The Development and Evolution of the Burdekin River Estuary Freshwater Plume During Cyclone Debbie (2017)

The development and evolution of the Burdekin River estuary freshwater plume during Cyclone Debbie (2017)

Yuanchi Xiao

A thesis in fulfilment of the requirements for the degree of

Master of Philosophy

School of Physical, Environmental and Mathematical Sciences

The University of New South Wales

Canberra, ACT, 2600, Australia

August 2018

THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet

Surname or Family name: Xiao

First name: Yuanchi Other name/s:

Abbreviation for degree as given in the University calendar: MPhil

School: School of Physical Environmental and Faculty: The University of New South Wales Mathematical Sciences Canberra

Title: The development and evolution of the Burdekin River estuary freshwater plume during Cyclone Debbie (2017)

Abstract 350 words maximum: (PLEASE TYPE) This thesis investigates the plume morphology and dynamics prior to and after the landfall of Cyclone Debbie (2017). The heavy rainfall and flooding produced a large buoyant coastal current, which moved southward after the cyclone made landfall then advected northward with the prevailing southerly wind. The plume is simulated using the eReef GBR1 1-km model and a passive tracer is used to investigate the plume behaviour. Based on the concentration of river tracers from the Burdekin River, the plume propagated over 100 km to the north during the 23 days after the cyclone made landfall.  Statistical analysis indicates that the longshore wind stress, x , is the dominant forcing for the freshwater plume from the Burdekin River. Under weak downwelling wind forcing (-0.1 Pa < < 0 Pa), the plume thickness is sensitive to river discharge and tides. With stronger downwelling wind forcing ( <= -0.1 Pa), vertical mixing is generated, the plume is restricted to the coast, and high river discharge affects the thickness of the plume, but not its width. After Cyclone Debbie made landfall, upwelling winds developed, and the freshwater plume reversed direction from northward to southward most likely due to the influence of the cyclonic northerly wind as well as the ambient current and topography. Based on an Empirical Orthogonal Function (EOF) analysis, the cyclone-induced river discharge is found to be the major cause for surface salinity variation with a lag time equal to 17 hours. In summary, this thesis highlights and compares the influence of both wind and river discharge on the river plume during the passage of cyclone and its analysis method for the river plume can be used in other case studies.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only). 29/08/2018 ……………………………………………… ……………………………………..…… ……….……………………... …………… ………… …….… Signature Witness Signature Date

The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.

FOR OFFICE USE ONLY Date of completion of requirements for Award:

i INCLUSION OF PUBLICATIONS STATEMENT

UNSW is supportive of candidates publishing their research results during their candidature as detailed in the UNSW Thesis Examination Procedure.

Publications can be used in their thesis in lieu of a Chapter if: The student contributed greater than 50% of the content in the publication and is the “primary author”, ie. the student was responsible primarily for the planning, execution and preparation of the work for publication The student has approval to include the publication in their thesis in lieu of a Chapter from their supervisor and Postgraduate Coordinator. The publication is not subject to any obligations or contractual agreements with a third party that would constrain its inclusion in the thesis

Please indicate whether this thesis contains published material or not.

This thesis contains no publications, either published or submitted for publication ☐ (if this box is checked, you may delete all the material on page 2) Some of the work described in this thesis has been published and it has been documented in the relevant Chapters with acknowledgement (if this box is ☐ checked, you may delete all the material on page 2)

This thesis has publications (either published or submitted for publication) ☒ incorporated into it in lieu of a chapter and the details are presented below

CANDIDATE’S DECLARATION I declare that: I have complied with the Thesis Examination Procedure where I have used a publication in lieu of a Chapter, the listed publication(s) below meet(s) the requirements to be included in the thesis. Name Signature Date (dd/mm/yy) Yuanchi Xiao 29/08/2018

Postgraduate Coordinator’s Declaration (to be filled in where publications are used in lieu of Chapters) I declare that: the information below is accurate where listed publication(s) have been used in lieu of Chapter(s), their use complies with the Thesis Examination Procedure the minimum requirements for the format of the thesis have been met.

ii Originality Statement

I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have 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.

Signed ……………………………………………...... ………………

Date ……………………………………………..…………...... 29/08/2018

Yuanchi Xiao

August 2018

iii Copyright Statement

I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act

1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in

Dissertation Abstract International (this is applicable to doctoral theses only).

I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.

Signed ……………………………………………...... ………………

Date ……………………………………………..…………...... 29/08/2018

iv Authenticity Statement

I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.

Signed ……………………………………………...... ………………

Date ……………………………………………..…………...... 29/08/2018

Abstract

This thesis investigates the plume morphology and dynamics prior to and after the landfall of Cyclone Debbie (2017). The heavy rainfall and flooding produced a large buoyant coastal current, which moved southward after the cyclone made landfall then advected northward with the prevailing southerly wind. The plume is simulated using the eReef GBR1 1-km model and a passive tracer is used to investigate the plume behaviour.

Based on the concentration of river tracers from the Burdekin River, the plume propagated over 100 km to the north during the 23 days after the cyclone made landfall.

Statistical analysis indicates that the longshore wind stress, x , is the dominant forcing for the freshwater plume from the Burdekin River. Under weak downwelling wind forcing (-0.1 Pa < < 0 Pa), the plume thickness is sensitive to river discharge and tides. With stronger downwelling wind forcing ( <= -0.1 Pa), vertical mixing is generated, the plume is restricted to the coast, and high river discharge affects the thickness of the plume, but not its width. After Cyclone Debbie made landfall, upwelling winds developed, and the freshwater plume reversed direction from northward to southward most likely due to the influence of the cyclonic northerly wind as well as the ambient current and topography. Based on an Empirical Orthogonal Function (EOF) analysis, the cyclone-induced river discharge is found to be the major cause for surface salinity variation with a lag time equal to 17 hours.

In summary, this thesis highlights and compares the influence of both wind and river discharge on the river plume during the passage of Cyclone Debbie and the methodology used to analyse the river plume can be used in other case studies.

Acknowledgement

This thesis includes work carried out as an MPhil student majoring in Oceanography in the Sino-Australian Research Centre for Coastal Management (SARCCM), School of

Physical, Environmental and Mathematical Sciences (PEMS) of the University of New

South Wales (UNSW) at Canberra, Australia.

Firstly, I am very grateful to my supervisor Prof. Xiao Hua Wang and Joint supervisor A/Prof Liz Ritchie-Tyo. Even though they are very busy, they manage to give me suggestion and support for my research. They give me so much help in both academic and language issues when I am writing the paper and this thesis.

Guandong Gao, Haifeng Zhang, Zhibin Li, Fanglou Liao, Yuan Yuan, Zhixin

Chen, Sengyang Xie, Isabel Jalon Rojas, Anna Maggiorano, Md Wasif Elahi, Difei Deng and Clair Stark and Gang Yang from SARCCM are acknowledged for giving me support to overcome problems I faced during my study. We also have very vivid and happy moment together in Canberra.

A gratitude goes out to the School of PEMS in UNSW Canberra for supporting my travels to Tasmania and Queensland to attend academic conferences. I am thankful to the head of our school Professor Warrick Lawson who approved these trips. Besides, acknowledgment is to Elvira Berra from the Research Student Unit of UNSW Canberra,

Annabelle Boag the former academic support officer of PEMS and the current officer

Julia Wee who gave me help in the last one and a half year.

I want to say thanks to Mark Baird, Farhan Rizwi and Nugzar Margvelashvili in

CSIRO Oceans and Atmosphere. During my visit to Hobart, they gave me a lot help for the clear concept of the eReefs. I am also grateful to Fanghua Xu from Tsinghua

University and Lulu Qiao from Ocean University of China who help me a lot in the concept of freshwater plume and the structure of my paper and thesis.

Finally, I want to say thanks to my parents. Although they are not in Canberra, I am glad to share my life whether it makes me happy nor not. When I am upset about my life and study they always try to encourage me and help to analyse the problem.

iii

Table of Contents

Abstract ...... i

Acknowledgement ...... ii

Table of Contents ...... iv

List of Tables ...... vii

List of Figures ...... viii

Chapter 1. Introduction ...... 1

1.1. Importance of riverine freshwater plumes ...... 1

1.2. Current problem ...... 3

1.3. Study area ...... 4

1.4. Aims and objectives of the thesis ...... 9

1.5. Research innovations ...... 10

1.6. Thesis structure ...... 11

Chapter 2. Literature Review...... 13

2.1. Introduction ...... 13

2.2. Freshwater plume morphology research in estuaries ...... 14

2.3. Wind influence on the freshwater plume...... 15

2.4. Cyclone impact on the freshwater plume ...... 17

2.5. River discharge and tidal influence on the freshwater plume ...... 18

2.6. The Burdekin River plume ...... 18

Chapter 3. Data and Methodologies ...... 20 iv

3.1. Definition of a Freshwater Plume ...... 20

3.2. GBR1 1km Model and river tracer ...... 21

3.3. The Moderate Resolution Imaging Spectroradiometer (MODIS) true-colour

images ...... 22

3.4. Empirical Orthogonal Function (EOF) Analysis ...... 23

3.5. Period of Study ...... 23

3.6. Cyclone Debbie and the river discharge ...... 24

Chapter 4. The influence of Cyclone Debbie’s wind field on the Burdekin River freshwater plume ...... 26

4.1. Introduction ...... 26

4.2. Comparison between the modelled surface salinity and satellite true-colour

images ...... 27

4.3. Evolution of the plume and Cyclone Debbie ...... 29

4.4. Plume morphology and wind ...... 35

4.5. Wind mixing of plume ...... 41

4.6. Conclusion ...... 44

Chapter 5. River discharge ...... 46

5.1. Introduction ...... 46

5.2. Fate of the freshwater...... 47

5.3. Comparison to a 2017 flooding event ...... 48

5.4. Conclusion ...... 51

Chapter 6. EOF analysis and Wind Strength Index- competitions between different forcing ...... 53 v

6.1. Introduction ...... 53

6.2. EOF analysis during Cyclone Debbie ...... 53

6.3. Comparison between wind and buoyancy forcing ...... 57

6.4. Conclusion ...... 59

Chapter 7. Conclusions and future work ...... 61

7.1. Conclusions ...... 61

7.2. Transient wind field and plume evolution during Cyclone Debbie ...... 61

7.3. Cyclone-induced flood and freshwater transport ...... 62

7.4. Dominant forcing during the passage of Cyclone Debbie ...... 63

7.5. Suggestion for future work ...... 64

References ...... 65

List of Tables

Table 6.1: Variance explained by the 10 EOF modes...... 54

vii

List of Figures

Figure 1.1: (a) Map of the Burdekin catchment. Image source: Blanchette and Pearson,

2012; and (b) Location and bathymetry of the zoomed research area indicated by the red box in (a). The red lines in (b) show the three cross-sections labeled c1, c2, and c3, respectively, used in later calculations and indicates the location of the 13-m-depth

Station A used for later analysis...... 5

Figure 1.2: (a) The north branch of the main channel between Rita Island and Peters

Island of Burdekin River in the Burdekin Delta on June 19 2018 when daily-averaged river discharge recorded at Clare Station was 29 m3/s; and (b) an aerial view of the

Burdekin Delta...... 6

Figure 3.1: Structure of a freshwater plume. Image Source: Lentz and Largier, 2006. 20

Figure 3.2: Track of Cyclone Debbie. Image Source: Bureau of Meteorology ...... 25

Figure 4.1: Comparison between true-colour Images from MODIS, salinity from the

GBR1 in: (a), (c); MODIS: 10:40 AEST March 31. GBR1 model: 11:00 AEST March

31; and (b), (d); MODIS: 14:05 AEST April 1. GBR1 model: 14:00 AEST April 1. ... 27

Figure 4.2: (a) Comparison between measured salinity at Yongala National Reference

Station and the GBR1 model at different levels in the ocean during 2017. (b) and (c) compare the measured near-surface temperature with the modelled temperature from the

GBR4 and GBR1 models at Townsville and Abbot Point from 2014 to 2015. Source: eReefs Modelling report (Herzfeld et al., 2016)...... 27

Figure 4.3: Surface plume and currents prior to, and after the passage of Cyclone Debbie, which made landfall on March 28 2017: (a) Prior to Debbie’s landfall; (b)-(f) After

viii

Debbie’s landfall. The compass shows the wind direction. Arrows show the surface current...... 30

Figure 4.4: (a) Surface longshore wind stress; and (b) Modelled surface longshore current at Station A; and (c) River discharge recorded at Clare station, which is 40 km upstream from the mouth of the Burdekin River. Vertical dash lines show the time nodes used to discuss the plume structure under different wind forcing in section 3.2. The dash- dot lines in (b) shows the low-frequency surface longshore residual current...... 32

Figure 4.5: (a) Longshore wind stress  x at Station A. Hovmöller diagram of surface salinity at: (b) c1; and (c) c2. White lines indicate when the plume width, Wp, is larger than 10 km on c1...... 34

Figure 4.6: Sea surface salinity distribution under different wind forcing conditions: (a)

Weak downwelling ( = -0.003 Pa); (b) After strong downwelling ( = -0.087 Pa); (c)

Weak upwelling ( = 0.054 Pa); and (d) Strong upwelling ( =0.214 Pa)...... 35

Figure 4.7: Salinity distribution on c1 under different wind forcing conditions: (a) Weak downwelling ( = -0.003 Pa); (b) After strong downwelling ( = -0.087 Pa); (c) Weak upwelling ( = 0.054 Pa); and (d) Strong upwelling ( =0.214 Pa)...... 37

Figure 4.8: Scatter plot between: (a) plume width, Wp, and downwelling wind stress ,;

and (b) Plume thickness, hP, and downwelling wind stress , along c1. Wind stress is recorded at Station A. Coloured points indicate the level of the river discharge...... 39

Figure 4.9: Scatter plot along c1 between Wn (=Ws/Wb) and wind stress calculated at

Station A. Coloured points indicate the amount of river discharge occurring for each point.

...... 41

Figure 4.10: Scatter plot between Bulk Richardson number and longshore downwelling wind stress at Station A...... 42 ix

Figure 4.11: Conditions at Station A during the passage of Cyclone Debbie: (a) wind stress; (b) river discharge; (c) longshore current (m/s); (d) salinity; and (e) sea temperature (°C). The wind stress (a) and hydrodynamic conditions (c-e) are from the

GBR1 model at Station A. The river discharge was recorded at Clare station 40 km upstream from the mouth of the Burdekin River. In (c) – (e) the surface (0-m) data are represented with a solid line, the 5-m (middle) data are represented with a dashed line, and the 13-m (bottom) data are represented with a dotted line...... 43

Figure 5.1: Tracer concentration distribution at the sea surface on April 20, 23 days after the cyclone landfall. The tracers were released into the model at the mouth of the

Burdekin River...... 47

Figure 5.2: Conditions during the flooding event in May 2017: (a) wind stress; (b) river discharge; (c) surface longshore current; and (d) surface salinity. The wind stress (a) and hydrodynamic conditions (c-d) are from the GBR1 model at Station A. The river discharge was recorded at Clare station 40 km upstream from the mouth of the Burdekin

River...... 49

Figure 5.3: Accumulated percent of fresh water transport at the three cross sections and accumulated river discharge during Cyclone Debbie (solid line) and the 2017 non- cyclone flooding event (dash line). The percentage freshwater transport is calculated as the amount of freshwater that propagated across the cross section compared to the accumulated river discharge at the end of each period (day 30). A negative value indicates that the freshwater came from outside the ‘box’ area in Figure 1.1...... 51

Figure 6.1: (a) Spatial pattern of EOF mode 1; (b) the expansion coefficient of EOF mode 1 and river discharge; and (c) Correlation between river discharge and Expansion coefficients from EOF mode 1...... 55

Figure 6.2: (a) Spatial pattern of EOF mode 2, is the station to show the ambient current; and (b) the expansion coefficient of EOF mode 2 and longshore ambient current recorded at ...... 56

Figure 6.3: (a) Magnitude of the Wind Strength Index ( Wsi ) and longshore wind stress at station A during the passage of Cyclone Debbie. and (b) Wind-driven flow uwind and buoyancy-driven flow udis...... 58

Chapter 1. Introduction

1.1. Importance of riverine freshwater plumes

Freshwater plumes are the river outflows from estuaries or river channels into the coastal ocean. Compared to the ambient coastal water, they have different characteristics such as low salinity, high turbidity and biological productivity. Plumes play an essential role in coastal regions, and drive sediment transport, contamination exchange, and coastal ecology (Zheng et al., 2004; Stukel et al., 2016).

The characteristics of river plumes, such as sediment and contaminant concentration, are highly related to land-based human activities, including farming, dredging, discharge of contamination, and construction of dams. For example, sediment exported into the sea from the Yellow River in China has decreased because of human activities including an increasing use of water for agriculture and reservoir projects

(Wang et al., 2007). In contrast, in the Great Barrier Reef (GBR), coral records show that the annual sediment exported from the land to the reef has increased 5-10-fold since the

1850s (McCulloch et al., 2003). The fate of riverine sediment transport to the sea from land clearing and farming is controlled by freshwater plume dynamics. Other biological influences such as the bloom of toxic algae may also be related to river plume dynamics

(Franks and Anderson, 1992). In addition, climate change will increase the likelihood of extreme weather with associated rainfall and river runoff and may directly influence the magnitude, morphology and dynamics of river plumes. Thus, investigating the behaviour of river plumes especially under extreme weather events, is important to understand how human activities and climate change influence the oceans.

River systems, including the Amazon River, Columbia River and Chesapeake

Bay, develop a discernible plume structure associated with the continuous high discharge.

Research into these types of river systems are typically focused on the long-term variation of the plumes and their tidal influence (Guo and Valle-Levinson, 2007; McCabe et al., 2009; Molleri et al., 2010). However, river basins such as arid or semi-arid, and monsoonal systems with very low average rainfall (~0) in the dry season and high rainfall in the wet season can only generate significant plumes during strong rainfall events from cyclones or the monsoon. Some typical examples include the Santa Clara River in

California and the Burdekin River in Queensland (Wolanski and Jones, 1981; Warrick et al., 2004). Strong rainfall events can cause a dramatic increase of the river flow from 0 to over 103 m3/s with a corresponding large freshwater river plume that develops along the coast. Furthermore, because these arid or semi-arid catchments lack the ability to conserve water and soil, considerable sediment along the river will be carried during the strong rainfall to the river and then exported to the ocean by the river plume.

Weather systems that generate strong rainfall such as tropical cyclones can affect river plume dynamics because of the transient high river discharge that results from heavy rainfall in catchment basins. Such heavy rainfall events can also contribute considerable sediments to the river flow. A tropical cyclone is particularly interesting because along with the heavy rainfall, the cyclonic structure of the wind field also makes the wind forcing on the ocean surface transient. Based on previous studies, the wind field and associated forcing plays a very important role in the dynamics and morphology of plumes. The wind field can affect the longshore transport, thicken or thin the plume, narrow or widen the plume and increase the vertical mixing. Because of the combination of the high river discharge and the changing wind magnitude and direction in a tropical

cyclone, it is important to investigate the influence of the tropical cyclone on the resulting freshwater plume dynamics.

Freshwater plumes are influenced by external forcings including wind, river discharge, topography and tides. Among these, both observation and modelling studies suggest that wind stress changes the morphology of the plume because the surface plume is the place where the large-scale wind-driven freshwater transport and mixing occurs. In addition, its characteristics, such as salinity and density, is highly different from the saltier ambient current and may have different response to wind (García Berdeal, 2002;

Fong and Geyer, 2001).

1.2. Current problem

The Burdekin River is the largest source of sediment for the Great Barrier Reef with an increasing amount of sediment output because of land-based human activities (Brodie et al., 2010, Andutta et al., 2011, Bainbridge et al., 2012, Alibert et al., 2003). Although considerable research has been carried out using observations and numerical simulations of the plume morphology and dynamics (Wolanski and Jones, 1981; King et al., 2001), there is still a lack of research that quantifies the influence of different forcings on the

Burdekin River plume especially under cyclone conditions. Since the Burdekin River is the largest sediment source to the Great Barrier Reef, a study of the development and evolution of the Burdekin River estuary freshwater plume during cyclones will be significant contribution to the existing literature and will enhance the GBR management system.

The aim of this study is to use eReefs hydrodynamic data to investigate the influence of Cyclone Debbie on the Burdekin River plume of which the dynamics and

morphology will be discussed. Different forcing related to the cyclone which may affect the plume dynamics will be compared.

1.3. Study area

1.3.1. Topography of the Burdekin River Estuary

The Burdekin River, located in north and far north Queensland (Figure 1.1), is the largest source of sediment to the GBR lagoon (Kroon, 2012). It begins in the Seaview Range and runs for 886 km until it reaches the coast in Upstart Bay adjacent to the Coral Sea.

The catchment area of the Burdekin River is 129,700 km2. Although the headwaters and the river mouth are located in a moderate wet climate with less than 1600 mm annual rainfall, most of its basin is semiarid with 500-700 mm annual rainfall and the total annual rainfall is 633 mm/year. Generally, the rainfall in the eastern coastal area is higher than the western inland region. The Burdekin catchment consists of six regions: the Upper

Burdekin, Cape River, Belyando, Suttor, Bowen and Bogie and the lower East Burdekin

(Lewis et al., 2007).

Figure 1.1: (a) Map of the Burdekin catchment. Image source: Blanchette and Pearson,

2012; and (b) Location and bathymetry of the zoomed research area indicated as the red box in (a). The red lines in (b) show the three cross-sections labeled c1, c2, and c3, respectively, used in later calculations and indicates the location of the 13-m-depth

Station A used for later analysis.

The Burdekin Delta is one of the largest deltas in Australia. The south part of the delta is active with supplied fluvial sediment, while the north part of the delta is inactive and less sediment is deposited in this area. There are three major entrances to the Upstart

Bay: The north branch from the north of Rita Island (Figure 1.2a), the middle branch between Rita Island and Peters Island, and Groper Creek (Figure 1.2b) along the south bank of Peters Island. The south coast of the river mouth is protected from waves by

Cape Upstart (Figure 1.1b) and the north coast is protected by a sand spit called Cape

Bowling Green (Figure 1.1b). The north branch of the river mouth is the major and newly formed branch, while the middle and south branch has been abandoned and only little

water go through it. The width of the north branch entry is about 500 m. The deposition of the sand bar within the river mouth is determined by short events such as floods.

Figure 1.2: (a) The north branch of the main channel between Rita Island and Peters

Island of the Burdekin River in the Burdekin Delta on June 19 2018 when daily-averaged river discharge recorded at Clare Staion was 29 m3/s; and (b) an aerial view of the

Burdekin Delta.

The average river discharge is approximately 380 m3/s and is highly variable from year to year and season to season depending on both the number of cyclonic events and the structure of the monsoon. The discharge varies from 0 in the dry season to 40,000 m3/s during the passage of cyclones. The recurrence intervals for a large flood event over

20,000 m3/s is 5-10 years, which is highly related to the passage of landfalling cyclones bringing heavy rainfall to the catchment, which causes bulk river discharge. Human activities including the construction of the Burdekin Dam has changed the downstream river discharge significantly. Previous estimates suggest that the river transports a mass of sediment ranging from 2.7 × 106 to 9.0 × 106 tonnes every year from the Burdekin catchment region to the ocean (Neil et al., 2002, Belperio, 1983) with an increasing trend since European settlement.

The Burdekin Delta is a shallow estuary with an average depth less than 5 meters.

Within Upstart Bay the water depth is shallower than 20 meters and the bathymetry increases seaward. The 20-m isobath is 11 km away from the river mouth. The deepest place within the study domain as shown in Figure 1.1 is over 60 meters near the reefs.

1.3.2. Oceanography of the Burdekin River Estuary

During the wet season with an active monsoon and the passage of cyclones, the salinity along the coast will decrease significantly and intrusion of freshwater may reach the GBR coral region. At the end of a large fresh water plume event, the salinity distribution can be inhomogeneous and patchy (Wolanski and Jones, 1981, Wolanski and Van, 1983).

The salinity increases seaward and away from the coast the salinity is generally over 35

PSU and can become hypersaline (>37 PSU) during the dry season.

The Burdekin Region is mainly controlled by a semi-diurnal tide. The mean tidal range in Townsville, the nearest standard measurement point, is 3.1 m. During the spring tide, the mean tidal range is 3.5 m and it is 0.5 m during the neap tide. Records from secondary ports at Cape Upstart and Cape Bowling Green show that the mean tidal range is about 2.5 m and 2.6 m, respectively. Generally, the tidal current is less than 0.2 m/s with an along-coast direction (Andrews and Bode, 1988). During spring tides, the tidal current increases to over 0.7 m/s in shallow embayment (Belperio, 1978). The current weakens during slack and ebb tides. In the middle shelf, the tidal current at the surface is normally less than 0.3 m/s (Church et al., 1985; Wolanski and Pickard, 1985).

The residual current in the region is mainly controlled by a prevailing longshore ambient current driven by the wind. Generally, based on long-term observations, there is a low-frequency background current less than 0.2 m/s with periods over 2 days. The peak can reach 1 m/s, depending on the wind strength (Wolanski and Pickard, 1985). From

March to October, because of prevailing southeasterly or southerly winds, the base

current is northward along the coast. Strong rainfall and high river discharge can also affect the coastal sea water by altering the sea level and generating a baroclinic current.

According to published wave data (Queensland Department of Environment,

1997), the mean wave period is 4.9 s with a mean wave height of 0.66 m at Townsville and 3.9 s with 0.53 m at Abbot Point. During the passage of cyclones, the mean wave height reaches 2.34 m and 1.38 m at Townsville and Abbot Point, respectively and the wave period increases. However, because of headland protection, the wave height in

Upstart Bay remains relatively lower.

There are a number of rivers and creeks that may produce freshwater plumes in the research area. Among them the river plume from the Burdekin River is an important influence affecting the local salinity, temperature, and density of the ambient saltwater.

During high rainfall periods, the higher river discharge will produce a large quantity of buoyant freshwater both along the coast and intruding offshore. Like other river systems, the Burdekin plume is transient and may be sensitive to different factors, including wind, ambient current, river discharge, tide and topography. Commonly, the plume water propagates northward and is restricted within the continental shelf by a prevailing southerly wind but it may turn southward with cyclones (King et al., 2001). Since the plume itself has a different density to the surrounding sea water, the accumulation of freshwater may also change the sea surface height, and when it reaches a steady condition, it can produce a geostrophic current along the coast (Wolanski and Senden, 1983).

1.3.3. The climate of the Burdekin River Estuary

The Burdekin Drainage has a dry and tropical climate, which consists of distinct wet and dry seasons. Most of the rainfall occurs during the wet season, from November to April.

A large percentage of the rainfall is contributed by the humid coastal region whereas the

inner region is semi-arid or arid and the rainfall is low. According to long-term data from the Bureau of Meteorology (BoM) from 1911 to 2017, the total annual rainfall for the

Burdekin River Basin is 643 mm (658 mm in 2017). The highest monthly rainfall occurs in January and February with 135 mm, and the lowest rainfall occurs in September (13 mm). In 2017, the highest rainfall occurred in March with 148 mm during the passage of

Cyclone Debbie. Based on the record for over 60 years (1951-2018) at Ayr Department of Primary Industries (DPI) Research Station, the mean annual maximum air temperature is 29.1 °C. The highest maximum air temperature occurs in January with 31.8 °C while the lowest value occurs in July (25.2 °C).

According to the Ayr DPI Research Station the annual average wind speed in the morning (9 am) is 3.5 m/s, while it is stronger in the afternoon (3 pm) at 4.8 m/s. The highest wind speed occurs during October to November. The prevailing average morning wind in March is southerly and southeasterly at 3.4 m/s. In the afternoon, the prevailing wind is easterly and northeasterly at 4.8 m/s.

In Queensland, there are an average 4.7 cyclones per year (Alharbi, 2014).

Cyclones pass the study region once every 1-2 years (Fielding et al., 2005). In the last 10 years (2007 - 2017), there have been four cyclones (Debbie, 2017; Yasi, 2011; Anthony,

2011; Ului, 2010) that affected the Burdekin Region directly. Cyclones Debbie and Yasi were the more severe of the four and made landfall with Category 4 and Category 5 winds, respectively, bringing heavy rainfall into the catchment area and nearby sea.

1.4. Aims and objectives of the thesis

The aim of this thesis is to use the GBR1 1km model from eReefs data developed by the Commonwealth Scientific and Industrial Research Organisation (CSIRO) Division

of Oceans and Atmospheres to investigate the river plume morphology, dynamics, and how the plume is affected by the landfall of Cyclone Debbie (2017).

The following objectives will be achieved:

1) Discuss Burdekin River plume evolution during the passage of Cyclone Debbie and wind influence on the plume dynamics and morphology;

2) Discuss the influence of high river discharge on the Burdekin River plume and compare the freshwater transport with a small flood in 2017; and

3) Compare the influence of wind forcing to river discharge on the river plume.

1.5. Research innovations

The innovations of the research are listed below:

(1) The research investigates the plume morphology for a freshwater plume generated by

a severe, landfalling tropical cyclone, and its relationship with wind stress and other

elements. This type of investigation has never been done in this region, especially

under a severe landfalling cyclone condition;

(2) Empirical Orthogonal Function (EOF) analysis and Wind Strength Index is applied

to investigate the spatial and temporal variation of a freshwater plume influenced by

a severe tropical cyclone; and

(3) The eReefs data provides the first overall real-time high-resolution modelling result

of the entire GBR. These data have not been closely analysed during a cyclone event.

Therefore, this research provides an example of how these data can be used to analyse

these important events.

1.6. Thesis structure

The thesis is organized as follows:

The research topic is introduced in Chapter One. First, some background on freshwater plumes and current research gaps are discussed. Next, the study area is described and finally, the objectives and the innovations of the thesis are explained.

In Chapter Two, previous research on plume morphology and evolution is described. Next, the influences of wind on plume structure and mixing is highlighted.

Thirdly, research on cyclone impacts on freshwater plumes is reviewed. Studies on the impact of the river discharge and tides are then discussed. Finally, previous work on the

Burdekin River plume is summarized.

Definitions, data, and methodologies used in the thesis are described in Chapter

Three. Specifically, a definition of a freshwater plume, a description of the eReefs GBR1

1km model, and the Moderate Resolution Imaging Spectroradiometer (MODIS) satellite true-colour images, is provided. The theory of EOF analysis is then introduced and finally, the landfall of Cyclone Debbie is described.

In Chapter Four, the wind influence on the Burdekin river plume is discussed.

Firstly, an introduction of wind influence in other regions is discussed. Then results from the eReefs GBR1 1km hydrodynamic model are compared with MODIS true-colour satellite images. The river plume evolution during the passage of Cyclone Debbie is analysed under different wind forcings and the changes in plume structure are discussed.

A comparison between wind forcing and buoyant forcing is made, and finally the wind mixing and its relationship to the downwelling wind is highlighted.

In Chapter Five, the distribution of the modelled freshwater plume from the

Burdekin River and how it evolves with time in the GBR region is analysed using tracers.

In addition, the freshwater transport is calculated using three cross-sections that represent

the propagation to the north, south and offshore directions during Cyclone Debbie and compared with another smaller non-cyclone flooding event.

In Chapter Six, EOF analysis is used to discuss the dominant forcings for the river plume evolution. In addition, a Wind Strength Index is applied to compare the influence of wind forcing and buoyancy on the freshwater plume.

A summary and conclusions of the thesis are provided in Chapter Seven. Besides, recommendations for future research are provided.

Chapter 2. Literature Review

2.1. Introduction

Investigating the evolution and the morphology of riverine freshwater plumes is important because rivers contribute waters with different properties, including sediment, solutes, and contaminants, into the coastal and offshore regions. In some areas human activities have caused river discharge, and the accompanying materials, to change significantly in the last 150 years (McCulloch et al., 2003). The river discharge and its properties also vary from day to day and from year to year, depending on variations in the climate. The dominant forcings for freshwater plumes in various regions are also different, so that it is very important to investigate the influence of forcing on river plume structure and evolution.

Generally, the major forcings controlling the freshwater plume are complex, and include wind, river discharge, ambient current, tides, and topography. Local winds advect freshwater plumes by Ekman transport (Csanady, 1978). With downwelling (downstream) wind, the freshwater plume is horizontally narrower and thickens vertically. With upwelling (upstream) wind, the width of the freshwater plume increases and the vertical thickness decreases (Lentz and Largier, 2006). The ambient current promotes the downstream propagation of the freshwater plume and the plume width can reach a steady state in an ambient current (Fong and Geyer, 2002). Tidal forcing can change the propagation direction of a freshwater plume by increasing mixing (Wu et al., 2011).

Changes in topography such as the bottom slope also determines if the freshwater plume will float on the surface or be constrained to the bottom (Lentz and Helfrich, 2002).

The objective of this chapter is to provide a background literature review of previous studies on the fresh water plume dynamics. The review consists of five parts: 1) previous studies of plume structure; 2) an introduction of the wind influence on the freshwater plume; 3) How cyclones influence the freshwater plume; 4) River discharge and tidal influence on freshwater plumes; and 5) research on the Burdekin River plume.

2.2. Freshwater plume morphology research in estuaries

The morphology of freshwater plumes from estuaries has long been observed in different studies. The freshwater plume has been identified as either a surface-advected (surface- trapped) or a bottom-advected (bottom-trapped) plume, depending on the percentage of fresh water that can reach the bottom and shape of sea bottom (Yankovsky and Chapman,

1997). Abottom-trapped plume extends to the sea bottom and is controlled by the bottom boundary layer dynamics. Surface-trapped plumes, which expand further offshore, have an obvious ‘bulge’ structure in the near field which is close to the river mouth with anticlockwise or clockwise velocity field depending on the rotation of the Earth. The width of the ‘bulge’ is wider than the plume when it extends far from the river mouth. In an idealised model without wind and ambient current, the freshwater plume may not reach a steady condition and the bulge grows continually (Fong and Geyer, 2001). The shape of the bulge is also dependent on the Rossby number determined by the river mouth width and river flow speed.

The northward extension of freshwater plumes is noted in different studies, which conclude that longshore wind and buoyancy are the major driving forces along the Shore

Bay (Fong and Geyer, 2002; Bainbridge et al. 2012; Delandmeter et al., 2015; Wolanski and van Senden, 1983). However, results from a three-dimensional model experiment in the Northern Hemisphere (Fong and Geyer, 2002) suggests that, the existence of ambient

currents such as alongshore current is the key point for coastal freshwater transport.

Furthermore, there is a negative correlation between coastal current transport and the

Rossby number, which is related to both river flow velocities and the width of the river mouth. The simulations ignored some external forcings including surface waves. Similar numerical experiments in different regions suggest that wind stress, tide, and ambient current are the key aspects for the coastal freshwater transport (García Berdeal, 2002;

Guo and Valle- Levinson, 2007; Schiller et al., 2011).

2.3. Wind influence on the freshwater plume

Local winds play important roles in the morphology and dynamics of the river plume. A theoretical model of how wind influences the plume is that offshore and onshore advection of the plume front is related to the Ekman transport caused by the alongshore wind (Csanady 1978). Upwelling (upstream) winds, which oppose the original direction of the plume depending on the Coriolis force can produce offshore advection, which thins the plumes. In contrast, downwelling (downstream) winds, which follow the original plume direction, generate an onshore advection, which thickens the plume. A number of early studies including numerical (Chao, 1987, 1988) and observational (Fong et al., 1997;

Lentz et al., 1999) studies have confirmed this initial concept. Because other forcings exist such as the buoyancy that drives the flow, a wind influence index was developed by Whitney and Garvine (2005) to compare the effect of wind and buoyancy on the plume.

Since there is no general model to test the dependence of the morphological characteristics and mixing of plumes on wind stress, Lentz and Largier (2006) developed a framework to discuss their relationship using the example of the Chesapeake Bay.

The response of plumes to upwelling wind is discussed by Fong and Geyer (2001) using an idealised model, which indicates that the freshwater plume is separated from the

coast during a period of upwelling wind. The simulation result also suggests that the thinned plume by the upwelling wind is sensitive to mixing caused by shear instabilities.

With several days of moderate upwelling wind, the plume tends to reach a quasi-steady and uniform thickness. The mixing processed happened at the first twelve hours after a constant upwelling wind field is imposed. In their study, a conceptual model is also built to estimate a depth where the plume is controlled by Ekman transport but without the consideration of important entrainment process. Lentz (2004) developed a two- dimensional model with entrainment and estimated various characteristics of plumes including thickness and width.

Compared to the upwelling wind, theoretical research on the impact of downwelling wind on freshwater plumes is lacking. Rennie et al. (1999) suggest that the stratification of water will be reduced by downwelling wind since it steepens the isopycnals. The longshore transport of freshwater is also intensified by the downwelling wind (Lentz 2004; Whitney and Garvine 2005). A conceptual model to estimate the properties of a freshwater plume such as width and thickness suggests that the morphology of a bottom-trapped plume will change less than a surface-trapped plume under downwelling wind forcing (Moffat and Lentz 2012). However, this conceptual model might still be too simple as the mixing processes are simplified.

However, wind clearly plays a significant role in the freshwater plume mixing.

Both upwelling and downwelling winds produce mixing (Fong and Geyer, 2001). It is the mixing that tends to be stronger under an upwelling wind because the plume thins and widens, and more water is exposed to the influence of the wind forcing. According to Hetland (2005), wind-induced vertical mixing is stronger in the near field, which is close to the estuary mouth, but weaker in the far field where the salinity is high and turbulence mixing by wind is suppressed.

2.4. Cyclone impact on the freshwater plume

Landfalling tropical cyclones generate a relatively short-lived, transient wind field, along with sometimes heavy rainfall in the catchment area. There are many studies focused on the influence of cyclones on freshwater plumes. Dong et al., (2004) and Schiller et al.,

(2011) suggest that the ocean current outside of the estuary is sensitive to the transient wind field associated with a tropical cyclone and the propagation direction of the freshwater plume is dominated by wind forcing during the passage of cyclones. Devlin and Brodie (2005) mapped the freshwater plumes in the GBR during the passage of various cyclones and suggested that the dispersion of the plume is governed by wind.

Cyclone-induced ocean eddies have been observed that injected the associated freshwater plume with high chlorophyll into the open shelf (Yuan et al., 2004).

Cyclones also cause cooling on the sea surface due to both rainfall and induced deep water upwelling in the ocean, which can affect the freshwater plume (Hu and

Muller-Karger,2007). However, compared to the surrounding ocean water, cooling of the plume water by cyclones is reduced (Reul et al.,2014).

Significant increase of the river discharge because of rainfall during the passage of landfalling cyclones was found (Wolanski and Jones,1981; Wolanski et al., 2003;

Geyer et al., 2000; Zhao et al., 2009), which finally led to the bloom of Phytoplankton

(Zhao et al., 2009). In the western tropical Pacific, freshwater from rivers generated during cyclone events can be found extending to the deep ocean (Kao et al., 2010).

Although the sediment in these freshwater plumes is constrained to the coast firstly by deposition, it can be resuspended by waves generated by cyclones (Wolanski et al., 2003).

2.5. River discharge and tidal influence on the freshwater plume

The magnitude of the river discharge directly influences the magnitude of freshwater plumes. Freshwater from land contributes to the seaward acceleration of the plume

(McClimans, 1986). Kourafalou et al. (1996) classified the plume as supercritical

(subcritical) when the bulk Richardson number is higher than 1 (less than 1), and the width of the plume bulge is wider (narrower) than the coastal current.

Water exchange between freshwater and saltwater occurs during the ebb tide

(Luketina and Imberger, 1989). Estuaries where the mean tidal range is higher than 3 m are dominated by tides (Mehta, 1989). During ebb tides the freshwater plume is enhanced and expands offshore while it dissipates during the flood tide and saltwater intrudes into the river (Kingsford and Suthers, 1994). To the nearshore of the estuary, the balance of the momentum equation is sensitive to the tidal phase (Hench and Luettich, 2003).

Observations (Orton and Jay, 2005) and simulations (McCabe et al., 2009) suggest that mixing and nutrient exchange occurred on the plume front during ebb tides. For estuaries, convergence of tidal forcing is the dominant forcing in the momentum equation

(MacCready et al., 2009). Some modelling results suggest that the existence of tides changes the direction of the freshwater plume and spreads it offshore with a weaker bulge

(Wu et al., 2011).

2.6. The Burdekin River plume

Since the Burdekin River is the largest sediment source to the Great Barrier Reef (Kroon,

2012), there have been many studies of its freshwater plume. Belperio (1978) reports that sediment from the Burdekin River was generally trapped in bays and did not spread far offshore. Wolanski and Jones (1981) observed that freshwater patches existed offshore and they also suggested that tidal forcing in this region was not the dominant forcing 18

compare to buoyancy during floods. Low-frequency currents initiated by the Burdekin

River plume was dominated by fluctuating wind forcing. Simulations suggested that significant plumes occur during floods and were able to travel hundreds of kilometres northward within two or three weeks (King et al., 2001). Devlin and Brodie (2005) mapped the plume produced by floods using aerial surveys in the Great Barrier Reef from

1991 and suggested that most plumes spread less than 50 km offshore and were governed by the wind field. However, they also showed that the dissolved nutrients within the plume distributed further offshore and downstream than the plume itself. This raised concerns about how and where the materials within the plume propagated. The sediment exported from the Burdekin River during a dry season (2010-2011) with cyclones was found over 100 km northward from the river mouth (Bainbridge, 2012). Brodie et al.

(2010) mapped by satellite images and suggested that nutritious water could be found

150 km offshore after flooding events. Devlin et al. (2012) has classified the Burdekin plume with three kinds of water: primary, secondary and tertiary by total suspended sediment, chlorophyII-a and coloured dissolved and detrital matter. However, from the perspective of plume structure which divides the plume into different part such as near and far field, the properties of the Burdekin River plume have not been well discussed.

Chapter 3. Data and Methodologies

3.1. Definition of a Freshwater Plume

For the Burdekin Region, a freshwater plume is defined as the coastal region where the salinity is lower than 33 PSU. Figure 3.1 shows the schematic of a freshwater plume

(Lentz and Largier, 2006) using a rectangular coordinate system with the x axis parallel to the coast, and the positive y axis pointing in the offshore direction. Wb and Ab define the width and cross-shore area of the plume where it extends to the sea bottom. Ws and

As are the width and area from the foot of the plume to the place where the 33 PSU isopycnal reach the sea surface (Figure 3.1). The thickness of the plume water column is

3 given by h(y), hp is the plume thickness, and 0 =1023 kg/m is the density of ambient water.

Figure 3.1: Structure of a freshwater plume. Image Source: Lentz and Largier, 2006.

According to Yankovsky and Chapman (1997), a plume can be categorized as either surface-advected (surface-trapped) or bottom-advected (bottom-trapped), depending on the percentage of the plume that reaches the sea bottom. In this study, a

freshwater plume is defined as surface-trapped if Ws is larger than Wb and as bottom- trapped if Ws is less than Wb.

3.2. GBR1 1km Model and river tracer

The eReefs provides three-dimensional real-time simulations of hydrodynamic, sediment, and biogeochemical processes for the entire GBR at 4- and 1-km resolution, and has been successfully used to analyse freshwater plume morphology and dynamics (Schiller et al.,

2015; Benthuysen et al., 2016; Condie and Condie, 2016). The GBR1 1km model is nested within the GBR4 4-km resolution modeland both models have been validated with observations and show similar levels of skill (Herzfeld et al., 2016; Robson, 2017). The hydrodynamic model is based on the Sparse Hydrodynamic Ocean Code (SHOC) model which is a finite difference Arakawa C-grid hydrodynamic model developed by the

Coastal Environmental Modelling Team at CSIRO Oceans and Atmosphere (Herzfeld,

2006). The model uses a curvilinear orthogonal grid system in the horizontal, which can account for complex topography in the Great Barrier Reef region. The resolution is 1 km with a domain size of 510 x 2390 grid points. A z-coordinate system is used in the vertical with 44 layers and a vertical resolution at the sea surface of 1 m to better capture the behaviour of the surface. Momentum, heat, and freshwater fluxes between air and the seawater on the sea surface layer are obtained from the ACCESS-R model. Boundary conditions are provided by the OceanMAPS model developed by the Bureau of

Meteorology from the BLUElink (Oke et al., 2008). Tides are implemented from a Center for Space Research (CSR) global tidal model (Eanes and Bettadpur, 1995) and the

Burdekin River flow data are obtained from the Queensland Department of Natural

Resources and Mines. As a near-real-time system, the GBR1 model has been run

continually from December 2014 to present (2018). Only a small part of the total model domain is used in this research to discuss the plume dynamics.

The freshwater is tracked using passive tracers, which are released at the river mouth and may be transported hundreds of kilometres from the river mouth. Thus, a larger domain is selected to study the transport of the freshwater from the river and this is described in section 3.4. In the GBR1 model, the tracers (1 kg/m3) are released from the river boundaries and are transported using a conservation flux scheme (Gillibrand and Herzfeld, 2016). The percentage of freshwater from the Burdekin River is then represented by the tracer concentration. If the concentration of the tracer at a grid point is 0.5, then this suggests that the percentage of Burdekin River water is 50%, and the rest of the water is either from the ocean or from other rivers.

3.3. The Moderate Resolution Imaging Spectroradiometer (MODIS) true-colour images

The Moderate Resolution Imaging Spectroradiometer (MODIS), which is a multi- spectral sensor with 36 spectral bands, is carried aboard the NASA Terra and Aqua satellites providing a twice-daily view of any location on the Earth. MODIS is able to provide 250-m resolution satellite true-colour images made up of a composition of three bands from the spectroradiometer: 645 nm mapped into red; 555 nm mapped into green; and 469 nm mapped into blue. MODIS observations of the ocean surface are obscured when clouds are present, such as during the passage of cyclones, and the plume cannot be observed by MODIS during these periods of cloudiness. In this study, the true-colour images are used after the passage of Cyclone Debbie to map the freshwater plume. The images are obtained from https://worldview.earthdata.nasa.gov/.

3.4. Empirical Orthogonal Function (EOF) Analysis

The Empirical Orthogonal Function (EOF) analysis is a statistical method also known as the Principle Component Analysis (PCA) and was first used to separate temporal and spatial variation in meteorological fields (Lorenz, 1956). Its application in oceanography can be found in many studies (e.g., Kaihatu, 1998; Servain and Legler, 1986; Chen et al.,

2017). The theory and calculation of EOFs is described in Hannanchi (2007). Based on the review of Hannachi et al. (2007), there are five kinds of EOF including the conventional EOFs, rotated EOFs, simplified EOFs, extended EOFs, and complex/Hilbert EOFs that are useful depending on the type and structure of the climate data being analysed. As a statistical analysis tool, EOF analysis is useful to reduce the dimensionality of high-dimensional data and separate out the patterns that explain the variance of the dataset (Monahan et al., 2009). In this thesis, EOF analysis is applied to the model data using a MATLAB EOF package developed by C. A. Greene, which can be accessed by https://au.mathworks.com/matlabcentral/fileexchange/61345-eof. The

North test (North et al., 1982) is applied to evaluate the statistical independence of the resulting patterns.

3.5. Period of Study

To highlight the influence of Cyclone Debbie, data are used from March 21 through April

19, which is one week before the landfall of the cyclone for a total 30 days. This period is chosen since it covers the period of the bulk river discharge after the passage of the cyclone, which lasts from March 29 to April 11. To evaluate the influence of Cyclone

Debbie, another non-cylone-forced flooding period in 2017 from May 15 to June 15 is chosen as well.

Three cross sections (Figure 1.1b) are used to discuss the vertical structure of the freshwater plume on the north, south, and seaward side of the river mouth. Their straight- line distances to the mouth are approximately 20 km, 17 km and 20km, respectively. The

13 m deep Station A as a numerical grid located on c1 in GBR1 model is chosen to represent a water column within the river mouth.

3.6. Cyclone Debbie and the river discharge

Forcing conditions during the passage of Cyclone Debbie are shown in Figure 4.4. To investigate the effect of Cyclone Debbie on the Burdekin River discharge, wind field data from the ACCESS-R model are utilized in the GBR1 model. Figure 3.2 shows the track of Cyclone Debbie. Based on data from the Bureau of Meteorology, the cyclone made landfall near Airlie Beach as a Category 4 tropical cyclone at 12:40 pm Australian

Eastern Standard Time (AEST) on March 28 2017. The averaged 10-m wind speed at c1

(location shown in Figure 1.1b) was initially more than 10 m/s and turned from southeasterly to west-southwesterly and then to northwesterly as the cyclone approached land and finally turned to southeasterly again after the passage of the cyclone. After landfall, it weakened into a tropical low by early on March 29.

Figure 3.2: Track of Cyclone Debbie. Image Source: Bureau of Meteorology

The Burdekin River discharge data recorded at Clare station which is 40 km upstream from the river mouth were obtained from the Department of Environment and

Resource Management (DERM) gauging network and freshwater released from the river boundary near the river mouth. In the model, unlike the real condition, there is no delay between Clare station and the river boundary cell. With the strong rainfall from Cyclone

Debbie in the Burdekin catchment, the discharge of the Burdekin River reached a peak discharge of 11,900 m3/s early on March 30. After that, the river flow rapidly decreased, but did not reach its previous flow of < 300 m3/s until April 11.

Chapter 4. The influence of Cyclone Debbie’s wind field on the Burdekin River freshwater plume

The work from Chapter 4-6 has been submitted as a journal article.

Y. Xiao, X.H. Wang, E. A. Ritchie, F. Rizwi, L. Qiao, 2018, The development and evolution of the Burdekin River estuary freshwater plume during Cyclone Debbie (2017),

Submitted to Estuarine, Coastal and Shelf Science.

4.1. Introduction

The prevailing wind plays an important role in freshwater plume dynamics and morphology. Previous studies suggest that the Burdekin River plume is dominated by a prevailing southerly and southeasterly wind (Wolanski and Jones, 1981). However, the passage of transient cyclones may change the freshwater plume dynamics because: 1) the heavy rainfall from the cyclone affects the river discharge; and 2) the strong clockwise cyclonic wind field will provide transient forcing on the freshwater plume dynamics.

Here, the GBR1 1-km model output is used to: 1) describe the plume dispersal during the passage of Cyclone Debbie; 2) find the relationship between the cyclonic wind stress, plume width, and thickness, and identify the impact of downwelling and upwelling wind forcing during the cyclone passage; and 3) discuss the wind-induced mixing.

4.2. Comparison between the modelled surface salinity and satellite true-colour images

Figure 4.1: Comparison between true-colour Images from MODIS, salinity from the

GBR1 in: (a), (c); MODIS: 10:40 AEST March 31. GBR1 model: 11:00 AEST March

31; and (b), (d); MODIS: 14:05 AEST April 1. GBR1 model: 14:00 AEST April 1.

A comparison between the Moderate Resolution Imaging Spectroradiometer (MODIS) images and salinity plots from GBR1 is shown in Figure 4.1. Note for future reference that all times are AEST using a 24-hour clock. On March 31 10:40 AEST, the plume in both MODIS imagery and the modelled data extended eastward (Figure 4.1a, c). The turbid water in Bowling Green Bay may be due to wind-induced sediment resuspension.

On April 1 14:05 AEST, the freshwater plume was located along the coast in both satellite image and plume plots (Figure 4.1b, d). From these plots, it appears that the river plume is captured well by the model.

Figure 4.2: (a) Comparison between measured salinity at Yongala National Reference Station and the GBR1 model at different levels in the ocean during 2017. (b) and (c) compare the measured near-surface temperature with the modelled temperature from the GBR4 and GBR1 models at Townsville and Abbot Point from 2014 to 2015. Source: eReefs Modelling report

(Herzfeld et al., 2016).

Information on Yongala National Reference Station used in Figure 4.2a can be found on: https://data.aims.gov.au/moorings/IMOS-NRS/IMOS-NRStoc.html. From the comparison, for the both surface and bottom layers, the GBR1 model captures the tendency of the salinity variation quite well but has a negative bias of about 0.3 PSU except in April. The validation done by Herzfeld et al (2016), at Townsville and Abbot

Point shown in Figure 4.2b and 4.2c suggests that the model matches the observations quite well.

4.3. Evolution of the plume and Cyclone Debbie

Figure 4.3 shows the surface salinity and current at the sea surface from the model. Before the passage of the cyclone (March 21, Figure 4.3a), most of freshwater was restricted to the coast and advected northward by a southeasterly wind. At this stage, only a small amount of freshwater was advected southward into Upstart Bay. By noon of March 29, the winds had turned northwesterly with the passage of the cyclone and the plume reversed its direction to southward (Figure 4.3b). The river discharge began to rapidly increase on March 29 due to the heavy rains associated with Cyclone Debbie and peaked at 11,900 m3/s just prior to March 30, which produced a freshwater plume in the ocean extending to the east. Driven by buoyancy and initial momentum from the Burdekin

River, the plume water moved faster than the ambient water. The freshwater extended beyond Cape Upstart on March 30, and into Abbot Bay. Restricted by the topography of

Cape Upstart, the bulge did not grow in a more regular circular shape and instead bulged around the end of the peninsular (Figure 4.3c). On March 31 (Figure 4.3d, e) the northwesterly winds turned back to southeasterly and the plume turned northward along the coast. With relatively high river discharge compared with the period before March

29, the plume moved along the coast (Figure 4.3f, g) and beyond Cape Bowling Green by April 2 (Figure 4.3h). Divergence of freshwater can be found near the Cape Bowling

Green in Figure 4.3h. Based on the definition of a freshwater plume in section 2.1

(salinity <= 33), the plume finally became restricted in Bowling Green Bay. A similar result was observed by aerial observation in a previous flood plume event (Wolanski and

Jones, 1981). The river flow into the ocean decreased gradually through April 12 (Figure

4.4c), the width of the freshwater plume became narrower with northwestward transport and the area of its bulge decreased with advection and mixing.

Figure 4.3: Surface plume and currents prior to, and after the passage of Cyclone Debbie, which made landfall on March 28 2017: (a) Prior to Debbie’s landfall; (b)-(f) After

Debbie’s landfall. The compass shows the wind direction. Arrows show the surface current.

The longshore wind stress and current measured at Station A are shown in Figure

4.4a and 4.4b along with the river discharge recorded at Clare Station in Figure 4.4c for the entire 30-day period. Prior to Debbie’s passage, the longshore wind stress was near zero and slightly positive and the longshore current was oscillating semi-diurnally. After

Debbie’s passage, the longshore wind stress increased to over 0.5 Pa, the longshore residual current became southward for approximately 3 days with a maximum speed of

0.2 m/s. Wind stress had a high correlation with the magnitude of the residual current.

When the wind stress was weak (~0 Pa) both well prior to Debbie’s passage (March 21) and well after her passage (April 10), the current oscillated around 0.1 m/s with a magnitude of less than 0.2 m/s. This result is similar to Wolanski (1983), which suggests that the tidal current is generally less than 0.2 m/s on the shelf of the Great Barrier Reef.

River discharge increased after the landfall of Cyclone Debbie (Figure 4.4c), so that the bulk of the freshwater moved northward with southerly wind at first.

Figure 4.4: (a) Surface longshore wind stress; and (b) Modelled surface longshore current at Station A; and (c) River discharge recorded at Clare station, which is 40 km upstream from the mouth of the Burdekin River. Vertical dash lines show the time nodes used to discuss the plume structure under different wind forcing in section 3.2. The dash-dot lines in (b) shows the low-frequency surface longshore residual current.

Time series of surface salinity at c1 (north cross section) and c2 (south cross section) are shown in Figure 4.5. From March 21 to April 20, the freshwater at c1 was mostly restricted to within 10 km offshore and only floats on the shallow shelf (depth <

20 m). When the plume was wider than 10 km on March 23, there was weak wind stress

(see Figure 4.5a, ~ 0 Pa) similar to April 10. The longshore salinity was between 25 and

30 PSU prior to, and after the passage of the cyclone, and there was an apparent

oscillation in the width of the fresh water in response to flood and ebb tidal currents. This suggests that river discharge may not have been the major cause for changes in the plume width when the downwelling wind was strong. The plume disappeared from c1 between

March 29 and April 1 and the area where salinity was less than 33 PSU decreased with the upwelling wind forcing. During the periods highlighted by the white line on Figure

4.5b, the plume width was more than 10 km. The plume was first produced on c2 after

March 30 when the river flow increased and the winds turned to a weak northerly wind on the backside of the cyclone. Subsequently, the wind turned back to southerly but the plume still expanded over 30 km on March 31 and April 1 because of high river discharge.

On April 2, the plume was wide (Figure 4.5b) with both large river discharge and weak wind stress. In contrast, there was no plume evident on the south cross section (c2) until the passage of Cyclone Debbie (Figure 4.5c). As the river discharge dropped and the prevailing southerly wind re-established, the plume width on c2 decreased and completely disappeared after April 3.Under similar downwelling wind forcing from April

3 to April 10 and again from April 12 to April 18, in the second time period (April 12 to

April 18) the river discharge decreased to less than half while the plume width did not decrease noticeably.

Figure 4.5: (a) Longshore wind stress  x at Station A. Hovmöller diagram of surface salinity at: (b) c1; and (c) c2. White lines indicate when the plume width, Wp, is larger than 10 km on c1.

4.4. Plume morphology and wind

Figure 4.6: Sea surface salinity distribution under different wind forcing conditions: (a)

Weak downwelling ( x = -0.003 Pa); (b) After strong downwelling ( = -0.087 Pa); (c)

Weak upwelling ( = 0.054 Pa); and (d) Strong upwelling ( =0.214 Pa).

The wind forcing may change not only the direction, but also the shape of the freshwater plume. Such a phenomenon was discussed by Fong and Geyer (2001) and Lentz (2004) based on numerical experiments. It has also been observed in nature for the western Gulf of Maine plume (Fong et al., 1997) and Chesapeake Bay (Lentz 2004 and 2006). Because of onshore Ekman transport, in moderate downwelling winds condition (Wind stress

< 0), the isopycnals become more vertical because of mixing and the freshwater plume is thick and narrow. The plume will be thin and wide with offshore Ekman transport when the wind favours upwelling ( > 0). Figure 4.6 and 4.7 show the plume at the surface and c1 under the different wind forcings, respectively. For downwelling wind, when the wind stress was weak ( = -0.003 Pa), the plume extended to 12 km offshore and showed a typical surface-trapped structure since the plume could extend to the ocean bottom only within 2 km offshore. Little vertical mixing occurred on the front of the 35

plume similar to Fong et al., (1997) and Lentz, (2004; 2006). When the wind stress was

stronger on March 27 (see Figure 4.6b and 4.7b,  x = -0.087 Pa), the plume width decreased to about 6 km and its thickness increased to about 8 m. At the same time, the vertically mixed area at the plume front also increased especially for the 35-PSU isohaline (Figure 4.7b). The horizontal structure of the plume in Figure 4.5b also shows that the plume was restricted to the coast under the strong downwelling wind. For the upwelling event, since the time period of the upwelling wind was short, and the wind stress changed rapidly, March 28 11:00 (0.054 Pa) and March 28 23:00 ( = 0.214 Pa) are chosen to represent weak and strong upwelling wind conditions, respectively. With weak upwelling winds, there were only some vertical isopycnals between 33 PSU and 35

PSU shown in Figure 4.7c. From the surface salinity distribution, the freshwater less than

33 PSU reversed direction with the upwelling wind and moved into Upstart Bay. Under weak upwelling wind, the plume was narrow off the coast (Figure 4.7c). When the upwelling wind was stronger, the plume on surface reversed (Figure 4.7d) and there was no plume evident on c1 (Figure 4.7d), and more freshwater reached Cape Upstart.

Furthermore, the 35 PSU ispycnal moved farther from the coast during this period.

Figure 4.7: Salinity distribution on c1 under different wind forcing conditions: (a) Weak

downwelling ( x = -0.003 Pa); (b) After strong downwelling ( = -0.087 Pa); (c) Weak upwelling ( = 0.054 Pa); and (d) Strong upwelling ( =0.214 Pa).

Scatter plots show the relationship between plume width and longshore downwelling wind stress, and plume thickness and longshore downwelling wind stress are shown in Figure 4.8. The colour in figures shows the river discharge of each point.

The plume width and thickness with upwelling (Not shown) remains low because of the reverse of the plume itself and this part has not been considered. For the no-wind

condition ( x =0), the width of the plume varies from 6 to 12 kilometres and the thickness is between 4 and 6 m (Figure4.8a and Figure 4.8b). Under these conditions other parameters may influence the variability of the plume structure. Onshore (offshore) tidal currents may push the plume toward (away from) the coast. For moderate downwelling winds (-0.2 Pa < < 0 Pa), there is a ‘triangle’ distribution. Generally, the plume width decreases with increased longshore wind stress. When the wind is weak, the variation of the plume width is large (4-18 kilometres) and may be influenced by other parameters such as tide and river discharge. For example, onshore (offshore) tidal currents could stretch (squeeze) the plume. Larger river flow could increase the volume of freshwater and change the cross-section area and morphology of the plume. The widest plume with yellow dots was found with weak downwelling wind less than 0.05 Pa and relatively high river discharge over 1000 m3/s. However, the thickest plume was found with 0.1 Pa downwelling wind and river discharge around 1000 m3/s (Figure 4.8b). When longshore wind stress was stronger than 0.1 Pa, the plume width decreased to about 5km and its depth varied from 4 m to 10 m. Dots in light blue and green had higher river discharge, wider and thicker plume than deep blue dots. It shows the plume was squeezed with stronger downwelling wind but the influence of higher river discharge cannot be ignored.

Figure 4.8: Scatter plot between: (a) plume width, WP, and downwelling wind stress,  x ; and (b) Plume thickness, hP, and downwelling wind stress, along c1. Wind stress is recorded at Station A. Coloured points indicate the level of the river discharge.

For c2 (figure not shown), because of bulk river discharge with a peak over

10,000 m3/s and transient northerly wind, the plume moved southeastward, and its largest

width (>30 km) was higher than on c1. However, since the southward wind lasted less than 2 days and the river discharge dropped quickly, the plume did not reach a steady state under these forcing conditions.

To determine if the plume is surface-trapped (Buoyancy-controlled) or slope-

WS controlled (Wind-controlled), the parameter Wn = is used to describe the structure of Wb

the plume. If Wn >1, the plume is surface trapped and controlled by buoyant forcing. If

<1, the plume is slope controlled by wind forcing. Figure 4.9 shows changes of with wind stress. Colour shows the river discharge of each point. Under downwelling wind, when the wind stress is larger than -0.1 Pa, is generally larger than 1, which

indicates that the plume is surface trapped. When  x < -0.1 Pa, the plume is pushed onshore and decreases to less than 1. Under downwelling wind, the amount of river discharge do not have obvious influence on . The plume is squeezed, which could also explain the ‘verticalization’ of isopycnals and vertical mixing.

Figure 4.9: Scatter plot along c1 between Wn (=Ws/Wb) and wind stress x calculated at

Station A. Coloured points indicate the amount of river discharge occurring for each point.

4.5. Wind mixing of plume

Strong downwelling winds can generate vertical mixing, which can be evaluated using the Bulk Richardson number. The Bulk Richardson number RiB is given by Lentz (2004)

gh RiB = 2 , (4.1) 0 ()u where 훥휌 and 훥푢 are the density and velocity difference between the top and bottom

3 layer at station A on c1, 0 = 1023 kg/m is the density of ambient saltwater, g is the gravitational constant, and h is the distance between the top and bottom layer. Since only the condition when the freshwater plume exists is considered, surface salinity larger than

33 is neglected. The bulk Richardson number variation with wind forcing is shown in

Figure 4.10. For  x larger than -0.1 Pa, RiB is larger than 1, which indicates that wind- 41

induced vertical mixing may not exist. For longshore wind stress  x smaller than -0.12

Pa (downwelling wind), the value fluctuates about 1 suggesting that mixing is occurring.

Similar results were also shown in Delaware Bay and Chesapeake Bay, which suggests that vertical mixing occurred in these areas with strong downwelling wind (Lentz and

Largier, 2006; Sanders and Garvine, 2001).

Figure 4.10: Scatter plot between Bulk Richardson number and longshore downwelling wind stress at Station A.

The vertical structure of the plume indicates whether there has been vertical mixing. For both downwelling and upwelling wind forcing in Figure 4.7, there has been vertical mixing occurring in the far-field (outer front) of the plume where the plume is dominated by wind (Fong and Geyer, 2001). Even when the wind forcing is weak (e.g., weak upwelling wind), there is strong vertical mixing (Figure 4.7c). When the upwelling wind lasts for a longer time and is stronger, the isopycnal is vertical and salinity increases seaward, which suggests that the vertical mixing has occurred through the entire water column. In addition to wind mixing, a southward ambient current during an upwelling wind also increases the water salinity. For the near-field, inertial shear mixing discussed 42

by Hetland (2005) is not evident, suggesting that the model may lack the ability to simulate the near-field plume structure.

To further discuss the influence of the cyclone, a time series plot for various parameters from March 27 to April 2 is shown in Figure 4.11. Data are obtained from

Station A. With downwelling wind forcing prior to the cyclone landfall (e.g., March 28), seawater on the 13-m station water column was stratified. Both salinity and temperature were higher at the ocean bottom than the surface. The surface salinity is likely influenced by other factors including the riverine freshwater, which could dilute salt water. In addition, the sea-surface temperature may be cooled by strong surface winds, which extract sensible heat to the atmosphere. By April 1 Cyclone Debbie had moved inland, the wind direction returned to southeasterly (downwelling wind forcing), and the longshore current turned to a northward direction. The surface current responded most quickly to the change in wind forcing, and velocities at different depths then gradually changed to northward after one day of the upwelling wind forcing. Between March 29 and March 30 both salinity and temperature were well mixed (Figure 4.11d and e).

4.6. Conclusion

Using the GBR1 1km hydrodynamic model, the transient Burdekin River plume during the passage of Cyclone Debbie is investigated. The model reflects the behaviour of the transient plume well compared with MODIS true-colour images and observation data.

Under weak downwelling winds, the plume (salinity < 33 PSU) was sensitive to river discharge and periodical tidal forcing. With large river discharge from the rainfall catchment area after the landfall of the cyclone, the thickness and width of the freshwater plume increased. With stronger downwelling wind forcing, the plume was restricted to the coast (< 10 km) but still influenced by river discharge reflected by its width and thickness. Its cross-sectional shape was highly depended on longshore wind stress and

tends to be bottom trapped when longshore downwelling wind was stronger. For upwelling wind forcing, the plume reversed southward even with weak wind stress.

Wind mixing in the far field zone of the freshwater plume is captured by the model.

However, the impact of the southward ambient current during upwelling wind forcing and tidal mixing is only transient and subject to further discussion.

Chapter 5. River discharge

The work from Chapter 4-6 has been submitted as a journal article.

Y. Xiao, X.H. Wang, E. A. Ritchie, F. Rizwi, L. Qiao, 2018, The development and evolution of the Burdekin River estuary freshwater plume during Cyclone Debbie (2017),

Submitted to Estuarine, Coastal and Shelf Science.

5.1. Introduction

In addition to the high winds, cyclones also bring heavy rainfall upon landfall, with associated runoff, which can lead to a large increase in river discharge to the ocean. For a semi-arid catchment like the Burdekin River, most of the river discharge occurs during the monsoon or transient rainfall events such as cyclones. The river discharge is the direct source of the freshwater plume. Previous studies described the freshwater plume evolution during floods (Wolanski and Jones, 1981; King et al., 2002). While these previous studies have investigated the sediment deposition regions processes during flood events (e.g., Bainbridge et al., 2012; Delandmeter et al., 2015; Kroon et al., 2012), the fate of the Burdekin River freshwater has not been discussed.

In this chapter, the freshwater transport during the passage of Cyclone Debbie is discussed. Using Burdekin River Tracers released into the GBR1 hydrodynamic model, the evolution and distribution of the Burdekin River freshwater during Cyclone Debbie will be examined and compared with a smaller non-cyclone flooding event.

5.2. Fate of the freshwater

Figure 5.1: Tracer concentration distribution at the sea surface on April 20, 23 days after the cyclone landfall. The tracers were released into the model at the mouth of the

Burdekin River.

Tracers released from the Burdekin River in the model are used to investigate the fate of the freshwater discharge. Figure 5.1 shows that, until April 20, with prevailing southerly wind, the tracers had moved north beyond Hinchinbrook Island, which is 100 km away from the river mouth. The coastal zone from Upstart Bay was filled with freshwater with concentrations higher than 5%. A freshwater patch with concentrations of about 8% were located around Cape Cleveland. It is notable that the freshwater plume was very narrow near Upstart Bay after the release from the river and its width increased as it propagated north. This suggests that the freshwater plume may not have reached the reefs close to the mouth of the Burdekin River. However, considerable diffusion can be seen on the edges of the plume and it broadened its surface area extent as it moved northward from its source. This suggests that under the cyclone forcing conditions, the freshwater plume 47

may have reached reefs further north around Orpheus Island (Figure 5.1). Most of the freshwater existed in the area where the water depth was less than 30 m close to the coast.

North of Townsville, the slope of the sea bottom decreases, and the area containing more than 1% freshwater expanded on the shelf. When the plume passed Orpheus Island, where the bottom slope is steeper, the width of the plume was narrower. This suggests that the offshore expansion of freshwater is related to the sea bottom slope and water depth.

5.3. Comparison to a 2017 flooding event

To determine whether the effects of the river discharge were different under the cyclone- induced conditions compared with non-cyclone forcing, another flooding event is used as a comparison. The period chosen is from May 15 to June 14, 2017. During this period the river discharge reached a peak of 3000 m3/s before decreasing, there were prevailingly downwelling (southeasterly) winds (Figure 5.2a), and the longshore current was frequently less than 0, which suggests that the residual current flowed northward.

River.

Accumulation of freshwater transport at the three cross sections during Cyclone

Debbie and the 2017 flood event are shown in Figure 5.3. Both time periods are 30 days.

The accumulated river discharge during the passage of Cyclone Debbie (2.3 x 109 ton) was higher than the non-cyclone flood event (0.8 x 109 ton). The freshwater volume through c1 was higher during the cyclone (1.5 x 109 ton) than the non-cyclone event (7.8 x 108 ton). However, the percentage of freshwater that passed through c1 during Cyclone

Debbie (66.0%) was lower than that during the non-cyclone flooding event (98.1%). This is because after the passage of Cyclone Debbie the fresh-water plume propagated southward for a short period. A similar result can be found at c2 where 1.6 x 108 ton

(6.9%) freshwater passed through the cross section during the cyclone. The amount of freshwater that passed through c2 was higher with northerly wind forcing but it decreased significantly after April 1 due to a combination of the southerly wind and the freshwater from other rivers outside the domain that also crossed c2. In contrast, the freshwater that passed c2 during the non-cyclone flood was negative (-10.3%) indicating that it had a source from the south other than the Burdekin River. Finally, the non-cyclone flooding event produced very low transportation of fresh water (5.5 x 106 ton, 0.6%) at c3 compared with 3.2% (7.0 x 106 ton) for Cyclone Debbie. This indicates that a small percentage of the freshwater propagated seaward during the cyclone event and may have reached into some environmentally sensitive areas such as the offshore coral reefs.

However, since the amount was small, the impact of Burdekin River flooding water on the nearshore reefs such as Old Reefs was likely limited.

5.4. Conclusion

Using the river tracers released from the Burdekin River, the propagation of freshwater during Cyclone Debbie has been examined. The cyclone-induced flood contributed bulk freshwater inputs into the Burdekin Estuary, which propagated northward due to the coastal current. The majority of the freshwater was within Upstart Bay constrained on the shallow continental shelf with depth less than 30 meters. After 23 days, the freshwater plume had transported over 100 km northward beyond Cape Cleveland and reached the 51

nearshore reefs near Orpheus Island. The freshwater was restricted to the coast until Cape

Cleveland. After that, because of a decreased sea-bottom slope, the freshwater plume expanded offshore but became narrower again after passing Orpheus Island (Figure 5.1).

The freshwater transport across three cross sections, c1, c2, and c3 during Cyclone

Debbie and another flooding event in 2017 were compared. In both events, most of the freshwater propagated northward. The major difference was that during the cyclone event, some freshwater initially transported southward with the northerly wind forcing on the back side of the cyclone. The freshwater then reversed direction to propagate northward as the prevailing southerlies re-established after the cyclone passage. In addition, more freshwater propagated offshore during the passage of the cyclone through c3 compared with the non-cyclone event. However, the amount and percentage of freshwater was limited and would not likely pose a measurable impact to the offshore reef systems such as the Old Reefs.

Chapter 6. EOF analysis and Wind Strength Index- competitions between different forcing

The work from Chapter 4-6 has been submitted as a journal article.

Y. Xiao, X.H. Wang, E. A. Ritchie, F. Rizwi, L. Qiao, 2018, The development and evolution of the Burdekin River estuary freshwater plume during Cyclone Debbie (2017),

Submitted to Estuarine, Coastal and Shelf Science.

6.1. Introduction

The spread of freshwater plumes is determined by different forcings, including wind, buoyancy forcing, tide, topography/bathymetry, earth rotation and ambient current

(Horner-Devine et al., 2014). For different regions, the dominant forcing will be different and understanding the pattern of river plumes and associated salinity variation under these different forcings is important. EOF analysis is applied to the hydrodynamic model output to identify the statistically significant patterns that underpin the ocean hydrodynamics. The Wind Strength Index developed by Whitney and Garvine (2005), is also applied to the model data to identify the relative importance of wind and buoyancy forcing within the freshwater plume.

6.2. EOF analysis during Cyclone Debbie

Using EOF analysis the GBR1 hourly surface salinity data from March 21 00:00 AEST to April 20 00:00 AEST were divided into 10 statistical modes. Among them, the first eight modes are well separated (North et al., 1982). The variance of the data explained by each mode is shown in Table 6.1.

Table 6.1: Variance explained by the 10 EOF modes.

Among the 10 EOF modes, the first and second mode explains 56.1% and 17.0% of the variance, respectively. Thus, only these modes are considered here. For the first mode, the spatial pattern shows a cohesive positive maximum close to the mouth of the

Burdekin River (Figure 6.1a). Since the expansion coefficient was negative during the passage of the cyclone, the dilution of salinity was strong especially near the river mouth.

There are high frequency signals in the expansion coefficient, which may be produced by tidal forcing. Similar frequencies occur in EOF mode 2 (Figure 6.2b). The lowest expansion coefficient for EOF mode 1 (-402.2) occurred when peak river discharge occurred few hours before. on March 31 as shown in Figure 6.1b, with a drop of salinity on the sea surface compared to the mean salinity of the whole time period. When the expansion coefficient reached its peak of approximately 100 on Mar 29, the salinity was higher than average, and the river discharge was only about 200 m3/s. It suggests that the first mode responds to river discharge, and the river discharge is the major cause for the sea surface salinity variation during the passage of Cyclone Debbie. With higher river discharge, the expansion coefficient was lower, and there was more diluted seawater on the sea surface especially around the river mouth. With the spread of riverine freshwater, the value shown by the spatial pattern decreased gradually from the river mouth to outside 54

closing to the coast. However, there is a lag between the expansion coefficient and the river discharge. To examine the relationship, the correlation coefficient between the river discharge and the expansion coefficients from EOF mode 1 for different lag times is shown in Figure 6.1c. The maximum correlation occurs at a lag time of 17 hours suggesting a response time of mode 1 to river discharge of about 17 hours with r = 0.65.

It suggests that the spread of freshwater had a 17-hour delay to the river discharge. Given that the river boundary cell of the GBR1 model is at the river mouth, the delay might be caused by the spread of freshwater on the sea surface. In addition, the diurnal variation

(fluctuation) of the expansion coefficient might be produced by tides.

Figure 6.1: (a) Spatial pattern of EOF mode 1; (b) the expansion coefficient of EOF mode

1 and river discharge; and (c) Correlation between river discharge and Expansion coefficients from EOF mode 1.

The second mode (17%) indicates that in the longshore area, there is a negative relationship between the north side (positive value) and south side (negative value) of the

river mouth (Figure 6.2a). On March 29, the most negative value of the expansion coefficient (-135) results in the lowest salinity off the north shore, which is closely related to the surface salinity distribution. On April 2, the highest expansion coefficient led to the lowest negative variation in the south side region, which is closely linked to the maximum southward extension of freshwater (Figure 6.2b). Meanwhile, the largest positive variation on the north side of the river mouth is associated with vertically mixed water off the north shore. In summary, mode 2 is related to the movement of the plume which may be driven by wind forcing or the ambient current. This result is similar to the study of Chen et al. (2017) in the Pearl River estuary that the opposite value in the both side of the river mouth related to the spread of the freshwater plume. The time series of mode 2 is compared to the longshore ambient current recorded near Abbot Point (Figure

6.2a) and longshore wind stress at station A, with correlation coefficients equalling to

0.36 and 0.18, respectively. Compared to wind, the ambient current is more likely to affect EOF mode 2. When the ambient current was directed northward along the coast, the north side of the Burdekin River mouth had lower salinity compared to the average value during the time period. Salinity on the south side of the Burdekin River mouth was lower with a southward longshore ambient current. In addition, similar to the first mode, there is a diurnal variation (fluctuation) in mode 2, which may be produced by tide.

Figure 6.2: (a) Spatial pattern of EOF mode 2, is the station to show the ambient current; and (b) the expansion coefficient of EOF mode 2 and longshore ambient current recorded at .

In general, the first two modes emphasise the influence of the river discharge and ambient current. EOF mode 1 is highly related to the spread of freshwater on the sea surface, with a maximum correlation of 0.65 when mode 1 is lagged by 17 hours to the river discharge.

The highest correlation is near the river mouth, suggesting that the diluting effect of river water in this region is stronger. EOF mode 2 suggests that compared to the direct influence of wind, the longshore ambient current is more likely to change the surface salinity directly.

6.3. Comparison between wind and buoyancy forcing

To evaluate the effect of wind compared with buoyancy forcing, the Wind Strength Index

uwind Wsi = , (6.1) udis

developed by Whitney and Garvine (2005) is used. The wind-driven flow is given by

-2 uUwind ~ 2.65 10 , where U is the longshore wind speed. The buoyancy-driven flow is

1 given by u= (2 g' Qf ) 1/4 , where K is the dimensionless current width, disK r

' a− river ggr = , is the reduced gravity, 휌푟푖푣푒푟 is the river water density, 휌푎 is the a ambient coastal current water density (Salinity = 35 PSU and temperature = 28.5 °C), Q is the river discharge (m3/s), f is the Coriolis frequency, and K=3 for a flooding event for the Burdekin River (Wolanski 1994). If Wsi is larger than 1, the plume water is controlled by wind forcing. If Wsi is less than 1, the plume water is controlled by buoyancy forcing.

Figure 6.3: (a) Magnitude of the Wind Strength Index (Wsi ) and longshore wind stress

 x at station A during the passage of Cyclone Debbie. and (b) Wind-driven flow uwind and buoyancy-driven flow udis. 58

From Figure 6.3, the plume was firstly controlled by buoyancy forcing. As the wind increased, Wsi also increased reaching a peak of ~5 in conjunction with the peak upwelling wind. This result is similar to the surface salinity distribution on March 29

12:00 (Figure 4.1b), which reflects that buoyancy-driven northward propagation of the freshwater plume is limited. After March 29, the longshore wind stress decreased, the river discharge increased, and Wsi decreased to almost 0, which indicates that the freshwater plume was controlled by buoyancy during this period. After April 2, downwelling wind stress increased and the river discharge decreased, and so Wsi increased with values generally higher than 1 except on April 10, which suggests that the plume was dominated by wind forcing after April 3. During this cyclone-induced flood, udis varied from 0.09 m/s to 0.2 m/s but uwind varied from 0 to 0.5 m/s indicating that the variation of wind strength can have a stronger influence on Wsi.

A typical condition is that when the river flow remains low, udis ~ 0.1 m/s, and if the wind speed is higher than 3.7 m/s (magnitude of wind stress ~ 0.03 Pa), the longshore propagation of the freshwater plume is controlled by wind forcing. For high river discharge such as the peak of the cyclone-induced flood, udis ~ 0.2 m/s, suggests that the freshwater plume will be driven by wind forcing if it is over 7.5 m/s (wind stress magnitude ~ 0.06 Pa).

6.4. Conclusion

A comparison between the first 2 modes from the EOF analysis suggests that the first mode (51.7% of the explained variance) corresponds to river discharge, and is the major cause for the variation in the surface salinity distribution. By calculating the correlation between river discharge and the time series of mode 1, it has been found that with 17 hours lag time, the mode 1 has the highest correlation (r=0.65) to the river discharge.

Wind stress and ambient current may cause the change in the propagation direction of the freshwater plume from northward to southward during the passage of Cyclone Debbie and back again after its landfall. The freshwater plume reversal is reflected in the second mode (17% of the explained variance), which has a pattern reflecting the northward and the southward spread of the freshwater plume which may be controlled by longshore ambient current (r= 0.36). Both modes have high frequency fluctuations, which may be produced by tide or the fluctuation of wind.

The Wind Strength Index compares the impact of wind forcing and buoyant forcing on the Burdekin River plume. This suggests that when the river flow is low the freshwater plume is controlled by wind forcing. When the wind is weak the freshwater plume is controlled by buoyancy forcing especially after the peak of the river discharge.

Chapter 7. Conclusions and future work

7.1. Conclusions

The aim of this study is to understand the plume dynamics of the Burdekin River in

Queensland, Australia during the passage of Cyclone Debbie in 2017. The eReefs GBR1

1km 3D hydrodynamic data from CSIRO and MODIS true-colour images were applied to illustrate the plume behaviour and calculate the amount of freshwater transport. First, surface and horizontal plume distribution under different wind condition was discussed.

Then, the relationship between the plume width, thickness, mixing and, longshore downwelling wind were discussed. The fate of the freshwater and freshwater transport amount was analysed. Finally, EOF analysis was applied to determine the dominant forcing for the freshwater plume during the passage of Cyclone Debbie.

7.2. Transient wind field and plume evolution during Cyclone

Debbie

Because Cyclone Debbie had a clockwise rotating wind field, its passage and subsequent landfall brought a strong, but transient wind field through the Burdekin River Estuary in a relatively short period. The region is typically under a prevailing southerly

(downwelling) wind regime, which influences the freshwater plume. However, during the passage of the cyclone, a relatively brief period of northerly (upwelling) winds impacted the region. The combination of both upwelling wind and flooding from the heavy rainfall after Debbie made landfall resulted in a massive freshwater plume that propagated southward beyond Cape Upstart. The upwelling wind lasted only a short time

and the freshwater plume subsequently reversed its direction to northward again as it came under the influence of the prevailing downwelling wind.

More generally, under weak downwelling winds, the plume (salinity < 33 PSU) was sensitive to river discharge and periodic tidal forcing reflected by the periodic diurnal variation. As the river discharge rapidly increased from the rainfall catchment area after the landfall of the cyclone, the thickness and width of the freshwater plume increased.

With stronger downwelling wind forcing, the plume was restricted to the coast (< 10 km) and higher river flow contributed to relative higher value of its depth and thickness. The cross-sectional shape of plume was more likely to be controlled by longshore wind. For upwelling wind forcing, the freshwater plume reversed southward even with weak wind stress. However, the current research is based on the eReefs modelling results and only considered the impact of different wind directions. In reality, the period of the wind forcing is also important because the freshwater plume cannot reach a steady condition under changing wind conditions. Thus, the use of real wind forcing conditions here makes it hard to compare the influence of different wind magnitudes and directions under the same wind period. Wind mixing in the far field zone of the freshwater plume was captured by the model. With stronger downwelling wind, the plume water was well mixed. In addition to wind mixing, the tidal mixing may also be important. However, due to lack of sensitivity experiments, its level of influence cannot be quantified here.

7.3. Cyclone-induced flood and freshwater transport

Using river tracers released from the Burdekin River into the GBR1 hydrodynamic model, the high concentration freshwater was traced hundreds of kilometres from the Burdekin

River mouth. The flood water finally reached some environmentally sensitive reef regions such as Orpheus Island to the north of the Burdekin River 21 days after the initial

discharge into the estuary. Although not modelled here, turbid flooding water can bring contaminants into these regions and may threaten the sensitive ecological systems. In addition, some freshwater patches restricted to the capes and estuaries may also have been present as well.

During the passage of the cyclone, the total amount of freshwater propagating north, south, and offshore was much higher than the comparable non-cyclone flooding event because of high river discharge and reversed wind field. Even though the amount and percentage of the freshwater propagating offshore was greater for the cyclone event, the amount was still limited to less than 10% of the total water. This indicates that for the coral reefs directly offshore from the Burdekin River mouth, the impact of freshwater might not be significant. However, the freshwater influence farther along the coast such as around Orpheus Island is significant.

7.4. Dominant forcing during the passage of Cyclone Debbie

Using EOF analysis, it was found that more than half of the salinity variation is contributed by mode 1, which was shown to be highly related to the river discharge, with correlation coefficient over 0.6.

EOF mode 2 explains 17% of the surface salinity variation captured the pattern of northward and southward extension of the river plume. The correlation coefficient between mode 2 expansion coefficient and the ambient current is higher (0.36) than the wind (0.18), which suggests that mode 2 is more likely to be controlled by the ambient current.

In summary, with weak or moderate wind forcing, the plume is controlled by the buoyant forcing from the river, while the cyclone wind forcing and the ambient current contributed to the southward movement of the freshwater plume.

7.5. Suggestion for future work

Sensitivity experiments for tides and ambient current: The study highlights the

impact of wind and the buoyant forcing. However, other forcings such as tide and

ambient current have not been discussed. Because there exists non-linear

relationships between different forcings, it is hard to separate their influence

especially during the passage of a severe cyclone. Numerical simulations can be used

to investigate their influence since it is possible to control the forcings separately in

the model. Thus, sensitivity experiments based on an improved GBR1 model or other

models such as the Princeton Ocean Model (POM) should be carried out to test the

influence of tides and ambient current on freshwater plume dynamics.

Sediment dynamics using a coupled sediment model: This study only discusses

the hydrodynamic characteristics of the Burdekin River plume. However, the bulk

flood water carries a large amount of sediment and other dissolved or undissolved

matter into the Great Barrier Reef. These potential contaminants can be deposited

close to the coast, and resuspended again by forcings such as wind, wave, and tides.

If re-suspended, the sediment and contaminants can be transported further from the

river mouth and finally have a greater impact on the GBR than the freshwater plume

itself. It is important to investigate the sediment dynamics in the Burdekin River

region. Compared to satellite images and in-situ observations, numerical simulations

can provide long-term high-resolution three-dimensional output, which can be used

to investigate the sediment dynamics of the region. The Relocatable Coastal Model

(RECOM) model coupled to GBR1 1km model may be used to simulate the sediment

transport process in the Burdekin region.

References

Alharbi, A. (2014). Risk communication discourse: A content analysis of some

Australian media coverage of cyclones in Queensland, Australia in 2011. International

Journal of Human Sciences / Uluslararası İnsan Bilimleri Dergisi, 11(1), pp.1019-1036.

Alibert, C., Kinsley, L., Fallon, S., McCulloch, M., Berkelmans, R. and McAllister, F.

(2003). Source of trace element variability in Great Barrier Reef corals affected by the

Burdekin flood plumes. Geochimica et Cosmochimica Acta, 67(2), pp.231-246.

Andutta, F., Ridd, P. and Wolanski, E. (2011). Dynamics of hypersaline coastal waters in the Great Barrie Reef. Estuarine, Coastal and Shelf Science, 94(4), pp.299-305.

Andrews, J. and Bode, L. (1988). The tides of the central Great Barrier Reef. Continental

Shelf Research, 8(9), pp.1057-1085.

Bainbridge, Z., Wolanski, E., Álvarez-Romero, J., Lewis, S. and Brodie, J. (2012). Fine sediment and nutrient dynamics related to particle size and floc formation in a Burdekin

River flood plume, Australia. Marine Pollution Bulletin, 65(4-9), pp.236-248.

Belperio, A. (1978). An inner shelf sedimentation model for the Townsville region, Great

Barrier Reef province. James Cook University of North Queensland.

Belperio, A. (1983). Terrigenous sedimentation in the central Great Barrier Reef lagoon: a model from the Burdekin region. BMR Journal of Australian Geology and Geophysics,

8, pp.179-190.

Blanchette, M. and Pearson, R. (2012). Dynamics of habitats and macroinvertebrate assemblages in rivers of the Australian dry tropics. Freshwater Biology, 58(4), pp.742-

757.

Brodie, J., Schroeder, T., Rohde, K., Faithful, J., Masters, B., Dekker, A., Brando, V. and

Maughan, M. (2010). Dispersal of suspended sediments and nutrients in the Great Barrier

Reef lagoon during river-discharge events: conclusions from satellite remote sensing and concurrent flood-plume sampling. Marine and Freshwater Research, 61(6), p.651.

Csanady, G. (1978). Wind effects on surface to bottom fronts. Journal of Geophysical

Research, 83(C9), p.4633.

Chao, S. (1987). Wind-driven motion near inner shelf fronts. Journal of Geophysical

Research, 92(C4), p.3849.

Chao, S. (1988). Wind-Driven Motion of Estuarine Plumes. Journal of Physical

Oceanography, 18(8), pp.1144-1166.

Chen, Z., Gong, W., Cai, H., Chen, Y. and Zhang, H. (2017). Dispersal of the Pearl River plume over continental shelf in summer. Estuarine, Coastal and Shelf Science, 194, pp.252-262.

Chen, Z., Pan, J., Jiang, Y. and Lin, H. (2017). Far-reaching transport of Pearl River plume water by upwelling jet in the northeastern South China Sea. Journal of Marine

Systems, 173, pp.60-69.

Church, J., Andrews, J. and Boland, F. (1985). Tidal currents in the central Great Barrier

Reef. Continental Shelf Research, 4(5), pp.515-531.

Delandmeter, P., Lewis, S., Lambrechts, J., Deleersnijder, E., Legat, V. and Wolanski, E.

(2015). The transport and fate of riverine fine sediment exported to a semi-open system.

Estuarine, Coastal and Shelf Science, 167, pp.336-346.

Devlin, M. and Brodie, J. (2005). Terrestrial discharge into the Great Barrier Reef

Lagoon: nutrient behavior in coastal waters. Marine Pollution Bulletin, 51(1-4), pp.9-22.

Devlin, M., McKinna, L., Álvarez-Romero, J., Petus, C., Abott, B., Harkness, P. and

Brodie, J. (2012). Mapping the pollutants in surface riverine flood plume waters in the

Great Barrier Reef, Australia. Marine Pollution Bulletin, 65(4-9), pp.224-235.

Dong, L., Su, J., Ah Wong, L., Cao, Z. and Chen, J. (2004). Seasonal variation and dynamics of the Pearl River plume. Continental Shelf Research, 24(16), pp.1761-1777.

Eanes, R. and Bettadpur, S., 1995. The CSR 3.0 global ocean tide model. Center for

Space Research, Technical Memorandum, CSR-TM-95-06.

Fielding, CR, Trueman, JD and Alexander, J (2005) Sedimentology of the Modern and

Holocene Burdekin River Delta of North Queensland, Australia – controlled by river output, not by waves and tides. In: Deltas, New and Old. SEPM special Publication no.

83, pp. 467-496.

Fong, D., Geyer, W. and Signell, R. (1997). The wind-forced response on a buoyant coastal current: Observations of the western Gulf of Maine plume. Journal of Marine

Systems, 12(1-4), pp.69-81.

Fong, D. and Geyer, W. (2001). Response of a river plume during an upwelling favorable wind event. Journal of Geophysical Research: Oceans, 106

Franks, P. and Anderson, D. (1992). Alongshore transport of a toxic phytoplankton bloom in a buoyancy current: Alexandrium tamarense in the Gulf of Maine. Marine

Biology, 112(1), pp.153-164. (C1), pp.1067-1084.

García Berdeal, I. (2002). Influence of wind stress and ambient flow on a high discharge river plume. Journal of Geophysical Research, 107(C9).

Geyer, W., Hill, P., Milligan, T. and Traykovski, P. (2000). The structure of the Eel River plume during floods. Continental Shelf Research, 20(16), pp.2067-2093.

Gillibrand, P. and Herzfeld, M. (2016). A mass-conserving advection scheme for offline simulation of scalar transport in coastal ocean models. Ocean Modelling, 101, pp.1-16.

Guo, X. and Valle-Levinson, A. (2007). Tidal effects on estuarine circulation and outflow plume in the Chesapeake Bay. Continental Shelf Research, 27(1), pp.20-42.

Hannachi, A., Jolliffe, I. and Stephenson, D. (2007). Empirical orthogonal functions and related techniques in atmospheric science: A review. International Journal of

Climatology, 27(9), pp.1119-1152.

Herzfeld, M. (2006). An alternative coordinate system for solving finite difference ocean models. Ocean Modelling, 14(3-4), pp.174-196.

Herzfeld, M. (2015). Methods for freshwater riverine input into regional ocean models.

Ocean Modelling, 90, pp.1-15.

Herzfeld, M. and Gillibrand, P. (2015). Active open boundary forcing using dual relaxation time-scales in downscaled ocean models. Ocean Modelling, 89, pp.71-83.

Herzfeld, M., Andrewartha, J., Baird, M., Brinkman, R., Furnas, M., Gillibrand, P.,

Hemer, M., Joehnk, K., Jones, E., McKinnon, D., argvelashvili, N., Mongin, M., Oke, P.,

Rizwi, F., Robson, B., Seaton, S., Skerratt, J., Tonin, H. and Wild-Allen, K. (2016). eReefs Marine Modelling: Final Report, pp.290-310.

Hench, J. and Luettich, R. (2003). Transient Tidal Circulation and Momentum Balances at a Shallow Inlet. Journal of Physical Oceanography, 33(4), pp.913-932.

Hetland, R. (2005). Relating River Plume Structure to Vertical Mixing. Journal of

Physical Oceanography, 35(9), pp.1667-1688.

Horner-Devine, A., Hetland, R. and MacDonald, D. (2015). Mixing and Transport in

Coastal River Plumes. Annual Review of Fluid Mechanics, 47(1), pp.569-594.

Hickey, B., Pietrafesa, L., Jay, D. and Boicourt, W. (1998). The Columbia River Plume

Study: Subtidal variability in the velocity and salinity fields. Journal of Geophysical

Research: Oceans, 103(C5), pp.10339-10368.

Hu, C. and Muller-Karger, F. (2007). Response of sea surface properties to Hurricane

Dennis in the eastern Gulf of Mexico. Geophysical Research Letters, 34(7).

Kaihatu, J., Handler, R., Marmorino, G. and Shay, L. (1998). Empirical Orthogonal

Function Analysis of Ocean Surface Currents Using Complex and Real-Vector Methods*.

Journal of Atmospheric and Oceanic Technology, 15(4), pp.927-941.

Kao, S., Dai, M., Selvaraj, K., Zhai, W., Cai, P., Chen, S., Yang, J., Liu, J., Liu, C. and

Syvitski, J. (2010). Cyclone-driven deep sea injection of freshwater and heat by hyperpycnal flow in the subtropics. Geophysical Research Letters, 37(21).

King, B., Mcallister, F., Wolanski, E., Done, T. and Spagnol, S. (2001). River plume dynamics in the central Great Barrier Reef. Oceanographic Processes of Coral Reefs:

Physical and Biological Links in the Great Barrier Reef, 145-159.

Kingsford, M. and Suthers, I. (1994). Dynamic estuarine plumes and fronts: importance to small fish and plankton in coastal waters of NSW, Australia. Continental Shelf

Research, 14(6), pp.655-672.

Kroon, F., Kuhnert, P., Henderson, B., Wilkinson, S., Kinsey-Henderson, A., Abbott, B.,

Brodie, J. and Turner, R. (2012). River loads of suspended solids, nitrogen, phosphorus and herbicides delivered to the Great Barrier Reef lagoon. Marine Pollution Bulletin,

65(4-9), pp.167-181.

Lentz, S., Guza, R., Elgar, S., Feddersen, F. and Herbers, T. (1999). Momentum balances on the North Carolina inner shelf. Journal of Geophysical Research: Oceans, 104(C8), pp.18205-18226.

Lentz, S. and Helfrich, K. (2002). Buoyant gravity currents along a sloping bottom in a rotating fluid. Journal of Fluid Mechanics, 464.

Lentz, S. (2004). The Response of Buoyant Coastal Plumes to Upwelling-Favorable

Winds*. Journal of Physical Oceanography, 34(11), pp.2458-2469.

Lentz, S. and Largier, J. (2006). The Influence of Wind Forcing on the Chesapeake Bay

Buoyant Coastal Current*. Journal of Physical Oceanography, 36(7), pp.1305-1316.

Lewis, S., Shields, G., Kamber, B. and Lough, J. (2007). A multi-trace element coral record of land-use changes in the Burdekin River catchment, NE Australia.

Palaeogeography, Palaeoclimatology, Palaeoecology, 246(2-4), pp.471-487.

Lorenz, E. N., (1956) Empirical orthogonal functions and statistical weather prediction.

Technical report, Statistical Forecast Project Report 1, Dept. of Meteor., MIT, pp.49

Luketina, D. and Imberger, J. (1989). Turbulence and entrainment in a buoyant surface plume. Journal of Geophysical Research, 94(C9), p.12619.

MacCready, P., Banas, N., Hickey, B., Dever, E. and Liu, Y. (2009). A model study of tide- and wind-induced mixing in the Columbia River Estuary and plume. Continental

Shelf Research, 29(1), pp.278-291.

Margvelashvili, N., Herzfeld, M., Rizwi, F., Mongin, M., Baird, M., Jones, E., Schaffelke,

B., King, E. and Schroeder, T. (2016). Emulator-assisted data assimilation in complex models. Ocean Dynamics, 66(9), pp.1109-1124.

McCabe, R., MacCready, P. and Hickey, B. (2009). Ebb-Tide Dynamics and Spreading of a Large River Plume*. Journal of Physical Oceanography, 39(11), pp.2839-2856

McClimans, T. (1986). Laboratory modeling of dynamic processes in fjords and shelf waters. The Role of Freshwater Outflow in Coastal Marine Ecosystems. pp 67-84

McCulloch, M., Fallon, S., Wyndham, T., Hendy, E., Lough, J. and Barnes, D. (2003).

Coral record of increased sediment flux to the inner Great Barrier Reef since European settlement. Nature, 421(6924), pp.

Mehta, A. (1989). On estuarine cohesive sediment suspension behavior. Journal of

Geophysical Research, 94(C10), p.14303.727-730.

Moffat, C. and Lentz, S. (2012). On the Response of a Buoyant Plume to Downwelling-

Favorable Wind Stress. Journal of Physical Oceanography, 42(7), pp.1083-1098.

Molleri, G., Novo, E. and Kampel, M. (2010). Space-time variability of the Amazon

River plume based on satellite ocean color. Continental Shelf Research, 30(3-4), pp.342-

352.

Monahan, A., Fyfe, J., Ambaum, M., Stephenson, D. and North, G. (2009). Empirical

Orthogonal Functions: The Medium is the Message. Journal of Climate, 22(24), pp.6501-

6514.

Neil, D., Orpin, A., Ridd, P. and Yu, B. (2002). Marine and Freshwater Research, 53(4), p.733.

North, Gerald R., Thomas L. Bell, Robert F. Cahalan, and Fanthune J. Moeng. "Sampling

Errors In The Estimation Of Empirical Orthogonal Functions". Monthly Weather Review

110, no. 7 (1982): 699-706. doi:10.1175/1520-0493(1982)110<0699:seiteo>2.0.co;2.

Oke, P., Brassington, G., Griffin, D. and Schiller, A. (2008). The Bluelink ocean data assimilation system (BODAS). Ocean Modelling, 21(1-2), pp.46-70.

Orton, P. and Jay, D. (2005). Observations at the tidal plume front of a high-volume river outflow. Geophysical Research Letters, 32(11).

Queensland Department of Environment (1997). Conservation Report W03.3. Wave recording program, Townsville Region, 1995–1997. Brisbane: Queensland Government, p.4

Rennie, S., Largier, J. and Lentz, S. (1999). Observations of a pulsed buoyancy current downstream of Chesapeake Bay. Journal of Geophysical Research: Oceans, 104(C8), pp.18227-18240.2.

Reul, N., Quilfen, Y., Chapron, B., Fournier, S., Kudryavtsev, V. and Sabia, R. (2014).

Multisensor observations of the Amazon-Orinoco river plume interactions with hurricanes. Journal of Geophysical Research: Oceans, 119(12), pp.8271-8295.

Robson, B. (2017). Evaluating the eReefs Great Barrier Reef marine model against observed emergent properties. In: 22nd International Congress on Modelling and

Simulation. Hobart: Modelling and Simulation Society of Australia and New Zealand, pp.1976-1982.

Sanders, T. and Garvine, R. (2001). Fresh water delivery to the continental shelf and subsequent mixing: An observational study. Journal of Geophysical Research: Oceans,

106(C11), pp.27087-27101.

Schiller, R., Kourafalou, V., Hogan, P. and Walker, N. (2011). The dynamics of the

Mississippi River plume: Impact of topography, wind and offshore forcing on the fate of plume waters. Journal of Geophysical Research, 116(C6).

Schiller, A., Herzfeld, M., Brinkman, R., Rizwi, F. and Andrewartha, J. (2015). Cross- shelf exchanges between the Coral Sea and the Great Barrier Reef lagoon determined from a regional-scale numerical model. Continental Shelf Research, 109, pp.150-163.

Servain, J. and Legler, D. (1986). Empirical orthogonal function analyses of tropical

Atlantic sea surface temperature and wind stress: 1964–1979. Journal of Geophysical

Research, 91(C12), p.14181.

Stukel, M., Coles, V., Brooks, M. and Hood, R. (2013). Top-down, bottom-up and physical controls on diatom-diazotroph assemblage growth in the Amazon River Plume.

Biogeosciences Discussions, 10(8), pp.13931-13976.

Van Woesik, R., De Vantier, L. and Glazebrook, J. (1995). Effects of Cyclone 'Joy' on nearshore coral communities of the Great Barrier Reef. Marine Ecology Progress Series,

128, pp.261-270.

Wang, H., Yang, Z., Saito, Y., Liu, J., Sun, X. and Wang, Y. (2007). Stepwise decreases of the Huanghe (Yellow River) sediment load (1950–2005): Impacts of climate change and human activities. Global and Planetary Change, 57(3-4), pp.331-354.

Warrick, J., Mertes, L., Washburn, L. and Siegel, D. (2004). Dispersal forcing of southern

California river plumes, based on field and remote sensing observations. Geo-Marine

Letters, 24(1), pp.46-52.

Whitney, M. and Garvine, R. (2005). Wind influence on a coastal buoyant outflow.

Journal of Geophysical Research: Oceans, 110(C3).

Wolanski, E. and Jones, M. (1981). Physical properties of Great Barrier Reef lagoon waters near Townsville. I. Effects of Burdekin River floods. Marine and Freshwater

Research, 32(3), p.305.

Wolanski, E. and Bennett, A. (1983). Continental shelf waves and their influence on the circulation around the Great Barrier Reef. Marine and Freshwater Research, 34(1), p.23.

Wolanski, E. and Van, S. (1983). Mixing of Burdekin River flood waters in the Great

Barrier Reef. Marine and Freshwater Research, 34(1), p.49.

Wolanski, E. and Pickard, G. (1985). Long-term observations of currents on the central

Great Barrier Reef continental shelf. Coral Reefs, 4(1), pp.47-57.

Wolanski, E. (1994). Physical oceanographic processes of the Great Barrier Reef. Boca

Raton: CRC Press, p.194.

Wu, H., Zhu, J., Shen, J. and Wang, H. (2011). Tidal modulation on the Changjiang River plume in summer. Journal of Geophysical Research, 116(C8).

Yankovsky, A. and Chapman, D. (1997). A Simple Theory for the Fate of Buoyant

Coastal Discharges*. Journal of Physical Oceanography, 27(7), pp.1386-1401.

Yuan, J., Miller, R., Powell, R. and Dagg, M. (2004). Storm-induced injection of the

Mississippi River plume into the open Gulf of Mexico. Geophysical Research Letters,

31(9).

Zhao, H., Tang, D. and Wang, D. (2009). Phytoplankton blooms near the Pearl River

Estuary induced by Typhoon Nuri. Journal of Geophysical Research, 114(C12).

Zheng, Q., Clemente-Colón, P., Yan, X., Liu, W. and Huang, N. (2004). Satellite synthetic aperture radar detection of Delaware Bay plumes: Jet-like feature analysis.

Journal of Geophysical Research: Oceans, 109(C3).