Diamond Head oscillates between flowing strongly southward to weakly northward and back again. Figures 3.10 - 3.13 show the reversals at each current meter and at all depths. In fact the current regime shifts from being a strong baroclinic southerly flowing jet to oscillating in a near barotropic manner. During this time, changes in the current direction occurred with a period of,...., 3 days. Figure 3.14 shows the surface currents at Smoky Cape and the fluctuations in sea level elevation at Port Macquarie. These fluctuations are highly coherent with each other.

From Figure 3.15 it is possible to calculate the phase speed of the fluctuations in sea level that propagated northward along the coast, from 28 December 1998 - 3 January 1999. The elevation anomalies appear to move 435 km in,...., 24 hours giving a speed of ,...., 5 ms-1 . Thus it is possible that the changes in sea level elevations CHAPTER 3. EXPERIMENTAL OBSERVATIONS 49

30.-----,-----,,--....,.....-----.------r------r--/~~--~\-.----.----,20 // \ : // \ .. )''· ...... - 0 -~ : 15 ., -~ . _11: \: '"' I 1~-30 ...... II : .. -'~- ... I \, I / : -~ ~Cl) > -60 ..... ; . i._ I_ 5 : 1· I .. I . __ /. I -90 L-....:....:...._LL_ ___----1... ______.______.1. ______::==::::::ic:::==::.Jo 26 28 30 3 Jan 1999

Figure 3.14: Time series of the sub-inertial alongshore surface current velocities (cms-1 ) at Smoky Cape {SCA, SCB) and 'f/, the sea level elevation (cm) at Port Macquarie (PMQ), during the Reversal event. propagating up the coast which coincide with the observed oscillatory barotropic current motion could be attributed to first mode CTWs.

Another conclusion from ACE was that the source of these CTWs was wind stress in Bass Strait. Figure 3.16 shows the mean atmospheric pressure at sea level, and sur­ face wind field for the 23 and 27 December 1998. The charts from the 23 December reveal nothing out of the ordinary, however by December 27 a massive low pressure system has developed in the southern Tasman Sea. The surface wind field shows mean wind speeds in excess of 30 knots (15 ms-1) off the coast of Victoria. The local wind observations, however, (Figure 3.5) show no hint of this massive wind event, that was eventually to claim the lives of six yachtsmen in the 1998 Sydney to Hobart Yacht Race. It is likely however, that this strong wind event in the southern Tasman Sea was the source of the CTWs that reversed the current flow in the study region some 1000 km to the north.

The satellite images of sea surface temperature (SST) in Figure 3.17 clearly show a warm jet flowing southward adjacent to the continental shelf to about 31 °S. The left image from 27 December 1998 shows the current separating from the coast CHAPTER 3. EXPERIMENTAL OBSERVATIONS 50

Adjusted sea level anomalies

20 E ~ 0F"---.;;:= :;,---...c::::------___::=--=---,f---~ $CC"' -20 17 19 21 23 25 27 29 31 2 4 6 8 Dec Jan 1998 1999

Figure 3.15: Sea level anomalies (cm) at Coffs Harbour (CH), Port Macquarie (PMQ), Crowdy Head (CRO), Port Stephens (PST) and Sydney (SYD), on the NSW coast from 17 December 1998- 9 January 1999. Sea levels are adjusted for atmospheric pressure variations, and are lowpass filtered with a 44 hour cut off. @ 12S 125 ::i:,.:

ISS ISS ~

,.. IIS ~ ~ 21S 21S ~ 245 245

27$ m ~

-335 -335 i 3IS 3IS ~

JIS SIS ~ 425 42S ~ t:"-1 211 23:002 23 DEC 1998 23:002 23 DEC 1998 GrADS:CXU/HEi ..... , COU/lGD

125 125 i ~ ISS ISS ~

IIS IIS ~ 0 21S 21S ~ 245 245

27$ 27S

-335 -335 ses

HS ... : . . . : .•.• , •· -·. ·1"J. , -.!;o~·-, " l':,11 ,,,,, ... : , ,. -- ...... 42S .,..~.'-.,1'.~;.l'.A . .A~.---+-· :~-.1.

:;:~ '5f1'.:0 517:E,-« 12K« 91.10£ 135[ •;J;JIl t&'t tiE, ,,__ f:_Kl __(_K • __:f' \ I __:J 4Sf1.. 115£ .~ 125£ I;.. ,... ,... I- To'" 23:002 26 DEC 1998 23:002 26 DEC 1998 -·- ...... COIA/IIID 01 I-' Figure 3.16: Mean Sea level Pressure (left) and surface wind field (right) for the 23 (top) and 27 (bottom) December 1998. CHAPTER 3. EXPERIAIENTAL OBSERVATIONS 52

between Smoky Cape and Diamond Head. flowing southward virtually along t he 153 °E parallel for 150 km. where upon it turns 90 ° and heads eastward. before meandering southward again. with the main jct more t han 300 km from the coast.

a b

28 30 S

27 3 1 S

26 32 S G' '<.... f-

25

33 S

24

34 S

23

35 S 151 E 152 E 153 E 154 E 151 E 152 E 153 E 154 E

Figure 3.17: Satellite images of sea surface temperature (° C) on 27 December 1998 (a) and 7 .Jant1 ary 1999 (b). Arrow, indicate the po, ition of the EAC separation point.

Figure 3.17a shows small coastal vortices in Stockton Bight immediately south of Port Stephens. A week later the SST image in Figure 3.17b shows that t hese fluctu­ ations have moved northward. It is hypothesised that these vortices are the result of interaction between a barotropic CTW pulse propagating northward with a south­ ward offshore current. The local current shear between the opposing flows may reatc instabilities resulting in such vorticities.

Freeland et al. (19 6) relate CT\Vs to the occurrence of upwclling. They state that CT\\'s can play a role in either suppressing or causing upwelling in coastal regions. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 53

It is interesting to note that there is no surface temperature signature associated with this current reversal, however from the time series of temperature, (Figure 3.3 and 3.4) we see that there are oscillations in the sub-surface temperatures across both mooring arrays. The oscillations are most pronounced in the coastal waters, in particular inshore at DHAA. It is recognised that second mode CTWs may cause noticeable oscillations in the thermocline (Griffin and Middleton, 1991).

3.5.4 Current Event

The Current event occurred from 2 - 15 January 1999. At the start of this period the alongshore currents were weak and northerly across both the northern and southern sections (Figures 3.10 and 3.11). Mid-shelf at Smoky Cape (SCB) the northerly flow was at speeds of up to 50 cms-1. This period is distinguished by a distinct change in current direction and a strengthening of the velocity that occurred throughout the 2 weeks. By 15 January the flow was again southward with a mean speed of 1 ms-1 in the surface waters at SCB. At Smoky Cape across-shore flow changed from offshore or almost negligible to onshore with a speed of up to 40 cms-1 in the surface waters (Figure 3.12).

What characterises this event is a significant decrease in bottom temperature and the distinct lack of wind forcing. The bottom temperature across the Smoky Cape array decreases more than 5 °C throughout the period, whilst the surface temperature shows little change. At Diamond Head the temperature decrease is seen initially in the bottom boundary layer, but by 9 January even the surface waters at DHA and DHB are considerably colder than at the start of the event. Figure 3.5 shows that from 1 - 8 January the alongshore wind stress was weak, building to a peak around 9 January. This peak however was a maximum of 0.07 Pa, and of less than 24 hours duration. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 54

The SST image from 7 January 1999 (Figure 3.17b) shows that the EAC separation point has moved further to the south. The eastward bend is less pronounced, and the turning point has been advected more than 150 km to the south. Locally at Diamond Head this has the effect of the current encroaching upon the coast with stronger velocities, which are evident in the velocity fields in Figure 3.11. The SST image (Figure 3.17b) shows colder water hugging the coast at Diamond Head and in the vicinity of Port Stephens. It is the process driving this cold water that is to be investigated in Chapter 5.

3.5.5 Mean Current Field

The spatial structure of the temporal mean current velocities are shown for six different periods at each of the moorings at Smoky Cape and Diamond Head in Figure 3.18. Each panel represents the mean current over a different time period, these being the total period, and each of the five events: EAC 1, Wind event, Reversal event, Current event, and EA C 2.

For the total period, the mean meridional flow at Smoky Cape is aligned with the isobaths to within 5 degrees, and decreases in strength with depth (Figure 3.18a). The only exception to this is at the mid-shelf mooring, where the surface waters are directed slightly more onshore. At Diamond Head fluctuations in the southerly current occur over the inner shelf throughout the experiment. The mean current over the inner shelf is very small in magnitude and directed equatorward, and is decidedly more complex than that at Smoky Cape.

Inshore at Diamond Head the mean flow is northward or negligible through every event. It is more variable in strength and sheared in direction. These results are consistent with satellite images which show that Smoky Cape is directly influenced by the warm EAC waters for the duration of the experiment, whereas generally the Total EAC 1 Wind @ ::i:,.: 31Si ~ '1:l 0.5ms -1 0.5 ms-1 0.5 ms-1 ~ ~ f.~ ~ l, ~ z/ ~ ~ t( ir ~ / -2om ~ - • 35/50m S2 I -aom t"i

~ Reversal Current EAC2 ~ 31S~ -1 > ~ 0.5ms 0.5 ms-1 0.5 ms-1 ~ ~ 0 I ..... ~ 01 0 ~(- ~~ D 1,?r .._ 32S~ ~ / 1 ~~ I - 153E 153E 153E

Figure 3.18: Spatial structure of the temporal mean of the current velocity (ms- 1) at the Smoky Cape and Diamond Head moorings. Each panel represents a different time period: Total (a), EAC 1 (b), Wind event (c), Reversal (d), Current event (e) and EAC 2 (f). Mooring Cll locations are indicated by the O. The depth of each meter is indicated in the legend and the 150 m isobath is marked by the solid line running Cll adjacent to the coast. Note: To highlight the variability in the across-shore direction, the map aspect ratio is not to scale. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 56 coastal waters at Diamond Head are not.

At Smoky Cape the total mean surface velocities are more than 1 ms-1 mid-shelf, and these velocities decrease both inshore, and with depth. Surface velocities are strongest at Smoky Cape during EAC 1 and EAC 2 (Figure 3.18b,f). During the Wind event velocities are strong and southward at Smoky Cape and weak and fluc­ tuating at Diamond Head (Figure 3.18c). During the Reversal event (Figure 3.18d) the mean surface currents at Smoky Cape are "' 20cms-1 southward, which is almost an order of magnitude smaller than the currents during a strong EAC event (Figure 3.18b,f).

As the current encroaches upon the shelf in the Current event the mean velocities at both Smoky Cape and Diamond Head increase. This is especially evident at the outer mooring at Diamond Head (Figure 3.18e).

At Diamond Head the maximum mean velocity in the surface waters is 40 cms-1, this occurs during EAC 2 as the current moves onshore (Figure 3.18f). Across both arrays the currents have a slight across isobath flow, which is directed onshore at Smoky Cape. South of the separation point the mean flow at the two outer moorings is directed offshore.

Table 3.4 shows the mean and standard deviations of the alongshore and across­ shore currents at each current meter during each event described above. At Smoky Cape the mean southward currents are comparable during both EAC events, with the inshore means being slightly weaker (EAC 1: v = -67 cms-1, EAC 2: v =

-80 cms-1 at SCA15 ) but slightly stronger offshore and at depth, (EAC 1: v = -101 cms-1, EAC 2: v = -76 cms-1 at SCC). During EAC 2 the across-shore currents in the surface waters are up to 25 cms-1, directed onshore. This is an order of magnitude larger than the onshore currents during any other event (Table 3.4). During this event at Diamond Head all the across-shore flow is directed seaward. CHAPTER 3. EXPERIMENTAL OBSERVATIONS 57

The alongshore currents are strongest at DHB and DHC during EAC 2, but fairly similar in magnitude both inshore, and in the across-shore direction.

The temporal means of the alongshore currents are small during the Reversal event and the standard deviations are high, indicating that alongshore oscillations dom­ inate the event. In many instances the standard deviations in both the along and across-shore directions are more than double the mean flow, and this is particularly noticeable during both the Reversal and the Current events. ~ ~ ""Cl ~ ~ t"-1 0 ~ ~ C11 ~ 00

~ ""Cl ~ ;:i::.: c.:> ~ ~ ~

'.:2!

@

current

each

by

V

7.7

5.0

5.5 6.6

3.2 8.4

17.7

11.3 18.0

15.7

15.3 12.4

14.2 21.1

11.1

2

-66.6 -77.0 -30.0

-69.9

-52.3

-79.4

-56.2 -68.5

measured

-102.5

as

u

EAC

1.9

3.4

5.0

4.6 6.8 3.7 3.5 2.1 2.8 4.0 3.5 3.0 0.3 4.9

0.9

1.17

11.9 12.4

-1.3

-2.2

-3.5

-17.9

velocities

V

8.7

10.3 17.8

11.9

21.5 23.9 23.4 21.5

20.6

31.2

33.5

30.6

-4.7

-1.9 -6.5

-43.5 -24.3 -17.4 -26.5 -20.9

-11.0 -30.5 -41.8 -18.4 -19.4 -22.1

current

u

Current

2.3

1.6

2.0 2.4 2.4

8.4

5.5 9.6

4.2 8.1 5.9 4.4

8.2

4.9

(u)

-8.8 -3.7 -9.5

-2.6 -1.5

-2.7 -1.3 -0.3 -1.2

-10.2

V

7.5 2.9

2.2 2.9

9.8

0.1

4.6

8.5

19.7 14.4

10.3 29.4 14.1 28.5

25.6

31.2 24.6

35.6

instrument.

-4.6

-4.0

-0.9

-14.6

-29.4

the

across-shore

u

of

Reversal

2.6 1.3

5.2 7.6 2.4

3.1 0.1 3.5

5.6 4.5 4.4 5.1

9.0

13.3

10.9

and

-3.3

-4.0

-1.9

-4.9 -4.8 -2.5 -1.4 -0.5 -0.3

-13.4

(v)

depth

V

3.6 5.2 8.0

0.4 4.8

16.6

16.0 the

16.5 19.6 12.2 20.9 26.5

19.2 23.7

26.9

-74.5

-55.7

-42.8

-66.7 -56.1

-43.9

-25.0

-15.5

-14.7

u

Wind

alongshore

1.6

2.6

1.1

5.9 2.8 5.8 4.8 5.3 5.4

3.2 3.7

3.4

0.9

0.9

8.7 0.5

-3.0

-2.4 -8.8

-1.3

-2.1

indicates

-0.03

-16.4

-17.6

the

of

V

7.9

5.8

0.6

6.5

8.8

(a)

11.1

12.6 16.8

12.6

11.3 10.5

19.5

14.0 21.7

1

-66.9

-46.2

-71.5

-53.2 -25.9

-17.4

-41.6

subscript

-106.4

-100.9

u

EAC

2.8

2.8 7.7 2.5 3.0

3.3 3.9 2.5 2.0 3.4

9.6 5.2

3.1 4.5 0.1 3.1 0.9 6.2

4.4

deviation

-2.2

-0.9

-0.2

-2.1

numeral

-25.8

-12.4

a x

a

a x

a x x x x a x The x x a x a x a

a x a

a

a

standard

events.

and

5

(x)

SCA15

SCB20

SCBso

DHA20

DHA3s

DHB20

DHBso

SCBso DHAA1s DHCso SCA3s

SCCso

the

of

Mean

each

3.4:

for

meter

Table CHAPTER 3. EXPERIMENTAL OBSERVATIONS 59

3.5.6 Mean Temperature Field

The mean temperatures at each thermistor during each event are shown as a function of depth in Figure 3.19. At SCA, DRAA, DHA, (i.e. inshore) the temperature range between events is high, particularly in the BBL. The coldest mean temperatures at depth occurred during the Current event, and the warmest mean temperatures were during the EAC 1 and EAC 2. During the Current event the surface temperatures at Smoky Cape (SCA,SCB) remained warm Trv 24 °C, (which was also the total mean temperature), however in the BBL, temperatures at Smoky Cape were 3 °C below the mean, and 5 °C less than the mean temperatures during EAC 2. At Diamond Head the mean temperatures in the surface waters dropped during the Current event, Trv 19 °C, which is 2 °C below the total mean temperature, and up to 4 °C below the mean during EA C 2 and the Reversal event. Offshore at SCC and DHC the warmest mean temperatures at depth(> 20 °Cat 80 m) occur during EAC 1 and EAC 2.

The temperatures in the BBL during the Wind event are consistently 2-3 °C warmer than those during the Current event. This is despite the fact that the wind event occurred early in the summer season, and the Current event occurred at the height of summer. Of note is the fact that the surface waters at SCA cooled considerably during the wind event, whereas during the Current event the surface waters were not affected.

3.6 Summary

The observations presented in this chapter were partitioned into specific events, these being 2 EA C events, a Wind event, a Reversal event and a Current event. The current regime observed across the Smoky Cape and Diamond Head mooring

0 0

O') O')

~ ~

~ ~

~ ~

~ ~

0 0

to to

Cl) Cl)

t-t t-t

0 0

~ ~

~ ~

~ ~

~ ~

~ ~

~ ~ --0 --0

~ ~

;:i:.: ;:i:.:

~ ~

~ ~

@ @

l l

25 25

period period

DHC DHC

sec sec

V V

time time

20 20

~x ~x

(°C) (°C)

o+-x o+-x

total total

T T

Oe Oe

0 0

the the

D D

•01 •01

for for

15 15

~x ~x

x+~ x+~

l l

l l

: :

moorings moorings

V V

CIX) CIX)

x• x•

OlEBD OlEBD

V V

DHB DHB

SCB SCB

-t< -t<

Head Head

-tO< -tO<

+IX) +IX)

20 20 25

<:o <:o

X X

X X

Diamond Diamond

T(°C) T(°C)

O,x O,x

-ICJ'vX -ICJ'vX

0 0

and and

~ ~

OWJ OWJ

~ ~

WOO WOO

15 15

Cape Cape

~ ~

~ ~

~ ~

~ ~

25 25

Smoky Smoky

()o ()o

the the

V V

x+ x+

DHA DHA

SCA SCA

at at

~D ~D

X X

-HX) -HX)

20 20

C) C)

X X

series series

(° (°

0 0

+oxv +oxv

~x ~x

T T

+ +

o o

time time

(a';(Z)+ (a';(Z)+

15 15

legend. legend.

0 0

temperature temperature

l l

the the

-60 -60

-40 -40

-20~ -20~

-80 -80

-160 -160

-140~ -140~

-120~ -120~

-100 -100

in in

the the

Q) Q)

a. a.

0 0

-5 -5

......

s s

of of

DHAA DHAA

X X

-evo -evo

marked marked

mean mean

' '

-eix -eix

20 20 25

as as

a:)+-

00 00

1 1

T(°C) T(°C)

0 0

events events

Reversal Reversal

EAC2 EAC2

EAC EAC

Wind Wind

Current Current

Total Total

5 5

Temporal Temporal

X X

D D

+ +

o o

0 0

V V

15 15

the the

of of

3.19: 3.19:

each each

-80 -80

-60 -60

-40~ -40~

-2:1 -2:1

-160 -160

-140~ -140~

-120 -120

-100 -100

Q) Q)

a. a.

0 0

-5 -5

......

s s

and and Figure Figure CHAPTER 3. EXPERIMENTAL OBSERVATIONS 61 arrays reveals some interesting features. For the duration of the experiment the two instrument arrays spanned the separation point of the EAC. This allows for an examination of the dynamics surrounding the separation point which is described in Chapter 4.

Across the northern array currents generally flowed swiftly southward, however there were deviations from this. Barotropic reversals in the strong southward flowing current were seen to result from the northward propagation of coastal trapped waves along the NSW coast. At the southern array, velocity fluctuations are dominant, and at times these are significantly correlated with the alongshore wind stress and at other times the variations in sea level elevation appear to drive a northward counter current.

From the results presented it is evident that different processes affect the tempera­ ture throughout the water column in different ways. Throughout the summer, two significant decreases in bottom temperature were observed. The first decrease coin­ cided with upwelling favourable winds ( Wind event), and the second was associated with the encroachment of the axis of the EAC upon the continental shelf ( Current event). Of note is the fact that the coldest mean bottom temperatures occurred during the current event. These two upwelling events are the subject of further in­ vestigation described in Chapter 5. The analyses presented in the following chapters are described with due regard to the specific events outlined in this chapter. Illimitable ocean, without bound, without dimension, where length, breadth and highth, and time and place are lost.

From Paradise Lost by John Milton

Variability about the EAC separation

It is well recognised that the EAC tends to separate from the coast in the Smoky Cape region. From historical records and evidence of bottom sediments Godfrey et al. (1980b) showed that the EAC typically separated from the coast around Sugar Loaf Point (32°30'8). Cresswell et al. (1983) found that warm southward flowing wa­ ter separated from the coast at Point Plomer and veered eastward into the Tasman Sea. In each instance a rapid change in current speed was associated with the sepa­ ration and inshore of the southward flowing jet there was often a cooler northward coastal current. As the current separates a marked temperature front of 2 - 5 °C is observed.

62 CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 63

From the current meter time series presented in Chapter 3 we see that the currents are highly variable, in both strength and direction, and that these variations tend to dominate the coastal environment. Furthermore it is evident that there are a number of mechanisms which could possibly explain the observed variability. The evidence presented in Chapter 3 showed that CTWs propagate through the region causing reversals in the current. The SST images and the time series of currents show that the separation of the EAC from the coast in this region has a profound affect on the coastal waters and occasionally local wind forcing plays a role. A modelling study by Marchesiello and Middleton (2000) shows that even with constant inflow into the domain, instabilities occur in the EAC system.

This chapter examines the dynamics about the separation point of the EAC with a view to explaining the inherent variability observed in the temperature and current records during the experimental period. It is the intention to better understand the dynamical balances that prevail north and south of the separation point under different conditions. Although several aspects of the data to be unveiled by the fol­ lowing analysis are very interesting and open many possibilities for further research, this chapter will emphasise those aspects relevant to the main theme of the present study. It is not the intention to explain why the current separates, rather to look at the dynamics about the separation point and explain the effects of separation on the inshore coastal waters.

These aims are addressed by examining the temporal evolution of both velocity and temperature during the experimental period, as well as for each of the main events as described in Chapter 3. The salient features of sub-inertial frequency fluctua­ tions of current, temperature, sea level and wind stress observed north and south of the EAC separation are interpreted in both the horizontal and vertical. Firstly, a short analysis of the current meter time series in the frequency domain is presented, then Empirical Orthogonal Functions (EOFs) are used to statistically examine the CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 64 temporal and spatial variability north and south of the separation point. The depth averaged currents are examined and are used to delineate the term balances the alongshore momentum equations. The terms, which are calculated from the obser­ vations north and south of the separation point, are examined both physically and statistically to determine the dominant process balances in the system.

4.1 Temporal Variability

To examine the variability north and south of the EAC separation point, an analysis of the current meter time series is conducted in the frequency domain, as well as in the time domain.

4.1.1 Frequency Analysis

Power Spectra Power spectra of across-shore (u) and alongshore (v) velocities from the raw current meter time series as measured at each mooring are presented in Figure 4.1. The spectra have been averaged over five adjacent frequency bands, giving 10 degrees of freedom. Energy peaks in the power spectra occur at the frequency of both the diurnal and semi-diurnal tides at all moorings and the .energy of these tides is amplified towards the coast. At Diamond Head where the shelf width (30 km) is almost double that at Smoky Cape, there is more tidal energy at the inshore moorings (DHAA15 , DHA20 ). However at 35 m depth above the 50 m isobath there is more energy at Smoky Cape than at Diamond Head at both the diurnal and semi-diurnal frequencies. At the semi-diurnal frequency the across-shore component dominates the spectra. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 65

104

~c. et:_ 102 -'!2 I ai :;: 0 a. 100

SCA1sm SCA35m SCB2om 104

g u

~c. -!2 102 ~1 ai 95% ~ + 0 a. 100

SCBsom SCBBOm SCCBOm

104

~c. N.,!2 102 ~ I ai :;: ~ 0 a. 100

DHAA15m DHA2om DHA35m 104

g u

~c. -!2 102 ~ -'!2 I 7 ai :;: + 95% + 95% 0 a. 100

DHB20m DHBsom DHCBOm

10-2 10-1 100 10-2 10-1 100 10-2 10- 1 100 Frequency (cpd)

Figure 4.1: Power spectrum of the alongshore and across-shore currents at Smoky Cape and Diamond Head averaged over 5 adjacent frequency bands. Each plot is labelled by mooring, where the subscript refers to the depth of the instrument of each particular mooring. Refer to Figure 2.3 for the schematic of the mooring arrays. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 66

At low frequencies the energy in the alongshore direction is consistently two orders of magnitude larger than that in the across-shore direction, both inshore and offshore. Mid-shelf over the 100 m isobath, at SCB and DHB across-shore energy is however only one order of magnitude smaller than the energy in the alongshore direction. From the current meter time series the across-shore currents are strongest at the mid-shelf moorings and are generally directed onshore at Smoky Cape and offshore at Diamond Head. This is possibly due to a funnelling effect resulting from the narrowing of the shelf at Smoky Cape and then the eastward flow of the current after it has separated from the coast.

At the most easterly moorings (SCC, DHC) which are located at the 150 m isobath the low frequency fluctuations in the EAC signature are revealed and again the alongshore component dominates the spectra. There is still evidence of the semi­ diurnal tide in both the along and across-shore components of the velocity.

Cross Spectra To further examine the coherence of the current in the alongshore direction, cross spectral analysis is performed on the detrended sub-inertial time series of the depth averaged currents (u, v) along the 50, 100 and 150 m isobaths. The squared coher­ ence, phase and gain for each location are shown in Figure 4.2. From the 70 day record it is only feasible to examine variability with periods of up to about 10 days. The comparison across the shelf is of interest. Along the 50 m isobath, significant coherence peaks occur at periods of between 2.5 and 5 days and are 90° of out phase, with the southern array leading the northern array by between 0.8 and 1.25 days. The magnitude of the low frequency fluctuations at the southern array are half those at the northern array as indicated by the gain. Hence inshore reversals in the current lead from the south and make their way north and it takes longer to reverse the stronger faster flowing currents across the northern array. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 67

0.8 VSOm

N 0.6 .s::. 0 O 0.4

0.2

1a8

90 0 Q) Illea 0 .s::. a. -90 -18q

0.8

c: 0.6 "iii c, 0.4

0.2

0

0.8 u50m

N 0.6 .s::. 0 o 0.4

0.2

1a8

90 0 CD Illea 0 .s::. a. -90

-1aq

0.8

c: 0.6 "iii c, 0.4

0.2

0 10-1 Frequency (cpd)

Figure 4.2: Coherence, Phase and Gain between the depth averaged currents at Smoky Cape and Diamond Head, along the 50, 100 and 150 m isobaths, averaged over 10 adjacent frequency bands. The 95% confidence limit is indicated by the dashed line on the coherence plots and is calculated based on the effective number of degrees of freedom. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 68

Along the 100 m isobath, the currents remain out of phase in the 2 - 5 day period band, with the southern array leading the northern array, however the lag is less than that observed along the 50 m isobath, being only 45° out of phase. The gain has increased, indicating that the mid-shelf currents at DHB are similar in magnitude to those in the north. At the outer mooring there is no indication of the current reversal with no significant coherence in the 3 - 5 day period band. At lower frequencies however the coherence is again significant, with a slightly positive phase lag and a higher gain. This indicates the low frequency fluctuations in the EAC which occur when the axis of the current has moved onshore and the effects of the strong jet are felt at depth across the southern array. Throughout the experiment period the EAC generally separated from the coast between the two arrays which means that the EAC signature is not often seen at DHB and often the currents are northward at DHB, when they are southward at SCB.

There is little coherence in the across-shore component of the currents. At low frequencies the most significant coherence is between the mid-shelf moorings, where the northern array leads the southern array by 120 - 180°, again indicating the onshore and offshore movement of the axis of the current. At the offshore mooring the coherence between the across-shore component of the current occurs at periods of less than 10 days which indicates the large scale strengthening of the EAC as it re-attaches itself to the coast.

An analysis of the World Ocean Circulation Experiment (WOCE) PCM3 array along 30°S by Mata (2000) revealed that the EAC tended to dominate the flow in this region, especially at the shelf break. Furthermore they found that the main EAC signal was modulated by alternating strengthening and weakening events and at times even reversals. The main low frequency oscillations were found to be in the upper meso-scale band with periods of 40 - 70 days. This coincides with our findings that show the EAC is attached to the coast at both the start and end of CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 69 the current meter deployment however, for a 50 day period in the middle the main EAC flow deviates from the coast somewhere between Smoky Cape and Diamond Head.

4.1.2 Time Series Correlations

Cross correlations were calculated between the currents at different depths to ex­ amine how the along and across-shore currents co-varied within the different events. In the alongshore direction the currents were positively correlated with the upper current leading the lower current. Currents in the alongshore direction were gener­ ally more correlated at Smoky Cape than those at Diamond Head, with correlation coefficients of greater than 0.9 having more than 95% confidence.

Figure 4.3 shows the correlations between the across-shore currents at different depths, mid-shelf at Smoky Cape (SCB) during the Wind event, the Reversal event, the Current event and EA C 2. During the Wind event the across-shore current at 20 m depth leads that at 50 m by 38 hours, the current at 50 m lags the current at 80 m by nearly 23 hours and the surface current is not significantly correlated with the bottom current at all.

During all the other events the surface current actually lags the current at 50 m by up to 14 hours. During the Reversal event the fluctuations in the current are barotropic and the across-shore currents are correlated at all depths, with the deepest current leading the reversal. During the Current event the only significant correlation occurs between the surface current and that at 50 m, where the deeper current leads the surface current by 15 hours

The correlations between the surface waters (20 m and 50 m) mid-shelf in both the along and across-shore directions are shown in Figure 4.4 for both Smoky Cape CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 70

Wind Reversal

0.5 0.5 ., ~ ·, : 0 0 .\.. I

-0.5 -0.5 1 ·

-1 -1 -4 -2 0 2 4 -4 -2 0 2 4

Current EAC - u20:50 ·-· u 50:80 - - u 20:80

0.5

0 0

-0.5

-1 -1 -4 -2 0 2 4 -4 -2 0 2 4 Lag (Days) Lag (Days)

Figure 4.3: Cross correlations between the across-shore currents at SCB during the Wind event, the Reversal event, the Current event and EAC 2. Reversal Current

-1 '----~--~------' -1 -4 -2 0 2 4 -4'----~-----e------' -2 0 2 4

0.5

O·

-1 '----~--~--~-~ -4 -2 0 2 4 -2 0 2 4 Lag (Days) Lag (Days)

Figure 4.4: Cross correlations between the along and across-shore currents at SCB and DHB during the Reversal event and the Current event. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 71

(SCB) and Diamond Head (DHB). Alongshore currents are highly correlated at the northern array with no lag and although they are also correlated at the southern array, the correlations at Diamond Head are slightly lower. At Diamond Head the across-shore currents are correlated with 95% certainty during the Reversal event, however they are not significantly correlated during the Current event. During the Reversal event at the southern array, the subsurface currents lead in the alongshore direction, but lag in the across-shore direction. Similarly at Smoky Cape there is no lag in the alongshore direction, although the across-shore current at 50 m leads that at 20 m. During the Current event there is no lag at DHB, but further to the north the subsurface current leads the surface waters in the alongshore direction.

4.2 Spatial Variability

Empirical Orthogonal Function analysis is a statistical technique which is useful in explaining variability within a spatially or temporally distributed data set. In the past Empirical Orthogonal Functions (EOFs) have been used for many oceano­ graphic applications, primarily, the decomposition of temperature and velocity fields, (Kundu et al., 1975; Lippert and Briscoe, 1990, inter alia) and the examination of the spatial structure of sea level pressure and sea surface temperature (Davis, 1976). More recently EOF analysis has been used in data assimilation techniques to assess the performance of a numerical model (Oke et al., 2000).

Lippert and Briscoe (1990) used EOFs to explain the low frequency variability in the Gulf Stream recirculation. They examined the vertical structure of the variability using various types of EOFs and found that the most suitable was an EOF represent­ ing a flow that was unidirectional and non rotating with depth. Gulf Stream eddy variability was also studied by Owens (1985) who compared different EOF forms to glean information about both the vertical and horizontal variability. Owens weighted CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 72 the EOFs with the depth intervals between instruments, giving an estimate of the depth integrated energy.

A number of different EOF methods may be used to study the statistical variability of a data set. These are outlined in Davis (1976) and compared in Lippert and Briscoe (1990). Here we chose two forms of the EOF to examine the sub inertial current fluctuations in the vertical and one form for the horizontal investigation. In the simplest case, the matrix of covariance coefficients of a scalar representation of the velocity in the time domain is formed where U = v'u2 + v2 , and the resulting eigenvalues and eigenvectors are found. The second form of empirical mode calcu­ lations were for a covariance matrix in which the velocity components (u and v) represent independent scalars, which reveal directional changes.

The sub inertial time series were detrended and then weighted so as to be able to give physical meaning to the statistical analysis. As the instruments are unevenly distributed in space, the velocities are weighted according to the horizontal and vertical distance to the other instruments following the method of Denbo and Allen (1984). For the vertical data set, each weighted velocity was multiplied by the square root of the depth interval over which that measurement is representative (Bi)- Furthermore the eigenvectors were normalised by the standard deviation of the time series and rotated to align with the principal current axes (Emery and Thomson, 1998).

It is important however to remember that being a statistical technique, EOFs are non-dimensional and thus there is not necessarily a direct physical or mathematical relationship between the statistical EOFs. Consequentially, as was pointed out in Kundu et al. (1975) absolute direction can not be determined from a mode of an EOF. The only directional information we can get from vertical EOFs is the relative change in direction with depth. Whilst dynamical modes conform to physical CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 73 constraints through the governing equations and associated boundary conditions, EOFs are simply a method for partitioning the variance within a group of concurrent time series (Emery and Thomson, 1998). For a thorough derivation of the technique, the reader is referred to Kundu et al. (1975).

For the purposes of this exercise, only four moorings are considered, SCA5o, SCB100 ,

DHA50 and DHB95. For SCB100 all three current meters are used at depths of 20, 50 and 80 m, whilst for the remaining moorings, only 2 current meters are used, 20 m and 50 m.

4.2.1 Vertical Structure

From the eigenvectors calculated from scalar representation of the velocity field at Smoky Cape and Diamond Head (Figure 4.5) we see an alongshore barotropic current with very little shear. The first mode accounts for more than 80% of the variability in all four moorings and shows a strong barotropic nature. The second mode representing a maximum of 16% of the variability at DHB, shows a change of sign with depth in all cases indicating vertical shear. Of more interest however are the depth weighted EOFs calculated from the individual components of velocity as shown in Figures 4.6 and 4. 7 for Smoky Cape and Diamond Head respectively. These figures show the eigenvectors calculated by taking the alongshore and across-shore components of velocity as independent variables.

For both locations we see that the first eigenvector accounts for more than 60% of the variance in the across-shore velocity components, increasing to more than 90% in the alongshore components. The alongshore velocity shows a strong barotropic com­ ponent indicating the dominance of the meridionally flowing EAC, even in shallow waters within 2km of the coast. The across-shore velocity components show slightly more variability with an observed current shear increasing with depth. In the case CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 74

o...... --~--~--~---, 0

-20 -20 ~ ,,. ,,. ,,. ,,. I -4o -40 ,,. ~,, t ,, ,, ~ -60 -60 ,, / <, -80 -80 -81% SCA IVI 1---_ :~7° I SCBIVI -- 15% -100L-~__ _;__---=::::::::==:::::'.J -100 -1 -0.5 0 0.5 -1 -0.5 0 0.5

o...... -----~--~-~ 0...... ------~-~ -20 : --~ -20 : ~ '~\------~ ---- I -4o -40 l .c --- : 15. ~ -60 -60

-80 -80 -86%1 -84%1 DHAIVI 1-- 14% DHBIVI 1-- 16% -100~-~--~--~-~ -100~-~--~--~-~ -1 -0.5 0 0.5 -1 -0.5 0 0.5

Figure 4.5: Scalar EOFs for both the Smoky Cape and the Diamond Head Moorings. Mode 1 (solid) and mode 2 (dashed) The percentage of variance explained by each mode is indicated in the legend for each plot. of DHA in the first mode (which accounts for 60% of the observed variability) we see a distinct surface intensification and a current reversal below 30 m depth.

For all the cases, (with the exception of the across-shore current at DHA) the second mode depicts a change in current direction with depth and is reminiscent of the first baroclinic theoretical mode. ( One must remember that derivation of the theoretical modes is based on a series of assumptions such as the flat bottom criterion, which do not necessarily hold true here.) Physically this can be interpreted as an onshore flow balanced by an offshore flow. Typically the variance explained by the second mode of variability is between 2 - 8% for the along shore component (v), which is not considered statistically significant. For the across-shore component (u) the CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 75

0 0

- --0 -20 -- --0 -20 ~ --~-. - l- _L-- -- ~ -- : :[ -40 -40 .J:. 15. 2l -60 -60

-80 -80 1-62%1 1-98%1 SCAu -- 38% SCAv -- 2% -100 -100 -1 -0.5 0 0.5 -1 -0.5 0 0.5

0 0 -68% 1-90%1 -- 23% -- 8% -20 ~ -20 \ \ :[ -40 -40 \ ~ =Q. &. Q) " 0 -60 -- ': ..... -60 : ------.... ------80 --0 -80 --~ SCBu SCBv -100 -100 -1 -0.5 0 0.5 -1 -0.5 0 0.5

Figure 4.6: Vertical EOFs for the Smoky Cape moorings. Mode 1 (solid) and mode 2 (dashed). The percentage of variance explained by each mode is indicated in the legend for each plot. 0 0

_,-0 -20 . -20 l- - - :-:-- - : () . -- :[ -40 ~ -40 t Q) 0 -60 -60

-80 -80 1-60%1 1-94%1 DHAu -- 40% DHAv -- 6% -100 -100 -1 -0.5 0 0.5 -1 -0.5 0 0.5

0 0

-20 - --0 -20 - --0 ..... - . - :[ -40 --. -40 --··-. -.,,,,. - : --- : =Q. l -- l -- 2l -60 -60

-80 -80 1-n%1 DHBu -- 23% DHBv I=-=:~~ I -100 -100 -1 -0.5 0 0.5 -1 -0.5 0 0.5

Figure 4. 7: Vertical EOFs for the Diamond Head moorings. Mode 1 (solid) and mode 2 (dashed). The percentage of variance explained by each mode is indicated in the legend for each plot. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 76 variance explained by the second mode is between 20 - 40%.

In the case of the two inshore moorings (SCA and DHA) the depth of the zero crossing of the baroclinic mode lies between 20 and 30 m towards the base of the mixed layer. Further offshore at the mid-shelf moorings this zero crossing deepens

( rv 50 mat SCB and rv 35 mat DHB) indicating a current reversal occurring further towards the bottom. At both locations the percentage explained by the second mode in the across-shore direction decreases with distance offshore. This implies that as the water column is influenced more by the EAC, the waters become more baroclinic and the variance in the across-shore direction increases.

4.2.2 Horizontal Structure

The horizontal structure of the low frequency variability is described by the EOFs computed from the complex representation of the 36 hour lowpass filtered velocity time series as measured at each mooring. In Figure 4.8 and 4.9 the spatial map is presented of both the first and second mode eigenvectors (respectively) for the 20 m depth layer and the 35 - 50 m depth layer, for the total period as well as the 5 events. The flow in a western boundary current is generally assumed to be temporally steady, hence the data presented here, which is separated horizontally is not lagged in time. Furthermore south of the separation point, where the flow is spatially variable, a temporal lag is meaningless.

Denbo and Allen (1984) recommend a significance criterion whereby an EOF mode is assumed to be significant if at least 50% of the total variance is explained by mode 1, or 40% by mode 2. Greater than 91 % of the variance is accounted for at Smoky Cape by the first mode eigenfunctions, however mode 1 accounts for only 59% at Diamond Head. Using the criterion of Denbo and Allen (1984) the second modes are found to be statistically insignificant at both locations, but are included here for

~ ~

~ ~

--.:i --.:i

'< '<

--.:i --.:i

~ ~ trl trl

~ ~

0 0

(f) (f)

~ ~ trl trl 0 0

~ ~

tJ:, tJ:,

~ ~ c:: c::

0 0

~ ~

>-< >-<

~ ~

;:i:.. ;:i:..

~ ~

~ ~

,1::,.. ,1::,.. ~ ~

~ ~

tJ:, tJ:,

~ ~

'"-I '"-I "U "U

0 0

I

e e

or or

h

t

35 35

· ·

is is

,-.::.

I I

/ /

<> <>

J J

I I

either either

t t

I I

isobath isobath

f f

00 00

(-, (-,

s s

m m

and and

~

f f

vJr

50 50

. .

id) id)

1

l

I

sc

/' /'

153E 153E

(so

~ ~

The The

.. ..

m m

: :

~ ~

',f ',f

% %

% %

,;

20 20

58

53% 53%

61

52(;.c 52(;.c

array. array.

of of

DH DH

h h

s s

h

-

-

eac

EAC2 EAC2

Wind Wind

ept

to to

d

-

-

-

-

at at

ext ext

I I I I

n

es, es,

series series

ur

g

/ /

' '

fi

-' -'

~ ~

,,,, ,,,,

/ " "

time time

(~ (~

the the

J J

\ \

eter eter

j j

~ ~

in in

! !

I I

f!Jj; f!Jj;

., .,

(_, (_,

s s

m

d d

.Ii

J J

( (

ent ent

cate

I I

( (

( (

~ ~

~f ~f

~0 ~0

urr

s s

s~~

c

indi

153E 153E

\ \

s s

i

. .

the the

e e

~ ~

% %

' '

of of

68% 68%

49% 49%

51

mod

1 1

h h

"~ "~

-

eac

EAC EAC

Current Current

by by

nfunction nfunction

--

-60% -60%

::-: ::-:

d d

ge

l l

i

e

e

11 11

I I I

in

I I I

I I I I

Jl

J J

a

l

exp

argest argest

l

,) ,)

,,J ,,J

the the

. .

. .

) )

of of

§ §

variance variance

coast. coast.

e e

~

& &

: :

,.__ ,.__

of of

ur

the the

( (

( (

to to

~1

~f

struct

s s

s0?

nt nt

l l

153E 153E

percentage percentage

o o

1

% %

% %

adjace

~~ ~~

61

47',·o 47',·o

73

66~

The The

Horizonta

OH

Reversal Reversal

Total Total

running running

4.8: 4.8:

e e

-

-

(dashed). (dashed).

lin

s~ s~

s~ s~

m m

31 31

31 31

50 50

Figure Figure solid solid

j j

~ ~

~ ~

~ ~

---J ---J

~ ~

~ ~

;:i::J ;:i::J

~ ~

(X) (X)

tr:l tr:l

f---3 f---3

0 0 \J} \J}

0 0

~ ~

~ ~

to to c:::: c::::

~ ~

0 0 f---3 f---3

~ ~

'"O '"O

'"< '"<

>---; >---;

~ ~

SS SS

~ ~

;:i::J ;:i::J to to

;:i::J ;:i::J

R R

h h

er er

h

at

t

b

i

e

so

i

d d

I I

,,,, ,,,,

~ ~

) )

t":> t":>

i i

m m

an

J J

/ /

j j

l l

0 0

I I

" "

I I

I I

ic,, ic,,

?, ?,

5

d) d)

~~ ~~

r r

~0 ~0

'" '"

1

_,/ _,/

li

l) l)

~,~ ~,~

r r

.. ..

e e

~ ~

h

(so

r r

T

/ /

.) .)

. .

0 0 m

ay

153E 153E

2

rr

f f

a

h h

s s o

% %

% %

h

t

eac

"14

21% 21%

p

34

3T0 3T0

e

o o

d

t

OH OH

at at

ext ext

EAC2 EAC2

Wind Wind

n

es es

-

-

ri

es, es,

se

ur

e e

g

m

fi

ti

e e

h

t

eter eter

in in

4" 4"

m

; ;

( (

d d

t t

e

n

0 0

t

e

0 0

I I

l.t) l.t)

I I

ca

-

urr

I I

\ \

~,r.-,'• ~,r.-,'•

indi

< <

,rtr· ,rtr·

. .

i

e e c

s s

h

i

t

e e

i i

f f

y y

sc~1

d

o

,

153E 153E

o

n n

m

o

i

h h

% %

ct

eac

32~,,; 32~,,;

24% 24% 25~:'.') 25~:'.')

23

1 1

y y

nfun

OH OH

ge

i

d d b

e

e

EAC EAC

Current Current

in

a

--

-

--

-

pl

argest argest

l

ex

ce ce

nd nd

n

. .

a

ri

seco

.,, .,,

( (

va

e e

coast

-

J J

f f

h

f f

;,_) ;,_)

( (

e e

t

o

A,-

h

,

f f

t

f f

( (

0

e e o

to to

tage tage

I I

& &

,_ ,_

n

t t

; ;

ur

n

ct

& &

I I

" "

erce

ru

ace

~ ~

p j

st

e e

ad

l l

153E 153E

g g

ta

Th

n

. .

)

% %

% %

zo

d

~ ~

1/; 1/;

ri

e

19% 19%

"16

~10~,0 ~10~,0

26

o

runnin

h

H

e e

(das

lin

9: 9:

.

Reversal Reversal

Total Total

d d

4

m m

li

-

e e

so

50 50

e e

h

or or

s s t

31S 31S

31S 31S

i

Figur 35 35 CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 79 completeness.

In the surface water the first mode accounts for between 47 - 66% of the variability, with the highest contribution being during the barotropic Reversal event and the lowest contribution coming from the total time series. In the middle of the water column, (35 - 50 m depth) the first mode accounts for 51 - 73% of the variability. This larger component at depth indicates that there is high current shear and that the water column is essentially baroclinic throughout every period.

For Smoky Cape, Figure 4.8 shows that the variability in the flow is aligned per­ pendicular to the 150 m isobath, except during EAC 2 when the current encroaches strongly upon the shore. The maximum across-shore variability at Smoky Cape occurs during the EAC 2 when the current encroaches toward the coast. Although the currents tend to co-vary in the same direction, the magnitude of the variance changes with events. The exception is during the current reversal event when the deviations from the mean flow are in a northward direction. The greatest variance is across the southern array as the current meanders on and off the continental shelf, from time to time impinging on the outer mooring. The large across-shore component of the variability in the surface waters reflects this.

The structure of the second eigenfunction varies substantially between events. The second mode reflects the baroclinic nature of the current at each mooring and during each event, particularly at Diamond Head where there is a large across-shore com­ ponent in the surface waters. The lack of any clear pattern indicates the variable nature of the EAC in this region, especially as it reverses or impinges upon the coast. However it is also possible that the small number of degrees of freedom in the shorter record lengths inhibits the ability of the EOF to find statistical independence. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 80 4.3 Depth Averaged Velocity Field

The depth averaged velocity fields are calculated for the mid-shelf and inshore moor­ ings along the Smoky Cape array (SCA, SCB) and the Diamond Head array (DHA, DHB). It is unfortunate that on both arrays the mooring positioned at the shelf break failed at a depth of 50m, as the current meter would have provided vital in­ formation about the surface currents in the centre of the EAC jet. The sole current meter that remained on each mooring (DHCsom and SCCsom) was positioned in the middle of the water column, so they provide a potential proxy for the depth aver­ aged currents at these two locations. The depth averaged velocities (u, ii), (where the overbar denotes the current has been vertically averaged), were rotated so as to minimise the variance in the across-shore direction. The principal angles of ro­ tation were small (Table 4.1) with the largest angle 5.5° being at SCB. Table 4.1 also shows the mean of the depth averaged along-shore currents and the major and minor principal axes.

Mean Principal Axes Site ii ii Major Minor Orientation SCA -2.7 -43.8 26.9 2.9 -0.66 SCB -5.4 -51.3 29.5 5.0 +5.5 DHA 0.9 2.1 13.4 3.7 -0.08 DHB 2.7 -15 23.8 6.6 +2.65

Table 4.1: Temporal mean the depth averaged across-shore (u) and alongshore (v) velocities at SCA, SCB, DHA and DHB in cms-1 . The major and minor axes of the 'principal ellipses are shown as well as the angle of rotation (0 ) clockwise from north.

Inshore at Diamond Head (DHA) the mean depth averaged flow is northward (ii=

2.1 cms-1). At the mid-shelf mooring (DHB) ii = -15 cms-1 which indicates a southward flow. At both moorings the mean across-shore flow is positive (i.e. offshore). At Smoky Cape the across-shore gradient in mean alongshore flow is small, with a mean alongshore flow of -44 cms-1 above the 50 m isobath and CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 81

-51 cms-1 above the 100 m isobath, which is indicative of the swift flowing southerly EAC jet. The mean across-shore flow is negative, indicating a mean onshore flow

(u = -5.4 cms-1 ).

In a two-dimensional environment where the depth averaged current u = 0, the onshore flow in one part of the water column is balanced by the offshore flow in another part of the water column. Figure 4.10 shows the time series of the sub­ inertial depth averaged alongshore v and across-shore u currents at the inshore (SCA, DHA) and mid-shelf moorings (SCB, DHB) from both arrays. For a large part of the time the flow is across the isobaths, i.e. u # 0. However at SCA there are times where the flow is almost barotropic, with velocity deviations of< 0.05 ms-1 in the across-shore direction.

At SCB the depth averaged across-shore flow is predominantly negative (i.e. on­ shore) where as at DHB (the southern mid-shelf array) the across-shore flow is mainly offshore. This indicates a funnelling of the current at Smoky Cape prior to separation and an offshore flow at Diamond Head as the current moves away from the coast.

4.4 Momentum Balance

To theoretically explore the relative importance of the various dynamical processes occurring on the continental shelf, the depth integrated momentum equations are examined. Current measurements obtained from the mooring arrays are used to examine characteristics of the depth integrated momentum balance to assess the relative importance of various terms and hence the role played by each of the pro­ cesses in the balance. The analysis is performed on a sub-inertial time scale in order to examine the low frequency motions occurring on the shelf. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 82

50

V

0

'Ill E u -50 -DHA - - DHB -SCA - • SCB -100

20 u E1 w R C 10

Ill E 0 ' u

-10

-20 " 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 4.10: Time series of the sub-inertial depth averaged alongshore and across-shore current velocities at the inshore (SCA, DHA) and mid-shelf (SCB, DHB) moorings at Smoky Cape and Diamond Head. As in subsequent figures, each event is marked EAC 1 (El), Wind event (W), Reversal event (R), Current event (C) and EAC 2 (E2). CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 83

4.4.1 Depth Averaged Momentum Equations

In the following analysis, a right hand coordinate system (x, y, z) is adopted, with the y-axis alongshore the x-axis positive eastward, in the across-shore direction and the z-axis is vertical. The corresponding velocity components are (u, v, w). The rigid lid approximation is assumed at z = 0, and the bottom is at z = -H(x, y).

The barotropic form of the momentum equations can be found by vertically averag­ ing the equations:

av 1 aP T8 y Tby at+ fu+u.Vu ---+--- (4.1) Po ay PoH PoH au 1 aP Tsx Tbx -fv +v.'\lv ---+--- (4.2) at Po ax PoH PoH where (u, v) are approximations to the barotropic currents where the over bar de­

3 notes the vertically averaged current. The reference density, p0 = 1023 kg m- •

(aP/ax, aP/ay) are the depth averaged horizontal pressure gradients (rsx, Tsy) are the surface wind stress in the x and y directions and (rbx, Tby) represent the bottom stresses due to friction at the sea bed.

On a sub-inertial time scale, order of magnitude arguments (where v >> u), im­ ply that the depth averaged across-shore momentum equation is dominated by the geostrophic balance of the alongshore velocity. Calculations from the data veri­ fied that the geostrophic assumption is valid in the across-shore direction hence the following analysis is only performed on the alongshore momentum balance.

4.4.2 Estimation of terms

To estimate the terms in the alongshore momentum balance, the depth averaged currents are used, as described in Section 4.3, as well as the lowpass filtered records of wind and sea level elevation as described in Chapter 3. Once the terms were CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 84 calculated they were lowpass filtered with a 60 hour cut off period to focus on the low frequency variability. Finally, as 8v / 8t is centred in time, all other terms were also calculated as a centred difference in time.

Tendency The time derivative of velocity Vt was calculated using a first-order difference scheme from the sub-inertial barotropic velocities.

Coriolis The Coriolis term Ju was calculated for each array using Coriolis parameters of !sc =

-7.07 x 10-5 s-1 and !dh = -7.33 x 10-5 s-1 for Smoky Cape and Diamond Head respectively. The depth averaged across-shore velocity u as shown in Figure 4.10 was used.

Pressure Gradient The pressure gradient term Py was approximated by differencing measurements of adjusted sea level elevation (Section 3.4) from tide gauges at different locations along the coast as: P, T/2 - T/1 y ~ g t:..y where g = 9.8 m s-2 is the gravitational acceleration, (ry 1, ry2) are the relative adjusted sea level elevations at coastal stations and t:..y is the alongshore distance between the two stations. This approximation does not take into account across-shore vari­ ation in sea level elevation nor does it account for the baroclinic contribution to Py. Furthermore the alongshore distance between the tide gauges is small compared to the wavelength of a typical CTW, thus in this analysis the impact on sea level elevation of the propagation of a CTW is negligible.

Wind Stress Measurements of wind speed and direction at Smoky Cape were used to calculate CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 85

surface wind stress (Figure 3.5) in the alongshore and across-shore direction (Tsx, T8 y):

T = CvPairlulu (4.3) where Pair is the density of air taken to be 1.3 kg m-3 , CD is the dimensionless drag coefficient, which increases with wind speed (Gill, 1982). For low wind speeds

Cv ~ 1.1 x 10-3. For speeds between 6 - 22 ms-1 a linear relationship between Cv and u is used; 103Cv = 0.61 + 0.063u.

Bottom Friction The kinematic bottom stresses can be parameterised as a linear function of velocity:

with a drag coefficient, r = 5 x 10-4 ms-1 (Lentz and Winant, 1986) and (ub, vb) are the bottom currents. The bottom friction term was calculated using the current measurements from the lowest meter on each mooring. SCA= 15 m from the bottom, SCB= 20 m from the bottom, DHA= 20 m from the bottom. In the case of DHB where the lowest meter failed after just one week, the alongshore time series was reconstructed using a regression coefficient obtained from the relationship between the shallower current meters on the same mooring. This extrapolation has only been applied for this specific purpose and only in the alongshore direction.

Non-linear terms In many applications the non-linear terms in the momentum equations are ignored however, in a rapid flowing western boundary current, advection is important. This is indicated in Chapter 6 of this study where Ross by numbers (which highlight the non-linearity of a system) are shown to be important in the Smoky Cape region.

The horizontal non-linear terms in the alongshore momentum equation v.'vv are given by av av u-+v- (4.4) ax 8y CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 86

Because of the limited nature of the two mooring arrays, several assumptions had to be made in calculating the non-linear terms. Firstly it was assumed that alongshore advection is constant across both arrays, i.e. 8v / 8y inshore at Smoky Cape is the same as 8v / 8y inshore at Diamond Head. Secondly to calculate the across-shore term 8v/8x at the inshore mooring at Smoky Cape (SCA), the term as calculated from the mid-shelf mooring (SCB) was extrapolated inshore linearly.

For ease of notation in the following section the terms in the depth averaged along­ shore momentum equation will be denoted as:

Vt + f U + Py - 7 8 + Tb + V. \7 V = 0 (4.5)

4.4.3 Uncertainties

The estimates presented below are subject to a certain degree of error arising from both instrument error and data availability. Using a limited data set various as­ sumptions must be made when calculating the terms in the momentum equations. Uncertainties in the estimates cannot be quantified accurately however as instru­ ment performance in the field is not understood well and spatial scales of variation are not well known (Lentz et al., 1999).

Current measurements have errors in the order of 2 - 3 cm. Errors occur when calculating the depth averaged velocities from only 2 - 3 current meters spaced vertically through the water column and when extrapolating the current profiles to the surface and the bottom. The across-shore velocity is at times strong across the arrays, corresponding with the lateral movement of the axis of the EAC either onshore or offshore. At these times it is difficult to estimate the principal axis of the flow and the across-shore terms become large. Hence uncertainties in the orientation of the principal axis reduce the reliability of estimates of the depth averaged across- CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 87 shore flow.

Uncertainties in the pressure term come from inaccuracies in measurements of sea level elevation and atmospheric pressure used in the adjustment process. Due to lack of measurements it is assumed that the pressure gradient does not vary across-shore.

In an attempt to improve the accuracy of the balance, the estimate of the alongshore pressure gradient was examined more closely. The magnitude of the gradient was estimated from a model simulation (described in Chapter 6). Although the residual was reduced at times across the northern array, at the southern array where the current regime is complex, (shifting from wind driven to EAC driven and from a rapid southward jet to a northward coastal recirculation), sea surface elevation could not be estimated accurately from a static model simulation. The simulated scenario used shows a northward current adjacent to the continental shelf, driven by a negative pressure gradient. This scenario was however observed to change with time, whilst the sign of the modelled gradient does not change, hence the reliability of the estimate of the alongshore pressure gradient is questionable.

A final estimate of the contribution of the pressure field was made by adding the mean of the residual to the pressure field. If we assume that all the terms are on the left hand side of Equation 4.2, then the terms should sum to zero. The residual occurs as a result of the assumptions and uncertainties mentioned above and is found by summing all the components of the alongshore momentum balance. Assuming that the majority of the uncertainty lies in the estimate of the pressure gradient term, the temporal mean of the residual was added to the 8TJ/8y term and the balance was subsequently re-evaluated.

Errors in the wind stress calculations occur as it is assumed that the land based winds represent the maritime winds and again these are assumed to be uniform in the across-shore direction, across the coastal ocean. As with the bottom stress term, CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 88 uncertainties in the drag coefficients add to the level of inaccuracy.

Finally in the advection terms, errors are propagated through current measurements and linear extrapolation towards the coast at SCA as described previously.

4.4.4 Balance of Terms 1: Temporal Analysis

Each of the terms in the depth averaged alongshore momentum equation (4.2) are estimated at four locations (SCA, SCB, DHA, DHB). The time series of the es­ timated terms are shown in Figure 4.11. The figure shows that at various times each of the terms are of a similar magnitude hence they are each important in the alongshore balance. Across both mooring arrays the magnitude of the residual is large when compared to the other terms in the momentum equation, however it is not unusual for the momentum equation to be out of balance when the terms are calculated from observations (Allen and Smith, 1981; Griffin and Middleton, 1986, inter alia). This simply highlights the difficulties in estimating the individual terms in the equations of motion using time series from unevenly spaced observations.

Figure 4.12 shows the temporal mean of each term at SCA, SCB, DHA and DHB, with error bars depicting the standard error within each term. At each array the mean tendency is the smallest term, with a large standard deviation which implies that the current fluctuates in both directions almost equally: The mean wind stress is the second smallest term, however, in this case the standard deviation is up to an order of magnitude smaller than any of the other terms, which implies that wind stress is small and invariant. At the mid-shelf moorings (SCB, DHB) the standard deviation of the residual is large which implies that errors in the estimates are large. The mean residual is smallest at SCA which suggests that it is at this mooring that the balances have been resolved most accurately. SCA 0 NX 0 ~ ~I Cl) l 'v E - 1 ~ : ;::o 14 21 28 5 12 19 26 2 9 16 23 30 .i:,..

5S ~ ;' SCB - t s;: ~NX 0l - fu tv I Cl) p ~ - y '-; 1 E - 1 s f--3 >-< 14 21 28 5 12 19 26 2 9 16 23 I - 'b ::i:,.. v.V v tv - R 0 c::: DHA f--3

NX 0 ~ ~I tr:l Cl) l ] E - 1 ~ 0 en 14 21 28 5 12 19 26 2 9 16 23 30 tr:l ~ ~ ::i:,.. "'0 ~ X ~~DHB ~ N 0 I Cl) ~< E - 1

14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999 c.::)

Figur e 4.11: T he time series of the terms and residual iu (4 .2) at each of t he four moorings. Units arc m s- 2 x 105 . CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 90

0.5

"'o X N 0 -f- 'E riQ_~D [i] -[i]

-0.5 D I SCA SCB Dlu DP, D', tb 0.5 D ['ZJ v.V V Ill R

1? X N 0 ' E ',]~---, +~~-c:prl!!JI

-0.5 DHA DHB

Figure 4.12: The temporal mean of the terms and residual in (4.2) at each of the four moorings. Error bars are calculated as the standard deviation of each term.

Up stream Upstream of the EAC separation (i.e. at Smoky Cape) the advection term is large and variable. This indicates the transfer of momentum into the region from further to the north. Inshore Tb is large and opposes the non-linear advection as a result of a strong c urrent in shallow water. Mid-shelf f u is the dominant term which indicates a strong and highly variable across-shore flow. The standard deviations of the tendency, bottom stress and advection are high upstream of the separation, where the currents are strong and non-linear. Figure 4. 13 shows the time series of

Tb and v.Vv at SCA and SCB. Clearly at SCA for a large part of the time series, the non-linear advection is balanced by bottom stress. At SCB the balance is not as dominant, however the terms are still evidently anti-correlated, particularly towards the end of the time series ( EA C 2).

Downstream Downstream of the EAC separation (i.e. at Diamond Head) , Py is t he dominant CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 91

1 " "b SCA w J \ C _J:2v-- - 'tb .... I X - - v.Vv N 0 I 1/) E -1

1 '\ "b.... SCB I ..... X N 0 I 1/) E -1 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 4.13: Time series of Tb and v.'vv at SCA and SCB. Units are ms-2 x 105 • term with a large standard error. The term Ju is again highly variable over the mid-shelf and this can be attributed to large variations in u as the axis of the EAC moves laterally across the shelf. The mean residual is largest at DHB which can be attributed to the imbalance between the two dominant terms Py and Ju. This obviously is a result of the mis-representation of the pressure term mid-shelf. Fur­ thermore, as the current meanders on and off the shelf at Diamond Head, it is difficult to ascertain the along and across-shore direction of the mean flow. (Recall the SST image in Figure 3.17 where the EAC bends eastward just south of Diamond

Head.) Figure 4.14 shows the time series ofi1t+Tb and Py at _DHA and DHB. Down­ stream of the separation point tendency and the pressure gradient are important contributors to the momentum balance. The circled regions point to periods when

Vt + Tb opposes Py at DHA. At other times Vt and Py fluctuate together. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 92

1 "b R C E2 - Vt+'t'b .... DHA p X - - y C\I 0 I (I) E -1 1 "b.... DHB X C\I 0 I (I) E -1 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 4.14: Time series of iit + Tb and Py at DHA and DHB.

Covariance

To aid an understanding of the covariance amongst the terms in the alongshore momentum equations, the cross correlation coefficients between terms are presented in Tables 4.2 - 4.4. The coefficients are calculated at each mooring and are shown for the entire record, as well as for each of the main events: the Wind event, the Reversal event and the Current event (Tables 4.2 and 4.3) and for comparison, EAC 1 and EAC 2 (Table 4.4). In general the correlation coefficients for the total period are very low implying that the dynamics at Smoky Cape and Diamond Head are both complex and transient. At SCA however the correlation between Tb and ii.Vii with 95% confidence is -0.84, which means that as well as being of the same order and opposite (Figure 4.13), that they also co-vary. This supports the result in Figure 4.13, that the dominant balance at SCA throughout the experimental period is evidently between non-linear acceleration and the retardation caused by contact with the sea floor. Mid-shelf at SCB Tb and ii.Vii also co-vary with 90% confidence, however the correlation (-0.45) is lower than at SCA. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 93

Downstream of the separation point at Diamond Head the Py and Vt are positively correlated to 95% significance throughout the period both inshore and mid-shelf.

Inshore at DHA Py is anti-correlated with Tb (-0.35) again to 95% significance.

Figure 4.14 that shows the time series of Vt + Tb and Py. From the figure and the correlations, it is evident that downstream of the separation point, tendency is more important than advection. Another significant correlation at both DHA and DHB is between non-linear advection and Ju. As the advection increases so does Ju which is consistent with the idea that as the EAC encroaches upon the shelf at Diamond Head the velocity increases are felt in both the along and across-shore directions.

For the duration of the time series Py is significantly correlated with Vt at each of the sites. As the pressure gradient increases so does the acceleration, which is consistent with coastal flow. This is perhaps indicative of flow driven by the propagation of coastal trapped waves through the region, which was investigated in Chapter 3.

The correlations when calculated for the individual events reveal interesting dy­ namical balances that vary in both the alongshore and across-shore directions i.e. upstream and downstream of the EAC separation point and from inside to outside the internal Rossby radius (2 - 16 km from the coast).

Wind event

During the Wind event, Ts is positively correlated with Ju at SCA, which is con­ sistent with wind driven across-shore transport. This is also·true at DHA although to a lesser extent. Although T 8 is more significant during this event, the dominant balances at SCA between Tb and v.v'v and between Py and Tb still hold.

At Diamond Head wind forcing is only significantly correlated mid-shelf, where T8 is anti-correlated with Vt (-0.73). Another correlation is between Ju and v.v'v, (DHA=0.78, DHB=0.78) as was also the case for the total period. Also of note is the correlation between Tb and v.v'v at DHA and DHB, indicating an increase in Smoky Cape A 50 m Smoky Cape B 100 m @ ;:i:.: Ju Py Ta Tb ii.Vii Ju Py Ta Tb ii.Vii '"Cl 0.04 -0.01 -0.13 -0.2 0.03 -0.01 -0.12 iit -0.04 0.4t 0.43t ~ Ju - 0.24 0.3* 0.09 0.01 - 0.25 -0.19 -0.01 0.26 ::0 Total Py - - -0.02 -0.64t 0.55t -- -0.02 -0.57t -0.03 ~ Ta - - - -0.2 0.20 - -- -0.14 0.05 ~ Tb - - - - -0.84t -- - - -0.45* ~ 0.78 0.69 0.2 -0.13 -0.28 0.37 0.82* 0.3 -0.32 -0.37 to iit t=: - -0.42 0.26 - 0.7 0.04 -0.83 0.2 ~ Ju 0.71* 0.73t ~ Wind event Py - - 0.54 -0.75* 0.42 - - 0.54 -0.74* 0.02 ~ - - --- ~ Ta - -0.72 0.8* -0.52 0.58 to Tb -- -- -0.91t ---- -0.51 0 c:: ~ iit -0.4 0.36 0.06 0.07 -0.22 -0.70* 0.42 0.15 0.01 -0.19 ~ Ju - 0.41 -0.54 -0.4 0.36 - 0.36 -0.64* -0.52 0.48 t_:rj Reversal event Py - - -0.16 -0.83t 0.72* -- -0.16 -0.81t 0.58 ~ Ta - -- 0.11 -0.28 - -- 0.16 -0.31 G Tb - - - - -0.78 -- - - -0.84 Cr.l ~ ~ iit -Q.7 -0.05 -0.28 0.39 -0.20 -0.82 0.05 -0.21 0.19 -0.05 Ju - 0.67* 0.77* -0.87t 0.58 - 0.18 0.06 -0.43 0.01 ~ ~ Current event Py -- 0.38* -0.77t 0.24 -- 0.38* -0.79t -0.02 0 Ta -- - -0.66* 0.72 - - - -0.48 0.61 2: Tb -- -- -0.50 - -- - -0.23

Table 4.2: Cross correlation coefficients between terms in the alongshore momentum equation at Smoky Cape for the total period and for the 3 events ( Wind event, Reversal event and Current event). The probability that the correlation coefficients would be obtained from uncorrelated variables is less than 0.05t and less than 0.1*. co .i::,.. Diamond Head A 50 m Diamond Head B 95 m @ Ju Py Ts Tb v.v'v Ju Py Ts Tb* v.v'v ::i:..: ~ iit 0.19 0.4t -0.22t -0.07 0.08 0.05 0.33t -0.15 -0.06 0.04 - -0.08 0.34t 0.07 0.55t - 0.21 0.21 -0.22 ~ Ju 0.58t ~ Total Py - - -0.27 -0.35t -0.02 -- -0.27 -0.14 0.23 ~ Ts - - - -0.07 0.27* - -- -0.21 0.15 Tb - -- - -0.49t -- -- 0.16* ~ ~ s;: iit -0.37 0.83t -0.43 -0.07 0.02 0.22 0.64 -0.73* -0.15 -0.08 0::, ~ Ju - -0.44 0.64 -0.31 0.78 - -0.26 -0.03 -0.59 0.71t "-I Wind event Py - - -0.63 -0.27 -0.03 -- -0.62 0.04 -0.37 1--3 ~ - Ts - - 0.06 0.30 -- - -0.25 0.46 ::i:a. - -- -0.80 - --- 0::, Tb - -0.9* 0 ~ iit 0.68 0.64 0.03 -0.06 -0.49 -0.05 0.73 0.21 -0.22 -0.5 1--3 Ju - 0.36 -0.37 0.3 -0.44t - 0.29 -0.13 -0.38 0.01 ~ trj Reversal event Py - - 0.07 -0.73 -0.52 - - 0.07 -0.07 -0.4 Ts - - - -0.12 -0.26 -- - -0.22 -0.04 ~ Cl Tb - --- -0.08 - - -- 0.16* ~ iit 0.24 0.81* -0.85t -0.42 -0.01 -0.45 0.54 -0.5 0.24 -0.28 ~ Ju - -0.11 -0.01 0.61 0.30 - -0.12 0.3 -0.54 -0.19 ~ Current event Py - - -0.87t -0.75 0.03 - - -0.87t 0.24 -0.74t ~ 0 Ts -- - 0.37 -0.31 --- -0.32 0.73t ~ Tb - - -- 0.30 ---- 0.16*

Table 4.3: Cross correlation coefficients between terms in the alongshore momentum equation at Diamond Head for the total period and for the 3 events ( Wind event, Reversal event and the Current event). The probability that the correlation coefficients would be obtained from uncorrelated variables is less than 0.05t and less than 0.1*. The bottom current meter failed at the mooring denoted by *, so the bottom stress term is an estimate only. ~ CJl ~

~

trj ~

~ ~ ~ ~

~ ~ ~ ~ s;: 0:, ~ Q ~ co ~ 0:, O')

~ ~ ~ 0 ~ ~

0 c:: ::i:..

@

the

The

for

0.1*·

Head

0.8

0.28 0.93 0.05 0.03

0.7*

0.15

0.33

-0.3

-0.2

than

0.53* 0.11*

-0.29 -0.12

-0.38

v.Vv

v.Vv

-0.82t

-0.54t

-0.51*

less

m

Diamond

-

m

and

Tb

0.44 0.89

0.46

0.81

0.09 0.13

0.31

0.17

Tb*

100

and

-0.02

-0.13 -0.35 -0.31 -0.28 -0.35

-0.55 -0.43

100

B

0.05t

B

-

------

Cape

0.4

0.01

0.44

0.27 0.77

0.29

Ts

Ts

than

Head

-0.09 -0.13 -0.33

-0.76

-0.65 -0.16

Cape

-0.77t

less

only.

Smoky

is

- at

-

0.38 0.32 0.41

Py

Py

Smoky

-0.3

0.78t

0.92t

Diamond

-0.14 -0.15

-0.32

estimate

variables

equation

an

------

is

Ju

Ju

-0.32 -0.03

-0.05

term

0

momentum

uncorrelated

stress

0.28

0.01 0.43 0.04

0.71 0.87 0.50 0.48 0.37 0.19 0.61

0.92t

-0.33 -0.23 -0.31 -0.22

v.Vv

v.Vv

from

-0.92t

-0.55t

-0.84t

-0.87t

bottom

alongshore

-

-

m

0.1

m

the

0.58

0.29

0.15

0.37

Tb

Tb

-0.4

the

0.67t

obtained

50

-0.07 -0.17 -0.26

-0.77 -0.31

-0.26

so

50

-0.021

-0.64*

-0.68*

in

be

A

A

-

by*,

terms

0.44

0.59 0.62

0.7*

would

Ts

Ts

Head

0.69*

0.71*

Cape -0.47

-0.09

-0.23

-0.33

-0.53*

-0.76*

denoted

between

- -- --

------

------

Smoky

0.09

Py

0.45 0.79

Py

-0.6

coefficients

0.64t

0.82t

Diamond

-0.29

-0.27

mooring

that

- -

------

-

the

coefficients

0.37

0.42

Ju

Ju

at

-0.48

-0.82t

failed

Probability

Ju Ju

Py

Py

Ju Ju Py

Py

Ts Tb

Ts Tb Vt

Vt

Tb

Vt

Tb Ts

Ts Vt

correlation

2.

1

1

meter

Cross

EAC

EAC

EAC2

EAC2

EAC

current

4.4:

and

1

EAC

bottom

Table CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 97 advection in this region during the Wind event. Hence although it is during this period that wind forcing is a maximum, it does not appear that the wind stress term is strong enough to drive the flow either at Smoky Cape or at Diamond Head.

Reversal event During the Reversal event the flow is essentially barotropic at the inshore moor­ ings, and across the shelf the current fluctuates and accelerates both northward and southward. This flow is driven by fluctuations in the barotropic pressure gradient which are seen in observations of sea level elevation (Figure 3.15). During this event

Tb is negatively correlated with Py at SCA, SCB an DHA which implies that this pressure driven flow is opposed by the bottom friction. At Smoky Cape the corre­ lation between Py and Tb is -0.83 inshore and -0.81 mid-shelf with more than 95% certainty. The balance between Py and iit + Tb in Figure 4.14 is evident at DHA where the anti-correlation to 95% significance is -0. 73. At DHB Py although anti­ correlated, appears to lag iit+Tb, which is probably a result of the mis-representation of Py across the shelf.

Inshore at Smoky Cape, advection is positively correlated with Py to 90% signifi­ cance, which is also one of the balances that holds throughout the total period. At Diamond Head the only significant correlation is a balance between Ju and ii.Vii.

Current event

The correlation coefficients show that Tb is anti-correlated to· 95% certainty with Ju and Py at SCA and with Py at SCB. Of note is the lack of significant correlation between ii.Vii and Tb, which is evident during all the other periods. Figure 4.13 however does show that by 9 January 1999 the balance between ii.Vii and Tbhas been re-established. At SCA Ts is anti-correlated with Ju, (-0.77) with 90% confidence.

At Diamond Head during the Current event, Py is negatively correlated with Ts, which implies that Py at Diamond Head opposes Py at Smoky Cape. Nonlinear CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 98 advection is anti-correlated with Py and and co-varies with Ts at DHB during the current event. At DHA the pressure gradient varies with the tendency, which was also evident during the Wind event.

EAC 1 and EAC 2 The correlation coefficients for the terms in the momentum equations reveal inter­ esting dynamics during EAC 1 and EAC 2. Across the northern array Py co-varies with iit with high correlation coefficients and more than 95% confidence both inshore and mid-shelf during EAC 1. During EAC 2 however at SCB the correlation is not significant, again this is most likely because Py is assumed to be constant across­ shore direction. During both events Tb balances ii.Vii, with correlation coefficients of -0.92 (EAC 1) and -0.84 (EAC 2) at SCA. However of interest is the fact that during EAC 2 when the full effects of the EAC are seen across the southern array as well (i.e. when the current is not separated) advection is again balanced by Tb at DHA (correlation = -0.87 to 95% confidence). During EAC 1 the current does not impinge directly on Diamond Head for the entire period. This is reflected in the importance of Ju, which co-varies with the non-linear advection at DHB.

In purely two dimensional flow there is no variation in the alongshore direction (i.e. 8/8y = 0) then Ju = 0, 8r,/8y = 0. At SCA during EAC 1 the depth averaged across-shore currents in Figure 4.10 are almost negligible, implying two dimensionality, however from the terms presented in Figure 4.11 the tendency is not balanced by the sum of the stress terms which should be the case in two dimensional flow. This indicates possible inaccuracies in the estimations of one or more of the terms.

Ageostrophy The time series of iit and Ju is shown in Figure 4.15. Regions where - Ju~ iit imply that - Ju =I= Py i.e. that the system is ageostrophic in the alongshore direction. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 99

1 LO ....0 X SCA (',I 0 I _vt II) E - - tu -1

1 LO ....0 X SCB (',I 0 I II) E -1

1 LO ....0 X DHA (',I 0 Ill) E -1

1 LO ....0 X DHB (',I 0 I II) E -1 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 4.15: Time series of Vt and Ju showing periods of ageostrophy, i.e. when Vt +Ju= 0.

Ageostrophic periods are circled in the figure. This is particularly noticeable at the mid-shelf mooring at the northern array (SCB), at tl;te end of the Reversal event where the terms are correlated to 90% significance. Inshore Vt ~ - Ju during EAC 2 where the terms are anti-correlated to 95% significance, with a correlation coefficient of -0.82. Downstream of the separation at no time does Vt ~ - Ju, and the correlations are consistently inconsistent. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 100

4.4.5 Balance of Terms 2: Statistical Analysis

In the previous discussion the possible balances between the terms in the depth averaged alongshore momentum equation (4.2) are assessed by referring to the time series, cross correlation matrices and standard deviations of each of the terms. To further examine the dominant balances between the terms, Empirical Orthogonal Functions (EOFs) are utilised to gain a statistical representation of the extent and type of covariance between time dependent terms in Equation 4.2. This is done by decomposing the time series of each term in the momentum balance into EOFs for each mooring. The method applied here follows that of Allen and Smith (1981), however, others have also successfully utilised this technique to explain the variance in a data set that is a function of both space and time (Kundu et al., 1975; Davis, 1976; Allen et al., 1995, inter alia).

For completeness, the mathematics of the procedure is outlined below using the notation of Allen and Smith (1981). Given N time series 9i(t), i = 1, ... N, remove the mean such that Yi = 0, where the over-bar denotes the temporal mean of the series. These terms can be written as the sum of N time series:

N 9i(t) = L En(t)cpf (4.6) n=l where An+l·

EOFs are a useful tool for evaluating the balance of terms for 3 reasons,

1. the amplitudes En(t) represent a common time dependence for all terms.

2. each empirical eigenfunction extracts the maximum amount of total variance possible. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 101

3. the En(t) are uncorrelated

In this application the practicality of EOFs comes from the fact that terms or com­ ponents of terms may balance each other when they are temporally co-dependent. Theoretically for an exact balance the sum of the eigenvector components would be zero, i.e. E: 1

EOFs are calculated for the six time series comprising the estimated terms in equa­ tion 4.2, (vt,fil,Py,Tb,Ts,V.Vv) for each of the four moorings, where the depth aver­ aged velocities can be calculated (SCA, SCB, DHA and DHB). This is done assuming that all terms are on the left hand side of Equation 4.2, as in Equation 4.5. The EOFs represent the variability about the mean of each term which are shown in Figure 4.12. Figures 4.16 and 4.17 show the EOFs of each term for each of the 6 modes.

The bars represent the magnitude of the eigenvector components (

i=l which represents the magnitude of the terms and errors that are not represented in the balance (Allen and Smith, 1981). The percentage of the variance in each term is shown above or below the bar in each mode, as is the total variance accounted SCA 59 @ 88 5 7 >­ 4 '"O 0.5 n 52 ~ ~ 0 Dm c _Doorn ooouo_ D~~HD~ 4 SS D ~ 18 s;: -0.5 313 6 69 to ~ 77 85 ,.___. -1 53% 28% 12% 5% 2% 0% f-3 '-< D · 1 >­ Ot u to D Py 0 c:: o -rs f-3 SCB o-rb ~ I•:<1 v Sl v tr:l 71 65 R ~ 22 .. 0 0.5 5 ~ 116 I I n 4 ~ - i--,c::]- 0 =[to~ 0 -- ~ ~UroiD2.n. >- 14 ~ ODD'°~ ~ @ 0 - 0.5 5 53 15 < 64 45 29 - 1 40% 34% 15% 10% 13 1% 0% 91 Mode 1 Mode 2 Mode3 Mode4 Mode 5 Mode6 ...... Figure 4 .16: EOFs of the terms in t he alongshore momentum equation (4.2) delineating the variability at SCA (top) and SCB (bottom). 0 1:-v The percentage contribution of each mode to the tota l variability is indicated, as well as the contribution of the variability with in each term. DHA 95 @ 70 1 ~14 ::i:,.. >-o 0.5 1~ I 38 1 ~ 1 8 13 ,4. 0 Do=D- Du=~-DDD~Dli =D=noc 2 ~ 20 :,;::, -0.5 $;: 44 D ·1 to Otu ,.__,t=1 0% -1 60% 25% 9% 5% 1% f--3 99 66 72 D Py '-< o ·s ::i:,.. D·b to v.V v 0 D C; B R f--3 DHB 76 ~ 91 69 tr:l ~ Q 0.5 3 8 ~ 4 8 :-4 gu0__! 131 2 = D ijl1 1 ___ L..Jo.., ~ 0 . D o= _e :,;::, ::i:,.. or--1 ~ 26 0 - 0.5 <

-1 55% 72 30% 10% 4% 51 81 1% 0%

Mode 1 Mode 2 Mode3 Mode4 Mode5 Mode 6

Figure 4.17: EOFs of the term s in the alongshore momentum equation (4.2) delineating the variability at DHA (top) and DHB (botto m). 0 -c.., The percentage contribution of each mode to the tota l variability is indicat ed , as well as the contribution of the variability with in each term. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 104 for by each mode. In many of the modes the residual is large and often it balances the largest term. This suggests that the terms have not been estimated accurately, as concluded in the previous time domain analysis. Bearing this in mind it is still useful to examine the processes which dominate each current meter record upstream and downstream of the separation point.

Upstream The eigenvalues of the EOFs at Smoky Cape (Figure 4.16) show that the first mode accounts for 53% and 40% of the variability, at SCA and SCB respectively. At the inshore mooring (SCA) the first mode is dominated by Py, with 88% of the variability within Py explained in that mode which does not appear to be balanced.

The time series of the modal amplitudes of the EOFs of each of the terms are shown for Smoky Cape and Diamond Head in Figures 4.18 and 4.19 respectively. The first two of a possible six modes are presented.

The amplitudes in Figure 4.18 reveal an apparent anti-correlation between Vt, v.v"v and Tb at Smoky Cape. In mode 1 at both SCA and SCB v.v"v is almost equal and opposite in sign to Tb. This indicates that these terms are in quasi-balance and although they are not always the dominant term, that the balance is manifest in mode 1 indicates its significance.

The second mode which contributes to 28% of the variabili_ty at SCA and 34% at SCB, is dominated by Vt (Figure 4.16) and advection is again balanced by the bottom stress at SCA. In mode 3 f fl is the dominant term and is not balanced, however

Py + Tb is balanced by Vt + v. v" v. Again the importance of advection and bottom stress is noted.

At the mid-shelf mooring (SCB) the first mode is dominated by Ju. This indicates high lateral variability as the EAC either encroaches on the shelf or separates from CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 105

SCA "'0 Mode 1: 53% X N 0 I CJ) E -1

SCA "'0 Mode 2: X N 0 fi~f!J- Af!/• I CJ) E -1

SCB "'0 Mode 1: 40% X N 0 - t I CJ) - fu E p - y -1 't - s _ 'tb SCB v.'i/ V "'0 Mode 2: 34% X N 0 I CJ) E -1 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 4 .18: R0constrnct0d time series from the a mplit udes ( x 105 ) of t he E OFs of terms in the alongshore monwntttm equation (4.2) at Smoky Cape. The first two modes arc s hown fo r S CA and SCB. The percentage contribution of each mode is marked. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION lOG

DHA "'0 Mode 1: 60% X N 0 I Cl) E -1

DHA "'0 Mode 2: 25% X N 0 I Cl) E -1

DHB "'0

X N 0 - t I Cl) - fu E p - y -1 ~-~--~--~~-~~-~~~~-~--~-~~~~_., 1: - s

r------.---,-----.-- -,--,----,----r---,----,--,------,---,---r----r----,1 -- ' b DHB V.'v V "'0 Mode 2: 30% X N 0 I Cl) E -1 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 4 .19: RC'co nstructC'd time seriC's from the a mplitudes (x l05) of the E OFs o f terms in tl1<' a longshore momentum equation (4.2) at Dia mond Head . The first t wo modes are shown for DHA a nd DHB. T he pC' rC'C' ntagC' contribut ion of each mode is marked. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 107 the coast. As Ju is not dominant at SCA (inshore) until the third mode (which accounts for only 12% of the total variability) this leads to the conclusion that Ju is not as variable inshore. This is self evident as in coastal waters, flows must be shore-parallel, while further offshore variations in the EAC may produce across-shore fluctuations.

At SCB, v.Vv and Tb are again almost equal and opposite in mode 1, however the two terms that dominate, Py and Ju, are of the same sign (Figure 4.18). In the second mode however Py and Ju are actually equal and opposite indicating a geostrophic component of the alongshore flow. In contrast to the inshore mooring, the second mode mid-shelf (SCB) accounts for 34% of the variability and represents a significant balance Ju+ v. V v + Tb ~ Py + Vt- In the third mode advection balances the sum of Vt +Ju+ Tb, but accounts for only 15% of the variability.

At both moorings T8 does not dominate until the sixth (final) mode and is not statistically significant representing < 1% of the total variability in the alongshore momentum equation. This emphasises the point that wind forcing is insignificant in the EAC system, however advection and bottom stress are important upstream of the EAC separation point.

Downstream South of the separation point the dominant balances are more difficult to resolve. In each mode there is a dominant term which differs from BHA to DHB and does not appear to be balanced by any other term. Inshore at DHA the pressure gradient dominates the first mode accounting for 60% of the variability and is out of balance. At DHA 99% of the variability within Py is explained by the mode 1. Mid-shelf at DHB the first mode is dominated by Ju and Py dominates the second mode and again these terms are not balanced.

The most obvious balance at DHB is between Vt and Ju in the second mode which CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 108 indicates an ageostrophic component of the flow. This is in contrast to the mid­ shelf mooring at Smoky Cape where mode 2 revealed that the flow in the alongshore direction was in geostrophic balance. Tendency dominates mode 3 inshore and mid­ shelf at Diamond Head, accounting for 9% of the variability at DHA and 10% of the variability at DHB. At Diamond Head the tendency stands out in many of the modes. This is in contrast to the northern array where the rate of change in alongshore velocity does not play a major role.

Across the southern array, advection is insignificant, not dominating until the 4th mode at both arrays accounting for less than 5% of the variability. Inshore to the north the most variability in advection is actually found in the first mode, with the first two modes contributing to 84% of the variability. Offshore, where variations in v.v'v are less prolific, advection does not dominate until mode 3 which accounts for 14% of the variability at SCB. In each case the wind stress term T 8 dominates the sixth mode which accounts for the least variability which indicates that wind forcing was of little significance during this period.

That the non-linear terms are significant in all modes across the northern array implies that alongshore acceleration is important before separation. However, across the southern array advection does not dominate until mode 4 which highlights the difference in the flow field after the current has separated from the coast.

To further examine the dynamics of specific events the eigenfunctions for the first two modes are shown for four events ( Wind, Reversal, Current and EA C 2) at each mooring (Figures 4.20 and 4.21). When compared with the entire time series, different terms dominate the balance during different events. For the individual events the terms that dominate are generally consistent across the shelf, but differ from the northern array to the southern array. CHAPTER 4. VA RIABILITY ABOUT THE EAC SEPARATION 109

SCA SCB

1 84 94 73 0.5 0.5 378~ 24 "O C 0 0 D 1 0 ~ ~dl D'r1c10~ 55 2 48 -0.5 41 -0.5 45 91 89 -1 63% 32% -1 68% 25%

1 78 0.5 0.5 cij ~ 10 Ql ~ > 0 0 Ql ~ o--•'"926 om~Dl a: JD.~~I 34 38 -0.5 -0.5 88 -1 99 69% 24% ov1 -1 71% 23% D tu D Py D\ D\ 86 ~ v.V V fll R 0.5 81 0.5 41~ c 5~ ~ 0 0 ::, OD u ~0-~0il ~o- ~D-~=i10 l 55 ~D~"D!l -0.5 24 -0.5 21 90 -1 93 70% 19% -1 97 73% 20%

1 62 0.5 0.5 62 C\I u ~ 33 5

Figure 4.20: EOFs of the terms in the a longshore momentum equation (4.2) delineating the variability at Smoky Cape. The first two modes are shown for each of four events: Wind event, Reversal event, Current event and E A C2. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 110

DHA DHB

-1 62% 29%

1 1 76

0.5 0.5 cii 53 1 ~ 1027 6 ~ 0 0 Q) a: 40 36 D V -0.5 42 t 0.5 Otu 95 -1 98 64% 94 32% 0 Py -1 63% 32% ~------. D \ 01:b ~ V.v' V JII R 1 83

0.5 0.5

c 36 113 25 13 0 0 ~::, 0 9 58 11 2 32 -0.5 -0.5 78 86 -1 56% 90 34% - 1 63% 24%

98 90 95 0.5 0.5 ~79 2 ~ 94 C\I 0 J, - - ~ ' <( 0 o 0~ 0 0 I 2 d LU ~J~D 23 62 ~D0~9 -0.5 -0.5

- 1 78% 73 12% -1 83% 14% Mode 1 Mode 2 Mode 1 Mode 2

Figure 4.21: EOFs of the terms in the alongshore momentum equation (4.2) delineating the variability at Diamond Head. The first two modes are shown for each of four events: Wind event, Reversal event, Current event and EA C2. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 111

Wind event During the Wind event up to 65% of the variability in the wind is resolved in the first two modes at the Smoky Cape moorings. This is compared with less than 2% for the entire period and only 2% at SCA during the Current event. During this event however, inshore at SCA, the residual, R, is large (Figure 4.20) which suggests that the dominant terms have not been accurately resolved. During the Wind event iit and Py account for most of the variability at the mid-shelf moorings, with wind stress contributing to the balance.

At SCA during this event Tb is opposed by ii.Vii with 24 and 55% of the variability within each term accounted for by mode 1. At SCB the majority of the variability in these two terms is in mode 2. At Diamond Head these terms oppose each other in mode 2.

Allen and Smith (1981) found that where (or when) the flow is strongly wind driven, one empirical mode carries most of the wind stress variance. The results presented here support this, in that all the variance in the wind stress is confined to the sixth mode at each location. Furthermore the sixth mode, the modal amplitudes show that the largest contribution to the wind stress term occurs during the Wind event. However, the sixth mode represents only a small percentage of the total variability, and it is not likely to be considered statistically significant.

Reversal event During this event tendency dominates at Smoky Cape. The variability in both advection and bottom friction are absolutely minimal (Figure 4.20). At SCA iit is not balanced, however at SCB iit is opposed by Ju. At Diamond Head the pressure gradient dominates the variability in mode 1 at both DHA and DHB (Figure 4.21). In mode 2 iit opposes Ju at DHB.

Current event CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 112

At SCA during the Current event the residual is small which implies that the balance has been well resolved. The dominate balance is between iit and Ju with bottom stress balancing the other three terms. This is indicative of the local acceleration of the flow as the axis of the EAC moves across the shelf and is evident in the time series of the amplitudes

At DHB Ju and Py are the dominant terms, the residual is large and r8 and Tb are positively correlated, however during the Current event the main balance is unresolved. This is more than likely attributable to errors in Ju, as ft becomes large and the across-shore alignment of the array becomes unsure combined with an inaccurate representation of Py.

EAC2 During EA C 2 Py dominates the variability across the shelf at Smoky Cape and is the largest term. The current is quasi-steady at Smoky Cape and advection is small. At Diamond Head however advection is more significant with more than 94% of the variability within the advection term accounted for in mode 1 and both DHA and DHB.

4.4.6 Discussion

From the above analysis it is clear that dynamical balances can be resolved from the time series of current measurements both within the EAC system and outside the EAC. When the terms were extracted directly from the recorded observations, balances were hard to find because of the noise within the system. However, when the time series of the EOFs were examined closely the dominant modes do emerge. The benefit of the EOF analysis is that it removes the noise component of the system so that only the dominant terms remain. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 113

There is no one single balance that can be used to describe the flow regime in this highly dynamic region both upstream and downstream of the point where the EAC separates from the coast. However, despite the limitations in the estimates, dominant forces have been elucidated in each region, as well as balances that are evident during the individual events. It is evident that at different times different forces are more important than others.

Upstream of the EAC separation point variability in the pressure gradient tends to dominate, however, at times advection is important and is balanced by bottom stress. This is evident in both the temporal mean and the time series of the terms, as well as the EOFs. However, this balance between advection and bottom stress is not evident south of the separation point except during EAC 1 and EAC 2, when currents flow parallel and adjacent to the coast. During EAC 1 and EAC 2 bottom stress balances the non-linear terms at SCA and SCB and also at DHA during EAC 2. Thus the non-linear transfer of momentum that dominates EAC 1 and EAC 2 is balanced by bottom stress. This is in keeping with the findings of Allen and Smith (1981) who found that bottom friction was a dominant term over a gently sloping shelf in northwest Africa.

There are very few times where bottom stress actually dominates the EOFs. The exception to this is during the Current event where the maximum variability in bottom stress occurs at the two moorings along the 50 m isobath, (SCA, DHA) where mode 1 and 2 account for 93% and 71 % of the variability in bottom friction respectively. During the Wind event bottom stress also opposes the non-linear terms. Of note is the fact that bottom stress is insignificant during the Reversal Event and when the flow at Smoky Cape is quasi-steady ( EA C 2). Furthermore the times when bottom friction and advection were important ( Wind event and Current event) coincide with the decreases in bottom temperature delineated in Chapter 3. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 114

At the mid-shelf moorings (SCB, DHB) the flow was found to be ageostrophic where Ju is highly variable. Allen and Smith (1981) found that off the Oregon coast both the time-dependent behaviour and the magnitude of Ju were not clearly related to the other estimated terms. They attributed high variability in Ju to either baroclinic alongshore pressure gradients or with non-linear advection of momentum, which they did not represent in their analysis. Here we hypothesise that as the axis of the EAC moves on and offshore the flow becomes ageostrophic, which is perhaps associated with funnelling and acceleration as the EAC moves on and off the shelf. Furthermore as the current fluctuates the principal axis of the flow can change from being alongshore to offshore. This is evident in the SST image in Figure 3.17 where the core of the EAC bends eastward.

Across the southern array, tendency is large and often it is not balanced by any one single term. As the flow moves offshore the slope of the pressure gradient and hence the sign of Py can actually reverse. The estimates of the terms in the momentum equations presented here were not able to reflect this change in the pressure gradient across the continental shelf.

For the duration of the time series and through each individual event wind stress was highly insignificant. The EOF analysis showed that variability in the wind stress was only significant in the sixth mode, which itself was statistically insignificant. Furthermore it is only for a 3-4 day period during the Wind event that wind forcing plays a possible role. Despite the fact that wind forcing is insignificant, the Smoky Cape region is known for its prolific nutrient enrichment events, which must therefore be driven by other processes. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 115 4.5 Summary

The time series of current vectors examined in this chapter show that the Smoky Cape region is highly dynamic. In Sections 4.1-4.3 temporal and spatial variability upstream and downstream of the EAC separation was examined. Frequency analyses highlight the different current regimes north and south of the separation point as well as from inshore to offshore.

An EOF analysis of the vertical structure of the current revealed that the major­ ity of the variability was in the alongshore component of the current, with only a small percentage in the across-shore direction. However, as the water column becomes more influenced by the EAC the flow becomes more baroclinic and the variance in the across-shore flow increases. Low frequency variability across the shelf was partitioned by examining horizontal EOFs at each mooring. During each event the currents tend to co-vary in the same direction, however the magnitude of the variability changes depending on the event. Mode 1 fluctuations about the mean were bigger at Diamond Head than at Smoky Cape. The largest fluctuations occurred during the Reversal event as the current oscillated between a northward and southward flow and were smallest at SCA during EAC 1 when the current was steady.

The balance of terms in the alongshore momentum equa;tions was examined using both time series and statistical techniques. One of the key findings of this section is the insignificance of wind forcing. Throughout the experimental period, there is no one time when wind forcing dominates or drives the flow. The region immediately south of Smoky Cape has long been recognised for its persistent upwelling and it is known that preferential nutrient uplift is a maximum during the Spring and Summer. That wind forcing was insignificant throughout the 1998-1999 Spring and Summer suggests that there must be other mechanisms which drive this persistent upwelling. CHAPTER 4. VARIABILITY ABOUT THE EAC SEPARATION 116

Throughout the analysis, the times when bottom stress was one of the important terms coincided with a significant decrease in bottom temperature. This is consistent with findings of the numerical study of Oke and Middleton (2000), who revealed that bottom stress was a dominant feature in this region and could contribute to the uplift of dense water south of Smoky Cape. Observations of high bottom stress at Smoky Cape is a new result and leads to the next question of 'Does the EAC cause upwelling?' OCEAN, n. A body of water occupying about two-thirds of a world made for man - who has no gills.

Ambrose Bierce (1842-1914)

Current Driven Upwelling

A pertinent question posed by Tranter et al. (1986) relevant to many facets of oceanography and biology is: 'How (do) western boundary currents increase pro­ ductivity of the adjacent coast'? In the EAC region, different hypotheses have been given for the occurrence of upwelling. As well as local wind forcing, other possible explanations for upwelling are associated with the western boundary current itself and off the east coast of Australia these include:

• A rapid flowing EAC near the continental shelf break (Boland and Church, 1981; Blackburn and Cresswell, 1993);

117 CHAPTER 5. CURRENT DRIVEN UPWELLING 118

• The separation of the EAC from the coast (Tranter et al., 1986);

• Eddies associated with the EAC system (Boland and Church, 1981; Cresswell, 1994; Gibbs et al., 1997); and

• Topographic acceleration of the EAC (Oke and Middleton, 2000).

The observations presented in Chapter 3 revealed the highly dynamic nature of the EAC in the Smoky Cape region. From the current meter time series it is evident that the currents are highly variable in both strength and direction. In Chapter 4 a study of the balance of terms in the alongshore momentum equation indicated the competing role of advection and bottom stress at Smoky Cape.

Several upwelling events that were observed in the region influenced by the rapid flowing EAC and downstream of the EAC separation point, are investigated in this chapter. The time series of temperature presented in Chapter 3 indicated two significant decreases in bottom temperature which occurred during the Wind event and the Current event. Together with these two events, additional observations of upwelling are presented from data obtained during the two scientific cruises aboard the RV Franklin described in Chapter 2.

Four different nutrient enrichment mechanisms appear to be active in the Smoky Cape region, these being, wind driven upwelling and three current driven mecha­ nisms related to the encroachment, topographic acceleration and separation of the EAC. These observations are now discussed and the mechanisms behind them are examined critically. Specifically it is the intention of this chapter to examine the mechanisms that drive cold nutrient laden water into the coastal region south of Smoky Cape.

The primary aim of this study is to examine the question of current driven upwelling with regard to the EAC by identifying and describing the key processes responsible CHAPTER 5. CURRENT DRIVEN UPWELLING 119 for upwelling and to investigate the frontal nature and three dimensional structure of the EAC in order to pinpoint the mechanisms driving upwelling. In this Chapter some of the mechanisms which determine the strength and coastal proximity of the EAC are examined, as are the concurrent uplift and upwelling events. The different upwelling mechanisms occurring on the mid-NSW coast are examined to elucidate the role of the EAC in current driven upwelling.

5.1 Nutrient Enrichment

The surface waters of the EAC are typically low in nutrient concentrations. Hal­ legraeff and Jeffrey (1993) found that in the vicinity of Smoky Cape, EAC surface waters have low nitrate concentrations ( < 2 µmolL - 1) increasing immediately west of the current towards the coast. Although generally low, nutrient concentrations in the NSW shelf waters have been found to vary both seasonally and spatially along the coast. Coastal nutrient concentrations in surface waters remain low throughout the year, with mean nitrate values of less than 1 µmolL- 1 . Mean nitrate levels at depths of 150 - 200 m were found to peak at 8 - 9 µmolL- 1 towards the end of winter (October) and decrease throughout the summer to a minimum of 2 µmolL- 1 in May (Rochford, 1984).

Upwelling in the Smoky Cape/Laurieton region has been seen to be a persistent feature during spring and summer. Rochford (1975) defined 'Laurieton Upwelling' as a rapid decrease in temperature, a decrease in salinity, a lowering of the surface oxygen concentration and in particular, an increase in surface nutrients such as nitrates. This upwelling was observed to begin approximately 9 km to the north of Laurieton moving 18-28 km in 5 days. The seaward extent of the upwelling (i.e. its effect on nutrients) did not extend more than 10 - 12 km offshore. Often a marked front is visible at the surface, between the nutrient rich upwelled water and the near CHAPTER 5. CURRENT DRIVEN UPWELLING 120 surface waters offshore, which are nutrient poor. It was found that newly upwelled water had temperature and salinity properties consistent with those found in deeper water off the continental slope. Rochford (1984) later observed that spring upwelling in the Smoky Cape/Laurieton region increases the surface nitrate concentrations to

4 - 12 µmo11- 1. Further to the north, at Evans Head, this spring increase also occurs although on a lesser temporal and spatial scale (Rochford, 1984).

5.2 Wind Driven Upwelling

The classic conceptual model of wind driven upwelling states that water in the surface layer is driven offshore by the alongshore component of the wind stress (rY) and that water upwells near the coast to replace the offshore directed Ekman transport (Smith, 1981). As a response to persistent alongshore winds (poleward on a western boundary), surface waters undergo offshore Ekman transport, followed by upwelling of deeper oceanic waters onto the continental shelf. Water is thus moved onshore up to the surface hence satisfying continuity. This replacement is confined to the coastal zone with an offshore scale given by the baroclinic radius of deformation. The cross-shore limit of coastal upwelling is defined by the internal Rossby radius NH f'V -- (5.1) ! Here H is the water depth, f is the local value of the Coriolis parameter and N is the vertically-averaged buoyancy frequency defined by:

(5.2)

where p0 is the mean density of the water column and p(z) is the density at any specific depth z. Calculations for the study region give values of Roi between 7-8 km at Smoky Cape and ,...., 5 km at Diamond Head. CHAPTER 5. CURRENT DRIVEN UPWELLING 121

The resultant across-shore Ekman Volume Transport (V) per unit length alongshore, is defined as: 7Y V=- (5.3) pf and is confined to the surface Ekman layer (He)- Here rY is the alongshore com­ ponent of wind stress and p is the mean density of the surface mixed layer. The average vertical upwelling velocity (w) can be approximated by:

7Y w=-- (5.4) pJRoi

The across-shore velocity (ue) of water, (depth averaged throughout the Ekman layer, He) attributed to an alongshore wind stress (rY) is given by:

7Y V Ue=--=- (5.5) pHef He

A schematic representation of wind driven upwelling is shown in Figure 5.1.

Persistent summer upwelling occurs off Oregon on the west coast of the USA for the duration of the summer season where strong equatorward winds drive surface waters offshore and an equatorward geostrophic current of up to 0.8 ms-1. Upwelling of cold saline water occurs near the coast, lowering surface temperatures by 3 - 4 °C over an 8-12 day period (Smith, 1981). Another region of persistent upwelling is on the Peruvian coast where seasonal southerly winds winds drive continual upwelling. With the onset of winter and a strengthening of the southerly winds, the surface temperature is observed to drop 5 °C in the space of a week and this upwelling continues until the arrival of spring where the upwelling favourable winds cease (Fahrbach and Brockmann, 1981).

In contrast to this, upwelling favourable (northerly /poleward) winds are rarely present for long periods along the NSW coast. Typically, local winds are highly variable in strength and direction. From the observations presented in Chapter 3, it is evident that the mean of the alongshore wind stress is small compared to its CHAPTER 5. CURRENT DRIVEN UPWELLING 122

Q-tY

Wind Driven Upwelling

Figure 5.1: Schematic representation of wind driven upwelling. rY is the alongshore wind stress, Ue is the offshore velocity component in the Ekman layer of thickness He and w is the vertical upwelling velocity that occurs within one internal Rossby radius, Roi from the coast. standard deviation and the wind direction rarely remains constant for any signifi­ cant length of time. Furthermore, from the analysis of the terms in the alongshore momentum equation (Equation 4.2) in Chapter 4 it is apparent that Ts is a relatively insignificant force. Hence wind driven upwelling is not a dominant process in the Smoky Cape region. It does play a limited role, however, and this is now discussed.

5.2.1 Observations of wind driven upwelling

Lowpass filtered wind records measured at Smoky Cape Light House are used to calculate the alongshore wind stress from 2 - 26 November 1998 (Figure 5.2). CHAPTER 5. CURRENT DRIVEN UPWELLING 123

0.15

0.075 as !!::... 0 -0.075

-0.15 2 4 6 8 10 12 14 16 18 20 22 24 26

Nov 1998

Figure 5.2: Lowpass filtered wind stress from winds recorded at Smoky Cape Lighthouse at South West Rocks, 2 - 26 November 1998. Circles indicate three periods. 1) Pre-experiment upwelling favourable wind stress 12 - 14 November; 2) Downwelling favourable wind stress 18 - 21 November; 3) Absence of significant wind stress 22 - 23 November 1998.

Three noteworthy periods are indicated: the first is from 12 - 14 November (i.e. in the two days prior to the first cruise), when persistent northerly (upwelling favourable) winds dominate, with a wind stress maximum of -0.1 Pa, (-0.4 Pa unfiltered). The second period occurs during the first cruise (18 - 21 November 1998) where the wind stress is strongly downwelling favourable. Winds recorded shipboard reached a maximum of 20 ms-1 from the southwest during this period. The third period is towards the end of cruise 1 (21- 23 November 1998) concurrent with an EAC encroachment event characterised by light northerly winds of little significance (Figure 5. 2).

Evidence of possible wind driven uplift at Urunga and Smoky Cape in response to upwelling favourable winds (Figure 5.2) is apparent 0in the hydrographic data obtained during cruise 1. Figure 5.3 and 5.4 show shore normal sections of (a) alongshore current (v), (b) nitrate and (c) chlorophyll-a at Urunga, Smoky Cape and Diamond Head (cruise 1, Figure 5.3) and Urunga, Point Plomer, Diamond Head Crowdy Head and Point Stephens (cruise 2, Figure 5.4) obtained during November 1998 and January 1999 respectively. ()

~ Urunga 16/11 Smoky Cape 1 6/11 Diamond Hd 20/11 Smoky Cape 22/11 Diamond Hd 23/11 'v 0 0 -5 0 ~ -0 .5 c..ri - 100 -1 '"'E ()

-150 V -1 .5 § , a a \ " I 1: a I 1· a ::a -200 \ . \\ l ~I • -2 0 ~ ~ t:, -50 :v I >---. £ - 100 a. ~ ,3 NO,&T - 150 I < b ~ -2 00 0 4 ~ t-< t-< -5 0 3

M ~ 2 'E G - 100 O> E l:Chl & T - 150 ' a C C + +++ + -2 00 1------~-~-' 0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 Distance offshore (km)

Figure 5.3: Row (a) shows cross sections of alongshore velocity (v) , where t he thick white line represents the zero velocity co11tour and negative velocities are southward (contour interval (Cl ) = 0.2 ms- 1). Rows (b) and (c) show cross sections of (NO3). uitrate (µmoJL - 1) aud Chl," chlorophyll-a (mg m- 3) respectively with over plots of temperature (T , white line, Cl =2 °C) during November 1998. The • in row (b) indicates the depth of the bottle samples and the + in row (c) shows the across-shore position of each CTD cast. T he locations and dates from lPft to right are: Uru11ga (lG/ 11/ 98), Smoky Cape (Hi/ 11/ 98), Diamoud Head (20/ 11/ 98), Smoky Cape (22/ 11/ 98) and Diamond Head (2:3/ 11/ 98)...... ,,.tv 0 ~ ~ ~ Urunga 21/1 Pt Plomer 23/1 Diamond Hd 24/1 Crowdy Hd 27/1 Pt Stephens 30/1 ::tJ 0 01 0 -50 I 0 -0 .5 § (/) -100 11 - 1 'E :u ~~ tr:l -150 ! V -1.5 a < r-"'.3 - 200 -2 t::J 0 15 ;::o >--,

~ -5 0 ~ 10 , E _, < £ -100 0 0.. E ::l. ~ o & T 5 - 150 IN 03 ID I' b ~ i 0 ~ -200 t-< 0 4 ()~ - 50 3 '" 2 'E -100 Ol E 1:Chl & T -150 ' a 1 ' c C C C + + + + + ++++ + + + ++ + + * + + + - 200 . 0 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 1 0 20 30 40 Distance offshore (km)

Figur e 5.4: As for Figure 5.3. during January 1999 at Urunga (21/ 01/ 99), Point Plomer (23/ 01/ 99) , Diamond Head (24/ 01/ 99), Crowdy Head (27 / 01/ 99) and Point Stephens (30/ 01/ 99).

...... l'v C)l CHAPTER 5. CURRENT DRIVEN UPWELLING 126

The ADCP transect at Urunga (16 November 1998) shows a strong southward flow with a maximum velocity of 1.6 ms-1, 40 km from the coast (Figure 5.3a). There is also a slight offshore component of the flow (0.1 ms-1) eastward of the 300 m isobath. The Urunga CTD transect (Figure 5.3b) shows the 14 °C isotherm uplifted by 100 m, from a depth of 200 m, approximately 40 km offshore to a depth of 100 m within 20 km of the coast. Associated with this is uplift of the nutrient isopleths. High concentrations of nitrate (10 µmolL - 1) are uplifted 80 m at the shelf break (from 180 m to 100 m). Concentrations at the base of the euphotic zone are significantly greater than background levels obtained from CARS (7.5 µmo11- 1 at 100 m depth, 10 µmolL - 1 at 200 m). The chlorophyll-a peak of ,..._, 4 mg m-3 over the shelf break (Figure 5.3c) overlies the nutrient maximum and is associated with a density of 25 - 25.25 kgm-3 (T = 20 - 21 °C, S > 35.6 psu). In the SeaWiFS images from 13 - 15 November 1998 (not shown, because of patchiness), significant chlorophyll-a blooms extend the length of the NSW mid-north coast.

The Smoky Cape transect has similar characteristics, with the 14 °C isotherm up­ lifted 100 m in the water column (Figure 5.3b). Nutrient values are similar to the previous transect, except that they extend further towards the coast, thus covering a larger area and are constrained by the 100 - 200 m isobaths. The chlorophyll-a peak (- 4.5 mgm-3), which is located further inshore, extends almost to the surface

(Figure 5.3c). Again, this peak is associated with densities of 25 - 25.25 kgm-3, overlying the nutrient maximum and the dissolved oxygen minimum and underlying the extent of the light penetration. Short term phytoplankton growth is restricted to the upper thermocline where in-situ light levels will support near maximum phy­ toplankton growth rates (Furnas and Mitchell, 1993). In a study centred at 32°S around the location of the EAC separation point, Hallegraeff and Jeffrey (1993) found the phytoplankton maximum mid-shelf at a depth of 30 - 40 m just below the mixed layer, where the nitrate concentration was 8 µmo11- 1. CHAPTER 5. CURRENT DRIVEN UPWELLING 127

Background (CARS) nitrate concentrations at Smoky Cape are only marginally higher at 100 m depth than those at Urunga (8.5 µmo11- 1), whilst at depths greater than 100 m they are the same. Observed nitrate concentrations are higher, with 15 µmo11- 1 uplifted from 250 m to less than 100 m, indicating the presence of significant uplifting. Surface nutrient concentrations, however, are negligible, which could be attributed to the rapid uptake of nutrient in the euphotic zone. This would account for the chlorophyll-a bloom in the surface waters evident in Figure 5.3c.

Remnant wind driven uplift is also present further south at Point Plomer (not shown). Very high fluorescence levels occur in two distinct pockets, one of which is high up on the shelf adjacent to the coast in less than 50 m of water. The other is overlying the shelf break and somewhat shallower than the nutricline. Maxi­ mum chlorophyll-a values are 4 - 4.5 mgm-3• The offshore chlorophyll-a pocket is found at the same depth as the Urunga pocket in warmer EAC waters (T = 20 °C, S = 35.6 psu), which indicates poleward advection of the bloom. Conversely, the coastal pocket of high chlorophyll-a is found in cooler, less saline waters (T = 18 - 19 °C, S < 35.6 psu) indicating the occurrence of uplift at the coast.

Rochford (1972) cites a southward spread of cold water intrusions at a rate of ap­ proximately 3.5 km per day. However, from the observations it appears that rather than spreading, the fluorescence is actually being advected southward at a more rapid rate as it is entrained into the EAC and dispersive diffusion is enhanced by the current shear.

5.3 Current Driven Upwelling

Rochford (1975) recognised that the proximity of the EAC to the coast plays a role in nutrient uplift. As the current moves from offshore to within an effective distance of CHAPTER 5. CURRENT DRIVEN UPWELLING 128 the coast, upwelling has been observed. Encroachment is defined here as the onshore movement of a current at any latitude as time progresses. This onshore movement can be delineated by the inshore edge of the temperature front, which is most evident in satellite images of sea surface temperature, or by the distance from the coast to the core of the current as is evident in cross sections of alongshore velocity. Bringing the current within closer proximity of the coast aids pre-conditioning of the isotherms and if the strength of the current is sufficient, enhanced Ekman pumping will occur through the bottom boundary layer.

To examine the relationship between alongshore current and bottom temperature, simple flat bottom theory can be used to investigate the across-shore mass transport. The mass transport in the bottom boundar:r layer in the across-shore direction (Mx) can be approximated by: (5.6) where v9 is the average geostrophic current, p is the density of the water and <5 is the boundary layer thickness, given by:

(5.7) where vis the kinematic viscosity of the fluid. However from Gill (1982, p332):

(5.8)

where Cd is a drag coefficient and v9 is the average geostrophic velocity. Furthermore the current driven vertical velocity (we) across a length scale, taken to be the internal Rossby radius (Roi), can be defined through the volume flux conservation equation:

1 WeRoi = 2lv9 l<5 (5.9) which upon substitution of (5.7) and (5.8) gives: _ -2Cdlv lv 9 9 (5.10) We - 1/IRoi CHAPTER 5. CURRENT DRIVEN UPWELLING 129

Current Driven Upwelling

Figure 5.5: Schematic representation of current driven upwelling. The average alongshore geostrophic current is v9 , Mx is the mass transport through the BBL of thickness 8 and We is the current driven uplift that occurs within one internal Rossby radius Roi of the coast.

A schematic representation of current driven upwelling is shown in Figure 5.5.

Onshore advection due to encroachment is readily distinguishable from onshore ad­ vection as a response to downwelling favourable winds. Whilst downwelling causes homogenisation and depression of the isotherms, encroachment causes uplift. In the encroachment case the entire axis of the current moves onshore as a jet. Rather than reversing the direction of the barotropic current as in the downwelling case encroachment can act to accentuate the alongshore current and cause nutrient en­ richment. CHAPTER 5. CURRENT DRIVEN UI'vVELLING 130

5.3.1 EAC Encroachment, Pre-conditioning and Upwelling

Satellite imagrs of Sra Surface Temperature (SST) are particularly useful for obtain­ ing a synoptic view of the EAC system. SST imagrs of the EAC pre-encroachment (15 November 1998) and post-encroachment (21 November 1998) arc shown in Fig­ ure 5.6a and Figure 5.6b respectively.

b

31 S

23

32 S u

22

33 S

21 34 S

151 E 152 E 153 E 154 E 151 E 152 E 153 E 154 E

Figure 5.6: SatC'llitf' images of sea surface temperature (°C) on a) 15 November 1998 and b) 21 NovC'mbcr 1998. The EAC is represented by the warm (24 °C) water flowing southward. parallel to tlw coast. overlying the 200m isobath (black line).

On 15 overnber 1998 the edge of the current dclinratcd by the thermal front ovcrlirs the shelf brea.k along the 200 m isobath at Smoky Cape (Figure 5.6a). Figure 5.6b shows that one week later the current is more defined and has moved closer to the coast. There is a marked front intersecting the coast between Smoky Cape and Diamond Head where the warm EAC encroaches upon the cooler coastal waters. CHAPTER 5. CURRENT DRIVEN UPWELLING 131

Surface currents measured by ADCP along the four transects from 14-23 November 1998 reflect this pattern. Figure 5. 7a shows the velocities before the encroachment. At the two northern transects (Urunga and Smoky Cape) the current flows parallel to the 200 m isobath at a speed of more than 1.5 ms-1. At the two southern transects (Point Plomer and Diamond Head) the current flows diabathically with an eastward (offshore) component. Inshore of the southward flowing current there is evidence of a weak northward flow extending up the coast.

14-19 November 1998 21-23 November 1998

152E 153E

Figure 5. 7: Vector plot of the current velocity {ms-1 ) at 16m depth, measured underway using the shipboard ADCP for a) 14 - 19 November 1998; and b) 21 - 23 November 1998.

From the ADCP transect at Smoky Cape on 15 November 1998 (Figure 5.3) it is evident that prior to the encroachment of the EAC, the current's core flow lies off- CHAPTER 5. CURRENT DRIVEN UPWELLING 132 shore, with a maximum speed of 1.6 ms-1 positioned over the 500 m isobath and an across-shelf flow of,....., 0.2 ms-1 eastward of the 300 m isobath. After encroach­ ment on 21 November 1998 (Figure 5.3) the core of the current has moved onshore significantly, with very strong coastal currents of 0.4 ms-1 at 2 km from the coast in a depth of 50 m, where previously the alongshore current was negligible.

After the strong southerly winds (18 - 20 November 1998), significant homogeni­ sation of the surface waters in the coastal region is evident at Diamond Head (20 November 1998), accompanied by a deepening of the mixed layer (Figure 5.3b). As­ sociated with the depression of the isotherms is a weak northward current in the near shore zone, with a maximum speed of 0.4 ms-1. The maximum southward cur­ rent of 1.2 ms-1 is found 45 km offshore. Diabathic flow is a positive maximum in the surface, associated with the maximum southward flow. Although concentrations of surface nutrient and chlorophyll-a are low due to homogenisation and depression of the surface waters, nitrate concentrations at depths of 100 - 200 m remain high (Figure 5.3b,c).

Onshore advection resulting from the alongshore wind stress is calculated using Ekman theory (Equation 5.5). Using the 3-hourly lowpass filtered wind records to calculate the alongshore wind stress at Smoky Cape and assuming the Ekman layer depth, He = 75 m and a mean Ekman layer density p = 1024.5 kgm-3 , onshore advection occurs at a rate of 0.01 ms-1 , which is equivalent to 3 km in 2 days. However, from the ADCP transects, the core of the current actually moved more than 6 km towards the coast in the two day period. Using the maximum (3-hourly filtered) wind stress of 0.2 Pa, the surface waters would move 6 km onshore in 2 days. However this instantaneous maximum, did not persist over the 2 day period, hence the onshore movement of the current cannot be attributed to local wind forcing.

By 21 November 1998 the onshore movement of the current rapidly overwhelms CHAPTER 5. CURRENT DRIVEN UPWELLING 133 the downwelling scenario. Surface currents are a maximum at Point Plomer (Fig­ ure 5.7b) and are again parallel to the shelf break. In the Smoky Cape section (22 November 1998), homogenisation resulting from the southerly winds is still ev­ ident in the surface mixed layer (Figure 5.3), yet as the current moves onshore the isotherms are again uplifted parallel to the bottom topography.

In the space of only four days the isotherms are also uplifted at Diamond Head and stratification occurs in the temperature profiles (Figure 5.3). As the core of the EAC moves onshore at Smoky Cape, uplift and hence pre-conditioning of the coastal waters is re-established. The 14 °C isotherm is uplifted over a vertical distance of 130 m whilst the surface temperatures have increased by 2 °C as the warm surface waters move onshore. Hence it is clear that as the core of the current moves closer to the coast, the coastal currents are accelerated and the isotherms are uplifted. This allows nutrients to be brought to the surface which results in a phytoplankton bloom.

The size and extent of the resultant fluorescence plume is larger by comparison than that observed previously at Diamond Head and those further to the north at Smoky Cape. A chlorophyll-a maximum of 2.5 mgm-3 has formed, associated with the thermocline, immediately below the pycnocline (Figure 5.3c), where 4 days previously there had been no evidence of biological activity.

From 15-20 November 1998 the entire current system encroaches towards the shore. The weak and variable winds are neither strong enough nor persistent enough to counteract such a strong downwelling event which occurred prior to the onshore encroachment. Thus it follows that the observed uplift of the isotherms is a rapid response to an EAC encroachment event which results in a bloom of phytoplankton. A schematic representation of encroachment driven upwelling is shown in Figure 5.8. CHAPTER 5. CURRENT DRIVEN UPWELLING 134

\\

Before encroachment

After encroachment

Figure 5.8: _Schematic representation of before and after encroachment showing how current driven upwelling occurs in the BBL as the current encroaches upon the coast. The perpendicular lines represent a southward flowing current. The solid line represents the line of zero velocity and the dashed lines represent a weak northward counter current. The oval represents a region of high bottom stress. Also shown are the alongshore current (v), mass transport (M.,) through the BBL of thickness 8 and the resultant current driven uplift (we) that occurs within one internal Rossby radius (Roi) of the coast. CHAPTER 5. CURRENT DRIVEN UPWELLING 135

The Current Event

Another example of encroachment driven uplift occurred during the Current event. As outlined in Chapter 3, the SST images in Figure 3.17 show that the core of the EAC moves closer to the coast over a 10 day period. Figures 5.9 and 5.10 show time series of the alongshore wind stress (7Y), the alongshore currents and the surface and bottom temperature during the Reversal event and the Current event mid-shelf at Smoky Cape and Diamond Head respectively.

The time series of the alongshore wind stress (7Y) shows that the winds are generally weak. The only increase in the (upwelling favourable) wind occurs towards the end of the Current event, when the maximum alongshore wind stress is little more than 0.05 Pa. This event is of only 48 hours duration and occurs more than 6 days after the temperature decrease begins in the BBL.

The dependent relationship between the current and temperature in the BBL is evident across the narrow shelf at Smoky Cape during both the Reversal event and the Current event (Figure 5.9). During the reversal, the barotropic alongshore cur­ rent fluctuates with a period of 2 days. Coincident with this is a 1 °C temperature change which lags the current by half a day in the BBL. At the end of the Reversal event (2 January 1999) the current reverses for the final time and again turns south­ ward and the temperature begins to decrease, lagging the current by 24 hours in the BBL. As the current continues to accelerate the temperature decrease continues, until after 8 days the current ceases accelerating and finally weakens by 30 cms-1 in the surface waters. Following this weakening the temperature in the BBL again increases, presumably as the mass transport through the BBL decreases and water sloshes down the slope again.

Mid-shelf at Smoky Cape, the current changes direction from a northward flow of 0.25 ms-1 · to a southward flow of more than 1 ms-1 in the space of 9 days CHAPTER 5. CURRENT DRIVEN UPWELLING 136

30 b 0 2

-II) 6 ' E -30 0~ -2. I-.. > -60 ········:• ...... :,. ····· ••:• . ····· .. -2 . . -90 ...._ _ __,_ ____,______,_ __..__ ...... __.__ _,_ __ _,__ __....___,__.______,_4 30 r--.---,---,-----i-:7,i;:--,--.----,---r--r-;::::c:==:::::::;i4 -am C ·-· 35m 0 - ....: ...... :...... - - 48m 2 . . . . I . . --II) . . 6 i-30 0~ I-.. > -60 -2

-90 ...._ _ __,_ ____,______,_ __..__ ...... __.__ _,_ __ _,__ __....___,___...._ _ ___,_4 24 26 28 30 3 5 7 9 11 13 Jan 1999

Figure 5.9: Time series of the alongshore wind stress (rY), the alongshore depth averaged cur­ rents {ii) {bold line) and the temperature anomaly {T0 ) during the Reversal event (R) and the Current event (C) at Smoky Cape. The thin lines show the temperature at various depths as indicated in the legend. CHAPTER 5. CURRENT DRIVEN UPWELLING 137

~;,r 6±± :11 et ±;tEj 30 1 --r----,r----.----r--=r-i----.--.------.-r;::::::i:===::::;i 4 b 0 2 '0- 6 E -30 ...... 0 't.,. ,£. I-.. > -60 -2 . . ; VDHB & TDHA -90'------'----'---'--...... -'-....._ __ .__ ...... __.....____.__.__~ -4 30 1 --r----,r----.----r--=r-i----.--.------.-r;::::::i:===::::;i 4 b 0 2 6 0 't.,. I-.. ----~,. /-i-----" . .! -60 -2

-90'-----'----'---'--...... ~--....._-~.__ ___._ __.....____.__.__~ -4 24 26 28 30 3 5 7 9 11 13

Jan 1999

Figure 5.10: Time series of the alongshore wind stress (rY), the alongshore depth averaged currents (ii) (bold line) and the temperature anomaly (Ta) during the Reversal event (R) and the Current event ( C) at Diamond Head. The thin lines show the temperature at various depths as indicated in the legend. CHAPTER 5. CURRENT DRIVEN UPWELLING 138

(2 - 11 January 1999). In this time the temperature in the BBL drops 4.5 °C.

To the south at Diamond Head (Figure 5.10), the reversals actually lead those seen at Smoky Cape, however during the Current event the decrease in temperature lags that at Smoky Cape and is not only seen in the BBL, but also in the surface waters. At Smoky Cape the decrease in temperature is confined to the deeper waters (Figure 5.9c), with the surface waters actually increasing in temperature as the axis of the current moves onshore. At Diamond Head, initially the temperature in the surface waters increases with the onshore movement of the current. However with time the cold water is actually upwelled all the way to the surface (Figure 5.10c). This possibly indicates an uplifting of water at Smoky Cape and an advection of colder water southward which is eventually upwelled further to the south.

As there is a phase lag between events at the two arrays, especially at the outer mooring, the decrease in temperature at Diamond Head begins before the current has reached full strength at Diamond Head. This can be attributed to downstream advection of colder water from Smoky Cape. As the core of the current moves onshore the current over the continental shelf accelerates resulting in uplift of colder water via the BBL further to the south on the coastal side of the temperature front.

5.4 Comparison of the Wind and Current events

As described in Chapter 3, there are two prolonged periods where the temperature in the BBL decreases significantly. The first of these (the Wind event) coincides with strong northerly (upwelling favourable) winds that blow for a 3 day period. The second event coincides with the encroachment of the EAC upon the coast.

Cross correlations between the alongshore wind stress (,Y) and the surface currents at Smoky Cape and Diamond Head during the Wind event and the Current event CHAPTER 5. CURRENT DRIVEN UPWELLING 139 are shown in Figure 5.11.

Wind Current scv:

-0.5 i,-,-.-,--~~.,..,-+-,~~~~:,:±,,:z=::c:::al~ - SCA15m ·-· SCA35m '----~--~Y -- SCB 20m -1 L..---~---~-----' -1a 2 4 .... SCB50m 0 2 4 6

DHU:

0.5 ...... , .... ·

- 0-5 r-~-~,;--,,,...... ':.-__~- :,7 .. ~.. .,-;,.:-;:,: ..:: ..::·""· o""'·H::,:,AA-15-m_, ·-· DHA20m -- DHA35m ~2-----"-14 .... DHB 20m -1 ,___--~---~-----' -1a,___ __ 0 2 4 6 Lag (Days) Lag (Days)

Figure 5.11: Cross correlations between surface currents and alongshore wind stress (rY) at Smoky Cape and Diamond Head during the Wind event and the Current event.

The alongshore currents at Smoky Cape are positively correlated with the alongshore wind stress with little or no lag. At Diamond Head the across-shore currents lag the wind stress from 2.4 days in 15 m of water at DHAA, up.to 5 days at depth (35 m) at DHA. For comparison the correlations from the Current event are also shown. Towards the end of this period a northerly wind event occurs (Figure 3.5) which appears to have little effect on the current regime. At Smoky Cape the correla­ tions show that the alongshore current is negatively correlated with the alongshore wind stress, which represents a southerly current and an opposing (downwelling favourable) wind. After a lag of 4 days the alongshore wind is again positively correlated with the alongshore surface currents. Across the southern array the cur- CHAPTER 5. CURRENT DRIVEN UPWELLING 140 rents are inversely correlated with the alongshore wind stress, however not with any degree of significance.

5.4.1 Vertical Uplift

The average wind driven vertical velocity (w) was calculated from time series of the mean hourly wind stress (Equation 5.4). This uplift velocity was then compared with the temperature anomalies at each mooring. The results for the most inshore mooring of each array are shown in Figure 5.12.

The maximum negative uplift, (indicating downwelling) occurred between 18 - 20 November 1998, which was before deployment of the moorings. The CTD transect at Diamond Head reflects this downwelling and homogenisation of the water column (Figure 5.3b,c). Figure 5.12 clearly shows two extended periods of positive uplift. The first, which occurred during the Wind event, is centred around 13 December 1998 and shows homogenisation of the water column and a temperature anomaly of -3 °C for a 3 day period at both SCA and DHA. Cross correlations were calculated between temperature and uplift (not shown). Maximum correlations occur in the bottom boundary layer with a time lag of approximately two days. The correlations decrease in significance moving up through the water column.

The second of the two events occurred during the Current event, starting from 2 January 1999. The event occurred over a more extended period, resulting in a greater decrease in temperature (tlT = -4 °C at SCA). However the magnitude of w was less than that during the Wind event and the temperature began to drop before the wind driven uplift commenced. This suggests that although northerly winds did occur during the Current event, there was another mechanism driving the decrease in temperature. CHAPTER 5. CURRENT DRIVEN UPWELLING 141

6 ~-~---,--~~-~-~-~--..-----,--..--,-----.----,r---,10 ;. a E1 W E2 ; SCA: 3 f ,,. . ... '.. 5 I-~:: : . . 6 . . C?..... 0 I-OI . . . . - am ...... , ...... , ...... , ...... , . -3 . . -- 35m -5 , ... 48m - w L__lJLL__----1.__ 1-L_....1,__j____..1. _ _.l_L__ _L_-----1_ _.l_...L__ ___l_.:::::::::::Jc::::::...J_10 -6

6.---~-~--~~-~-~--,~--..-----,--..--,-----.-----,--,10

b DHAA 3 . '..... 5

. . - am -3 ...... -- 16m -5 .... 28m - w -6 L__jjlL__ ___..1. __L_L__L_L___.l _ _i_J___ _i__ __J_ _j_..L__---1....::::::==c:::::::'.J _ 10 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 5.12: Time series of wind driven uplift (w) in metres per day (Equation 5.4) and (T0 ), temperature anomaly (°C) at a) Smoky Cape A and b) Diamond Head AA. As in subsequent figures, each event is marked EAC 1 (El), Wind event (W), Reversal event (R), Current event (C) and EAC 2 (E2).

Estimated Isothermal Displacement

To further examine the relationship between alongshore wind and current and resul­ tant upwelling, a simple dynamical model is utilised. To a first order, the wind driven upwelling velocity (w) can be calculated by the simple coastal upwelling model given in Equation 5.4, (Figure 5.12). The current driven upwelling velocity (we) can be approximated by Equation 5.10. These estimates of upwelling velocity can then be used to calculate the expected temperature anomaly Tw and Tc respectively using a CHAPTER 5. CURRENT DRIVEN UPWELLING 142 regression based on the dynamical model:

8T -W- (5.11) 8z 8T -w­ (5.12) e 8z where T = T(z, t) is an estimate of the measured temperature field as a function of depth (z) and time (t) (Kelley and Bourque, 1997). In this case the estimate of the background stratification 8T/ 8z is calculated from the temporal mean of the temperature time series presented in Chapter 3.

The estimated temperature anomalies Tw (wind driven temperature anomaly) and Tc (current driven temperature anomaly) are calculated during the Wind event and the Current event and are shown in Figure 5.13. As the temperature anomaly is calculated as a temporal integral, there is no allowance for relaxation once the wind or current event subsides. Hence, using this model, a strong reversal is needed to change the temperature in the opposite direction. For this reason the estimated tem­ perature anomaly is calculated discretely over the two events. In general, estimates of Tc were not found to be as accurate as estimates of Tw. As We is dependent on the square of the average current any misrepresentation of the current driven uplift is amplified. Furthermore the underlying theory is based on the flat bottom assump­ tion, hence across a sloping bottom, over-estimates are to be expected. However, a brief comparison of the results of the two methods is worthwhile.

During the Wind event there is good agreement between the estimated tempera­ ture Tw and the mean bottom temperature anomaly Ta as measured at the inshore moorings at both Smoky Cape and Diamond Head. The agreement is best at DHA where the flow appears to be wind driven. A temperature anomaly of up to 4 °C was recorded at DHAA. Tw accurately estimates the magnitude of the temperature anomaly, but slightly underestimates the timing. Inshore at SCA the temperature does not drop as much as Tw predicted. This could either be because the currents CHAPTER 5. CURRENT DRIVEN UPWELLING 143

Wind Current 4 / / / / / .,,, 2 I / / I ,... - - 6 0 ··-·- ~ ' ·- I-"'-2 '\ ' '\ I ' I -4 - '\ _/ SCA SCA

-i I / a I I 2 I I I I C / l1J 6 0 -·-·'t"·-·-··-·- ~ I --- "' ' I- -2 - -4 DHM DHAA -6 8 10 12 14 2 4 6 8 Dec Jan 1998 1999

Figure 5.13: Observed temperature anomalies Ta, wind driven temperature anomalies Tw and current driven temperature anomalies Tc, at Smoky Cape A (top) and Diamond Head AA {bottom).

are still affected by the EAC, or because the estimation of the background stratifi­ cation was not realistic at this time. Tc, which increases steadily at both moorings does not, however, represent T0 • This occurs as Tc is calculated as the temporal integral of a steady current. Furthermore the nature of the calculations is such that they are very dependent on the starting point.

During the Current event, Tc decreases rapidly as a response to a sudden current acceleration. Although the timing of the temperature decrease is more sudden than that actually observed, the magnitude of the response is comparable. At Smoky CHAPTER 5. CURRENT DRIVEN UPWELLING 144

Cape, Tc > Ta by ,...., 1 °C, whereas at Diamond Head, Tc < Ta by ,...., 1 °C. These discrepancies could be due to a mis-representation of either the stratification or Roi (which changes inversely proportionally to the stratification) during this period of strong upwelling. There is a slight wind driven decrease in the temperature (Tw) towards the end of the period, however at both moorings the measured temperature anomaly Ta is up to 2 °C greater than Tw,

From comparisons with the actual isothermal uplift measured at the Smoky Cape and Diamond Head moorings, it is apparent that local wind forcing was responsible for the uplift of the isotherms during the Wind event. It was not, however, a major contributor to the uplift of isotherms in this region during the Current event.

5.5 Topographic Acceleration

After examining correlations between wind direction and upwelling, Rochford (1975) suggested that the narrowing of the continental shelf to the north of Laurieton was a contributing factor to occasional (approximately 5 times per year) upwelling events. Numerical simulations presented by Oke and Middleton (2000) indicate that alongshore topographic variations can impact the shelf circulation by accelerating the alongshore flow through a narrowing shelf cross section. As a result, near bottom currents increase, thus increasing bottom stress on the shelf and slope which acted to drive an onshore flow through the bottom boundary layer. The simulations showed that as the EAC accelerates due to the funnelling effect of the narrow continental shelf, the gradient Richardson number Ri:

Ri= N2 (5.13) u2z + v2 z is reduced to below the critical value (Ri = 0.25). A reduction in Ri enhances vertical mixing, thus decreasing the buoyancy frequency (N, Equation 5.2), which CHAPTER 5. CURRENT DRIVEN UPWELLING 145 in turn reduces the Slope Burger number: N2a2 Bu=p (5.14) where a is the bottom slope. This enhanced mixing and the subsequent reduction of the Slope Burger number acts to lengthen the time it takes for the bottom bound­ ary layer to shut down (MacCready and Rhines, 1993). Hence deeper waters are transported up the slope for a longer period of time resulting in upwelling.

A paper by Oke and Middleton (2001), which examines nutrient enrichment off Port Stephens, does not make the distinction between separation induced upwelling and topographically induced upwelling. In an idealised numerical simulation, they show the narrowing of the shelf topography at Smoky Cape causes uplift of slope water onto the continental shelf, which is then upwelled to the surface as the EAC separates from the coast. In this study the two scenarios are isolated and examined independently, however it is possible that they are actually related at times and it is the topographic effects that make the separation so clean at Smoky Cape. Furthermore, encroachment of the core of the EAC may contribute to topographic acceleration of the current which can result in upwelling. From the time series obtained it is possible to calculate the Burger number in order to investigate the relationship between encroachment, topographic acceleration and upwelling.

Burger Number Time Series

The time series of Burger number calculated at each mooring is presented in Fig­ ure 5.14. Mid-shelf at both Smoky Cape and Diamond Head that the Burger number is reduced below the critical value of Bu = 0.2. This is because Bu is dependent on both the buoyancy frequency and bottom slope. At Diamond Head Bu is con­ sistently lower than that at Smoky Cape, which is related to less stratification and a more gentle bottom slope. CHAPTER 5. CURRENT DRIVEN UPWELLING 146

102 ... ·...... , ...... /··· ...... ···.\: ... /\.. r·--· ..... ·-._ .. ·.. ·.. -.-· ...... 10' 0 C: !ii 100 ::,e ID 10-'

10-2 14 21 28 5 12 19 26 2 9 16 23 30

102

' ,".,,., ,''· ·"'- i'...... 10' ·""""" I . . ~ \" r . I· ,\ /~·,r. • ,..1 I v'.. J I~ \ 0 ' 11 I .\ ., ·' ."\_ ., .. , ."· I C: ,. i \ .. 1'. 11:1) / 1/ If·! ·,,. 1 !ii 100 ...... ·...... :•: ·. ...-·"\..\;-··",~·\.\i.. :·} .. _/\ .... ~- ::,e ID :: : .'·: 10-'

10-2 14 21 28 5 12 19 26 2 9 16 23 30

Figure 5.14: Time series of Burger number (Bu) for each mooring, at Smoky Cape (top) and Diamond Head (bottom). The horizontal black line represents the critical value Bu = 0.2.

The buoyancy frequency (N) was calculated at Smoky Cape and Diamond Head using an average of four repeat hydrographic stations at each location. In Figure 5.15 the mean values of N (cph) are plotted at the 50 m and the 100 m isobaths at both locations during November 1998 and January 1999.

The figure shows that the vertical structures of the buoyancy frequency at the 100 m isobath are fairly similar in shape at Smoky Cape and Diamond Head, with maxi­ mum values at about a depth of 50 m. This indicates that mid-shelf conditions were similar at both locations, which is to be expected as the two cruises coincide with EA C 1 and EA C 2, where the EAC is flowing adjacent to the coast at both mooring arrays, as described in Chapter 3. Inshore at the 50 m isobath however, where the CHAPTER 5. CURRENT DRIVEN UPWELLING 147

0 0 ~ ············· -20 · .... -20 ...... ··.·._·.·············· ········. . g -40 ? ··-_·:::::::::::. -40 .t:: a -60 -60 2l -80 -80 SCA DHA -100 -100 0 5 10 15 20 0 5 10 15 20

0 0

-20 -20

-40 -40 g •,. i -60 -60 2l -80 -80 SCB DHB -100 -100 0 5 10 15 20 0 5 10 15 20 N(cph) N(cph)

Figure 5.15: Buoyancy Frequency (N) in cycles per hour (cph) at Smoky Cape and Diamond Head, at both the 50 m (SCA, DHA) and 100 m (SCB, DHB) isobaths, during November 1998 (-) and January 1999 (:). The profiles are calculated from an average of four hydrographic stations at each location. effects of the EAC at Diamond Head are less during EAC 1 the shapes of N differ, i.e. the waters are less stratified and N is lower (Figure 5.15).

Figures 5.16 and 5.17 show time series of alongshore .velocity (v), temperature anomaly (Ta) and Burger number (Bu) mid-shelf at Smoky Cape and Diamond Head respectively.

At Smoky Cape where the continental shelf is steeper and narrower than that at Diamond Head, Bu is often large. The only period when it is consistently below the critical value of 0.2 is from 12 - 19 December 1998, towards the end of the Wind event when the temperature anomaly is low, although stable (Figure 5.16). However, CHAPTER 5. CURRENT DRIVEN UPWELLING 148

Smoky Cape B

0 1 ~ -50 £. > -100

" 2 ", .. " ... l ... 6 b ~o ·-·-·- ·-·- I-"' /:. . : -88m -2 .. ,: .. · - · 98m

1 ...... 0.8 ...... C ...... 0 •,···· ...... • ...... ~ ...... -· ...... ~ ...... C 0.6 . . (I) .• ...... :. 0.4 . . . e:::, ID 0.2 0

-0.2 .___ __,_ __"-L,._ __ .__..____.___,__....__..L...... 1'-----'---...._- ...... '-----'-''---...._--' 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 5.16: Time series of; a) alongshore velocity (v); b) Temperature anomaly (Ta) and c) Burger number (Bu) at Smoky Cape. The horizontal black line represents the critical value Bu = 0.2. CHAPTER 5. CURRENT DRIVEN UPWELLING 149

Diamond Head B so~-~--~--~~-~~~-~~--~--~~~--~--~~

0 ~~ -50

> -100 .. ; ...... :.. - 20m E1 W: R: C E2 : ·-· 50m -1soL_ _L_ __JL__...1.__j_---1__JL..1..,__L...L __..1__----1._...1_.,L__----1.....:::::==I::::::::'....J

2 ...... b .E 0 I-., . ... : -80m -2 ...... d...... · · · · · · ·:-- · - · 93m

0.8 C ci 0.6 ...C Q) e> 0.4 :::, ID 0.2 0 ...... ~ . -0.2 ~-~----'---_._~_,___.__~_~__.___ _._ __ .._____.___. _____._ __....._____, 14 21 28 5 12 19 26 2 9 16 23 30 Nov Dec Jan 1998 1999

Figure 5.17: Time series of; a) alongshore velocity (v); b) Temperature anomaly (Ta) and c) Burger number (Bu) at Diamond Head. The horizontal black line represents the critical value Bu = 0.2. CHAPTER 5. CURRENT DRIVEN UPWELLING 150 across the wider shelf further to the south at Diamond Head (Figure 5.17), there are two significant periods where Bu < 0.2 for an extended period of time. The first period coincides with that at Smoky Cape, the second period however occurs during the Current event. It is during this period that the acceleration of the current to the north at Smoky Cape causes an increase in bottom friction, which drives an onshore Ekman transport through the bottom boundary layer. The lower Burger number indicates that the bottom boundary layer remains open for a longer period of time, thus allowing prolonged upwelling to occur. The temperature anomalies show significant decreases also indicating upwelling during these two periods. The latter period ( Current event) is coincident with a significant current acceleration to the north at Smoky Cape.

Vorticity

Standard manipulations of the equations of motion show that potential vorticity Q is conserved along streamlines when flow is steady i.e. /+( ~ = Q (constant) (5.15) where H is the depth of the water and ( is the relative vorticity given by:

(=av_au (5.16) ax 8y The time series of the depth averaged vorticity is calculated between the Smoky Cape and Diamond Head mooring arrays. To do this the assumption is made that

8u/8y is constant in the alongshore direction. Figure 5.18 shows time series of 7Y and vat SCB and(, R0 and Tat SCA and DHA.

There are two periods where ( becomes more negative, these occur during the Wind event and the Current event. During the Wind event the increase in negative vor­ ticity occurs two days after the decrease in bottom temperature commenced (Fig- CHAPTER 5. CURRENT DRIVEN UPWELLING 151

~ OJJDciE-~~~ ~ LafT-1~7,'1 -0.1 L---L-----''----'---l....--J..1..L-1---'---.L...... L--.....__-___.__ __.__.....______.___ ..,_____,

2 0 ~

Figure 5.18: Time series of a) alongshore wind stress (rY); b) depth averaged alongshore velocity

(ii) at SCB; c) vorticity ((); d) local Rossby number (R0 ); and e) temperature (T) in the BBL at SCA. ure 5.18b,e). During the Current event however the increase in negative vorticity precludes the decrease in temperature in the BBL at both Smoky Cape and Diamond Head. These two incidents coincide with higher Rossby number which indicates an increase in non-linearity. Although large, during EAC 1 and EAC 2, both ( and

R0 remain fairly constant. During these events the temperature in the BBL also remains constant.

From the data it is not possible to deduce that the actual narrowing of the Shelf at CHAPTER 5. CURRENT DRIVEN UPWELLING 152

Smoky Cape accelerates the flow, however it is evident from the data that encroach­ ment of the EAC onto the continental shelf coincides with a non-linear acceleration of the flow over the shelf proper. It is during non-steady flow that the BBL remains open for a longer period thus driving colder water through the BBL into the surface waters at the coast. As the current moves onshore at Smoky Cape, the isotherms are uplifted, bottom friction increases and the bottom temperature decreases at both Smoky Cape and Diamond Head. At Diamond Head the colder, nutrient rich water is actually uplifted all the way to the surface, resulting in upwelling.

5.5.1 Hydrographic evidence of encroachment driven up­ welling

Figure 5.19a shows the EAC flowing southward on the 18 January 1999 in a narrow jet (,....., 100 km) which separates from the coast just to the north of Diamond Head. The surface waters are anomalously warm (27 °C), related to heating occurring in the Coral Sea possibly resulting from the strong La Nina phase (Berkelmans and Oliver, 1999). In the coastal regions from Diamond Head south to Port Stephens the surface waters are significantly cooler (20 °C).

One week later (Figure 5.19b) the jet structure appears less defined (owing perhaps to cloud cover in the images), however the point of separation has moved > 90 km to the south as the EAC water encroaches upon the coast. The vector plot of surface ADCP currents from 21 - 30 January 1999 (Figure 5.20) shows an overview of the prevalent conditions. At Urunga, a weak northward flow occurs near the coast due to a small recirculation north of the point and a narrowing of the jet towards Smoky Cape. The current velocities at Point Plomer are stronger than those to the north. South of the separation there is a slight northward current inshore. A vertical profile of the alongshore current from a shore normal ADCP transect at Urunga CHAPTER 5. CURRENT DRIVEN UPVvELLING 153

b

29

28

27

26 >-

25

24

23

151 E 152 E 153 E 154 E 151 E 152 E 153 E 154 E

Fig ure 5.19: Satellite image. of sea surface temperature (0 C) on a) 18 .January 1999 and b) 26 .J anuary 1999.

out to 35 km offshore is shown in Figure 5.4a. Inshore the currents arc northward flowing at 0.2 ms- L. Surface current. ( top 50 m) over the shelf break d crease with depth from 1. 5 ms- 1 to 1 ms- 1.

The CTD transect at Urunga shows the signature of the EAC in the temperature 111casurcmcnts with a maximum temperature of 27 °C in th surface wat rs associ­ ated with the core of the current. overlying the shelf break. Salinities in the surface EAC waters arc 35.4 psu (not shown) with a salinity maximum (35.7 psu) at 100 m depth. The isopycnaL reveal a gentle downturn towards the coast reflecting the weak northward flowing coastal current evident in the ADCP plots. There is a peak in chlorophyll- a below the EAC waters. i.e. where T < 25 °C. S> 35.5 psu. The

maximtun concentration is ,.-..., 4 mg m -:l located at a depth of 30 - GO m over the CHAPTER 5. CURRENT DRIVEN UPWELLING 154

21-30 January 1999

31S

32S

33S...__ _._ _ ___.'--_ __._...... ______, 152E 153E

Figure 5.20: Vector plot of the current velocity (ms-1 ) at 16m depth, measured underway using the shipboard ADCP, during the period 21 - 30 January 1999. CHAPTER 5. CURRENT DRIVEN UPWELLING 155 shelf break.

The shore normal ADCP transect at Point Plamer (23 January 1999) (Figure 5.4a) shows that the EAC has increased in strength to a maximum of 2 ms-1 above the shelf break, at a distance of 20 km offshore. The width of the core is less than 5 km across and the maximum temperature is 27 °C. Uplift of the isotherms is parallel to the bottom bathymetry (more than in the previous transect). Water with a temperature of 15 °C is uplifted more than 150 m to shallower than 50 m, whilst velocities of 1 ms-1 are found approaching the bottom of the water column; in all, a very rapid coastal current. There is also a highly stratified BBL, a flow of nutrients up through the BBL and the beginnings of a chlorophyll-a bloom in the surface waters with trace concentrations (rv 1 mgm-3).

At Diamond Head (24 January 1999), 90 km further to the south (Figure 5.4a), there is a strong thermal gradient between the warm (27 °C) EAC water flowing at 2 ms-1 southward and the cooler coastal water. The horizontal temperature change of 7 °C occurs over 4 km at a depth of 25 m. Associated with this thermal front is a change in current direction, with maximum velocities offshore from this point being 2 ms-1, whilst inshore of this region currents reverse and flow northward at

0.2 ms-1. Maximum current speeds occur over the shelf break and maximum current shear is in the surface waters, 15 km from the coast.

At Point Plamer, the highest concentrations of chlorophyll-a are found inshore of the front, at depth underlying the southward current, and are associated with the highest nutrient concentrations mid-shelf (Figure 5.4b,c). At Diamond Head however, the bloom now appears in the surface waters. This reflects the process of topographic upwelling and southward advection, beginning at Smoky Cape and resulting in a surface chlorophyll-a peak at Diamond Head.

An ADCP transect at Smoky Cape (not shown) reveals a weakening of the cur- CHAPTER 5. CURRENT DRIVEN UPWELLING 156 rent. The SST image (Figure 5.19b) shows a broadening of the EAC jet, which is associated with a relaxation in velocity. Maximum current speeds are 1.2 ms-1 and located 10 km offshore. Despite the core being within close proximity of the coast, as the current weakens the isotherms are lower in the water column and the upwelling response attributed to the western boundary current pre-conditioning is reduced. There is distinct horizontal stratification, with the less saline surface waters of low density extending from 4 km inshore all the way to the end of the transect, 20 km offshore. This contrasts with the pre-conditioned scenario where the isotherms, although stratified, are uplifted parallel to the bottom bathymetry. Thus it appears that upwelling is only induced topographically under certain current regimes, defined by proximity and velocity.

This implies that the narrowing of the continental shelf at Smoky Cape is increasing the bottom stress and the effects are felt downstream with the formation of an onshore BBL flow. Evidence of this is shown in the transect at Diamond Head. The data thus support the predictions of the model of Oke and Middleton (2000, 2001), as there is a significant decrease in the Burger number as the bottom slope changes.

Sensitivity experiments by Oke and Middleton (2001) showed that the magnitude of the topographically driven upwelling is dependent on the strength of the current. Furthermore, it was shown that when the slope of the uplifted isopycnal matches the bottom slope, along isopycnal mixing is optimised, t~us allowing slope water to be upwelled to the surface without the opposing effects of buoyancy. This implies that nutrient enrichment of the coastal waters will also be enhanced under these conditions. The results show that as the current weakens at Smoky Cape, the effects of pre-conditioning are reduced, horizontal stratification returns, and nutrient supply ceases as does the biological response. CHAPTER 5. CURRENT DRIVEN UPWELLING 157 5.6 Western Boundary Current Separation and Enrichment

A theoretical study by Janowitz and Pietrafesa (1982) found that on the cyclonic side of a fast flow with low Rossby number (i.e. where rotation effects outweigh advection) isobath divergence should induce upwelling and associated onshore flow. They give the example of the western side of the Gulf Stream, where the onshore transport associated with the divergence of a jet 100 km wide, carrying 100 Sv, is in the order of 2 Sv/100 km (or 20 m2 s-1). In the Southern Hemisphere, upwelling should be seen on the inshore side of the EAC as it separates from the coast.

It is known that the latitude at which the EAC separates varies temporally and spa­ tially. It appears that this variability is linked to the near semi annual (150 - 180 days) westward propagation of baroclinic Ross by waves across the Pacific Ocean. These Rossby waves appear to travel towards the east Australian coast and then propagate poleward along the coast as sea level anomalies with a speed of rv 0.06 ms-1 (Mata et al., 2000). As these anomalies move southward, eddies are pinched off from the main EAC flow. The point of separation moves northward as the eddy is shed and the process starts all over again. Inshore of the separated current there is often a large cyclonic eddy which through Ekman pumping raises the thermocline to a point where the region is pre-conditioned to upwelling. ~or this reason upwelling is often observed inshore of the separated EAC.

Tranter et al. (1986) observed a relationship between the separation of the EAC and regions of high chlorophyll-a concentrations immediately south of the separation point. They were not however able to adequately establish how the separation of the EAC resulted in upwelling events. CHAPTER 5. CURRENT DRIVEN UPWELLING 158

5.6.1 Separation of the EAC

From the sub-inertial alongshore current meter time series an image of the structure of the EAC jet can be formulated as it moves laterally across the continental shelf. Across the northern array the strength of the alongshore current at 80 m depth at the outer mooring, is larger than that at 80 m depth at the mid-shelf mooring. This indicates a strong jet with a deep core and high current shear. At Diamond Head, the currents measured in the middle of the water column over the 150 m isobath are weaker than those measured mid-shelf at 50 m depth for a lengthy period. This indicates that the jet is much shallower and the current shear is less, implying that the axis of the current has separated from the coast and the full effect of the EAC is not felt at Diamond Head.

However, this is not the case for two 10 day periods (EAC 1 and EAC 2) indicating that the EAC has moved. The axis of the current extends down the coast past Diamond Head and current strength at the outer mooring is stronger than at 50 m mid-shelf. This again indicates a strong jet with a deep core and high current shear.

A series of transects (both ADCP and CTD) were taken in January 1999 south of the point where the EAC separated from the coast. These transects were taken across the thermal front that was created as the anomalously warm current moved away from the coast. Two of these transects are shown in Figure 5.4. The first is taken at Crowdy Head, immediately poleward of the separation point. The second is taken at Point Stephens almost 150 km downstream of the separation point.

The ADCP transect at Crowdy Head (Figure 5.4a) shows that the current strength has decreased since separation to 1.6 ms-1 and the core is located 20 km from the coast. There is a strong thermal gradient across the front from 27 °C in the core to 20 °C near the coast. Isothermal uplift is present in the BBL where water of 18 °C extends to within 5 km of the coast. There is a slight pooling of nutrients CHAPTER 5. CURRENT DRIVEN UPWELLING 159 occurring 5 - 10 km from the coast at a depth of 50 m where nitrate concentrations reach 5 µmo11- 1. Associated with this is a surface chlorophyll-a signature, inshore of the thermal front which underlies the warm (26 °C) less saline (35.3 psu) surface waters, coinciding with the region of highest dissolved oxygen. This chlorophyll-a bloom increases in size and concentration moving southward (not shown).

The strong frontal structure continues southward and is still present in the most southerly transect at Point Stephens (Figure 5.4). Current strengths remain high in the warm water, whilst inshore a current reversal occurs, where a weak north­ ward current (0.2 ms-1) extends 10 km offshore. The core of the current is more than 45 km from the coast, due to the fact that the coast bends westward south of Sugar Loaf Point. The vertical change in velocity (8v/8z) is large, associated with the strong temperature gradients across the front. The nutrient plume has expanded so that it now extends for 15 km across the continental shelf, with nitrate concentrations reaching 9 µmo11- 1 in the centre of the plume. Most significantly, however, is the expansion of the chlorophyll-a plume which now overlies the nutri­ cline, extending from the front to within 5 km of the coast. It should be emphasised that this chlorophyll-a bloom does not reach the coast, indicating a frontal process rather than a coastal process. Concentrations are highest(> 3 mgm-3) immediately inshore of the thermal front.

As the current separates from the coast (or as the directi\m of the coast curves away from the current), the current strength decreases downstream of the separation point, although only marginally. Nutrients, however, are brought to the surface as the flow proceeds downstream, with a strong bloom resulting inshore of the front. Although concentrations are lower than those associated with the previous scenarios, the nutrient pool covers a far greater region over the continental shelf. The chlorophyll-a concentrations are of similar magnitude to those seen previously, however their extent is again far greater. Both the nutrient pool and the chlorophyll- CHAPTER 5. CURRENT DRIVEN UPWELLING 160 a bloom are centred at a distance of - 18 km from the coast. This distance is greater than the internal Rossby radius (R 0 i) for this region, i.e. outside the direct area of influence of wind driven coastal upwelling. Furthermore, northerly (upwelling favourable) winds are not present. At the same time an EAC relaxation occurs as the current broadens and decreases in strength and the alongshore steady flow that results is not upwelling favourable. Despite a strong temperature gradient, the uplifting and upwelling effects are minimal.

5.7 The Magnitude of the Upwelling

The previous sections have provided a delineation of the different mechanisms which drive slope water into the surface waters. It is now desirable to identify and com­ pare the magnitude of these events. A review of the various contributions to shelf edge circulation by Huthnance (1995) shows that the circulation contribution from a western boundary current is at least one order of magnitude larger than either coastal currents or slope currents in other regions, regardless of the forcing. Huth­ nance (1995) then compares the magnitude of cross shelf exchange associated with the various physical processes occurring at the shelf edge. In his comparison cross shelf exchange resulting from upwelling attributed to the divergence of a western boundary current is 20 times greater than the exchange associated with wind driven upwelling. Thus intuitively it could be hypothesised that the associated upwelling should also be larger.

Rochford (1975) defined upwelling in the Smoky Cape/Laurieton region as a rapid drop in the sea surface temperature. Here there is a temperature decrease of 3 °C in the surface waters at Smoky Cape and at Diamond Head during the Wind event. During the Current event the temperature change is not seen in the surface waters at Smoky Cape, however at depth and in the surface waters at Diamond Head the CHAPTER 5. CURRENT DRIVEN UPWELLING 161 drop is greater (!:!,.T = 4 °C). Furthermore, the isotherms are depressed for a longer period of time.

To quantify the magnitude of the different upwelling events, it is helpful to examine the ratio of the wind driven vertical uplift (5.4) to the current driven vertical uplift (5.10):

Current 2Cdlv9 lv9 / f R (5.17) Wind 7Y/pfRoi 2pCdv 2 (5.18) 7Y

Which when substituting standard values (Cd = 0.002 rns-1, p = 1000 kg rn-3 , T =

0.1 Pa) gives a ratio of rv 20 x v2 • Hence it follows that in this region where wind stress is minimal and currents are strong (v rv 1 rns-1 ), current driven upwelling is much more substantial than wind driven upwelling.

Nutrient Concentrations

The magnitude of an upwelling event can be quantified by comparing the concentra­ tion of the nutrients that are brought to the surface. (Hallegraeff and Jeffrey, 1993; Furnas and Mitchell, 1993). Measured nitrate concentrations are shown for the Urunga (15 November 1998), Diamond Head (24 January 1999) and Point Stephens (30 January 1999) sections in Figure 5.21 as indication~ of the upwelling concen­ trations during a wind driven upwelling (Figure 5.21a), an encroachment driven upwelling (Figure 5.21b) and a separation induced upwelling (Figure 5.21c). For comparison, plots of background nitrate concentration derived from CARS data are shown for the same three sections (Figure 5.2ld,e,f).

The U runga cross section shows that measured subsurface nutrient concentrations after a wind driven uplifting event are 50% greater than background concentrations, (14 µrnolL- 1 at 150 rn depth). The increase in nutrient is generally restricted to the CHAPTER 5. CURRENT DRIVEN UPWELLING 162

Urunga Diamond Hd Pt Stephens

b I £c. -100 C1l 0

-150

15/11/98 24/01/99

-50 d I £ c. - 100 C1l 0

-150

-200 -""'--"""""'~-L-->----'---'L__J L,,__ _...... ,..,_...... ,_.L_~~ 153 153.2 153.4 153.6 152.8 153 153.2 ------152.2 152.4 152.6...L...J Longitude (0 E)

Figure 5.21: Observed nitrate concentrations during events at: (a) Urunga (wind driven); (b) Diamond Head (encroachment driven) ; and (c) Point Stephens (separation driven) and mean ni­ trate concentrations at each location, obtained from CARS (d,e,f). Maximum concentration is 10 µmolL - 1 . The contour interval = 2 µmolL - 1 and the thick black line represents 8 µmolL - 1 . shelf break and tends to follow the bottom bathymetry.

The Diamond Head section shows an example of uplifting induced by encroachment and possible topographic acceleration. Here the concentrations are slightly lower

( rv 10 µmolL - 1), however they are more widespread horizontally and extend fur­ ther up into the water column. This is indicative of uplift further to the north and downstream advection. The background concentrations at both Urunga and Diamond Head are similar in magnitude.

The final example from Point Stephens shows that higher background concentrations CHAPTER 5. CURRENT DRIVEN UPWELLING 163 are more widespread. The ("' 8 µmolL - 1) isopleth is uplifted to a depth of 75 m. This indicates a persistent pattern of nutrient uplift, most probably associated with the semi-permanent EAC separation feature. Measured nitrate concentrations are the lowest 'upwelling' concentrations observed (7 µmo11- 1 ), however the bloom extends across the entire continental shelf and is lifted throughout the entire water column. The presence of a persistent pool of nitrate rich water on the continental shelf immediately below the euphotic zone explains the regular occurrence of algal blooms at Point Stephens (Hallegraeff and Jeffrey, 1993).

5.8 Discussion

By combining time series of current velocity and temperature with hydrographic data, satellite images of sea surface temperature and information about the circu­ lation derived from ADCP measurements, a conceptual model is constructed of the three dimensional nature of current driven upwelling in the vicinity of the separation point of the East Australian Current.

The hydrographic sections from Smoky Cape and Diamond Head show clear evidence of alongshore variability over a short spatial scale ("-' 90 km), as well as uplift and advection of nitrates and subsequent phytoplankton blooms. This is reflected in the cross sections of chlorophyll-a where the regions of high fluorescence are associated with the colder water. At Smoky Cape, the fluorescence peak is at the base of the water column, at a depth of 80 m. Further to the south at Diamond Head, the fluorescence maximum has been upwelled to the surface, overlying the 80 m bathymetric contour.

The acceleration of the current at Smoky Cape, either through encroachment of the EAC across the shelf or through topographic processes, transports and lifts colder CHAPTER 5. CURRENT DRIVEN UPWELLING 164 nutrient rich water via the bottom boundary layer which is advected and upwelled to the euphotic zone in the Diamond Head region.

A higher abundance of zooplankton was found at Diamond Head associated with the higher fluorescence levels, whilst virtually no zooplankton was found at Smoky Cape (Dela Cruz pers. comm.). It is highly likely that this spatial variability is associated with the acceleration of the EAC, which increases upwelling, resulting in an increase in nutrients and a greater phytoplankton biomass and hence zooplankton.

Time series of temperature and current velocities showed that the encroachment of the EAC upon the continental shelf caused a significant decrease in bottom tem­ perature during the Current event. This is attributed to an increase in non-linear acceleration of the flow, which increases the negative vorticity in the region. The Burger number was reduced and Ekman pumping through the bottom boundary layer was induced. The temperature in the BBL decreased both at Smoky Cape and Diamond Head, however downstream at Diamond Head the temperature de­ crease also occurred in the surface waters. Thus indicating an uplift at Smoky Cape and advection and upwelling downstream at Diamond Head.

Gibbs et al. (1998) found that in the Sydney region, after the EAC had separated from the coast, wind driven upwelling was significantly more massive than upwelling driven by EAC eddy intrusions. However, the evidence presented here shows that in the vicinity of the separation point, this is not the case. From the temperature time series and concurrent hourly wind records it is evident that wind driven up­ welling does occur (during the Wind Event), albeit infrequently (1 episode in nearly 3 months). Furthermore, surface temperatures are not affected to the same extent as those in typical upwelling regions such as the Peruvian coast. The nutrient input relating to such an event is important however, as is the rapid biological response to the increase in nutrient concentrations. The data also shows that downwelling CHAPTER 5. CURRENT DRIVEN UPWELLING 165 events occurred as frequently as upwelling events during the summer months when the measurements were taken.

Generally on the east coast of NSW the southerly wind events are stronger and more persistent than the northerly wind events, hence downwellings must be larger than upwellings. However, the pre-conditioning effects of the EAC allow weaker northerly wind events to have a greater effect than southerly wind events of the same strength, which first have to counteract the pre-uplifted isotherms, i.e. the pre-conditioning effects of the EAC. Thus it follows that an encroachment event that overwhelms a downwelling event is more likely to be larger than any wind driven upwelling.

Ekman theory was used to estimate the response of the near-shore temperature field to wind and current forcing. This showed that in the Smoky Cape region where swift alongshore currents flow adjacent to the coast, current driven upwelling is significantly more massive than wind driven upwelling.

In a multi-year study of upwelling on the northern NSW shelf, Rochford (1972) showed that upwelling at Laurieton was more pronounced than that observed further to the north at Evans Head. At Laurieton there are the contributing effects of topographic upwelling related to the narrowing of the shelf and hence acceleration of the current at Smoky Cape (to the north of Laurieton), combined with the effects of the encroachment and separation of the current anywhere in this region, which also drive upwelling. At Evans Head however, encroachment driven upwelling is most likely to occur, with the occasional separation induced event, as the point of separation moves periodically equatorward and south again.

During November 1998 an EAC encroachment event occurred which totally over­ whelmed the existing downwelling conditions present on the shelf. This onshore movement of the core of the current is evident from the rapidly re-established uplift of the isotherms. During January 1999, the hydrographic data show the effects of CHAPTER 5. CURRENT DRIVEN UPWELLING 166 topographically induced upwelling, combined with the effects of separation of the EAC downstream from Smoky Cape were observed. The effects of topographically induced upwelling combined with encroachment are the most massive upwelling events observed. As the current comes within close proximity of the shelf at Smoky Cape, alongshore acceleration occurs. This makes it difficult to isolate the two cases downstream of Smoky Cape. Separation induced upwelling whilst being smaller in magnitude, appears to have a far greater effect on the waters further downstream of the separation point as blooms are distributed across the shelf. This is also a semi-permanent feature as seen in the nutrient climatology, and thus explains the historically regular occurrence of algal blooms downstream of the EAC separation.

5.9 Summary

The results of this study show evidence of four different upwelling mechanisms: wind driven upwelling and three current driven mechanisms. Current driven upwelling is induced through the encroachment, topographic acceleration and separation of the EAC in the Smoky Cape region.

From the data presented in Chapter 4, it is clear that non-linear acceleration and bottom stress are important in the vicinity of the EAC separation point. Here evidence was presented that showed the acceleration of the current at the shelf break causes upwelling. The EAC can be accelerated in the region either through encroachment of the current across the shelf, or through topographic acceleration where the continental shelf narrows at Smoky Cape. However, despite the mecha­ nism, if the proximity and acceleration of the current is adequate, then the bottom boundary layer shut down time is lengthened. Hence Ekman pumping through the bottom boundary layer is prolonged and current driven uplifting occurs, resulting in upwelling further downstream. CHAPTER 5. CURRENT DRIVEN UPWELLING 167

Once upwelling has occurred the question remains as to destination of the nutrient rich water. This leads to a modelling study to investigate the circulation and ad­ vection of upwelled water and nutrients in the vicinity of the separation point of the EAC. If I can't, I must!

Nelson Mandela

A Modelling Study about the EAC Separation.

Recently several studies have been undertaken using the Princeton Ocean Model to hind-cast specific oceanographic events that have occurred on the NSW continental shelf (Gibbs et al., 1997; Marchesiello et al., 2000; Oke and Middleton, 2000, 2001). Generally these applications have focused on the large scale features of the system. In each of these cases initialising the model has proved to be a challenge because of the coastal nature of the region, that is, the shallow water and steeply sloping bathymetry over the shelf cause numerical problems.

168 CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 169

The most commonly accepted initialisation technique has been the use of weekly compilation charts of sea surface temperature and temperature at 250 m depth (Marchesiello et al., 2000; Oke and Middleton, 2001). These charts are routinely compiled by the Royal Australian Navy using a combination of (often sparse) data points and sea surface temperature information derived from satellites (AODC., annual). The surface temperature and the temperature at 250 m depth was then interpolated onto the Levitus climatology which has a resolution of 1°. Salinity was derived from the interpolated temperature using a well documented relationship be­ tween temperature and salinity that exists for the Tasman Sea region. Thus the 3 dimensional density field was calculated from these fields of temperature and salin­ ity, and used to calculate the geostrophic currents assuming a commonly accepted 2000 m level of no motion. In the majority of these cases the model was initialised with 'features based on' the weekly Navy charts. This means that although the model appeared to replicate the observations, it was in fact 'tuned' in order to do so. Marchesiello et al. (2000) reported that their 70 day simulation captured 'fairly well' the large scale features (such as a meso-scale warm core eddy) when compared to observations.

It is the intention here to investigate the circulation which is derived from the climatological conditions and not, as done previously, to replicate a specific event. In this study, a new high resolution (1/2°) climatology of both temperature and salinity is used to initialise a numerical model in order to investigate the 3 dimensional velocity fields in the region of interest. The specific aims of this study are three fold:

• To assess various initialisation techniques, including ramping, nudging, and forcing with geostrophic inflow;

• To investigate the 3 dimensional velocity fields associated with the climatolog­ ical values of temperature and salinity on the NSW shelf, and in the adjacent CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 170

Tasman Sea; and

• To address the question of 'the origin of upwelled water'.

By initialising the model with a high resolution climatology of both temperature and salinity it is the intention that the capabilities of the coastal ocean model will be improved. Analysis of the various techniques is done by comparing the energy of the system, as well as by examining the various barotropic and baroclinic fields calculated in the model such as sea surface elevation, depth averaged velocity and actual velocity.

Observations of velocities measured by ADCP combined with hydrographic data and concentrations of chlorophyll-a, as described in Chapter 5, are crucial to the assessment of the model. The model results are then used to aid interpretation of the flow regime in the vicinity of Smoky Cape. Finally a series of off-line particle tracking experiments are run to investigate the possible origin and fate of upwelled 'particles'.

6.1 Model Configuration

In this study the Princeton Ocean Model (POM) is utilis~d. POM is a 3 dimensional, non-linear, primitive equation model. The advantages of POM are numerous, and it is freely available and widely used. The sigma coordinate system provides high resolution over the sloping bottom of the continental shelf and slope, which in this study is very important. Furthermore, a configuration of POM for the NSW shelf has been developed 'in house' at UNSW. The model equations used in POM are included in Appendix A for completeness, however for an in-depth description of the model equations and numerical techniques, refer to Blumberg and Mellor (1987) CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 171

and Mellor (1996).

The model domain extends from 28.5 - 37.5°S and 150 - 157.5°E. The grid is a curvilinear grid where the horizontal grid spacing increases from 2.5 km at the coast to 16 km at the eastern boundary. Alongshore the grid spacing ranges from 6-22 km (Figure 6.1), with the grid spacing increasing near the open boundaries to minimise boundary effects. Vertically the grid consists of 31 sigma levels that vary linearly over mid-depths and logarithmically over upper and near-bottom depths, in order to resolve the surface and bottom boundary layers. The maximum depth across the domain is 2000 m. In contrast to the simulations of the NSW shelf circulation described by Marchesiello et al. (2000) where the minimum depth was 50 m, in this study POM is implemented using a minimum depth of 15 m.

Boundary conditions are chosen so as to allow disturbances to propagate out of the model domain, whilst at the same time relaxing the boundary fields to their initial states (Palma and Matano, 1998). Hence at the southern and eastern boundary, a combination of radiation, zero gradient and relaxation conditions are implemented as described by Gibbs et al. (1997). The conditions at the northern boundary are clamped to their initial values except in the third case as described below. The western boundary (being the coast), is closed.

In this study the model is initialised using climatological data of temperature and salinity obtained from the CSIRO Climatology of Australian Regional Seas (CARS) as described by Ridgway et al. (2001). To obtain the atlas to 1/2° horizontal resolu­ tion, raw data is interpolated onto a uniform grid (Figure 6.2) using a 'Loess filter'. Vertically the atlas has 56 (unequal) depth layers from O - 5500 m, with maximum resolution being from O - 300 m. In this application, the climatological data is in­ terpolated linearly onto the model grid, also shown in Figure 6.2. Density is then calculated within the model from the initial fields of temperature and salinity using CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATI ON. 172

29S a

31S

33S

35S

37S

148E 150E 152E 154E 150E 152E 154E 156E

Figure 6.1: The location of the study area on the NSW shelf. The topography of the coast and the bottom bathymetry (a) and the model grid (b) . Isobaths marked are 100,200,500, 1000, 1500, 2000 m (the 200 m isobath is in bold). The across-shore grid spacing extends from 2.5 km at the coast to 16 km at the eastern boundary. Alongshore the grid spacing ranges from 6 - 22 km. Also marked are the four sections where observations were taken during November 1998, from north to south, Urunga (U), Smoky Cape (SC), Point Plomer (PP) and Diamond Head (DH). CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 173

CARS grid POM grid

28S

30S

32S

34S

36S

38S 149E 151E 153E 155E 157E 149E 151E 153E 155E 157E

Figure 6.2: Comparison of the CARS grid and the POM grid

the non-linear equation of state (Mellor, 1991).

In each of the simulations shown the model is run diagnostically for 25 days, where the temperature and salinity fields are constrained to their initial conditions, and only the baroclinic velocity field is allowed to adjust to the mass field with time. Af­ ter about 20 days the fluctuations in the energy fields are minimal and the simulation is assumed to be at a steady state.

The model was initially run for 60 days, however from the energy fields (not shown) it is evident that the imposed boundary conditions cause difficulties for periods longer than 30 days. The model remains in a quasi-steady state, however fluctuations in the sea surface elevation field occur as a result of constant relaxation at the open boundaries. The boundary conditions are not entirely passive which means that as CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 174 the velocity relaxes at the boundaries, the model becomes inadvertently forced. The circulation within the EAC is considered to be quasi-steady for a 2-3 week period, and prognostic simulations can adequately represent this state, however, after longer periods simulations become unrealistic.

In this idealised investigation of the mean current field associated with the climato­ logical values of temperature and salinity, external forcings such as heat flux, fresh water or wind stress terms are not imposed. Three methods are used to test the different initialisation techniques, these being: ramping, nudging and forcing. Each case is discussed below.

6.1.1 Case 1: Ramping

Five different experiments are run to test the period over which the baroclinicity of the model should be ramped. Ramping allows for a gradual adjustment from rest to the forced state so as to minimise spurious wave generation. Figure 6.3 shows the evolution of the kinetic energy fields for each of the different experiments where the model spin up is ramped over 1, 2.5, 5, 7.5 and 10 inertial periods.

The kinetic energy increases as the model relaxes to the final state, and after 20 days the difference in kinetic energy between four of the five experiments becomes sufficiently small, that the model is considered to have reached a quasi-steady state (Ezer and Mellor, 1994). It is only the first experiment that is significantly different, where the ramping was implemented over 1 inertial period (thick line, Figure 6.3). From these results it is concluded that the mean ramping time, i.e. a ramping of 5 inertial periods, is adequate for subsequent simulations. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 175

6.1.2 Case 2: Nudging

A second initialisation method investigated is a technique that is commonly known as nudging. In this case, nudging involves initialising the model with a horizontally uniform temperature and salinity field and then relaxing the initial conditions with time to a field that varies spatially. This has the benefit of allowing the velocity field to evolve gradually, thus reducing spurious noise and transients in the results. In this way, the velocity fields are ramped gradually as the density field evolves. The results from this technique are promising. Fluctuations in the kinetic energy field are minimal, and the potential energy field does not vary, which means that fluctuations in sea level elevation are minimal. The nudging period was tested with numerous model runs, and it was found that 5 inertial periods was sufficient for the fields to converge.

6.1.3 Case 3: Forcing

The third initialisation technique involves forcing the model at the northern bound­ ary with the corresponding geostrophic velocity calculated from CARS. In a geo­ physical environment it can be shown through scaling arguments that rotation is a dominant term in large scale ocean dynamics. In such a system the pressure gradient (resulting from a change in sea surface elevation TJ) is often balanced by the Coriolis force to first order. This geostrophic balance can be expressed as:

fv = ga'f/ox (6.1)

These simplified momentum equations can be solved for the horizontal velocity corn- CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 176 ponents u and v, giving: 1 8p 1 8p u=---, v=-- (6.2) Pof 8y Pof8x where the velocity vector (u, v) is perpendicular to the pressure gradient (Vp). In

Equations 6.1 and 6.2, f is the Coriolis parameter (s-1 ), g the gravitational accel­ eration (ms-2), 1J the surface elevation (m), and p0 the reference density (kgm-3).

Theoretically, in the absence of wind forcing, differences in sea surface elevation between two columns of water can be used to calculate the relative geostrophic velocity between the two columns. The difference in geopotential anomalies (~) between two hydrographic stations enables the estimation of the geostrophic current fields. The following equation is the practical form of the geostrophic equations.

2 2 (Vi - ½) = }1 [/.~ 0Bdp - f.~ 0Adpl 1 Lf [~B - ~cI>A] (6.3) where (Vi - ½) is the relative velocity (ms-1 ), L is the distance between stations

(m), and o is the specific volume anomaly, (m-3kg-1) (Pond and Pickard, 1983, p 73). Johns et al. (1989) tested geostrophy in the Gulf Stream and found that gradient velocity calculations were fairly accurate even for a high velocity, narrow, western boundary current, such as the Gulf Stream.

For this study the geostrophic velocities are calculated :tcross the domain from the CARS climatology of temperature and salinity and are presented in Figure 6.4. For this application the depth of no motion is taken to be 2000 m which is standard practice for the EAC region (Gibbs et al., 1997; Marchesiello et al., 2000). Over the continental shelf where the depth is shallower than 2000 m, the geopotential anomalies were calculated using the extrapolation technique of Reid and Mantyla (1976).

The general flow pattern (Figure 6.4) does represent that found off the NSW coast, CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 177 with a strong southward jet flowing adjacent to the coast (Hamon, 1965; Godfrey et al., 1980b). However, the core of the EAC is somewhat weaker than the obser­ vations (v < 1 ms-1), and the jet is broader and is located further offshore. This is attributed to the ageostrophic nature of the western boundary current, the paucity of data over the shelf and the smoothing effect of the climatology. Furthermore it suggests that the EAC has a barotropic component which is not resolved through geostrophic calculations.

The calculated geostrophic currents are used to force the model at the northern boundary, the idea being that the model will adjust more quickly, and be more accurate. To do this, the geostrophic current and the sea level elevation as calculated from the thermohaline field are imposed at the northern boundary of the model. To be consistent the forcing is also ramped up over five inertial periods. The density field for the northern boundary, as calculated using the CARS data is shown in Figure 6.5. Also shown are the alongshore and across-shore geostrophic currents calculated across the same section. For the reasons mentioned above, the magnitude of the geostrophic current is weaker than observed (Vmax = 0.45 ms-1) and the core of the jet is located further offshore.

6.1.4 Comparison

The three methods used to initialise the model are now considered. The forced model adjusts more quickly, however due to its forced nature and the associated time scales, it also becomes unstable more quickly. Figure 6.6 shows the kinetic energy fields from the three cases (ramped, nudged and forced). Whilst the first two cases come to equilibrium after 20 days, the forced case appears to be in equilibrium after just 15 days. Furthermore, in the forced case the kinetic energy field begins to increase steadily after day 20 until after about 25 days the fields are no longer valid. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 178

By contrast, for the first two cases the model does not begin to lose credibility until after 40 days. To ensure an accurate representation of the climatological velocity once the model has reached equilibrium, the fields are averaged over a 5 day period. As the simulations are strictly diagnostic, once the model is in equilibrium, the only fluctuations in the fields are inertial oscillations. For the first two cases the 5 day mean is calculated from day 17.5 - 22.5. For the third case which spins up more quickly, the 5 day mean is centred around day 15 instead. The mean, centred on day 20, is represented by the box in Figure 6.3, and the mean for the forced case, centred on day 15, is represented by the box in Figure 6.6.

To quantify the difference between the three model runs, the root mean square (RMS) difference is calculated for various fields, including the sea surface elevation and the magnitude of the depth averaged velocities. Elevation is a response to many factors including the velocity field and surface gravity waves, and hence is one of the more difficult variables to capture in the model. In each case the mean fields are used to calculate the RMS differences.

Figure 6. 7 shows a time series of the RMS differences in elevation and depth averaged velocity. The differences in elevation are greatest between the nudged and the forced case, and in velocity they are greatest between the forced and ramped case. While the RMS differences in elevation are between 0.13 - 0.22 m, the RMS differences in velocity are < 5% of the total velocity.

Table 6.1 shows a comparison of RMS differences between the three initialisation methods calculated from the five day mean. The fields shown are the elevation ("l), the total depth averaged velocity Vtot, and the velocity at the 4 sections introduced in Chapter 2: Urunga, Smoky Cape, Point Plomer and Diamond Head.

The difference in elevation is least between the ramped and the nudged case which is to be expected, as in these two cases elevation is not imposed at the northern bound- CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 179

Kinetic energy 1 ramp= 1 0.8 ramp =2.5 ramp= 5 N- I ramp= 7.5 Ill 0.6 N.s ramp= 10 >, 0.4 ei G> C w 0.2 0

-0.2 0 5 10 15 20 Time (Days)

Figure 6.3: Evolution of the kinetic energy field during the model spin up for 5 different ramping experiments. The model is spun up with different ramping periods, where ramp = 1,2.5,5,7.5 or 10 inertial periods. The kinetic energy increases at different rates as the model relaxes to the final state, until after 25 days when the difference in kinetic energy between experiments is insignificant. The black box indicates the five day averaging period centred on day 20.

,,, Vtot vu vsc Vpp VDH Ramped (1) Nudged (2) 0.045 0.061 0.065 0.059 0.06 0.053 Nudged (2) Forced (3) 0.12 0.064 0.052 0.054 0.055 0.057 Ramped (1) Forced (3) 0.09 0.078 0.084 0.081 0.071 0.057

Table 6.1: A comparison of RMS differences in rJ (m) and v (ms-1) between the three initialisation methods from a 5 day mean. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 180

Surface Geostrophic Velocity (ms- 1) Depth Averaged Geostrophic Velocity (ms-1) 29S

., \ \ 1 m/s > I \ 1 m/s > 30S ' \ \ ' \ \ 31S J \ J 32S t ' , ~ J Port Port 1 ~ ~ 'J 33S \ \ ~ ! ,I \ \ \ ~ ! ' ' 34S '\ \ '\ \ I I - ' ' 151E 153E 155E 151E 153E 155E

Figure 6.4: Surface geostrophic velocity field (left) and the depth averaged geostrophic velocity field (right) as calculated from CARS data. ary. Moreover, the differences in elevation are greatest at the northeastern corner and the southern boundary of the domain, both of which are a considerable distance from the sections of interest. Differences in velocity range between 2. 7 - 3.3% of the maximum velocities, with the greatest RMS difference at· the northern most section (Urunga). The velocity differences between case 2 and case 3 (nudged/forced) in­ crease moving southward, i.e. further from the forced northern boundary, however between case 1 and case 3 (ramped/forced) the converse is true. The largest differ­ ences in velocity are between the ramped and the forced cases, with RMS values of

0.057 - 0.087 ms-1. All the fields shown in Table 6.1 were also calculated for the forced case using a 5 day mean centred on day 17 and day 20, however the RMS differences were a minimum given a spin up time of 15 days. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 181

0 / ' / - 500 .s 27 £ -1000 0.. 27.25 Ql 0 -1 500 27.5

-2000 153 154 155 156 153 154 155 156 153 154 155 156 Latitude (0 E)

Figure 6 .5: The forcing fields imposed at the northern boundary. The density field (p) as calcu­ lated from the CARS data, (contour interval (Cl)= 0.25 kgm-3 ) and the alongshore (v) and across­ shore (u), geostrophic velocities calculated from the geopotential anomalies (Cl= 0.05 ms- 1 ). The maximum alongshore velocity is 0.45 ms- 1 southward, and positive velocities are indicated by the dashed contours.

Kinetic energy

- Ramping 0.8 -- Nudging Forcing N~ 'cn 0.6 N.s >, 0.4 . --- Ol

-0.2'------'------'------'------'------' 0 5 10 15 20 Time (Days)

Figure 6.6: Evolution of the kinetic energy fi eld during the model spin up for the three different cases, ramping, nudging, and forcing. The black box indicates the five day averaging period centred on day 15. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 182

RMS difference TJ 1 ~....---_-_-_- ...... - _-_-_-_-_..-..., ---..-,--~,--~,----.,---,.----.--,--..-- ,----, · - · R & N (0.13) - - N & F (0.22) 0.8 F & R (0.16) - e-;;o.6 ... - 0 ~ iii 0.4 ._ - iii .- :--:- ...... (. ":"'-. 02~·-~ - ..... ------' ,,,. - - .- , ...- · -·--· ~ ··~··"<,:,',,,- .':. - . ·- ..::: - -- - ...... ~...,,,...··::_·:~::.,...----...... / '· ,,,,, ·, ···········:-...:::..:·.-~·-·:_·::..::.:...:::...:·.···:········ ·-·-. ...,,,..,,,,...... ,. · ... 0 .____ ..__, __..... , ____,_, ____,_, ____,, _____,,'-- __..__ __.,__ __...,______, 17.5 18 18.5 19 19.5 20 20.5 21 21.5 22 22.5

RMS difference IVI 0.2 ...... ---_-_-_-~..-_-_-_-_-_-_ ..... _,--~,--~,------.----,,----,-----,-----.-----, · - · R & N (0.076) - - N & F (0.083) 0.15 · .. · F & R (0.099) 1(/) s ...... ~ 0.1 -...... ,..~ '·...... · · · · · · · · · · · · · · ·. ... ' ...... ;..,:.. · · · · · · .... . _ _,_ -c..· ~ -a;8 :-- - .,, ,,,,, ...... -~ -- .------.- :-_-:----: =...... _-.:::. -_ --- _-. .::::-· ------> ---· ·------·---·- 0.05 -

0 .____ ..__, __..... , ____,_, ____,_, ____,, _____,, ___,...._ __..__ __...,______, 17.5 18 18.5 19 19.5 20 20.5 21 21.5 22 22.5 Time (Days)

Figure 6. 7: Time series of the RMS differences in sea level elevation r, (top) and the magnitude of depth averaged velocity (bottom), for three model initialisation techniques R - ramping, N - nudging, F - forcing. Values in brackets represent the mean RMS difference in time.

Although the range of sea level elevations as calculate4 by the forced model are slightly higher than the first two cases, the current field is similar with differences within 4% of the total velocity. Furthermore the forced case reaches a steady state 5 days ahead of the other two cases, which substantially improves the efficiency of the model by reducing computing time needed. For these reasons the forced case is considered to be the 'scenario of choice'. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 183 6.2 Analysis of the Climatological Velocity Field

For this analysis currents have been extracted from the forced case. To recapitu­ late, the model is initialised using the climatological conditions of temperature and salinity, the geostrophic velocities and elevation are imposed at the northern bound­ ary and the model is ramped over five inertial periods. To obtain a representative current, the velocity fields are averaged from day 12.5 - 17.5, i.e. a five day average centred on day 15.

The barotropic fields are presented in Figure 6.8. The depth averaged velocity field (left) shows a dominant southward flowing jet extending down the east coast of NSW. The current flows adjacent to the continental shelf, to Smoky Cape (31 °S) and then separates from the coast. The current continues due south, whereas the coastal orientation bends away in a southwesterly direction. The current speeds are up to 2 ms-1 southward in the centre of the jet.

Inshore of the jet there is a weak northward flow adjacent to the coast between Port Stephens and Smoky Cape. Just to the north of Port Stephens the current tends to bifurcate forming a broader jet, which bends eastward, and a narrower jet that extends southwards along the coast, but further offshore. Both these scenarios have often been observed in the EAC, however at times this bifurcation is also evident. Observations show that when the main axis of the current bends eastward the remaining alongshore flow is generally weaker (Tilburg et al., 2001), however when the current remains attached to the coast the alongshore flow remains strong. The SST images presented in Chapters 3 and 5 show examples of the two different scenarios. In Figure 3.17 the EAC bends eastward, however a month later the flow is again southward (Figure 5.19). The equal bifurcation that is produced in the simulation would appear to be a trait of the climatological fields with which the model was initialised. As the CARS atlas is an amalgamation of data obtained over CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 184

a

30S -1 m/s

31S

1, • I

32S

Port Port

33S D

34S

151E 153E 155E

Figure 6.8: Plot of (a) the depth averaged velocity field (ms-1) and (b) the sea surface elevation TJ (m) for the forced case.

many years, it actually captures both scenarios concurrently.

Sea Surface Elevation

The sea surface elevation anomaly is shown for the Smoky Cape region in Figure 6.8 (right). There is a region of higher elevation to the east of the jet, and the jet is strongest where the slope in elevation is the greatest. From the figure it is evident that the pressure gradient is not constant in the across-shore direction, hence this explains why Py proved so problematic in the estimations of the term balances pre­ sented in Chapter 4. Across the shelf at Smoky Cape the sea level gradient between the inshore and mid-shelf mooring is 2 cm, and between the mid-shelf and offshore CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 185 mooring the gradient is 7.5 cm. Downstream at Diamond Head, the simulations show a northward flow which implies that the across-shelf pressure gradient is in the opposite direction. The sign of the gradient changes between the two inshore moorings at Diamond Head (DRAA - DHA) and the gradient is small("-' 0.5 cm), across the other three moorings the gradient is almost constant (3 - 3.5 cm). Thus the numerical simulations can be used to gain an indication of sea level elevation and hence the barotropic pressure gradient in the region.

Surface Velocities

The surface velocities as calculated by the forced climatological model are shown in Figure 6.9 along with observed currents obtained using a shipboard ADCP during the first cruise in November 1998. The similarities between the two are strong in that both the magnitude and direction of the currents are comparable. Both fields show strong southward flow at the shelf break, with a slight onshore component north of Smoky Cape and an offshore component further south. The jet is within close proximity of the coast at Smoky Cape, and has moved slightly offshore at Diamond Head. This indicates that the separation point of the simulated EAC lies between Smoky Cape and Diamond Head.

The most rapid coastal current in the simulation is immediately south of Smoky Cape, perhaps indicating a funnelling of the jet as the continental shelf narrows. Inshore of the jet is a slight northward flow, which is underestimated by the model at the southern most observation line (Diamond Head). However apart from this, the velocity field as approximated by the model closely resembles the observed flow field. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 186

b >

Smoky cape 31S

32S

I l

152E 153E 152E 153E

Figure 6.9: Comparison of near surface velocities obtained from (a) ADCP measurements along the four observational transects shown in Figure 6.1 and (b) a subset of the modelled near surface velocities. Velocities are in ms-1 .

Velocity Cross Sections

Figure 6.10 compares observed and modelled velocities across two sections. The first cross section shown is at Smoky Cape, where the current _hugs the coast in a narrow jet, and the second is at Diamond Head, after the current has separated from the coast. At both locations the model has accurately depicted the position of the jet over the shelf, the direction of the flow, and the strength of the jet (to within 10%). At Smoky Cape both the observations and the model show the core of the jet to be 20 km offshore with a maximum observed velocity of 1.6 ms-1, compared to a maximum simulated velocity of 1.8 ms-1. All the flow is directed southward, with speeds of 0.8 ms-1 within 10 km from the coast. The simulation differs from the CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 187

Smoky Cape Diamond Head 0 V.

-50

-S -100 £c. (l) 0

-150 v obs

-208

-50 E -100 £c. (l) 0

-150 v model

- 200 ~~---~~~~--~~ -~~ 0 10 20 30 400 10 20 30 40 Distance (km) Distance (km)

Figure 6.10: Comparisons of the observed (top) and modelled alongshore currents (bottom) up­ stream (Smoky Cape) and downstream (Diamond Head) of the EAC separation point. The solid line represents negative (poleward) alongshore flow the dotted line represents positive (equator­ ward) alongshore flow, and the bold solid line represents the zero velocity contour.

observations in the vertical extent of the jet, and the tig_htness of the front. In the simulation the core of the jet extends to a greater depth and the velocity front is more vertical. The observations show a velocity of 1 ms-1 below the core of the jet, at a depth of 200 m whereas the simulation shows a maximum velocity of 1.6 ms-1 at this depth. The weaker current at depth dictates a more gentle gradient across the observed velocity front .

Downstream of the separation point a northward counter current forms inshore of

the main jet. The strength of this northward flow is up to 0.3 ms-1 at 10 km from CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 188 the coast. The point of zero velocity in both the observations and the modelled results (solid line Figure 6.10) is located at about 20 km from the coast. Again the position of the jet is depicted accurately ("-' 40 km from the coast), however the strength of the modelled jet (1.6 ms-1) is slightly higher than the observed current

(1.4 ms-1 ). Of note is the fact that in both the simulation and the observations, the jet weakens and broadens after the current separates from the coast. Eastward of the jet is a northward return flow (Figure 6.8, left) which is seen in many of the ear­ lier observational studies (Ridgway and Godfrey, 1997). The present observational study, being coastal in nature, did not extend far enough offshore to capture the return flow.

The climatological currents for 4 cross sections are shown in Figure 6.11. The cross sections chosen are the same four sections described in previous chapters, Urunga, Smoky Cape, Point Plomer and Diamond Head. Of interest is the strengthening of the current as it moves southward. At Urunga, the core of the jet is flowing at a speed of 1.6 ms-1, and the maximum alongshore velocity is 1.8 ms-1. As the current moves further south, the jet tightens at Smoky Cape and actually accelerates, to a maximum speed of 1.9 ms-1 at Point Plomer. The actual core of the current is now subsurface with the maximum alongshore velocity found at a depth of 50 ms-1, located directly above the shelf break. Further to the south at Diamond Head, the jet weakens slightly as it widens, with the maximum alongshore velocity remaining subsurface with a speed of 1.6 ms-1. At Urunga and Srrioky Cape the across-shore component of the current is predominantly onshore, with a maximum of 0. 7 ms-1. This onshore flow weakens moving southward to Diamond Head downstream of the separation point, where the across-shore flow component is predominantly offshore. Inshore of the jet there is a slight eastward flow at the two northern most sections, associated with the tightening of the front. Downstream of the separation at Di­ amond Head, the inshore flow is directed northward and onshore. This creates a divergence and positive vertical velocities of up to 0.003 ms-1 at the shelf break.

~ ~

f--' f--'

~ ~

;:i::.. ;:i::.. ~ ~

~ ~

t:?:l t:?:l

c.o c.o

00 00 :< :< ~ ~

~ ~

0 0

;:i::.. ;:i::..

'"< '"<

b:l b:l ~ ~

0 0

0 0 t--3 t--3

~ ~

t-< t-<

~ ~

§ § t, t,

S2 S2

V)_ V)_

G G

~ ~

0 0

;:i::.. ;:i::..

'""Cl '""Cl

~ ~

~ ~

?'l ?'l

-' -'

40 40

0 0

·. ·.

(') (')

: :

:i,.

&

·.

Point Point

. .

30 30

.I> .I>

. .

oJ oJ

.

· ·

·.

contour). contour).

Cape, Cape,

(km) (km)

Head Head

20 20

velocity velocity

Smoky Smoky

Distance Distance

, ,

Diamond Diamond

zero zero

10 10

nga

the the

Uru

s s

i

0 0

at at

40 40

), ),

bold bold

1 1

and and

ms-

( (

30 30

(km) (km)

positive positive

model model

Plomer Plomer

s s

20 20

i

the the

ed ed

Point Point

Distance Distance

dott

10 10

from from

, ,

ive

0 0

egat

n

obtained obtained

40 40

s s

i

as as

w) w)

solid solid

30 30

u, u,

(v, (v,

6.10, 6.10,

(km) (km)

Cape Cape

20 20

Figure Figure

currents currents

Smoky Smoky

Distance Distance

for for

10 10

(As (As

0 0

I I

40 40

I I

depth. depth.

climatological climatological

·· ··

t t

·

m m

·

the the

IIIL.

-

200 200

of of

30 30

I I

s s

to to

". ".

11111)1 11111)1

·

(km) (km)

ion

20 20

'-1:''I '-1:''I

I I

l1Jll.L___J.-...._..,__...L.J l1Jll.L___J.-...._..,__...L.J

sect

Head, Head,

Urunga Urunga

Distance Distance

Cross Cross

10 10

Diamond Diamond

6.11: 6.11:

L...... =-----

and and

0 0

o o

0 0

o~---,,------;;;::--, o~---,,------;;;::--,

-50 -50

-50 -50

-50 -50

-2001 -2001

-150 -150

-200 -200

-100 -100 -150 -150

-100 -100

-200

-150 -150

-100 -100

Q) Q)

a. a.

Q) Q)

Q) Q)

a. a.

a. a.

0 0

£ £

£ £

0 0

£ £

0 0

E E

.s .s

E E

Plomer Plomer Figure Figure CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 190

Inshore of the jet at Smoky Cape, at a depth of 50 m there is a slight vertical velocity of 0.001 ms-1, and although this velocity appears weak, it equates to a persistent pumping of,...., 86 m per day into coastal waters. In a nutrient poor environment, this uplifting of deeper nutrient laden waters to the euphotic zone is biologically significant.

6.3 Density Surfaces and Chlorophyll-a Concen­ trations

From the observations of chlorophyll-a and nutrient concentrations off the NSW coast presented in Chapter 5 it is evident that the a = 25.25 kg m-3 isopycnal is of biological importance. Figure 6.12 shows chlorophyll-a concentrations as measured during November 1998 along the 4 transects off the NSW coast. The figure shows

26.5 ,------,------,------,-----,-----.-----;::====::::::1 ...... -u .,·.~ . . . . ,_, SC ....' .._ ...... :...... : ...... :..... -PP 26 · · · · · · · · · · · · · · .:,: · · .'.'\..,. · · · · ...... : : : : ·-· DH ...... :. :.:!'P~-~- ...... ' . . . . - - ...... - ...... - . . . . ' ' ' . ' : .~.. I - e - I ·-·-,Jr.•• . . . '' '.' ...... :, ' .. --~ ...... '. : ...... -."'•···············:················:··············· : ' . . :...... ' ....

25 ...... , ...... 24.5 .______.....::,....:::;;:-.&. __ __,______-'- ______.______.i....______, 0 0.5 1.5 2 2.5 3 3 Chlorophyll8 (mg m- )

Figure 6.12: Mean chlorophyll-a concentrations (mgm-3 ) versus density (o- kgm-3 ) as mea­ sured at Urunga, (U) Smoky Cape (SC), Point Plamer (PP) and Diamond Head (DH) during November 1998. the average density and chlorophyll concentrations from between 8 and 10 CTD casts taken from the coast out to 40 km offshore, in the depth range of 25 - 1000 m. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 191

Mean concentrations are a maximum at the most northerly section (2.6 mgm-3) and decrease with distance southward. Maximum concentrations however, were highest at Point Plomer (immediately south of Smoky Cape) where concentrations of up to

5 mgm-3 were observed (Figures 5.3 and 5.4). Of particular note however, is the observation that the maximum chlorophyll-a concentration (i.e. phytoplankton) is associated with the u = 25.25 kg m-3 isopycnal. This suggests that the organisms flourish in water of this density, perhaps because of temperature or buoyancy require­ ments, or simply because of adequate nutrient or light concentrations. Hence, this relationship between phytoplankton concentration and density allows the use of the u = 25.25 kg m-3 isopycnal as an indicator of where the maximum concentrations of nutrient or phytoplankton may be found.

Given that the u = 25.25 kg m-3 isopycnal is of biological importance in the Smoky Cape region, the u = 25.25 kg m-3 surface is chosen as a basis for a series of 3 dimensional Lagrangian experiments enabling examination of the advection of nutrient and planktonic organisms in a 3 dimensional sense. To do this it is necessary to calculate the velocity field along the isopycnal surfaces.

Velocity Along the u = 25.25 kg m-3 Isopycnal

Velocities were calculated along 5 isopycnal surfaces (u = 25, 25.15, 25.25, 25.35, 25.5 kg m-3; A weighted average of the velocity above and below each isopycnal was used to calcu- late the velocity along an isopycnal surface. The depth of the isopycnal as well as the thickness of the slab were also calculated. This was done in order to advect particles along the density surface. Only the velocity field along the u = 25.25 kg m-3 isopy- cnal surface is presented here, as the velocities calculated from the surfaces above and below (u = 25.15, 25.35 kgm-3) were not significantly different. The isopycnal is taken to represent the transport of a slab of water up to 25 m thick, (with a CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 192 mean thickness across the domain of 9.5 m), centred around the a= 25.25 kg m-3 isopycnal.

The velocity field along the a= 25.25 kg m-3 isopycnal and the depth of the isopy­ cnal are shown in Figure 6.13. The maximum depth of the isopycnal is 175 m in the north of the domain. In the region of interest (between Smoky Cape and Port

Stephens) the a = 25.25 kg m-3 isopycnal is found at a depth of 50 - 75 m in the

EAC jet, (where the maximum velocity is,.._, 1.5 ms-1) at a distance of 20 - 40 km offshore. Towards the south of the domain this isopycnal is actually uplifted to the surface. This also occurs in the coastal waters to the south of Smoky Cape. As a

Velocity,a= 25.25 Depth, er= 25.25

a 30S 1r- · .. , , 1 m/s 'it",,,,...... ' . ... / 31S ""''"· ·· \

\ ff? 32S ' \ .... \ I \ Port . \ J!: ... I .... .,, 33S I '

\ 34S ' ...... - -- - 153E 155E 151E 153E 155E

Figure 6.13: Plot of (a) the velocity field {ms-1 ) and {b) the depth {m) of the a= 25.25 kgm-3 isopycnal slab.

precursor to the Lagrangian experiments, the dynamics of the flow are examined in CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 193 the following section. As well as investigating the surface velocity field, these calcu­ lations are performed for the CT = 25.25 kg m-3 velocity field, and the geostrophic velocity field as calculated from CARS.

6.4 Flow Dynamics

To examine the effects of the non-linearity in the system, it is useful to consider the local Rossby number. In the equations of motion the Rossby number is the ratio of the non-linear term to the Coriolis term. Consider the horizontal equations of motion:

au a'f/ at v.v'u - fv -g-+ other terms (6.4) + ax av a'f/ -g-+ other terms (6.5) at + v.v'v + Ju ay

In the across-shore momentum equation (6.4) the balance between the Coriolis force and the pressure gradient dominates as in ( 6.1). However by scaling arguments it can be shown that in the alongshore direction (6.5), the non-linear terms v.v'v are also important. Furthermore the analysis in Chapter 4 showed that non-linear advection does occur in the Smoky Cape region.

Continuity in the horizontal is satisfied to first order by:

(6.6) which scales as U V (6.7)

Typically Lx/ Ly ,..._, 0.1 and the across-shore current components are smaller than the alongshore components by an order of magnitude. For the across-shore momentum CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 194 equation we can define a balance between the non-linear terms and the Coriolis term as:

_ U2 /Lx u R (6.8) ox - JV /Ly' and for the alongshore momentum equation we can define:

_ V 2/Ly V R (6.9) oy - JU fLx However because of the magnitude of the scales U /V and Lx/ Ly it follows that U V f Ly<< fLx

So that

Rox<< Roy

Thus although the alongshore current component is in geostrophic balance with the across-shore pressure gradient, non-linear terms may still play a role in the alongshore direction, and hence the Rossby number in the y direction, Roy, is the important parameter. To utilise these relationships to examine a flow field it is necessary to define a dynamical version of the Rossby number Ro. In this study, the local Rossby number in the y direction can be approximated by the vorticity divided by the Coriolis parameter as follows: 8v/8x - 8u/8y 8v/8x Ro (6.10) f f V (6.11) fLx Here the relative vorticity of a fluid is given by

(=8v_8u (6.12) 8x 8y Thus a vorticity Ross by number defined by Ro = ( / f is in effect the dynamical equivalent of Roy. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 195

If the horizontal continuity equation (6.6) is correct to first order then we can cal­ culate a horizontal streamfunction 'l/J which obeys: o'l/J o'l/J u=-- (6.13) V = ox' oy

To investigate flow balances within the EAC, calculations are made of the Rossby number for the depth averaged currents obtained from the model. High values of Ro are associated with strong advective accelerations. The local Rossby number across the Smoky Cape region is shown in Figure 6.14a. The maximum value (Romax) is

a

30S Ro(3) 0

31S

32S

Port 33S 0

34S

151E 153E 155E 151 E 153E 155E

Figure 6.14: Plot of (a) the local Rossby number Ro (CI=0.5, min=0.2, light grey= maximum) and (b) the depth averaged transport streamfunction '11, (CI=4 Sv) (dashed is negative (poleward) transport, solid is positive, bold is zero).

found to the north of Smoky Cape, Romax = 3, on the eastern side of the EAC jet where there is a strong northward flow (0.6 ms-1) and a small anticyclonic eddy CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 196

(Figure 6.14a, light shading). Another region of high Rossby number is found on the continental shelf beteen Smoky Cape and Diamond Head adjacent to the coast. The minimum contour on the plot is Ro = 0.2 which is still a considerable value implying that non-linearities play a role in the entire EAC jet region, particularly in the vicinity of Smoky Cape.

Transport

Transport in the EAC has been measured on a number of occasions and the re­ sults are quite varied. The most recent transport estimates come from the WOCE PCM3 current meter array positioned at 30°S, where over a three year period, Mata et al. (2000) observed transports of 22.1 ± 30 Sv towards the south. Others have reported transports of 22 - 35 Sv southward, between the coast and 155°E (Ridgway and Godfrey, 1994). Given these observations, estimates from the model of a max­ imum transport of 33 Sv are reasonable. Although the depth averaged velocities show a bifurcation of the current near Port Stephens, the transport streamfunc­ tion (Figure 6.14b) clearly shows separation from the coast, eastward, with minimal southward transport.

The transport streamfunction is calculated for a slab of water at the surface with a thickness of 20 m from both the modelled velocity field and the geostrophic velocity field as calculated from CARS (Figure 6.15). The slabs are of comparable thickness, however the velocities advecting the currents vary in each case causing the differences in transport magnitudes. It is encouraging to note that the general transport pattern is very similar. In the geostrophic case, as with the velocities, the transport is centred further offshore, and the jet is not as narrow as in the modelled case, however, this is a function of the smoothing inherent in any data atlas where the resolution is unable to resolve gradients. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 197

29S.------,------..---r-m-.-...... -:::_-_"""T'"_-=i_ ,---,------T"""---r-----,--.--U):--..---, C\/ 19 ,,, Surface Ill .. ' ., CARS tfc:,"":1 .,-o. I J ., I// i - .. ,,,,I / ,, .- - 30S / -- .... ' . ,11, ,,. ... \ ' 'I' (1.3 Sv) ,,,, ,.- 'I' (1.6 Sv) / I "1:I I I ., ,,,, '..> Ff.,,., -- ... I. ,,,,, -~.,/ I I' ·t '- '- I ·1 I I I 31S ,,,.,1, - ,_ ' ' ' I I I I \ ,, ,,,, l > ' ' ,,, '., ,,,,, ' ·s,- ' ' I I ' ,... '11 I \ \ .._ \ ,,,,,,'I I I '•1b', ', \ \ \ I I I f ' 32S ,, "cP ' ' ' ,_. f I I \ \ \ ' ,,~ ' I I I \ \ II I I \ -1.2 r Port II I ,, Port I I I oO I / ., J I I " I I 5=> I \ ,' '' I I I - '?~ ... I I I / \ I ,. \ '-' ', \ I I I I b< \ ' ... --- I --- 34S / ~- ...... ' I I I I I I/ ', ', --- I I I I / / I - ' .._ I ,, / '--o. ,; I I .... 151E 153E 155E 151E 153E 155E

Figure 6.15: Plot of \JI (Sv), the transport streamfunction (CI=0.2) along a surface slab from the modelled velocity field (left) and the geostrophic velocity field calculated from the CARS data (right). Maximum values are included in brackets. The solid line represents \JI= 0.

Richardson Number and Burger Number in the BBL

In Chapter 5 bottom boundary layer theory was introduced. It was noted that a reduction in the gradient Richardson number (Equation 5.13) below the critical value (Ri = 0.25), combined with a decrease in the Slope Burger number (Equation 5.14) acts to lengthen the time it takes for the BBL to shut down. This occurs as the shut down time is inversely proportional to Bu2 (Garrett et al., 1993). Thus it follows that as bottom slope (a) increases, mixing decreases, and less water is moved shorewards and upwards within the BBL over a shorter time scale. However, when mixing in the BBL is enhanced over a more gently sloping shelf, deeper waters are moved upwards within the BBL for a longer period of time. This could eventually result CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 198 in upwelling of slope waters to the surface.

In Chapter 5 it was noted that the value of Bu obtained from observations between Smoky Cape and Diamond Head decreased in mid-shelf regions. Calculations of Ri and Bu in the BBL across the model domain are shown in Figure 6.16. Values of

30S

31S

32S

33S

34S

151E 153E 153E

Figure 6.16: Plot of (a) the Richardson number Ri (CI=0.05, max=0.25) and (b) the slope Burger number Bu (CI=0.l, max=l). Darker contours represent lower values and the dashed line is the 200 m isobath indicating the shelf break.

Ri < 0.25 and Bu < 1 are plotted, and the 200 m isobath is represented by the dashed line. Immediately south of Smoky Cape there are two regions where low Ri values coincide with low Bu. The first of these regions is mid-shelf extending between Point Plomer and Diamond Head. It is here that the bottom slope is more gradual, and mixing is enhanced, thus facilitating movement of water shoreward through the BBL. The second region of low Ri and low Bu is at 32°S, again mid­ shelf. Although observations are not available for this area, the simulations suggest CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 199 that transport of slope waters shoreward through the BBL would be persistent here as well. The observations in Chapter 5 showed that current driven upwelling can occur, and the values as calculated from the mean climatological conditions support this view.

Divergence, Convergence and Rossby Number

The analysis of the observations presented in Chapter 4 found that the flow about the EAC separation point can be non-linear and ageostrophic. In order to further investigate the dynamics in the region the linearity and geostrophy of the flow are examined. The divergence (and hence convergence), and Rossby number have been calculated for the two layers which are used in the Lagrangian experiments, these being the surface slab and the a= 25.25 kgm-3 isopycnal surface. These are shown in Figure 6.17. Along both the surface slab and the isopycnal surface divergence is high eastward of the jet, interestingly however, divergence is also high within close proximity to the coast between Port Stephens and Smoky Cape (Figure 6.17). This positive divergence maximum inshore of the jet, south of the separation point shows the existence of upwelling. The maximum values range from 6.8 x 10-5 s-1 along the surface slab, to 6.6 x 10-5 s-1 along the a = 25.25 kg m-3 isopycnal. The amount of upwelled water resulting from the divergence can be calculated for a given volume. Assuming a slab of water (such as the surface slab) to be 20 m thick across an area of 1 km2 , then the amount of water upwelled would be in the range of 1340 - 1440 m3s-1 , across the slab. This is derived from an upwelling velocity of 0.0013 - 0.0014 ms-1 which is comparable to the vertical velocities obtained from the model.

Although the thickness of the isopycnal slab varies as it is uplifted, the maximum values of divergence (6.6 x 10-5 s-1 ) and convergence (8.5 x 10-5 s-1) are CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATI ON. 200

Surface

30S Div (6.Be-05 s- 1) Ro (4.7) Con (8.3e-05 s- 1)

31S

32S

Port Port

33S & 0 0 34S 0 0 I

29S

cr = 25.25

30S Div (6.6e-05 s- 1) Con (8.Se-05 s- 1) ~ 31S Smoky Cape 0

32S

33S

34S

151E 153E 155E 151E 153E 155E

Figure 6.17: Divergence (left , CI= l x 10- 5 s- 1 ) and the local Rossby number Ro (right, CI= 0. 5), at the surface (top) and along the (J' = 25.25 kg m- 3 isopycnal (bottom). Maximum values are included in brackets. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 201 comparable to the maximum surface values (6.8 x 10-5s-1 and 8.3 x 10-5s-1) where the slab is a constant 20 m thick. In fact the maximum convergence occurs at depth, across the isopycnal slab.

The Rossby number identifies the importance of the non-linear terms in the along­ shore flow. The Rossby numbers for each slab in Figure 6.17 are larger than those calculated for the depth averaged currents (Figure 6.14) The Romax = 4.7 along the surface slab Romax = 4.5 along the a= 25.25 kg m-3 isopycnal surface. Furthermore the regions of high Ro extend further along the coast than those calculated from the depth averaged velocity field. This implies that non-linearities are more prominent in the upper layers of the jet than at depth emphasising the baroclinic nature of the flow. The maximum values overlie the shelf break to the north of Smoky Cape and are associated with regions of convergence, where av / ax is the greatest. Fur­ ther eastward in regions of divergence Ro decreases, however it remains high where the current separates from the coast and moves into deeper water. As the current separates, initially vortex stretching occurs and the vorticity (Ro) increases. Vortex stretching will serve to tighten the front enhancing horizontal temperature gradients and maintaining strong horizontal and vertical velocity shear. However there comes a point when decoupling occurs and velocity is no longer constrained by contours of Coriolis parameter over depth (! / H). It is then that the vorticity decreases again, and both the front and the core of the current weakens and widens. The expansion of the jet is seen in the depth averaged velocity vectors of Figure 6.8, and the veloc­ ity cross sections shown upstream and downstream of the EAC separation point in Figure 6 .10.

It is encouraging that the flow dynamics calculated from the simulated velocity fields reproduce the observations obtained at Smoky Cape and Diamond Head. Further­ more the simulated fields have shed further light on the non-linear processes that occur in the Smoky Cape region. That non-linearities and current driven upwelling CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 202 are seen in the climatological current fields implies that these are persistent features along the coast between Smoky Cape and Diamond Head, thus explaining the com­ monly observed surface nutrient plumes in the region (Rochford, 1975; Hallegraeff and Jeffrey, 1993). Hence the modelled velocity fields can now be used to investigate the fate of the upwelled water.

6.5 Lagrangian Particle Tracking Experiments

Model experiments using Lagrangian particles are used to quantify meridional path­ ways and time scales of spreading within the EAC system in a similar fashion to that of Haines et al. (1999). A series of Lagrangian particle tracking experiments are conducted in an off-line fashion using the average climatological current field as obtained from the model. The experiments are conducted to address the question: 'Where does the water come from and where will it go?'. Specifically, where do upwelled waters originate in a 3 dimensional sense (i.e. depth and position), and once upwelled where do surface waters move to? These questions are of importance from a biological perspective, however knowledge of the movement of surface waters also has other practical applications, for example in search and rescue operations. In order to answer these questions, particles are advected both forward and backwards in time.

6.5.1 Particle Advection: Forward in time

The mean velocity fields obtained from the forced scenario are used to advect par­ ticles forward in time. Multiple trajectories are calculated where a white noise component is added to the mean current at each time step to obtain an estimate of the error in the mean. The white noise is constructed by multiplying the standard CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 203 deviation of the 5 day mean at each point by a normalised random number. This ensured that the error component, whilst being independent in time and space, is of the correct magnitude. A starting position is specified (t = 0) and both the particle and the noise are then advected for a 10 day period (t = 10) with an hourly time step. Assuming a maximum current of 2 ms-1 a time step of one hour can be justi­ fied, as the particle will not move outside the grid box in that time. Furthermore, other experiments were run with shorter time steps which are computationally far more expensive, but these revealed little additional information. Each experiment is run 50 times to provide an estimate of the reliability of the mean particle path. As the scenario modelled is only representative of the climatological currents, and not actually a reproduction of a particular event, it was not deemed necessary to run each experiment more than 50 times. Although performing many more runs would make the results statistically significant in a Monte Carlo sense, they would not aid in the interpretation of the dynamics. The surface velocities were used in order to investigate flow in the surface waters, which may represent the fate of water or particles that have been upwelled all the way to the surface.

Three types of experiments are conducted in this manner. In the first series of experiments, referred to as 'jet', particles are seeded across the EAC jet extending offshore from Smoky Cape, to examine how the particles move in and around the EAC jet. The separation of particles is examined in the second series of experiments, referred to as 'separation', where particles are seeded along a line very close to the separation point to see how particles are advected when the current separates. The final series of experiments, referred to as 'entrainment', is designed to investigate the possible advection of coastal particles in the northward cyclonic flow, before being entrained in the southward flowing EAC. To do this the particles are seeded in the coastal region where there is a northward flow, inshore of the jet. Figure 6.18 shows three representative cases (i.e. each with different release points) from each of the three different release experiments (jet, separation and entrainment). CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 204

29S.------r------, Jet 30S

31S Smoky Cape

32S

33S

34S

29S.------..------, .------r------, Separation 30S 2 3

Smoky Cape

29S.------r------. Entrainment 30S 2 3

31S Smoky Cape Smoky Cape

151E 153E 155E 151E 153E 155E 151E 153E 155E

Figure 6.18: Trajectories of particles advected forward in time for 10 days by surface velocities, showing three scenarios: jet (top), separation (middle) and entrainment (bottom). Each scenario shows three different release points (1-3). The particle origin (t=O) is indicated by t and the end-points (t=lO) are indicated by •. The paths indicate where nutrients or phytoplankton or a person swept from a yacht would be likely to travel. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 205

Jet The three jet experiments shown in the top row of Figure 6.18 show fairly similar scenarios. In the first two cases, each of the 50 particles within each experiment follow a similar path. In case 1 the maximum distance travelled by any of the 50 particles is only 191 km. The path followed is narrow, ensuring that there is very little spread in the destination points. In case 2, the maximum distance of any particle from the final mean position is only 22 km, with the minimum distance between particles only 100 m. In contrast to this is the third jet experiment where the maximum distance travelled is nearly 390 km, the end points are spread over 280 km from each other, and up to 257 km from the mean end point. If this scenario were real, the highly variable nature of the paths travelled would, for example, severely hamper search and rescue operations, and inhibit our ability to infer biologically important pathways.

Separation The results from the separation experiments (Figure 6.18 middle row) were less revealing, as the bifurcation of the current is not dominant. In the first two cases the end points were within 57 and 93 km of the mean end point respectively, having only travelled a distance of up to 230 km. The final separation experiment shown (case 3) reveals a tight initial path, which expands with latitude, leading to a final spread of 161 km, with 52% of the particles landing to the west of 153°E.

Entrainment Finally in the entrainment experiments the particles are released in the northward flowing waters inshore of the EAC jet (Figure 6.18 bottom row). The purpose is to investigate whether the particles will in fact get entrained in the jet and thus be advected southward once again, or if they remain caught in a constant eddy. In the first case all the particles were entrained rather rapidly, with 10% of the particles getting caught up in a swift flowing part of the jet, and travelling more CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 206 than 480 km with a mean speed of 0.6 ms-1 . The second and the third cases reveal initial northward movement before entrainment occurred. Speeds were slower, with 32% of the particles travelling less than 150 km in case 2, and a maximum distance travelled of 300 km. In case 3, 92% of the particles were entrained into the southward flow, however 8% travelled a maximum distance of 20 km northward, before landing 'on the rocks' at Smoky Cape.

Trajectories of particles advected forward in time by surface velocities give an indi­ cation of the path travelled by upwelled waters, or a person swept from a yacht. The paths, which are a result of the speed and distance travelled, are highly varied de­ pending on point ofrelease. In some cases the paths are tight, however, generally the results show that in a divergent/convergent situation the particles are not confined to any one particular path at all and simple conclusions are not appropriate.

6.5.2 Particle Advection: Backward in time

To further address the fundamental question of 'Where does the upwelled water come from'? the advection of particles is examined in a reverse fashion, both in the surface waters and along the a = 25.25 kg m-3 isopycnal surface. Particles are released in two directions, meridionally and zonally at t = 0 and then advected backwards in time for 10 days (t = -10).

Meridional Release Firstly particles are released meridionally, i.e. in the coastal waters extending the length of the coast at various distances from land. The results from three release distances (d = 9.5, 31 and 44 km from the coast) are displayed in Figure 6.19. By way of comparison, the three upper plots show the trajectories as determined by the surface currents, whereas the three lower plots show the trajectories of particles that have been advected along the a= 25.25 kg m-3 isopycnal. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 207

Surface trajectory

Release distance = 9.5 km Release distance = 31 km Release distance = 44 km 29S.-..-----,--..---~--, ...... ------~---~ ...... ----~~~~ 1-C.+ if'-'•

30S •

31S Smoky Cape

• t=-10 I • t=O .

151E 153E 155E 151E 153E 155E 151E 153E 155E lsopycnal trajectory a = 25.25

Release distance= 9.5 km Release distance = 31 km Release distance = 44 km

-150

# ·-1 • t=-10r • t=O . . . . . ,•

151E 153E 155E 151E 153E 155E 151E 153E 155E

Figure 6.19: Trajectories of particles released parallel to the coast at three offshore distances: 9.5 km (left), 31 km (middle) and 44 km (right). The release position (t=O) is indicated by •. The particles are advected backwards from their final position for 10 days (t=-10) to the particle origin which is indicated by+. Particles are advected at the surface, (top) and along the 25.25 a9 isopycnal, (bottom). The depth of the isopycnal is indicated by the dotted contours. The particle paths indicate where upwelled water may have originated. CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 208

At a distance of 9.5 km from the coast nearly all particles in the surface waters have a source immediately upstream, indicating simple southward advection. Particle trajectories further offshore indicate that particles have been entrained into the EAC from the coastal waters, with many of the particles undergoing initial northward movement before being caught up in the strong southerly flow. At the furthest distance from the coast north of Smoky Cape there is evidence of entrainment from inshore, as well as onshore flow eastward of the jet, and subsequent southward advection. South of Smoky Cape, both entrainment and advection occurs, as well as coastal recirculation. These trajectory experiments confirm that particles upwelled to the coastal waters at Smoky Cape would be advected to Port Stephens within 5 days.

The three lower panels of Figure 6.19 show the trajectories along the a= 25.25 kg m-3 isopycnal at the same initial distances from the coast. The isopycnal trajectories show that uplift from depth and subsequent entrainment play an important role along the isopycnal surface. North of Smoky Cape, uplift is occurring from east­ ward of the jet, and entrainment from inshore of the jet. Furthermore, at each of the offshore distances, there is a recirculation found to the north of Port Stephens centred around 32° 30'8.

Zonal Release The second method involved seeding particles zonally, i.e. along transects aligned perpendicular to the coast. Nine shore normal transects varying with latitude have been presented in Figure 6.20. The transects show that alongshore advection plays an increasing role southward of Smoky Cape. At the northern most section (north of Smoky Cape) the particles have travelled less than 180 km in the 10 day period, whereas further to the south, some of the particles that arrived at Port Stephens travelled well over 400 km. These particles are obviously caught up in the main EAC jet and are advected rapidly. At the northern sections, many of the particles CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 209

'

Diamond Hea ,<:,Q ,. Diamond Hea <:,() • . '

··1 .• t=-10I·· . t=O .. ..

0

.. .·· !(] . 30S --- ,1s· 175

31S

32S ~ .- "'o __ .

33S

· 100 . . ·· • t=-101 ·· 34S . 1 • t=O .. .. .

30S . - .. '175 175

31S

32S · 150 150

· 125. 33S

t=-10I · 34S . 1 t=O

151E 153E 155E 151E 153E 155E 151 E 153E 155E

Figure 6.20: Trajectories of particles released shore normal at nine alongshore locations. The release position (t=O) is indicated by • . The particles are advected backwards from their final position for 10 days (t=-10), along the 25.25 uo isopycnal surface, to the particle origin which is indicated by +. The depth of the isopycnal is represented by the dotted contours (maximum depth = 175 m) . CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 210 are uplifted from depth with the isopycnal, the depth of which is indicated by the dotted contour. Across the southern sections between Diamond Head and Port Stephens the particles originated in shallower water. The sections south of Smoky Cape clearly show that particles that are found inshore actually originated further to the south and are advected northwards by the inshore counter current. The particles that move northwards travel at a much slower rate and show evidence of a cyclonic recirculation.

Biologically the implications are important. Phytoplankton form the basis of the marine food chain and are eaten directly by herbivorous animals such as zooplankton and larval fish. For a fish stock to benefit from an algal bloom however, the plank­ ton cells must be present in sufficient quantity and quality, and in the right place in the water column at the right time. If this is not the case, it is possible that fish stocks will not develop (Jeffrey and Hallegraeff, 1990). Retention in a recirculating eddy allows greater utilisation of the nutrient by photo-synthetic organisms such as phytoplankton, as it increases the residence time in any one location. Moreover, the presence of a cyclonic eddy is likely to enhance upwelling through Ekman pump­ ing. Furthermore if the nutrient rich water is persistently being brought toward the surface algal blooms are able to develop frequently. If the particles represent phytoplankton (chlorophyll-a), then retention in a tight group (i.e. a bloom) for a longer period of time allows the phytoplankton to be taken up through the food chain, enhancing growth of zooplankton and larval fish.

It is well documented that different phytoplankton species have their own light preferences, however it is generally accepted that the lower limit of marine plants is at the 0.5 - 1% surface light depth (Jeffrey and Hallegraeff, 1990). In order to photosynthesise, phytoplankton must be within this euphotic zone. The a = 25.25 kg m-3 isopycnal surface was raised to a depth of 50 - 70 m in the EAC jet, and observations showed that light penetrated to 50 - 60 m in this region. At the CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 211 three most northern sections, particles which were found to lie offshore were uplifted from 175 m to a depth of less than 100 m, and were then advected southward. The observations showed an increase in the concentrations of nutrient in the surface waters at Smoky Cape, and subsequent alongshore advection towards Point Plomer, where chlorophyll-a concentrations were seen to increase. This is a consequence of the isopycnal surface being raised into the euphotic zone, where photosynthesis can occur.

6.6 Summary

The primary aim of this study was to implement the POM to gain insight into the climatological oceanographic conditions prevalent in the waters off the NSW coast. The model was initialised with the climatological temperature and salinity records obtained from CARS and different initialisation techniques were trialed. Ramping and nudging over 5 inertial periods proved to be effective initialisation techniques, however forcing the model at the northern boundary with a geostrophic current reduced the time taken for the simulation to spin up.

The findings of the observational study confirm that the simulated currents can in fact realistically represent actual conditions. Cross sections of the alongshore veloc­ ity compare favourably with measurements taken both upstream and downstream of the EAC separation point. Surface velocity profiles replicate accurately the strength and proximity of the EAC to the coast.

Dynamical investigations confirmed the findings of previous chapters that non­ linearities are important in the EAC system, furthermore divergence is prevalent along the coast between Smoky Cape and Port Stephens. This divergence perhaps contributes to the persistent upwelling observed in the region. Calculations of vol- CHAPTER 6. A MODELLING STUDY ABOUT THE EAC SEPARATION. 212 ume transports from the model simulations are within the commonly accepted range of measured volume transports.

Particle tracking experiments were undertaken to investigate the origin of upwelled water and to understand the importance of advection and entrainment. The uplift of the a= 25.25 kg m-3 isopycnal from depth combined with southward flowing cur­ rents explains how nutrient enriched water reach the euphotic zone south of Smoky Cape. Advection of particles in a Lagrangian fashion both forward and backward in time revealed that surface waters flowing northward along the coast were more often than not entrained into the EAC jet and advected southward. Biologically this would imply that any surface plume of nutrient rich water (upwelled from the persistent divergence) or a bloom of phytoplankton that occurred inshore of the southward flowing jet would more than likely be advected southward, with the pos­ sibility of some northward movement first. Phytoplankton blooms have historically been observed in this region (Rochford, 1975). However contrary to popular belief, the algal blooms observed did not necessarily originate to the north at Smoky Cape, but could also have been advected from Diamond Head northward. The relationship between nutrient concentrations, and the development of phytoplankton and zoo­ plankton blooms in the EAC environment is not well documented, and thus could form the basis for future work. Now is not the end. It is not even the beginning of the end. But it is perhaps, the end of the beginning.

Winston Churchill. 10 Nov 1942

Recapitulation

The results presented in this thesis have addressed the disparity in the literature regarding the role of the East Australian Current (EAC) in driving the uplift and upwelling of nutrient rich water into the near shore zone in the vicinity of Smoky Cape, NSW, Australia. In the past, direct observations of current driven upwelling have been both scarce and inconclusive, whilst highly idealised modelling studies have downplayed the importance of current driven upwelling. Furthermore, this is the first observational study to address the question of the dynamics about the separation point of the EAC.

213 CHAPTER 7. RECAPITULATION 214

Chapter 2

In Chapter 2 the design and implementation of the field experiment was presented. The experiment was designed specifically to address the key issues regarding current driven upwelling and separation point dynamics which were raised in Chapter 1. Previous studies have been suggestive and have lacked the details necessary to do more than infer conclusions regarding the role of the EAC in upwelling. The data set consists of high resolution hydrographic sections concurrent with velocity and temperature time series. The mooring arrays spanned the separation point of the EAC for a two month period and the data collected forms the basis of the first definitive study in the region of the EAC separation point.

Chapter 3

The time series collected from the mooring arrays enabled a partitioning of the data set into key events which formed the basis for the analyses in Chapter 4 and Chapter 5. These events are as follows:

• a Wind event where persistent northerly winds drove a significant upwelling event;

• a Reversal event where the swiftly flowing southerly EAC jet was positioned further offshore and the currents across the shelf fluctuated between a weak northward and weak southward flow. These oscillations in the current field were shown to be the result of northward propagating coastal trapped waves;

• a Current event which was distinguished by the encroachment of the EAC upon the continental shelf. This encroachment of the current was seen to drive an onshore flow of colder water through the bottom boundary layer at Smoky Cape which resulted in upwelling at Diamond Head; and CHAPTER 7. RECAPITULATION 215

• two EA C events where the EAC was flowing close to the coast at both Smoky Cape and Diamond Head, implying that the current separated from the coast further to the south.

These key features which have not previously been distinguished allowed for the examination of the dynamics about the EAC separation point.

Chapter 4

In Chapter 4, variability about the separation point of the EAC was examined. Var­ ious techniques were utilised to investigate the flow dynamics including time series analysis (spectra, coherence and cross correlations) which showed that the dynamics about the separation point are both complex and confusing. EOF analysis was used to examine both the horizontal and vertical structure of the currents upstream and downstream of the separation point. Upstream, the flow was mainly barotropic and unidirectional, whereas downstream of the separation point the across-shore com­ ponent of the flow increased as the axis of the current moved across the continental shelf.

The balance of terms in the alongshore momentum equation was investigated using both time series and statistical techniques. Whilst the time series were difficult to interpret, when they were decomposed into EOFs balances could be established. The balance of terms revealed that the alongshore pressure gradient dominated the flow upstream of the separation point and both bottom stress and non-linear advection were also important. Downstream of the separation point the flow was generally dominated by tendency inshore and fluctuations in the across-shore velocity mid­ shelf.

The EOFs of the terms were also calculated for each of the events described in CHAPTER 7. RECAPITULATION 216

Chapter 3 in order to examine the dominant balances driving each event. During the Wind event, despite causing the upwelling of colder water, wind stress did not drive the depth averaged flow. During the Reversal event tendency ( acceleration at a point) was important at Smoky Cape and the pressure gradient was driving the flow at Diamond Head. Throughout this event the contribution from advection and bottom stress was minimal. During the Current event, where the EAC encroached across the continental shelf, tendency dominated at Smoky Cape and alongshore the pressure gradient was again dominant at Diamond Head. However, inshore at Smoky Cape the balance between advection and bottom stress was re-established. During EA C 1 and EA C 2 the pressure term again dominates at Smoky Cape, whilst at Diamond Head advection is important.

Essentially the balances in the alongshore momentum equation are difficult to resolve however there are two significant aspects: Firstly, despite the fact that upwelling is a common feature of the Smoky Cape region, it is demonstrated that the contribution of wind stress to the horizontal flow is virtually negligible. However, during the Wind event significant upwelling does occur. Secondly, during the Current event when upwelling also occurs, bottom stress and non-linear advection become important.

Chapter 5

The previous chapter showed that the EAC was possibly responsible for upwelling that occurred during the Current event. In Chapter 5 additional evidence was presented that conclusively showed the occurrence of both wind driven and current driven upwelling in the Smoky Cape region.

Although wind stress was found to have little effect on the depth averaged flow, when circumstances were optimal, i.e. when the isotherms were pre-conditioned by the proximity of the EAC over the shelf break, upwelling can result from persistent CHAPTER 7. RECAPITULATION 217 northerly winds. The observations showed three ways that the EAC can induce upwelling: i) through encroachment of the axis of the current across the continental shelf; ii) through topographic acceleration of the flow; and iii) through the separation of the current from the coast. The results presented indicated that when bottom stress is increased either through encroachment of the current across the shelf, or through acceleration of the flow by the narrowing of the shelf at Smoky Cape, cold nutrient rich water is driven shoreward through the bottom boundary layer into the coastal waters. Finally, as the EAC separates from the coast, an increase in transport through the bottom boundary layer results in an increase in nutrient rich water being brought to the surface downstream of the separation point.

Current driven upwelling was found to be more effective than wind driven upwelling in uplifting the isotherms by an order of magnitude. As a consequence, concentra­ tions of nutrients resulting from current driven upwelling were greater than those which were uplifted by the upwelling favourable winds.

Chapter 6

Finally, a modelling study documented in Chapter 6 permits a cohesive interpreta­ tion of the results of previous chapters. The climatological records of temperature and salinity were used to initialise the model, which resulted in a very realistic rep­ resentation of the velocity field. Although it was not intended that the modelling study reproduce any particular event, the general circulation that was simulated replicated the conditions that were observed during November 1998. From the com­ parison between the climatological currents in the modelling study and the obser­ vations presented here, it can be assumed that the conditions observed during this time were similar to the mean current field off the east coast of Australia. Hence conclusions can be drawn concerning the mean nutrient enrichment processes that CHAPTER 7. RECAPITULATION 218 occur in this region.

The velocity fields obtained from the model were used to examine the dynamics across the shelf in the Smoky Cape region. It was found that over the shelf break the flow can be highly non-linear and that these regions coincided with regions of horizontal divergence and low Burger number. The implication of this is that the bottom boundary layer can remain active for a longer time, thus allowing the flow of nutrient rich water through the bottom boundary layer into the coastal region.

Extrapolating the results of the modelling study suggests that current driven nutri­ ent enrichment occurs regularly in the Smoky Cape region. This would explain why nutrient enrichment has been seen as a semi-permanent feature downstream of the EAC separation point. Hence the results presented here provide conclusive evidence for the hypothesis of Rochford (1975), that the EAC does in fact drive upwelling in the Smoky Cape/Laurieton region and for the suggestion of Oke and Middleton (2000), that an increase in bottom stress is important in causing the upwelling.

A relationship between the chlorophyll maximum and the a= 25.25 kgm-3 isopy­ cnal was identified in the hydrographic data set. This was investigated further through the results of the model simulations where the flow along the a = 25.25 kg m-3 isopycnal was used as a proxy for upwelled water. The mean velocity fields both at the surface and along the a = 25.25 kg m-3 isopycnal were then used in a series of Lagrangian particle tracking experiments which investigated both the source and fate of the upwelled water.

It was found that upwelled water originated at depth to the north of Smoky Cape, and at times well offshore. South of Smoky Cape, coastal recirculations were evident which suggested that upwelled water could be trapped in an eddy by the coast. This would allow the utilisation of upwelled nutrients by phytoplankton and further integration into the food chain. This explains the frequent phytoplankton blooms in CHAPTER 7. RECAPITULATION 219 the region south of Smoky Cape. Furthermore, an observed northward flow towards the separation point was entrained in the southward flowing EAC jet, which shows how plankton blooms could be advected from the coastal region out to sea along the EAC front.

7.1 Objectives Revisited

With regard to the objectives of this study that were outlined in Section 1.4 each issue has been addressed as follows:

1. To investigate the dynamics upstream and downstream of the EA C separation point.

The dynamics upstream of the separation point tend to be predictable whilst downstream the dynamics are more complex and are dependent on both the location of the separation point and the proximity of the EAC jet to the coast. The larger scale pressure field tends to dominate the flow which may be influenced by Rossby waves and time dependence in the EAC.

2. To examine upwelling in the Smoky Cape region and suggest possible mecha­ nisms driving the upwelling.

Upwelling in the region South of Smoky Cape can be driven by local wind forcing, however, more important is the acceleration of the flow which results from either encroachment of the current across the shelf, or through topo­ graphic acceleration of the flow. Through an increase in bottom stress both mechanisms activate the pumping of deeper waters shoreward through the bot­ tom boundary layer and this often results in upwelling south of Smoky Cape. Furthermore, separation of the EAC can also play a role in the upwelling of nutrient rich waters into the euphotic zone. CHAPTER 7. RECAPITULATION 220

3. To understand the advection processes near the EA C separation point.

Particle tracking experiments allowed for an understanding of the advection and circulation near the separation point and completed the three dimensional picture of advection of nutrient rich water in the Smoky Cape region. Upwelled water was found to originate offshore at depth, which upon reaching the sur­ face was then advected both northward inshore of the separation front and southward within the EAC jet.

7.2 Concluding Remarks

The comprehensive data set obtained in this study has the potential to address many more questions than those addressed in this study. For example, the dynamics of the separation front where a 7 °C change was observed across a narrow velocity front could be investigated. Upwelling was observed associated with this front, the mechanisms of which are not fully understood.

The success of the modelling study indicates that the model lends itself to further investigations, in particular in regard to the dynamics about the separation point. Whilst it is difficult to obtain a balance in the equations of motion in an observational study, a numerical study to investigate the balance of terms could be of benefit.

Finally additional analysis of the observational data in conjunction with a numer­ ical modelling study could provide further insight into the mechanisms driving the different upwelling scenarios observed. For example, the various upwelling scenarios presented could be investigated by prognostic model simulations. It may also be useful to compare the relative strengths of the various acceleration mechanisms and thus approximate the magnitude of the resulting upwelling. If encroachment of the current causes a more rapid acceleration, the upwelling response may be different in CHAPTER 7. RECAPITULATION 221 time, extent and magnitude. Thus there remains scope for further investigations of the individual western boundary current upwelling mechanisms identified here and the associated nutrient transport. But in your wonderful, secret way, my God, you had already taught me that a statement is not necessarily true because it is wrapped in fine language or false because it is awkwardly expressed.

Confessions Saint Augustine (354-430 AD)

Appendix A: The Princeton Ocean Model

The Princeton Ocean Model (POM) is a finite difference model with sigma co­ ordinates (topography following) that solves the non-linear, hydrostatic, primitive equations with a free-surface (Blumberg and Mellor, 1987). The model is freely available from the POM home page: http://www.aos.princeton.edu/WWWPUBLIC/htdocs.pom The main attributes of the model are as follows: Vertical diffusivities of momentum, salt and heat are calculated using a 2.5 level embedded turbulent closure sub-model (Mellor and Yamada, 1982). The horizontal grid is constructed with curvilinear coordinates and an Arakawa C grid. The horizontal diffusion coefficients are calcu­ lated using the Smagorinsky formula (Mellor, 1996). This formulation assumes that the horizontal diffusion coefficients are proportional to the horizontal grid resolution and to horizontal shears in velocity.

The horizontal time differencing is explicit whilst the vertical time differencing is implicit. Eliminating time constraints in the vertical allows for higher resolution in both the surface and bottom boundary layers.

The equations that are solved in the model are: arJ at+ 'v.(Dv) 0

222 APPENDIX A: THE PRINCETON OCEAN MODEL 223

8uD 877 at+ v.V(uD) - fvD + gD ax + gD2 1° + [8p' _ a' 8D 8p'] da' ~ [KM au] F. Po IT ax D ax 8a' 8a D 8a + u 8vD 877 at+ v.V(vD) + fuD + gD ay +

gD2 1° [8p' _ a' 8D 8p'] da' ~ [KM 8v] F. v Po IT 8y D 8y 8a' 8a D 8a + 80D at+v.V0D ~ [KH 80] +Fe 8a D 8a 8SD at+v.VSD ~ [KHasl +Fs 8a D 8a

Where:

V velocity vector ( u, v, w) in the x, y and a directions ( ms-1) gradient operator (i%x + j%Y + ktlT) 77 sea surface elevation (m) H mean depth (m) D total water column height (H + 77)

P,Po in-situ density and reference density (kg m-3) horizontal diffusion terms of momentum horizontal diffusion terms of heat and salt

vertical eddy diffusivity of momentum (m2 s-1)

vertical eddy diffusivity of heat and salt (m2 s-1) potential temperature salinity (psu)

g gravitational acceleration (m s-2)

f Coriolis parameter (s-1)

(7 z-;}' the sigma coordinate

A mode splitting technique is implemented for computational efficiency, whereby the depth averaged velocities are calculated every external time step whilst the three APPENDIX A: THE PRINCETON OCEAN MODEL 224 dimensional fields of velocity, temperature and salinity are only calculated at a longer internal time interval. This is possible, as baroclinic motions such as internal waves have slower phase speeds than the barotropic motions such as surface gravity waves, and hence less need for temporal resolution. Bibliography

Allen, J. S., P. A. Newberger, and J. Federick, 1995: Upwelling circulation on the Oregon continental shelf. part 1: Response to idealised forcing. J. Phys. Oceanogr., 25, 1843-1866.

Allen, J. S., and R. L. Smith, 1981: On the dynamics of wind-driven shelf currents. Phil. Trans. R. Soc. Lond., A 302, 617-634.

Anderson, D. M., 1989: Toxic algal blooms and red tides: A global perspective .. In Okaichi, T., D. Anderson, and T. Nemoto (Eds.), Red tides - Biology, Environmental Science and Toxicology., pp. 11-16. Elsevier., New York.

AODC., annual: Western Tasman oceanographic conditions. Hydrographic Office RAN North Sydney.

Berkelmans, R., and J. K. Oliver, 1999: Large-scale bleaching of corals on the Great Barrier Reef. Coral Reefs, 18, 55-60.

Blackburn, S. I., and G. Cresswell, 1993: A Coccolithophorid bloom in Jervis Bay, Australia. Aust. J. Mar. Freshw. Res., 44, 253-60.

Blumberg, A. F., and G. L. Mellor, 1987: A description of a three-dimensional

225 BIBLIOGRAPHY 226

coastal ocean circulation model. In N.Heaps (Ed.), Three-Dimensional Coastal Ocean Models, Vol. 4, p. 208. American Geophysical Union, Washingtion DC.

Boland, F. M., and J. A. Church, 1981: The East Australian Current 1978. Deep-Sea Res., 28, 937-957.

Boland, F. M., and B. V. Hamon, 1970: The East Australian Current 1965-1968. Deep-Sea Res., 17, 777-794.

Chapman, D. C., and S. J. Lentz, 1997: Adjustment of stratified flow over a sloping bottom. J. Phys. Oceanogr., 27, 340-356.

Church, J. A., 1987: East Australian Current adjacent to the Great Barrier Reef. Aust. J. Mar. Freshw. Res., 38, 671-683.

Church, J. A., and P. D. Craig, 1998: Australia's shelfseas: Diversity and complexity. In Robinson, A. R., and K. H. Brink (Eds.), The Sea, Vol. 11, pp. 933-964. John Wiley & Sons, Inc., New York.

Church, J. A., and G. R. Cresswell, 1986: Oceanographic features of Southeast Australian waters. Tech. rep., CSIRO Marine Labs, GPO Box 1538, Hobart, TAS 7001, Australia. 20pp.

Church, J. A., and H. J. Freeland, 1987: The energy source for the coastal trapped waves in the Australian Coastal Experiment region. J. Phys. Oceanogr., 17, 289-300.

Church, J. A., H. J. Freeland, and R. L. Smith, 1986a: Coastal trapped waves on the East Australian continental shelf. part 1: Propagation of modes. J. Phys. Oceanogr., 16, 1929-1943.

Church, J. A., H. J. Freeland, and R. L. Smith, 1986b: Coastal trapped waves on the East Australian continental shelf. part 2: Model verification. J. Phys. Oceanogr., 16, 1945-1957. BIBLIOGRAPHY 227

Cresswell, G. R., 1994: Nutrient enrichment of the Sydney continental shelf. Aust. J. Mar. Freshw. Res., 45, 677-691.

Cresswell, G. R., C. Ellyett, R. Legeckis, and A. F. Pierce, 1983: Nearshore features of the East Australian Current system. Aust. J. Mar. Freshw. Res., 34, 105- 114.

Davis, R. E., 1976: Predictability of sea surface temperature and sea level pressure anomalies over the North Pacific Ocean. J. Phys. Oceanogr., 6(3), 249-266.

Denbo, D. W., and J. S. Allen, 1984: Rotary empirical orthogonal function analysis of currents near the Oregon coast. J. Phys. Oceanogr., 14, 35-46.

Dewey, R. K., 2000: Mooring design and dynamics a matlab package for designing and analyzing oceanographic moorings. url=http://canuck.seos.uvic.ca/rkd/mooring/moordyn.html. Ma­ rine Models, 1, 103-157.

Emery, W. J., and R. E. Thomson, 1998: Data analysis methods in Physical Oceanography. Pergamon Press.

Ezer, T., and G. L. Mellor, 1994: Diagnostic and prognostic calculations of the North Atlantic circulation and sea level using a sigma coordinate ocean model. J. Geophys. Res., 99 C7, 14,159-14,171.

Fahrbach, E., and C. Brockmann, 1981: The northern Peruvian upwelling system during the ESACAN experiment. In Richards, F. A. (Ed.), Coastal Upwelling, Vol. 1 of Coastal and Estuarine Sciences, pp. 107-118. American Geophysical Union, Washington, D. C. 529 pp.

Freeland, H. J., F. M. Boland, J. A. Church, A. J. Clarke, A. M. G. Forbes, A. Huyer, R. L. Smith, R. 0. R. Y. Thompson, and N. J. White, 1986: The BIBLIOGRAPHY 228

Australian Coastal Experiment: A search for coastal trapped waves. J. Phys. Oceanogr., 16, 1230-1249.

Furnas, M. J., and A. W. Mitchell, 1993: Nutrient inputs into the central Great Barrier Reef from subsurface intrusions of Coral Sea waters: A two dimensional displacement model. Cont. Shelf Res., 16, 1127-1148.

Garrett, C., P. MacCready, and P. Rhines, 1993: Boundary mixing and arrested ekman layers: rotating stratified flow near a sloping boundary. Annu. Rev. Fluid. mech., 25, 291-323.

Gibbs, M. T., P. Marchesiello, and J. H. Middleton, 1997: Nutrient enrichment of Jervis Bay, Australia during the massive 1992 coccolithophorid bloom. Mar. Freshw. Res., 48, 473-478.

Gibbs, M. T., J. H. Middleton, and P. Marchesiello, 1998: Baroclinic response of Sydney shelf waters to local wind and deep ocean forcing. J. Phys. Oceanogr., 28(2), 178-190.

Gill, A. E., 1982: Atmosphere Ocean Dynamics. Academic Press.

Godfrey, J. S., 1973: Comparison of the East Australian Current with the western boundary flow in BRYAN and COX's (1968) numerical model ocean .. Deep-Sea Res., 20, 1059-1076.

Godfrey, J. S., G. R. Cresswell, and F. M. Boland, 1980a: Observations of low Richardson Numbers and undercurrents near a front in the East Australian Current. J. Phys. Oceanogr., 10, 301-307.

Godfrey, J. S., G. R. Cresswell, T. J. Golding, and A. F. Pierce, 1980b: The sepa­ ration of the East Australian Current. J. Phys. Oceanogr., 10, 430-440. BIBLIOGRAPHY 229

Griffin, D. A., and J. H. Middleton, 1986: Coastal trapped waves behind a large continental shelf island, Southern Great Barrier Reef. J. Phys. Oceangr., 16, 1654-1664.

Griffin, D. A., and J. H. Middleton, 1991: Local and remote wind forcing of New South Wales inner shelf currents and sea level. J. Phys. Oceanogr., 21, 304- 322.

Haines, M. A., R. A. Fine, M. E. Luther, and Z. Ji, 1999: Particle trajectories in an Indian Ocean model and sensitivity to seasonal forcing .. J. Phys. Oceanogr., 29, 584-598.

Hallegraeff, G. M., 1988: Three estuarine Australian dinoflagellates that can produce paralytic shellfish toxins .. J. Plankton. Res., 10, 533-541.

Hallegraeff, G. M., and S. W. Jeffrey, 1993: Annually recurrent Diatom blooms in Spring along the New South Wales coast of Australia. Aust. J. Mar. Freshw. Res., 44, 325-34.

Hamilton, L. J., 1996: Mesoscale variability in volume transport in the Tasman and Coral Seas. Mar. Freshw. Res., 47, 497-508.

Hamon, B. V., 1965: The East Australian Current 1960-1964. Deep Sea Res., 12, 899-921.

Hamon, B. V., 1968: Temperature structure in the upper 250 metres in the East Australian Current area. Aust. J. Mar. Freshw. Res., 19, 91-99.

Hamon, B. V., and J. D. Kerr, 1968: Time and space scales of variations in the East Australian Current, from merchant ship data. Aust. J. Mar. Freshw. Res., 19, 101-106. BIBLIOGRAPHY 230

Huthnance, J. M., 1995: Circulation, exchange and water masses at the ocean margin: the role of physical processes at the shelf edge. Prog. Oceanogr., 35, 353-431.

Huyer, A., R. L. Smith, P. J. Stabeno, J. A. Church, and N. J. White, 1988: Currents off South-eastern Australia: Results from the Australian Coastal Experiment. Aust. J. Mar. Freshw. Res., 39, 245-288.

Janowitz, G. S., and L. J. Pietrafesa, 1982: The effects of alongshore variation in bot­ tom topography on a boundary current - (topographically induced upwelling). Cont. Shelf Res., 1, 123-141. -

Jeffrey, S. W., and G. M. Hallegraeff, 1990: Phytoplankton ecology of Australasian waters. In Clayton, M., and R. King (Eds.), Biology of marine plants, pp. 310-348. Longman Cheshire.

Johns, E., D. R. Watts, and H. T. Rossby, 1989: A test of geostrophy in the Gulf Stream. J. Geophys. Res., 94, 3211-3222.

Kelley, D. E., and M. C. Bourque, 1997: Inferring coastal upwelling dynamics from a thermistor array. J. Geophys. Res., 102, 3439-3449.

Kundu, P. K., J. S. Allen, and R. L. Smith, 1975: Modal decomposition of the velocity field near the Oregon coast. J. Phys. Oceanogr., 5, 683-704.

Large, W. S., and S. Pond, 1981: Open ocean momentum flux measurements in moderate to strong winds. J. Phys. Oceanogr., 11, 324-336.

Lentz, S. J., R. T. Guza, S. Elgar, F. Feddersen, and T. H. C. Herbers, 1999: Momentum balances on the North Carolina Inner Shelf. J. Geophys. Res., 104, 18,205-18,226.

Lentz, S. J., and C. D. Winant, 1986: Subinertial currents on the Southern Califor­ nian Shelf. J. Phys. Oceanogr., 16, 1737-1750. BIBLIOGRAPHY 231

Lippert, A., and M. G. Briscoe, 1990: Observations and EOF analysis of low­ frequency variability in the western part of the Gulf Stream recirculation. J. Phys. Oceanog., 20.

MacCready, P., and P. Rhines, 1993: Slippery bottom boundary layers on a slope. J. Phys. Oceanogr., 23, 5-22.

Marchesiello, P., M. T. Gibbs, and J. H. Middleton, 2000: Simulations of coastal upwelling on the Sydney continental shelf. Aust. J. Mar. Freshw. Res., 51, 577-588.

Marchesiello, P., and J. H. Middleton, 2000: Modeling the East Australian Current in the Western Tasman Sea .. J. Phys. Oceanogr., 30, 2956-2971.

Mata, M. M., 2000: On the mesoscale variability of the East Australian Current at Subtropical latitudes. Ph.D. thesis, FIAMS Flinders University of South Australia, GPO Box 2100, Adelaide, SA 5001, Australia.

Mata, M. M., M. Tomczak, S. Wijffels, and J. A. Church, 2000: East Australian Current volume transports at 30°S: Estimates from the World Ocean Cir­ culation Experiment hydrographic sections PR11/P6 and the PCM3 current meter array. J. Geophys. Res., 105, 28509-28526.

Mellor, G. L., 1991: An equation of state for numerical models of Oceans and Estuaries. J. Atmos. Ocean. Tech., 8, 608-611.

Mellor, G. L., 1996: Users guide for a three-dimensional, primitive equation, numer­ ical ocean model. Tech. rep., Program in Atmospheric and Oceanic Sciences, Princeton University, Princeton, NJ 08544-0710, USA.

Mellor, G. L., and T. Yamada, 1982: Development of a turbulent closure model for Geophysical Fluid problems. Rev. Geophys. Space Phys., 20, 851-875. BIBLIOGRAPHY 232

Oke, P. R., and J. H. Middleton, 2000: Topographically induced upwelling off East­ ern Australia. J. Phys. Oceanogr., 30, 512-531.

Oke, P.R., and J. H. Middleton, 2001: Nutrient enrichment off Port Stephens: The role of the East Australian Current. Cont. Shelf Res., 21, 587-606.

Oke, P. R., R. N. Miller, J. S. Allen, G. D. Egbert, and P. M. Kosro, 2000: Assim­ ilation of surface velocity data into primitive equation coastal ocean models. in preparation.

Owens, W., 1985: A statisical description of the vertical and horizontal structure of eddy variability on the edge of the Gulf Stream recirculation. J. Phys. Oceanogr., 15, 195-205.

Palma, E. D., and R. P. Matano, 1998: On the implementation of passive open boundary conditions for a general circulation model: The barotropic mode .. J. Geophys. Res., 103 Cl, 1319-1341.

Pearce, G. L., 1980: Variability of currents off Cape Hawke, New South Wales 1978- 1979. Tech. rep. 128, CSIRO Division of Fisheries and Oceanography, Marine Laboratory, Cronulla NSW, Australia.

Pond, S., and G. L. Pickard, 1983: Introductory Dynamical Oceanography. Pergamon Press.

Reid, J. L., and A. W. Mantyla, 1976: The effect of the geostrophic flow upon coastal sea elevations in the northern North Pacific Ocean. J. Geophys. Res., 81, 31300-31310.

Ridgway, K. R., J. R. Dunn, and J. L. Wilkin, 2001: Ocean interpolation by 4- dimensional weighted least squares: application to the waters around Aus­ tralasia. J. Atmos. Ocean. Tech., in press. BIBLIOGRAPHY 233

Ridgway, K. R., and J. S. Godfrey, 1994: Mass and heat budgets in the East Aus­ tralian Current: A direct approach. J. Geophy Res-Oceans, 99, 3231-3248.

Ridgway, K. R., and J. S. Godfrey, 1997: Seasonal cycle of the East Australian Current. J. Geophy Res-Oceans, 102, 22 921-22 936.

Rochford, D. J., 1972: Nutrient enrichment of East Australian coastal waters. 1. Evans Head upwelling. CSIRO Aust. Div. Fish. Oceanogr. Rep., 33.

Rochford, D. J., 1975: Nutrient enrichment of East Australian coastal waters. 2. Laurieton upwelling. Aust. J. Mar. Freshw. Res., 26, 233-43.

Rochford, D. J., 1984: Nitrates in Eastern Australian coastal waters. Aust. J. Mar. Freshw. Res., 35, 385-97.

Rochford, D. J., 1991: 'upwelling': Does it need a stricter definition?. Aust. J. Mar. Freshw. Res., 42, 45-46.

Smith, R. L., 1981: A comparison of the structure and variability of the flow field in three coastal upwelling regions: Oregon, North West Africa, and Peru. In Richards, F. A. (Ed.), Coastal Upwelling, Vol. 1 of Coastal and Estuarine Sciences, pp. 107-118. American Geophysical Union, Washington, D. C. 529 pp.

Tilburg, C. E., H. E. Hurlburt, J. J. O'Brien, and J. F. Shriver, 2001: The dynamics of the East Australian Current system: The Tasman Front, the East Auckland Current, and the East Cape Current. J. Phys. Oceanogr., 31, 2917-2943.

Tranter, D. J., D. J. Carpenter, and G. S. Leech, 1986: The coastal enrichment effect of the East Australian Current eddy field. Deep-Sea Res., 33, 1705-1735.

Wunsch, C., and D. Stammer, 1997: Atmospheric loading and the oceanic inverted barometer effect. Rev. of Geophys., 35, 79-107.