Hydrodynamics of shallow mediterranean estuaries, and relevance to some biogeochemical processes affecting Nodularia blooms
Barbara Robson
A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy, Department of Geography and Oceanography, University College, University of New South Wales. September 1999
Abstract This thesis examines interactions between hydrodynamic characteristics of shallow mediterranean estuaries and biogeochemical processes affecting nuisance blooms of Nodularia spumigena in Harvey Estuary, Western Australia between 1978 and 1993.
Two biogeochemical models describing processes affecting Nodularia in Harvey Estuary are presented. The first incorporates most major relevant processes and (with calibration) could be used to model Nodularia in other systems. The second is a greatly simplified model, which allowed an examination of hydrodynamic factors controlling blooms in Harvey Estuary. The results support the view that Nodularia blooms in the estuary were controlled by the relationship between stratification and sediment phosphorus release. Interannual variability of bloom size is related to winter riverflow and can attributed to either the variability of stratification or sediment oxygen demand (which depended on the mass of diatom detritus). However, the model strongly suggests that the former process was the major factor controlling the bloom size. The results of this analytical model indicate that there was no special physical condition attached to the Nodularia blooms in Harvey Estuary apart from the existence of weak ocean exchange (which is common is shallow mediterranean estuaries), and the oxygen flux ratio (defined within) having risen above the critical value. Two- and three-dimensional hydrodynamic models (TRIM and the Princeton Ocean Model, respectively) are applied to two shallow mediterranean estuaries (Harvey Estuary and Tomales Bay, California). The three-dimensional model is driven by riverflow, meteorological conditions, and conditions along the open boundary. The model's reproduction of seasonal and interannual variability of salinity and temperature in these estuaries is shown to be satisfactory, although it overestimates vertical mixing and hence is not adequate to support a detailed coupled physicalbiological model of Nodularia.
rI
II
Ii
i
1
f
~
The effects of varying topography, winds and tides on hydrodynamics of shallow mediterranean estuaries are examined. Despite the relatively linear topography of the basins modelled, significant lateral variability in salinity and temperature is found. Salinity and currents are very sensitive to wind direction. Vertical mixing is predominantly due to tidal shears in Tomales Bay and wind mixing in Harvey Estuary. Salinity stratification is significant between late autumn and early summer.
Acknowledgements Many thanks are due to my supervisor, Dr Clifford Hearn, whose advice, support and thoughtful criticism has been essential throughout the progress of this work.
Thanks also to my co-supervisor Dr Graham Symonds who lent his support and experience in physical oceanography.
I am grateful to Dr George Mellor for making available the Princeton Ocean Model to the academic community.
The Supercomputing Facility at the Australian national University and the Information Technology service at University College, made available the computing facilities required for the hydrodynamic modelling work, and I am particularly grateful for the assistance of Roger Brown, of ANU, for his assistance in optimising my code for vectorisation.
Thanks also to Dr John Hunter for advice on calculating residual currents and to Dr Rod Lukatelich. of BP Oil for discussions on the biology of Nodularia in Harvey Estuary.
Thanks to my fellow research student, Mohamed Ali, for advice with maps and figures, and to my good friends, Stuart Barrow and Jenny Mason-Barrow for their comments on this thesis.
The Centre for Water Research, and the School of Environmental Science at Murdoch University made available observational data for Harvey Estuary.
Western Australia Waters and Rivers Commission provided data on water flow and water quality in Harvey River.
II
The Department of Transport, Western Australia, provided bathymetric charts for Harvey Estuary.
Dr Stephen V. Smith, University of Hawaii, provided observational data for Tamales Bay.
My thanks to Drs Ralph Cheng, Des Lord, and Arthur McComb, for their valuable comments.
An Australian Postgraduate Award (with stipend) and University College supplementary scholarship provided financial support during my research.
My thanks to all staff of the School of Geography and Oceanography, University College, for their support and help.
Not least, thanks to my family and friends for their support and encouragement, without which this work would not have been completed.
ii
For Rob, even so.
111
DECLARATION I hereby declare that this submission is my own work and to the best of my knowledge it contains no material previously published or written by another person, nor material which to a substantial extent has been accepted for the award of another 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 colleagues, with whom I have worked at UNSW or elsewhere, during my candidature, is fully acknowledged. 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.
IV
........
--------------------~~~
.I
Table of Contents Chapter I: Introduction ............................................................................................. 1 1.1 Objectives .................................................................................................................. 1 1.2 Thesis Structure ........................................................................................................ 5
Chapter 2: Study Sites ................................................................................................ 7 2.1 Harvey Estuary ......................................................................................................... 7 2.1.1
Location and topography ................................................................................................ 7
2.1.2
Social and Environmental Context... ............................................................................... 9
2.1.3
Climate ............................................................................................................................ 9
2.1.4
Tides ............................................................................................................................. 10
2.1.5
Hydrology ..................................................................................................................... 11
2.1.6
Eutrophication ............................................................................................................... 14
2.1.7
Sediments ...................................................................................................................... 16
i.1
2.1.7.1 Sediment properties .................................................................................................. 16 2.1.7.2 Sediment processes .................................................................................................. 18 2.1.8
Management. ................................................................................................................ 19
2.1.9
Review of Previous Work on Harvey Estuary .............................................................. 22
2.1.9 .1 Monitoring ............................................................................................................... 22 2.1.9.2 Process Studies ......................................................................................................... 23 2.1.9.3 Modelling ................................................................................................................. 25
2.2 Tomales Bay ............................................................................................................ 32 2.2.1
Location and topography .............................................................................................. 32
2.2.2
Environmental context .................................................................................................. 34
2.2.3
Climate .......................................................................................................................... 34
2.2.4
Tides ............................................................................................................................. 35
2.2.5
Hydrology ..................................................................................................................... 35
2.2.6
Previous Work .............................................................................................................. 36
2.2.6.1 Analysis of non-conservative fluxes of carbon, nitrogen and phosphorus ............... 37 2.2.6.2 Analysis of sediment input and accumulation in the Bay ........................................ 38 2.2.6.3 Analysis of water circulation, currents, and exchange between the Bay and the ocean.
38
2.2.6.4 Analysis of nutrient cycling through the major biotic components of the system .... 39 2.2.6.5 Examination of denitrification and nitrogen fixation within the system ................. .40 2.2.6.6 Modelling of watershed contributions to nutrient fluxes .......................................... 40
v
'jl,
Chapter 3: Modelling dynamics of Nodularia spumigena blooms ..................... .42 3.1 Background ............................................................................................................. 42 3.1.I
Taxonomy and basic physiology ................................................................................... 42
3.1.2
Distribution ................................................................................................................... 44
3.1.3
Marine and estuarine cyanobacterial blooms and nitrogen fixation .............................. 45
3.2 Constructing a biogeochemical model for Nodularia blooms in Harvey Estuary
so 3.2.1
Numerical Differencing Scheme ................................................................................... 52
3.2.2
Components and Processes Represented in the Model ................................................. 52
3.2.2.1 Nodularia .................................................................................................................. 52 3.2.2.1.1
Germination of akinetes .................................................................................. 52
3.2.2.1.2
Growth ............................................................................................................ 53
3.2.2.1.3
Wind-mixing ................................................................................................... 54
3.2.2. l.4
Mortality ......................................................................................................... 54
3.2.2.1.5
Effects of Phosphorus and Nitrogen ............................................................... 55
3.2.2.1.6
Salinity ............................................................................................................ 58
3.2.2.1.7
Temperature .................................................................................................... 64
3.2.2.1.8
Light ............................................................................................................... 67
3.2.2.1.9
Other factors ................................................................................................... 69
3.2.2.2 Sediments and nutrients ........................................................................................... 72 3.2.2.2. l
Binding of nutrients in sediments ................................................................... 72
3.2.2.2.2
Adsorption of phosphorus by suspended sediments ....................................... 73
3.2.2.2.3
Release of phosphorus from bottom sediments .............................................. 73
3.2.2.2.4
Absorption of phosphorus and nitrogen by bottom sediments ........................ 73
3.2.2.2.5
Suspension of free sediments .......................................................................... 74
3.2.2.3 Oxygen ..................................................................................................................... 75 3.2.2.3. I
Sediment oxygen demand ............................................................................... 76
3.2.2.3.2
Photosynthesis ................................................................................................ 77
3.2.2.3.3
Respiration ...................................................................................................... 77
3.2.2.3.4
Surface oxygen fluxes ..................................................................................... 77
3.2.2.4 Diatoms .................................................................................................................... 79 3.2.2.4.1
Growth ............................................................................................................ 79
3.2.2.4.2
Grazing ........................................................................................................... 80
3.2.2.4.3
Mortality of diatoms and zooplankton ............................................................ 81
3.2.2.5 Detritus ..................................................................................................................... 81 3.2.2.5.1
Sinking ............................................................................................................ 81
3.2.2.5.2
Decay .............................................................................................................. 82
3.2.2.6 Parametres ................................................................................................................ 83
Vl
3.3 ······························•······································•···········•··•································•··•··•·········· 84 3.4 Simulating Nodularia blooms in a simple box model ..•....................................... 85 3.4.1
Introduction ................................................................................................................... 85
3.4.2
Akinete germination ...................................................................................................... 89
3.4.2. l Introduction .............................................................................................................. 89 3.4.2.2 Methods .................................................................................................................... 89 3.4.2.3 Results ...................................................................................................................... 90 3.4.3
A simulated Nodularia bloom in a simple one-layer model ......................................... 91
3.4.3.1 Introduction .............................................................................................................. 91 3.4.3.2 Methods .................................................................................................................... 91 3.4.3.3 Results ...................................................................................................................... 92 3.4.3.4 Discussion ................................................................................................................ 98 3.4.3.4. l 3.4.4
Bubbling ......................................................................................................... 99
A Seasonal Nodularia bloom cycle ............................................................................. 101
3.4.4.1 Introduction ............................................................................................................ 101 3.4.4.2 Methods .................................................................................................................. 101 3.4.4.3 Results .................................................................................................................... 103 3.4.4.3.l
Salinity and Temperature .......................................................................... 104
3.4.4.3.2
Light...........................................................................
3.4.4.3.3
Decay ............................................................................................................ 106
..................... 105
3.4.4.4 Forcing senescence in a seasonal Nodularia bloom ............................................... 106
3.5 Closing comments ..•.....•.•..•.....•.........••..••....••.•........••.••.......•.•...............••..•..•........ 107
Chapter 4: Analysis of water quality data ............................................................. 109 4.1 Introduction .•.•.....•.•.....•....•....•.......•.••......•..•........••.•••.....••..........•..•...••.....•..•..•..•... 109 4.2 Relationships between observed variables - Harvey Estuary •.•••.....•..••.......••..• 109 4.3 Relationships between observed variables -Tomales Bay ....•...................••...... 114
4.4 Inter-annual variations ......................................................................................... 117 4.4.l
Stratification and minimum salinity ............................................................................ 119
4.4.2
Stratification and oxygen depletion ............................................................................ 120
4.4.3
Phosphorus release ...................................................................................................... 124
4.4.4
Stratification and chlorophyll a ................................................................................... 125
Chapter 5: An analytical model of interannual variability of Nodularia spumigena blooms
126
5.1 Abstract .................................................................................................................. 126
vii
.i
5.2
Introduction ........................................................................................................... 126
5.3 Phosphorus Release Rate ...•••............................•..••..•...•..•..•..•.•.......•...........•......•. 130 5.4 Time dependence of the Nodularia Bloom .....•......................................•............. 131 5.5 Sediment oxygen demand ...•••...•...............•......................................•....•..•............ 132 5.6 Basic Dynamics of Nodularia .................•..................•...•..••.......•..••..•.................•• 135 5.7 Stratification .•.••........•..••...............•..•.................•..•••......•..•.•.....................•............ 137 5.8 Model of Vertical Mixing ••.••••................•..........•......•..................•.......•...............• 138
'
I [
5.9 Model of stratification of Harvey Estuary •.....•..•..••.••...•..•..................•..••........... 140 5.10
Bottom Oxygen Concentration ........................................................................ 140
5.11
Oxygen flux ratio ......•.•••.•.••.............•.....•............•............•..•..•...••..............•.••..• 141
5.12
Ocean Salt Exchange ........................................................................................ 143
5.13
Integration of dynamic equation and values of growth and decay constants. 143
I
5.14
Determining the oxygen flux ratio A,. ............................................................. 144
5.15
Comparison of predicted and observed Nodularia blooms .......................... 145
5.16
Discussion and Conclusion ............................................................................... 149
Chapter 6: Hydrodynamics of shallow mediterranean estuaries ......................... 152 6.1 Abstract .................................................................................................................. 152 6.2 Introduction ........................................................................................................... 152 6.3 TRIM ...................................................................................................................... 153 6.3.1
Configuration .............................................................................................................. 154
6.3.2
Results ......................................................................................................................... 155
6.4 The Princeton Ocean Model ................................................................................ 155 6.4.1
Configuration .............................................................................................................. 156
6.4.2
Inclusion of River Inflow ............................................................................................ 166
6.4.3
Heat fluxes .................................................................................................................. 169
6.4.4
Residual Circulation .................................................................................................... 173
6.4.4.1 Verification ............................................................................................................ 174 6.4.4.2 "Winter Exercise" runs (Harvey Estuary) .............................................................. 174 6.4.4.3 Multi-year runs (Harvey Estuary and Tomales Bay) .............................................. 184
Vtll i
_j
6.4.5
Investigative Simulations ............................................................................................ 189
6.5 Results •...•...........•......•.....................•..••..••.•......................•..................••................. 191 6.5.1
Normal Conditions ...................................................................................................... 191
6.5.1.1
Harvey estuary ....................................................................................................... 191
6.5.1.2 Tamales Bay .......................................................................................................... 198 6.5 .2
Investigative Simulations ............................................................................................ 202
6.5.2.1
Harvey estuary ....................................................................................................... 202
6.5.2.2 Tamales Bay .......................................................................................................... 209
6.6 Discussion ........•.....••..•....••.•••..•...............••.•..•..•..........•.•....•......••...••...................... 213
Chapter 7: Applicability of the hydrodynamic model in modelling biogeochemical processes in shallow mediterranean estuaries ........................................................ 217 7.1 Introduction ...•.••........•..•.•..•••..••.................••....•.•..........•.•..•.•........•••..••.••..••..•........ 217 7.2 Vertical stratification in Harvey Estuary ..........•................•.•••..•••...................••. 217 7 .2.1
Possible sources of error ............................................................................................. 222
7 .2.1.1 Sigma coordinate pressure gradient errors ............................................................. 222 7.2.1.1.l
Trials ............................................................................................................. 223
7.2.1.2 Turbulent mixing scheme ....................................................................................... 227 7 .2.1.3 Model resolution of high frequency events ............................................................ 228 7.2.1.4 River inflow data .................................................................................................... 228 7 .2.1.5 Representation of bathymetry ................................................................................ 228 7 .2.2
Discussion ................................................................................................................... 229
Chapter 8: Conclusions .......................................................................................... 231 8.1 Requirements for a model of Nodularia in shallow mediterranean estuaries. 231 8.1. l
Physico-chemical factors in bloom formation ............................................................ 231
8.1.2
Hydrodynamic modelling ........................................................................................... 232
8.2 Closing Comments ................................................................................................ 234 8.3
Contribution of the present work ........................................................................ 236
8.4 Directions for further work .................................................................................. 237
IX
....
------~~~~~
f,
·1',
:
I
I Table of Figures
i' 8
I! i'
Figure 2-2 Mean daily rainfall (mm) for Mandurah, 1989-1996
JO
Figure 2-3 Mean minimum and maximum daily temperatures for Mandurah, 1989-1996.
10
I'
Figure 2-1 Map of the Peel Inlet and Harvey Estuary
I
Figure 2-4 Tide records from the Estuary, showing diurnal tides and larger variations due to barometric influences
10
Figure 2-5 Catchment areas of Harvey River and drains feeding Harvey Estuary
12
Figure 2-6 Seasonal and interannual salinity variations in suiface salinity at the southern end of
Harvey Estuary
14
Figure 2-7 Nitrogen to phosphorus ratio (by atom) in Harvey Estuary (average of suiface and bottom observations), 1978-1983. Figure adapted from Lukatelich ( 1987).
16
Figure 2-8 Location of the Dawesville Channel.
21
Figure 2-9 Tamales Bay location and t o p o g r a p h y . - - - - - - - - - - - - - - - - - 32 Figure 2-10 Tamales Bay catchment area. The three major subcatchments are identified as the Olema Creek watershed, the Lagunitas Creek watershed, and the Walker Creek watershed. _ _ _ _ _ 33 Figure 2-11 Mean daily rainfall at Dillon Beach ( 1987-1994)
34
Figure 2-12 Mean daily temperature at Dillon Beach (0511987-0511993)
34
Figure 2-13 Suiface elevaNon in Tamales Bay as calculated from tidal constituents.
35
Figure 3-1. A Nodulariafilamentfrom the Baltic Sea (Scale: 1:1000)
43
Figure 3-2 Conceptual diagram showing major biochemical processes influencing Nodularia blooms
in Harvey Estuary.
51
Figure 3-3 Effect of salinity on Nodularia growth (from Huber, 1984a).
60
Figure 3-4 Effect of salinity on nitrogenase activity (from Huber, 1986b). _ _ _ _ _ _ _ _ _ 61 Figure 3-5 Effect of salinity on akinete germination (from Huber, 1985).
62
Figure 3-6 Relationship between salinity and akinete germination
63
Figure 3-7 Effect of temperature on Nodularia growth (from Huber, 1984a).
65
Figure 3-8 Effect of temperature on akinete germination (from Huber, 1984a)
66
Figure 3-9 Relationship between temperature and akinete germination
67
Figure 3-10 Effect of light on Nodularia growth rate.
68
Figure 3-11 Oxygen dynamics.
76
Figure 3-12 An illustrative diagram of the bloom cycle (from Huber, 1984a).
85
Figure 3-13 Changes in phosphorus concentration in Harvey Estuary water samples under laboratory conditions. (Figure from McComb and Lukatelich, 1995). Figure 3-14 Predicted Nodularia concentrations (in phosphorus-equivalent mg
87 1 L' )
over a 200-day
simulation run. A: Under normal conditions. B: if it is assumed that all available akinetes within the top 5 cm of sediment will germinate. C: If it is assumed that light-limitation of Nodularia growth does not occur. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 90
x
'
i!
'Jc ••
Figure 3-15 Simulated Nodularia population (in phosphorus-equivalent mg L 1) over a 150-day trial. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 93 Figure 3-16 Simulated concentration of soluble reactive phosphorus (mgL 1) over a 150-day trial. _94 Figure 3-17 Simulated concentrations of detritus from Nodularia over a 150-day trial. _____ 95 Figure 3-18 Simulated bacterial population during the trial.
96
Figure 3-19 Simulated dissolved oxygen concentration (mgL 1).
97
Figure 3-20 Sediment phosphorus stores (in mgL·') during the J50-day simulation.
98
Figure 3-21 Response of theoretical saturation oxygen concentration to salinity when temperature is
15°C and to temperature when salinity is 15. ____________________ IOI Figure 3-22 PredictedNodulariapopulation over the course of the run.
103
Figure 3-23 A comparison of the results of the trial described in section 0 (A) with results obtained when the Nodularia growth rate is set to zero when the salinity exceeds 29 (B). _ _ _ _ _ _ _ 107 Figure 4-1 Observations of temperature and chlorophyll a concentrations near the surface at a site in
the middle of Harvey Estuary. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 112 Figure 4-2 Observations of organic nitrogen and nitrates near the suiface at a site near the middle of
Harvey Estuary. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 114 Figure 4-3 Minimum salinity and maximum chlorophyll a concentration observed during each
calendar year at site 12 in Toma/es Bay. ______________________ 117 Figure 4-4 Relationship between minimum salinity and maximum chlorophyll a concentration _118 Figure 4-5 Observed relationship between minimum salinity and spring stratification in Harvey
Estuary _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ll9 Figure 4-6 Observed dissolved oxygen concentration at the bottom of the water column at a site in the middle of Harvey Estuary against vertical density difference (kg m3).
_ _ _ _ _ _ _ _ _ _ 121
Figure 4-7 Observed oxygen deficit (mgL· 1) against observed difference between surface and bottom density (kg3m" 3).
122
Figure 4-8 Bottom oxygen concentration predicted from surface oxygen concentration and vertical
density difference against observed bottom dissolved oxygen concentration (mgL. 1) at a site in the middle of Harvey Estuary.
123
Figure 4-9 Relationship between observed deoxygenation in Harvey Estuary bottom water and spring stratification in the Estuary. The dashed line shows a linear best fit. _ _ _ _ _ _ _ _ _ _ _ 124 Figure 4-10 Spring stratification in Harvey Estuary against observed relationship between maximum chlorophyll a concentration. The dashed line shows a linear fit.
125
Figure 5-1 Mass of chlorophyll a at peak of spring bloom of Nodularia plotted against minimum salinity s 0 of Harvey Estuary for preceding winter.
127
Figure 5-2 Series of vertical profiles of salinity and oxygen at a station at centre of Harvey Estuary. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 129 Figure 5-3 Vertical profiles of salinity (upper boxes) and oxygen (lower boxes) at the southern (left hand boxes) and northern end (right hand boxes) of Harvey Estuary on three days in the spring of 1989. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 130
XI
Figure 5-4 The time dependence of the mass of the Nodularia bloom for the wet years 1989 lfull line) and 1984 (dashed line) compared with the predictions of the model (dotted line)
146
Figure 5-5 Repeat of Figure 5-1 showing the peak mass of chlorophyll a at against minimum winter salinity s0•_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 147 Figure 5-6 Variation of the model prediction shown in Figure 5-5 with the value of the oxygen flux ratio A1: (a) A1=1.5, (b) A1 = 2, (c) A1 = 5, (d) A1 = 50. Figure 5-5 corresponds to A1 = 2.6. _ _ 148 Figure 6-2 Harvey Estuary map with grid overlain and approximate location of sites shown _ _ 158 Figure 6-3 Toma/es Bay map with grid overlain and approximate location of sites shown
158
Figure 6-4 Harvey estuary inputs, showing a two-year subset of available data: a) wind speed and direction b) river flow (Harvey River) c) air temperature at Perth Airport d) dew point temperature e) cloud cover f) rainfall
162
Figure 6-5 Tamales Bay inputs, showing a two-year subset of available data: a) wind speed and direction b) river flow c) air temperature at Perth Airport d) rainfall
163
Figure 6-6 Derived inputs for Harvey Estuary: a) Evaporation (mmld) b) Short-wave (solar) radiation c) Long-wave radiation input d) Long-wave radiation output e) Sensible heatflux _ _ _ _ _ _ 164 Figure 6-7 Derived inputs for Tamales Bay: a) Evaporation (mm!d) b) Short-wave (solar) radiation c) Long-wave radiation inputd) Long-wave radiation output e) Sensible heatflux _ _ _ _ _ _ _ 165 Figure 6-8 Filter response for residual current calculations.
173
Figure 6-9 Horizantal slice of salinity 0.1 mfrom the suiface on 28July1983 (2:24pm), as predicted by the model.
174
Figure 6-10 Horizantal slice of salinity field on 27 July 1983 (9:30am), interpolated from data at the 35 sites of Lukatelich (1987).
175
Figure 6-11 Vertical cross-section of salinity on 28 July 1983 (2:20pm) as predicted by the model.175 Figure 6-12 Vertical cross-section of salinity on 27July1983 (9:30am), interpolated from the data of Lukatelich (1987).
176
Figure 6-13 Suiface elevation at 2:20pm, 28 July 1983, as predicted by the model.
176
Figure 6-14 Salinity at JO:OOam on 28 July 1983, as predicted by the model.
177
Figure 6-15 Salinity at 8:30am on 28 July 1983, as derived from the data of Lukatelich (1987). _177 Figure 6-16 Vertical slice of salinity at 10:00am on 28 July 1983, as predicted by the model. __ 178 Figure 6-17 Vertical slice of salinity at 8:30am on 28 July 1983, as derived from the data of Lukatelich (1987).
178
Figure 6-18 Salinity field from run b (flat bathymetry) at !Oam on 27 July 1983.
179
Figure 6-19 Vertical slice of salinity from model run bat 10am on 27 July 1983.
179
Figure 6-20 Vertical slice of salinity from model run bat 10am on 28 July 1983.
180
Figure 6-21 Vertical slice of salinity from model run (b) at 10am on 29 July 1983. _ _ _ _ _ 180
I
Figure 6-22 Vertical slice of salinity from model run (b) at 4pm on 29 July 1983.
181
'ii
Figure 6-23 Vertical slice of salinity at 9:30am, derived from the observations of Lukatelich (1987). -------------------------------~181
Xll
I
Ii
ii
Figure 6-24 Vertical slice of salinity at 3:30pm, derived from the observations of Lukatelich ( 1987).
Figure 6-25 Vertical slice of salinity at 9:30am, derived from the observations of Lukatelich ( 1987).
Figure 6-26 Vertical slice of salinity at !Oam on 31 July 1983, as predicted by model run b. _ _ 183 Figure 6-27 Vertical slice of salinity at !Oam on 31July1983, as predicted by model run c. _ _ 184 Figure 6-28 Harvey Estuary salinity at site A (1982 - 1993) as observed (crosses) and as predicted by the model (solid line). Figure 6-29 Tamales Bay salinity at site A ( 1988 - 1993) as observed (crosses) and as predicted by the model (solid line). Figure 6-30 Harvey Estuary temperature at site A ( 1982 - 1993) as observed (crosses) and as predicted by the model (solid line). _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 187 Figure 6-31 Tamales Bay temperature at site A (1988-1993) as observed (crosses) and as predicted by the model (solid l i n e ) . - - - - - - - - - - - - - - - - - - - - - - - - - - 187 Figure 6-32 Harvey Estuary longitudinal transects. Observations are shown as astricts, model predictions as solid lines. Surface values are in blue, bottom values in red. _ _ _ _ _ _ _ __ 188 Figure 6-33 Tamales Bay longitudinal transects. Observations are shown as astricts, model predictions as solid lines. Surface values are in blue, bottom values in red. _ _ _ _ _ _ _ __ 189
I::,:
Figure 6-34 Harvey Estuary horizontal and vertical cross-sections, 9th May 1984. Vertically averaged residual currents are overlain as arrows. - - - - - - - - - - - - - - - - - 192
:11:.
Figure 6-35 Harvey Estuary horizantal and vertical cross-sections, 26th September 1984. Vertically averaged residual currents are overlain as arrows. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __
193
Figure 6-36 Harvey Estuary horizontal and vertical cross-sections, 24th March 1987. Vertically
,,i
averaged residual currents are overlain as arrows. - - - - - - - - - - - - - - - - - 194
Figure 6-37 Harvey Estuary: a) Wind speed; b) Depth-averaged vertical mixing coefficient; c) Density difference (bottom - surface). _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ 195 Figure 6-38 Harvey Estuary horizontal cross-sections under normal conditions, on two days in
March, 1983. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __ 197
I
i'I 1',, ' ·i 11
Figure 6-39 Tamales Bay horizantal and vertical cross-sections, 4th December 1988. _ _ _ __ 199 Figure 6-40 Tamales Bay horizantal and vertical cross-sections JO August 1988. Vertically averaged residual currents are overlain as arrows. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 200
Figure 6-41 Tamales Bay: a) Wind speed; b) Depth-averaged vertical mixing coefficient; c) Density difference (bottom - surface). _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 201 Figure 6-42 Harvey Estuary: surface salinity at site A as observed and as predicted by the model 203 Figure 6-43 Harvey Estuary: surface temperature at site A as observed (crosses) and as predicted by the model (solid line): a)normal conditions (run I) b) no wind forcing (run 2) c) reversed wind direction (run 3) d) no tidal forcing (run 4) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 204 Figure 6-44 Harvey Estuary density difference (channel density minus estuary density): a) normal 205
conditions (run 1) b) no wind forcing (run 2)
xiii
ii·
Figure 6-45 Harvey Estuary horizontal and vertical (lateral) cross-sections, 29th March 1983, with wind directions reversed. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 207 Figure 6-46 Harvey Estuary horizontal cross-sections, 28th March 1985; a) under normal conditions; b) with wind directions reversed. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 208 Figure 6-47 Tamales Bay: swface salinity at site A as observed (crosses) and as predicted by the model (solid line): a)normal conditions (run 5) b) no tidal forcing (run 6) c) twice actual length (run 7) d) one quarter actual length (run 8) e) density gradient artificially prevented from becoming negative (run 9) f) half actual depth (run 10)
210
Figure 6-48 Tamales Bay suiface temperature at site A as observed (crosses) and as predicted by the model (solid line): a)normal conditions (run 5) b) no tidal forcing (run 6) c) twice actual length (run 7) d) one quarter actual length (run 8) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 211 Figure 6-49 Toma/es Bay density difference (channel density minus estuary density): a) normal conditions (run 5)
212
Figure 7-1 Relationship between observed and modelled stratification at a site in the middle of Harvey Estuary _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 219 Figure 7-2 Stratification in Harvey Estuary (September to November, 1989-1993). This shows the proportion of records for which the difference between density of bottom water and density of suiface water at a site in the middle of the basin was less than the value given on the X-axis (in kgm·'). The black line represents the (approximately weekly) observations, the blue line shows model results (with daily output).
220
Figure 7-3 Spring stratification in Harvey Estuary (September to November, 1989-1993). This shows the proportion of records for which the difference between density of bottom water and density of suiface water at a site in the middle of the basin was less than the value given on the X-axis (in kgm-1). The black line represents the (approximately weekly) observations, the blue line shows model results (with daily output).
221
Figure 7-4 Error in modelled difference between suiface and bottom water density.
222
'1', '
:·1
Figure 7-5 Stratification in Harvey Estuary, June 1989 -June 1993. This shows the proportion of records for which the difference between density of bottom water and density of suiface water at a site
,,
"',
in the middle of the basin exceeded the value given on the X-axis. The black line represents the (approximately weekly) observations, and coloured lines show modelled results (with daily output). _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 225 Figure 7-6 Spring stratification in Harvey Estuary, (September to November, 1989-1993). This shows the proportion of records for which the difference between density of bottom water and density of suiface water at a site in the middle of the basin exceeded the value given on the X-axis. The black line represents the (approximately weekly) observations, and coloured lines show modelled results (with daily output).
l:i
226
Figure 7-7 Variations in suiface salinity at a site near the head of Harvey Estuary. Observations are shown as crosses, results for a run with uniform flat bathymetry shown as a line. _______ 227
XIV
Introduction
Chapter 1: 1.1
Introduction
Objectives
Over recent decades, improved understanding of the hydrodynamics of coastal and ocean systems, facilitated by increasing computer power, has allowed the development of increasingly sophisticated hydrodynamic models. Models such as the widely used Princeton Ocean Model, or POM (Blumberg and Mellor, 1987), TRIM (Cheng and Casulli, 1993), DieCAST (Dietrich and Ko, 1994), SPECIES (Hearn, 1996) and others allow fully three-dimensional simulations of baroclinic systems to be conducted over relatively long intervals. While our ability to model physical aspects of aquatic systems is now in many cases very good, ecological systems are more complex, and ecological models are not yet as far advanced. Many good, predictive ecological models exist, but few can be generalised beyond the system for which they were developed without considerable calibration (J¢rgenson, 1994). Nevertheless, there is a clear need to develop coupled physical-biological models to facilitate the identification and management of vulnerable coastal systems. The rapid growth of the human population, together with increasing demands on our resources, is placing unprecedented pressure on aquatic ecosystems around the world, and the effects of this pressure depend on the way hydrodynamic and biogeochemical processes interact in these systems. While the open ocean is insulated by distance ? and size from some of the more immediate impacts, shallow coastal ecosystems near cities and towns have proven more susceptible. One of the most common environmental problems affecting coastal and inland aquatic systems is eutrophication (nutrient enrichment, usually caused by increased phosphorus and nitrogen loading due to urbanisation and rising use of fertilisers) (JS'lrgensen, 1994). Nutrient pollution of waterways can result in a variety of problems, the most evident of which is usually nuisance blooms of algae or cyanobacteria.
1
Introduction
One coastal system in which eutrophication has created severe problems is Harvey Estuary, a shallow lagoon in southwestern Australia. For many years (1974-1993), serious and extended nuisance blooms of the cyanobacteria, Nodularia spurnigena occurred almost annually in this estuary (e.g. McComb et al., 1981; Huber, 1984a; EPA, 1988). During the period 1978 to 1993, Nodularia blooms failed to occur in only three years - 1979, 1987 and 1990. Each of these followed an unusually dry winter, with low riverflow and consequently low phosphorns loads. In calm weather during blooms, Nodularia filaments floated to the surface, where
they formed a thick (up to 60 cm) scum and often drifted to the shore, where they decayed, producing a nauseating odour that affected residents in nearby areas (EPA, 1988). When the blooms eventually collapsed, decay in the body of the water produced the same problems. The blooms radically altered the ecology of the estuary (Rose, 1994), affecting every component from phytoplankton and macroalgae, to invertebrates, fish and bird life. Nodularia blooms greatly reduced light penetration (disadvantaging benthic
seagrasses, macroalgae and phytoplankton). Seagrasses Ruppia and Halophila are no longer found in the Estuary, probably primarily as a result of this competition for light from phytoplankton and epiphytic algae (Hillman, 1985). The toxicity of Nodularia was first noted by Francis (1878) in his observations of a bloom in a South Australian lake. Nodularin is secreted by Nodularia, accumulates in molluscs and other shellfish (Lukatelich and McComb, 1986) and has caused stock deaths in other parts of the world. At high concentrations, nodularin is also considered a potential human health-risk. Nodularia are unpalatable and possibly toxic to many zooplankton and invertebrates
and resulted in reduced diversity and invertebrate numbers on at least a seasonal basis (Lukatelich and McComb, 1986; Rose, 1994). Among other effects, the blooms contributed greatly to the nitrogen budget of Harvey Estuary (Huber, 1986a), altered 2
Introduction
the oxygenation status of the water (occasionally causing localised fish kills due to deoxygenation as the blooms collapsed), and caused a worrisome accumulation of the toxin, nodularin, in molluscs in the system (Falconer et al., 1992). Nodularin, a toxin from Nodularia blooms has been associated with animal deaths in other systems (Francis, 1878; Edler et al., 1985; Carmichael, 1994) and may represent a health risk to humans (Falconer et al., 1992). The aesthetic problems caused by the blooms greatly reduced the otherwise significant recreational value of the Estuary, inconvenienced residents of nearby areas and reduced property values (EPA, 1992). Further, it greatly diminished the commercial and recreational fishery in the Estuary (although much of this activity moved to the Peel Inlet) (Lenanton et al., 1985). A great deal of effort and expenditure was put into understanding and controlling this phenomenon, culminating in the construction of a channel to connect the Estuary directly to the ocean in 1994 (EPA, 1992; Bradby, 1997). The Dawesville Channel allowed increased tidal flushing and rendered the Estuary considerably more marine (McComb and Lukatelich, 1995). To date, this measure has been effective in -~
eliminating the blooms in the estuarine basin, however some important hypotheses regarding the relationship between Nodularia bloom dynamics and physical processes in the estuary (Lukatelich, 1987; McComb and Lukatelich, 1989) have remained untested. Nodularia, along with Aphanizomenon (Lehtimaki et al., 1997) and the non-
heterocystous Trichodesium, is unusual in being a nitrogen-fixing cyanobacterial taxa known to cause blooms in brackish or marine waters (Sellner, 1997), and Nodularia blooms have been observed in only a few estuarine systems around the
world. It is not clear why such blooms are not more widespread in eutrophic coastal and estuarine systems (Paerl et al., 1987; Paerl, 1990; Howarth and Marino, 1990; Howarth et al., 1988a, 1993, 1999; Marino et al., 1990).
3
~
.__
Introduction
This thesis focuses on shallow mediterranean estuaries (i.e., shallow estuaries that exist in areas with warm, dry summer and mild, wet winters (Hearn, 1995; Hearn et al., 1996; Largier et al., 1996; Largier et al., 1997; Hearn, 1998a)). Estuaries of this type are likely to alternate seasonally between classical (basin fresher than the channel) and inverse (basin hypersaline, relative to the channel) (Largier et al., 1996; Largier et al., 1997). Harvey Estuary is taken as an example of such an estuary that has had severe eutrophication problems in the past, while Tamales Bay, California, is an example of a relatively undisturbed estuary of this type. For both, extensive physical and biological data sets are available. In the chapters that follow, I will seek to answer the following questions: I.
What physical properties of shallow mediterranean estuaries are important in controlling their hydrodynamics?
2.
To what extent did hydrodynamics control Nodularia blooms in Harvey Estuary?
3.
What was unusual about Harvey Estuary, that caused nuisance blooms of a diazotrophic cyanobacterium to occur in this coastal ecosystem, but few others?
4.
Are the physical models presently in wide use adequate to support predictive ecological modelling of shallow estuarine systems? The Princeton Ocean Model (Mellor and Yamada, 1982) is taken as a test case in the present work.
5.
What are the requirements for a complete physical-biological model for Nodularia in shallow mediterranean estuaries (i.e. what would such a model
need to include)?
4
Introduction
1.2
Thesis Structure
This thesis is arranged in eight chapters. In this chapter (Chapter One), the background is set and the objectives of the work
described. In the second chapter, two shallow estuaries are introduced and compared; Harvey
Estuary in south-western Australia, and Tomales Bay, California. In Chapter Three, the behaviour, distribution and major physico-chemical factors affecting the bloom-forming cyanobacterial species, Nodularia spumigena are discussed. A preliminary model of the major processes believed to affect Nodularia in Harvey Estuary is presented, and its performance in reproducing key aspects of a typical Nodularia bloom cycle is examined. Observational data from Harvey Estuary is analysed and discussed in Chapter Four, in light of the host of previous work in this area (e.g. Lukatelich, 1987; McComb and Lukatelich, 1990) and with a view to verifying current understanding of the key factors controlling bloom formation in Harvey Estuary. The results are used to produce a simpler model, the implementation and application of which is described in Chapter Five. Chapter Six examines in more detail the hydrodynamics of the two systems (Harvey Estuary and Tomales Bay) and, more generally, of shallow mediterranean estuaries. The development and implementation of a sophisticated three-dimensional hydrodynamic model based on the widely-used Princeton Ocean Model (Blumberg and Mellor, 1987) is described. This model is then used to examine the effects of factors such as basin length and depth and wind speed and direction on seasonal variations in estuarine conditions. The results are discussed in the light of their implications for Harvey Estuary. The utility of this model for the examination of nutrient cycling and bloom formation in Harvey Estuary is considered in Chapter Seven. Possible directions for future work are discussed.
5
Introduction
Chapter 2:
Study Sites
2. 1 Harvey Estuary 2.1.1 Location and topography The Peel Inlet I Harvey Estuary system (illustrated in Figure 2-1) is located in southwestern Australia at 32°36' S, 115°42' E (70 km south of Perth). This study focuses on Harvey Estuary between 1978 and 1994. Harvey Estuary is a shallow, bar-built Estuary lying along a depression between the Spearwood and Bassendean dune systems of the Swan Coastal Plain (Birch, 1980). The Estuary covers an area of 53 km2 at low water and has an average depth of approximately one metre. In these respects, Harvey Estuary is similar to other estuaries in the region, including the Swan River Estuary in Perth, Hardy Inlet near Augusta, Nornalup Inlet, Wilson Inlet and Leschenault Inlet, all of which are among the 80 bar-built estuaries along the south-western coast of Australia (Hearn, 1998a). Until 1994, Harvey Estuary was connected to the ocean via Peel Inlet and Mandurah Channel (Figure 2-1). Until the construction of training walls in 1967, the entrance to Mandurah Channel was periodically blocked completely for the months between major winter storms (Hodgkin, 1986). Marginal shallows (with water less than 0.5 m deep) account for 14% of the area of Harvey Estuary (Hodgkin, 1986), and much of this area is covered by mangrove vegetation.
7
Introduction
J
1i ''"
,1J ·~
1 ·i
'l
'
l' ~
:I
Figure 2-1 Map of the Peel Inlet and Harvey Estuary (from Lord & Associates, 1998}. Note that the Dawesvil/e Channel was constructed in November 1994, after the time-period of primary interest for the present study.
~
)
~
'•
,fl ·~
,·,~
8
Introduction
2.1 .2 Social and Environmental Context The Peel-Harvey Estuarine System is a major drought refuge and breeding ground for many species of birds, and salt marshes along its southern shores provide habitat for these. The Estuary's presence enhances the worth of the nearby land, including canal estate developments, and the site is much valued for its active and passive recreational utility, including sightseeing, water-skiing, boating, crabbing and recreational fishing. The system also supports a sizeable commercial fishery (Lenanton et al., 1985; McComb and Lukatelich, 1995).
2.1.3 Climate The Estuary is located in the southern hemisphere sub-tropical zone, and the climate in the region is of the "mediterranean" type, with long, hot, dry summers and mild winters (Figure 2-3). Precipitation is extremely seasonal; over 70% of rainfall occurs between the (winter) months of May to August (Figure 2-2). Most fresh water inflow to the Estuary also occurs during this period. Another feature of summers in the region is the strong daily sea-breeze cycle in summer. During this season, easterly winds prevail, interrupted by the cooler southwesterly sea breeze each afternoon. Atmospheric patterns in winter are more variable, and characterised by a series of storms associated with low pressure systems, which bring regular cold fronts (often with gale strength westerly winds) from the Indian Ocean, interspersed with somewhat calmer conditions as atmospheric pressure rises. Completely calm wind conditions, particularly in winter, are uncommon.
9
Introduction
15b E f()Q E
50
J F MAM.J J A:SQN D
Figure 2-3 Mean minimum and maximum daily temperatures for Mandurah, 1989-1996.
Figure 2-2 Mean monthly rainfall (mm) for Mandurah, 1989-1996
(Data from the Australian Bureau of Meteorology).
2.1.4 Tides Astronomic tides in the ocean adjacent to the estuary have a predominantly diurnal signal with a range of 0.2 to 0.9 m (Lord and Associates, 1998), however tidal exchange between the coastal ocean and Harvey Estuary is limited. Until late 1994, Harvey Estuary interacted with the ocean only via a narrow (50 m wide and about 3 m deep) connection to the Peel Inlet, which in turn was linked to the ocean by the relatively long (5 km) and narrow (200 m) Mandurah Channel (Figure 2-1).
20·~-~-~~~~-~-~--
10 Cl
~
0
-40~-~-~-~~-~-~--
14/1 0 21 /10 28/10 04/11 11 /11 18/11 25/11 02/12 1990
Figure 2-4 Water level records from the Estuary, showing diurnal tides and larger variations due to barometric influences (AHO= Australian Height Datum).
10
Introduction
Exchange with the ocean was further limited by a sand bar and a shallow tidal delta, particularly before 1987, when dredging of the Mandurah and Sticks Channels improved exchange. As a consequence, the tidal signal due to astronomical forces was further attenuated to around 10% of the ocean tide in the region, producing a daily tide with a range of less than 10 cm (Black and Rosher, 1980. High tide levels in Harvey Estuary lagged behind those in Peel Inlet by approximately 3 hours, due to the Harvey's indirect connection to the ocean (Lord and Associates, 1998). Observations of water depth in the system show larger (approximately 50 cm (Hodgkin and Di Lollo, 1958)) variations with a period of five to twenty days (Figure 2-4). These are caused by changes in ocean level due to barometric pressure changes as summer high pressure systems and winter lows cross the coast (Black and Rosher, 1980). Barometric tides are especially large on the south-west coast of Australia, and these are not attenuated by the Mandurah Channel (Lord and Associates, 1998). Wind effects and more episodic weather events are also significant: the barometric effects of a 25-year storm are sufficient to produce a 0.6 m rise in water surface level. (Eliot, 1993). The opening of the Dawesville Channel in 1994 increased the mean tidal range in Harvey Estuary to around 62% of that in the adjacent ocean (Lord and Associates, 1998), however the impact of the Channel is beyond the scope of the present study.
2.1.5 Hydrology The largest single source of fresh water to Harvey Estuary is Harvey River (Figure 2-5). Mayfields Drain contributes approximately 13% of total inflow, and others (including the South Coolup and Mealup drains) together contribute less than one percent (Black and Rosher, 1980).
11
Introduction
CATCHMENTS HaNey Rive1 Mayfield Drain Other Agricultural drains
0
Skm
Figure 2-5 Catchment areas of Harvey River and drains feeding Harvey Estuary (Rivers feeding Peel Inlet are not shown).
Although Harvey River has been dammed for many years (Hodgkin et al. , 1981), this is not believed to have greatly influenced total fresh water inflow to the Estuary, as the water lost from the upper catchment area has been replaced by drainage from the agricultural plain (Black and Rosher, 1980). The Harvey River and drains flow from the largely agricultural catchment area on the coastal plain (see Figure 2-5). The response of the River and drains to significant rainfall events is rapid, with peak flow observed within 24 hours of heavy rain (Hodgkin, 1986). Over the course of an average year between 1979 and 1993, the
12
Introduction
total volume of fresh water flowing into the Estuary from the River and drains was about 5.8 times the volume of the Estuary itself (calculated from data supplied by the Water and Rivers Commission). Between 1977 and 1986, mean flow from the Harvey catchment was 221 m 3 , with a standard deviation of 35% (McComb and Lukatelich, 1995). Fresh water entering Peel Inlet also influences conditions in Harvey Estuary, although to a lesser extent. Two river systems (the Murray and Serpentine rivers) and a number of drains feed into the Peel, from a somewhat more varied catchment, which includes both agricultural coastal plain land similar to the catchment of the Harvey River and forested areas (Hodgkin et al., 1981). Because the rivers entering the Peel-Harvey System drain different subcatchments (Figure 2-5), storm peaks and year-to-year variations in total flow may differ for the Peel Inlet and Harvey Estuary. Direct precipitation, too, makes a significant contribution to the Estuary's fresh water budget (supplying around 15% to 23% of the total), while groundwater contributes less than 1% (Black and Rosher, 1980). Evaporation rates in summer are high, with rates of 8 to IO mm d. 1 not unusual in the area, and total annual evaporation has been estimated as 1.4 m, which accounts for more than the volume of the Estuary (Black and Rosher, 1980). Because of the small tidal range and limited exchange with the ocean, the flushing time of the Estuary before the construction of the Dawesville Channelwas long; Mcllveen (1995) estimated summer and winter flushing times for Harvey Estuary as 27 and 22 days respectively. This slow flushing rate, combined with the strong seasonal variation in meteorological conditions and the Estuary's high surface-area to volume ratio produced extreme seasonal variations in physical conditions in the Estuary (Hodgkin et al., 1981; Black and Rosher, 1980). With high evaporation rates in summer, the estuary became hypersaline (salinity in the main body of the water typically reaching a peak of 40 to over 501), while minimum winter salinities often fall below 5 as the estuary is flushed with fresh water. Summer temperatures of circa 25°C and winter temperatures of 12°C are typical.
1
Note that all salinities in this thesis are given in "practical salinity units", a pure
ratio approximately equivalent to parts per thousand by weight.
13
I,'
Introduction
Interannual variations are also considerable, as can be seen in Figure 2-6. During a dry winter (such as occurred in 1986), observed salinity in the estuary may not fall below 10, while maximum salinity the following summer is likely to be higher than average.
50 40 -\
~30
.,.., .....~
t
(rj 20 VJ
10 0
1985
1986
1987
1988
1989
Figure 2-6 Seasonal and interannual salinity variations in surface salinity at the southern end of Harvey Estuary {from data provided by the Murdoch University School of Environmental Science).
2.1.6 Eutrophication i I
The soils characteristic of the Harvey River subcatchment on the Swan Coastal Plain include deep, grey, podzolic sands and shallower (30-50 cm) sand over clay, with smaller regions of other soil complexes (Lukatelich, 1987). These soils are not naturally highly fertile, so agricultural development in the area over the last sixty years has been dependant on widespread application of phosphatic fertilisers (Birch, 1980). Coastal Plain sands are very poor at retaining the fertilisers applied, so large quantities of nitrogen and phosphorus are washed into the Estuary (Hodgkin et al., 1981). The soils in the forested portion of the Peel Inlet's catchment area are less sandy and more effectively retain nutrients. Hence, the concentration of phosphorus in the Harvey River and associated drains tends to be higher than in the rivers entering the Peel (Hodgkin et al., 1981). Point sources, including stock holding yards and a
14
~I
Introduction
piggery in the Serpentine Catchment (McComb and Lukatelich, 1995), have also contributed to the total nutrient load of the system. The winter influx of water from the Harvey River and storm-water drains brings with it an estimated 80 tonnes of phosphorus (Black and Rosher, 1980), around 80% of which is present in the form of filterable reactive phosphate (FRP), which is readily available to phytoplankton and algae (McComb and Lukatelich, 1995). Increasing urbanisation in recent years has provided another source of nutrients to the system. Some indication of the extent of nutrient loading from the Harvey River is given in Table 2-1. Table 2-1 Discharge and nutrient loading from the Harvey catchment (adapted from a table given by McComb and Humphries (1992)).
.
Min. Max. Mean Std. Dev.
DischargeJoad (billion litres per 'vear)
86 370 225 35%
Tota[ phosphorus. [oad (tonnes per year)
25 133 82
Total nitrogen )oad ·.(tonnes per year)
138 1,115 430
Although a hydrological and chemical survey in the late 1940's did not find unusually high nutrient levels, by the late 1960' s, signs of eutrophication were becoming obvious (Hodgkin, 1986). Initially, the major problem in the system was the accumulation of macroalgae (particularly Cladophora montagneana prior to 1979, Chaetomorpha linum more recently, and Ulva lactuca and Enteromorpha intestinalis since the mid-1980's), although this occurred to a lesser extent in Harvey Estuary than in the Peel Inlet. From the late 1970' s, an even more serious symptom of eutrophication was evident, and this time more strongly in Harvey Estuary. In 1973, 1974 (Eliot, 1993) and almost every year between 1978 and 1994, large blooms of the toxic (Francis, 1878; Runnegar et al., 1988; Bolch et al., 1999; Steffensen et al., 1999) cyanobacteria, Nodularia spumigena occurred in late spring and early summer.
The high ratio of available nitrogen to available phosphorus in the system (Figure 2- 7) suggested that phosphorus was usually the primary limiting nutrient, even
before Nodularia blooms were observed (Hornberger and Spear, 1980). Laboratory
15
Introduction
determinations of critical phosphorus concentrations (Gordon et al., 1981) and tissue nutrient concentrations in alga samples from the system (Lukatelich, 1987) confirmed this. A large proportion (up to 40% according to Humphries et al., 1981) of the total phosphorus load entering the system each year was trapped in the estuarine sediments, with the balance flushed to the ocean. This accumulated phosphorus might have sustained blooms in future years even if external phosphorus loading was considerably reduced, had other management measures not been taken. 10110 .
~~~M~n
ll .
$1lllll Jl'MA!iW.d(ll!li>J!'l!tAW~!>~ J~M#i!WMlllllloll'-#llU!lllll!I• ~fl!IAM!llliS
19111
111711
191111
1W:f
~H:2:
1'9il:l
Figure 2-7 Nitrogen to phosphorus ratio (by atom) in Harvey Estuary (average of surface and bottom observations), 1978-1983. Figure adapted from Lukatelich (1987).
2.1.7 Sediments 2. 1. 7. 1 Sediment properties The top few centimetres of the sediments of the central basin of Harvey Estuary are fine dark silty muds with a layer of decaying organic matter (mainly phytoplankton
.
and invertebrate faecal material), and are high in nutrients (including phosphorus). The sediments in shallower regions are also predominantly silty mud, but contain less organic material and lower concentrations of nutrients. Sediments along the eastern edge of the Estuary are sandy (Lukatelich, 1987).
16
.~
l Introduction
Lukatelich and McComb (1990) found that sediments from sites in the central basin of Harvey Estuary were up to 18.63% organic matter, while organic matter content from sediments from the eastern edge was as low as 2.3%. Sediment phosphorus concentrations ranged from 180 µg g -I in the less organic sediments to 67 6 µg g· 1 in the central basin. Sediment total phosphorus concentrations for Harvey Estuary reported by McComb et al. ( 1998) ranged up to 900 µg g - 1• Lukatelich (1987) also provides a comparison between sediment characteristics of Harvey Estuary and a nearby ocean site. Total phosphorus content in the relatively unpolluted ocean sediments (377 µg g- 1) was not vastly less than in the Harvey Estuary sediments (433 µg g-1), and this concentration of phosphorus is not considered particularly high for sediments in a eutrophic system (Lukatelich, 1987). A more substantial difference was found in the form this phosphorus took: 85% of phosphorus in the ocean sediment sample was apatite phosphorus (which is largely unavailable to algae and phytoplankton), while only 21 % of phosphorus in the Harvey Estuary surface sediments was in this form (Lukatelich, 1987). Further, the maximum phosphorus release rate observed for the Harvey Estuary sediment sample was 66.1 mg phosphorus m· 2d- 1 (Lukatelich and McComb, 1985). This is orders of magnitude higher than the maximum release rate observed for the ocean sediments, 0.04 mg phosphorus m· 2d- 1• These differences illustrate the much greater potential of Harvey Estuary sediments to contribute phosphorus for phytoplankton growth. As a further point of comparison, Lukatelich and McComb ( 1986) and Hill et al. (1991) found that the physical and chemical characteristics of the sediments of Harvey Estuary were similar to those of the nearby Swan-Canning Estuary. The latter estuary has also been subject to nutrient pollution but had not yet had such severe nuisance algal problems. Some smaller blooms have, however, been observed in sections of the Swan Estuary (e.g. Jernakoff et al., 1996). Nitrogen, rather than phosphorus, appears to be the primary limiting nutrient in the Swan (Thompson, 1998). Lukatelich (1987) also noted that there was little evidence of substantial long-term build-up of phosphorus stores in the sediments of the Estuarine system in response to
17
Introduction
anthropogenic increases in phosphorus since the 1950's, and the later results of Lukatelich and McComb (1990) are consistent with this point (i.e. sediment phosphorus concentrations did not appear to have risen substantially between the two studies). 2. 1. 7.2 Sediment processes
The sediments of Harvey Estuary act as both a source of and a sink for phosphorus and other nutrients. In oxic conditions the sediments adsorb phosphorus, helping to trap it in the system. When the water column becomes stratified, the bottom layer often becomes hypoxic. Under these conditions, phosphorus is released from the sediments (McAuliffe, 1986; McComb and Lukatelich, 1990). In a study of phosphorus release from intact sediment cores from Harvey Estuary, McAuliffe (1986) found release rates of 15.1 mg m"2d" 1 in anaerobic conditions (compared with only 0.3 mg m· 2d· 1 in aerobic conditions). The development of hypoxic conditions depends partly on low vertical diffusion of oxygen from the surface water and partly on the decay of organic matter in the water colunm and sediments. Variations in the degree of stratification and particularly in the amount of decaying organic material vary significantly from year to year, and it is believed that these variations, rather than fluctuations in sediment stores, that controls the amount of phosphorus released from the sediments in late spring (McComb and Lukatelich, 1990). Bioturbation and wind-induced turbulence produce a high rate of sediment resuspension (Lukatelich, 1987), producing elevated turbidity, to the detriment of benthic plants and macroalgae. Suspended sediments also interact more strongly with the water column than bottom sediments, enhancing phosphorus adsorption in windy conditions (McComb and Lukatelich, 1995) and for several days after storms. Exchange rates between sediments and the overlying water column are controlled by a number of factors. These include redox potential (influenced by concentrations of oxygen, nitrates, and other electron acceptors), pH, mineralisation rates of decaying detrital matter, the amount of sediment in suspension, and deposition rates of biomass (Kidby et al. 1984; McAuliffe, 1986).
.
18
Introduction
2.1.8 Management
Early management efforts were focused on eliminating the principal cause of the problem: nutrient pollution. A range of measures to reduce phosphorus inputs to the Estuary was considered. These included alteration of soil in the catchment area to reduce leaching of fertilisers, use of natural or artificial wetlands to absorb excess nutrients before they reached the system (Chambers, 1984), and alterations in land use in the catchment. Through the education of farmers and others in the catchment, as well as the development of fertilisers and approaches to their use more suited to the region, some success was achieved in reducing phosphorus loading from fertilisers (EPA, 1992). In addition, a ban was placed on further clearing of native vegetation in the catchment area. Despite this, the blooms continued. In 1984, phosphorus loading was significantly reduced, but the Nodularia bloom was on the same scale as observed in previous years, which was taken to be indicative of progressive eutrophication (Hodgkin, 1986). A program of weed harvesting from the Estuary was put into place, which (at some considerable expense) ameliorated the aesthetic impacts and adverse effects of eutrophication on fishing. Although this had some success in removing macroalgae, it was ineffective for the planktonic Nodularia and did not address the underlying problem (EPA, 1988). Between 1976 and 1981, the Peel-Harvey Estuarine Systems Study was conducted in an effort to formulate a better management approach. A major undertaking, this study included examination of the causes of eutrophication in the Estuary and Inlet, hydrology and meteorology, phosphorus loading from the catchment area, biological processes in the system (Huber, 1980), and the nature of the sediments (Gabrielson, 1981 ). It was evident from these studies that increased removal of phosphorus by increased flushing of the Estuary would improve matters. It was also recognised that increased flushing would result in higher spring salinities, which would reduce the viability of
19
Introduction
Nodularia. Options considered included deepening and widening the existing channel and construction of a new channel (EPA, 1988). At the same time, efforts to reduce phosphorus loading continued. By 1991, the application of phosphatic fertilisers has been reduced by some 30% (McComb and Humphries, 1992), however a considerable store of phosphorus remains in the soil, fertilisers continue to be applied, and increasing urbanisation in the area provides a new potential source of nutrients (McComb and Lukatelich, 1995). The Peel-Harvey Study Group was formed m 1984 to examine these and other options and formulate a management plan (Hodgkin et al., 1985). The Stage 1 Environmental Review and Management Plan subsequently released in 1985 recommended a range of measures, including further consideration, impact assessment and public review of the possible construction of a new channel at the northern end of the Estuary. In the meantime, the Mandurah Channel was dredged and deepened (Hodgkin et al., 1985). This improved flushing to some degree, but had disappointingly little impact on the annual blooms in Harvey Estuary. In 1988, the Western Australian Environmental Protection Authority completed a management strategy for the Peel Inlet and Harvey Estuary (EPA, 1988). The strategy combined catchment management to further reduce nutrient loading with the construction of a new channel to connect Harvey Estuary to the ocean at Dawesville. It was anticipated (Hart, 1995) that the Channel would increase tidal range, increase
flushing, reduce stratification, increase winter and spring salinity, increase phosphorus loss from the system, and increase oxygen concentrations. All of these factors would combine to increase the resilience of the system and great! y reduce the likelihood of Nodularia blooms occurring, although it was recognised that careful catchment management was also necessary to prevent the re-emergence of large macroalgal blooms in the system (EPA, 1988). The strategy was enacted in 1992 with the implementation of an Environmental Protection Policy by the Environmental Protection Authority (WA) (EPA, 1992) and the Western Australian government. Construction of the Dawesville Channel (Figure 2-8) commenced that year. The channel, 1.7 km long and 2 m to 6 m deep, was
20
Introduction
completed and formally opened in November 1994 and now connects the estuary at Dawesville directly to the Indian Ocean (Gerritse et al., 1998).
IND/AN OCQ
,
(3-8)
+ K Nod p
where P1n1ra is the intracellular phosphorus store (mg), Kp is the half-saturation coefficient for phosphorus with respect to Nodularia growth (Bowie et al. ( 1985) report a literature range of 0.0025 to 0.06 mg C 1 for KP for cyanobacteria), and [Pavallabte]
is the effective available phosphorus concentration (mg L-1); (3-9)
where [P] is the concentration of available phosphorus in the grid cell (mg L-1), [Peru] is the critical phosphorus concentration (mg L- 1). J0rgensen (1979) reports a literature range of
l
0
'
10
20
----~
'
30
-20 -30 Figure 3-7 Effect of temperature on Nodularia growth (from Huber, 1984a).
These observations are incorporated into the model through the temperature limiting function, h(T):
h(T)
l £,'
=
(3-13)
Tm,,,_ -T 10 '
where Tis the water temperature (°C) and Topr and T_ are constants given in Table 3-1. Lehtimtiki et al. (1994) found maximum growth at temperatures around 20°C for
Nodularia spumigena strains from the Baltic Sea. The results
~f
Lehtimaki et al.
(1997) agreed more closely with those of Huber (above). It is likely that the temperature response of Nodularia adapts to the prevailing climatic conditions (see, e.g. Steel, 1995), so it is appropriate where possible to rely on data for Nodularia from Harvey Estuary for this study.
65
Modelling dynamics of Nodularia spumigena blooms
Examining factors influencing akinete germination, Huber (1985) found that germination was slowed substantially at temperatures below 15°C. The response of akinetes in culture to different temperatures is shown in Figure 3-8.
Temperaluro
100 -,~ '-.,'
y
.
' ....... /- --.......:.
.........
'~
80 A.,'
..,' / r 1 /
,•"./~
~
c
..
0
60
c
E
.,
.....
~
c.J 40
, ,,
,
--
/
•
I I I I
-,.!! Q
I I
20
I I
I
0-1-~~i"-~~~.--~~~..-~-,..~.---.--..-,..,.--,
20
15
10
25
·c Figure 3-8 Effect of temperature on akinete germination (from Huber, 1984a)
This result is incorporated into the model with the following relationship:. 0, f(T)
=
0.8x (T
1
~
T < 12 12
12 s; Ts; 19.5
)
T > 19.5
1,
The shape of the curve produced by f(T) is shown below. )!
'
66
(3-14)
Modelling dynamics ofNodularia spumigena blooms
LO,---------.--------,,---------,
0.9
0.8 0.7 0.6
p
I
~ 0.5
0.4 0.3
0.2 0. .1 o~----~-~------~------~
lO
20
15
25
Temperature ('C)
Figure 3-9 Relationship between temperature and akinete germination, showing the relationship between temperature in the model and the proportion olNodularia akinetes that germinate.
3.2.2.1.8
Light
Photosynthesis, and hence Nodularia growth, increases with increasing light until a saturation level is reached and the rate of growth plateaus. At the very high irradiances found at the surface during summer, photoinhibition and bleaching (photo-oxidation) occur, reducing net growth rates for Nodularia. Nordin and Stein (1980, cited by Huber, 1984a) found an optimal light range of approximately l08µEm- 2s· 1 for Nodularia. In another laboratory study, Rhodes (1995) found maximal productivity of Nodularia isolated from blooms in Pyramid Lake, Nevada at around 200uEm- 2s· 1 with little reduction at higher light intensities, while Huber (1984a) found maximum growth rates in the much lower range of 20-30µEm- 2s· 1 for Nodularia isolated from the Peel-Harvey system, with severe inhibition of growth for
irradiances above this range and bleaching above lOOµEm- 2s· 1 (Figure 3-10).
67
Modelling dynamics of Nodularia sputnigena blooms
120
,.c: 0..
I
II
8
Ul
-a>
I
I
2-
)
)_ 0
100 50 1ime from start of simulation (days)
150
Figure 3-15 Simulated Nodularia population (In phosphorus-equivalent mg L-') over a 150-day trial.
Examination of simulated available phosphorus concentrations (Figure 3-16) shows a rapid initial decline and a levelling-off at a very low concentration approximately 40 days into the trial. This indicates that there was an initial rapid uptake of phosphorus by Nodularia during the exponential growth phase, but that the bloom was selfsupporting with respect to phosphorus by around day 40, as there was no immediate decline on Nodularia population after this point.
93
Modelling dynamics of Nodularia spumigena blooms
500 ' 450 400 -\
I!
350 . . \
\\
,.... 300
·~ 250
\
6
\
200 -
\
150 -
\ i
I
\
JOO 50 0
\
\ \
"---......
100 50 Time from start of simulation (days)
150
Figure 3-16 Simulated concentration of soluble reactive phosphorus (mgL·') over a 150-day trial.
Figure 3-17 shows the source of the phosphorus that sustained the bloom - detritus from dead Nodularia. The peak at around 22 days reflects the lag resulting from the time taken for bacterial populations (Figure 3-18) to build up to a level in equilibrium with Nodularia concentrations.
94
Modelling dynamics of Nodularia spumigena blooms
Nodularia dctri lus 4 ·'""' ...l
/\
i\
3.5 -
-~
tlJl
s
'j
"':;j " ';;J
2.5 -
'-'
>
Vl :;j
f
\
I
O"
"
\
I
~
·a
ii
2-
I..;
!
0 .£:
0. l.5 0 "' .£:
I
0.
I
1
I
,!
0.5 I
0
I
/ 100 time from start of simulaiion (days) 50
Figure 3-17 Simulated concentrations of detritus from Nodularia over a 150-day trial.
95
150
Modelling dynamics of Nodularia spumigena blooms
Bacteria 60
so···
I/
/
10·
/
/
~./
0
JOO 50 Time since start of simulation (days)
150
Figure 3· 18 Simulated bacterial population during the trial.
Model results show a water column that became supersaturated with dissolved oxygen during the initial exponential growth phase of the Nodularia (Figure 3-12), but thereafter declined. Once the bloom had stabilised, the oxygen concentration remained fairly constant and at a level close to saturation. This pattern reflects an oxygen budget dominated by during photosynthesis during the early growth phase of the bloom, by respiration once enough bacteria was present to break down the detritus which had built up, and balanced between the two once the bloom had stabilised.
96
Modelling dynamics ofNodularia spumigena blooms
Oxygen 90' i\
801·
i\
r
I
!\
i !
70 . i --l ----Of)
s
I \!
60' !I .I
\
50 -
I \
I i
40 - I I
20 ;I f
i
j
lO
I
\
•,
'""'"'
~~-.
'"'~-"""~~---·~
__J
0
100 50 Time since start of simulation (days)
150
Figure 3-19 Simulated dissolved oxygen concentration (mgL· 1).
Sediment phosphorus stores are shown in Figure 3-20. As illustrated, there was an initial rapid uptake of free phosphorus by the sediments, followed by a fairly constant increase in sediment phosphorus stores for the remainder of the simulation. This is because oxygen concentrations at the bottom of the water column were never low enough to cause the release of phosphorus by the sediments.
97
----------
-----
Modelling dynamics of Nodularia spumigena blooms
Sediment P
600 .. i,i
500
//''"""
f
1 ··
i
!
4001· 1
I1 Ij i!
300
1
1
200 JI
I! Ii
ii '
1000~1------L-----~--------'.1)•0 50 100 Time since start of simulation (days) Figure 3-20 Sediment phosphorus stores (in mgL- 1) during the 150-day simulation.
3.4.3.4 Discussion
Qualitatively, the behaviour of the Nodularia population is consistent with field observations (Figure 3-12). Huber (1984a) reports that Nodularia blooms in the Estuary generally occurred two to three weeks after conditions became suitable for germination of akinetes, and this is consistent with the results obtained. In addition, the declining availability of phosphorus would be expected to become limiting during a prolonged bloom (Hamel and Huber, 1985), as was the case in this simulation. The actual Nodularia concentrations reached were higher than expected, suggesting that light limitation was not adequately simulated. This is probably due to a combination of the low resolution of it one-layer model and the decision to assume mid-summer irradiance levels with an unclouded sky.
98
Modelling dynamics of Nodularia spumigena blooms
Similarly, although the behaviour of oxygen in the system is qualitatively appropriate, dissolved oxygen concentrations became unrealistically high. This is largely due to the lack of any mechanism in the model to limit supersaturation. In reality, oxygen tends to bubble off when water is supersaturated, and dissolved oxygen concentrations during the growth phase of Nodularia blooms in the Estuary have been found to be between 7 and 10.5 mgL- 1 (Huber, 1984a). Simulated bottom oxygen concentrations in particular were too high, as the system was assumed to be fairly well mixed and this simple model allowed little vertical resolution. The artificially high oxygen concentrations resulted in constant uptake of phosphorus by sediments, which does not occur in the field (Hamel and Huber, 1985). As a result, available phosphorus concentrations stabilised at a level below the 1-22mgL· 1 found in the field (Huber, 1984a). In conclusion, this trial suggests that the basic mechanisms considered by the model are correct, but that some improvements are required: in particular, some mechanism must be included to limit oxygen supersaturation. The proposed mechanism is described below: 3.4.3.4.1
Bubbling
When phytoplankton growth rates are high (as is the case during the exponential growth phase of a Nodularia bloom), photosynthesis is often great enough to supersaturate the water with oxygen. If the water becomes very supersaturated, bubbles are formed and the excess oxygen is lost. In this model, it is assumed that any oxygen in excess of 200% the theoretical saturation point is lost through bubbling. The equation used to determine the saturation point of water with respect to oxygen is an empirically derived (Hartman and Hammond, 1985) function of salinity and temperature, illustrated in Figure 3-21. The saturation oxygen concentration, C0 , is given by:
99
Modelling dynamics of Nodularia spumigena blooms
(3-34)
C 0 =0.2094·a(l-V)·l.1428 where V=l-9.0701xl0-4 ·CL·
l
18.1973. T!D + 3.1813x10-T (1- exp(26.1205 · T!D ))exp 373 16 2 T~ ) [ l.8726x 10- (!- exp(8.03945 · TID ))+ 5.02802 ·
alog(
(3-35)
where TK is the temperature in Kelvin (i.e. TK = T + 273.16).
where ci=-7.424, c2=4417, c3=2.927, C4=0.04238, cs=-0.1288, C6=53.44, c7 =0.04442, and cs=7.145x10-4. CL= (S -0.03) 1.805
(3-37)
T/D=l-373.16
(3-38)
TK
100
l Modelling dynamics ofNodularia spumigena blooms
13
salinity response at 15°C temperature response at a salinity of 1
,-,•
12
~11 ~
g"'
.,,§ 10 • "v" "9 " & u 0
u
~ 8
0
"
I•
7
'
6
salinily or temperature (0 C)
Figure 3·21 Response of theoretical saturation oxygen concentration to salinity when temperature is 15'C and to temperature when salinity is 15.
3.4.4 A Seasonal Nodularia bloom cycle 3.4.4. 1 Introduction In order to test the efficacy of the representation of Nodularia growth and mortality
responses in the model described in section 3.2, an attempt was made to duplicate this cycle over the course of a full season (from germination to decay) by imposing the expected changes in salinity, temperature and light on a simple one-layer implementation of the Nodularia model. 3.4.4.2 Methods
As described in section 3.4.3.2, the biogeochemical submode! was called from within a shell program set up to simulate a one-layer water column section 1 m wide and 2 m deep. All boundaries were closed.
IOI
I
Modelling dynamics of Nodularia spumigena blooms
Initially, the Nodularia population was assumed to be zero, akinetes were present in the top 5cm of sediments at a concentration of 5.0 x 108 m- 1, and physico-chemical conditions were set to values representative of estuarine conditions found by Huber and Hamel ( 1986a) during the germination phase of a typical Nodularia bloom (see Table 3-3). Salinity, temperature, and daily maximum photon flux were kept constant throughout the germination phase; other values were allowed to vary as the model predicted. At the end of twenty-one days, these parameters were reset to values representative of the exponential growth phase of a typical bloom, and the simulation continued. This was followed by a period representative of the stationary phase of a bloom, and finally a period during which conditions were typical for the bloom senescence and decay (Table 3-3). Table 3-3 Duration of and conditions at the start of each phase of the simulation. Note that the value given for light refers to the maximum total instantaneous PPFD reached over the course of the day. )oration (days): iemperature ("C): "ight (µEm- 2s- 1): Salinity: SRP (mgL- 1): IJN (mgL- 1):
Germination
0 xoonential
21. 21. 300. 12. 63. 1785.5
35. 1.5 J50. 14. 11.5 73.
Growth Stationarv Phase Bloom Senescence 56. 23.5 650. 21. 11.5 89.
102
60. 24. 1000. 30. 13.5 1764.5
,...,,
I
Modelling dynamics of Nodularia spurnigena blooms
3.4.4.3 Results
Nodularia
25
Stationary phase Ge.rmination Growth p1mse , ph~)-,,,.~~~ ! 20
I
I
/ I!
I\!
I
/
Senescence
I
·---------'
5
()
10
20
40 60 80 100 120 time from start of simulation (days)
140
160
Figure 3-22 Predicted Nodularia population over the course of the run.
The results of the simulation run are shown in Figure 3-22. In accordance with observational data, the Nodularia population gradually rises during the germination phase, grows exponentially during the second phase, and is fairly stable during the third phase. Using an estimated ratio (accurate perhaps to within an order of magnitude) of: Phosphorus in Nodularia (mgI; 1) : Nodularia density (µm mL- 1) = 1 : 3.6 x 103
103
Modelling dynamics of Nodularia spurnigena blooms
The concentration of Nodularia filaments during the stationary phase is roughly 7.4 x 104 µm mL- 1. This is a little below the expected range of Ix 105
-
3 x 106
µm mL- 1 (Huber, 1984a), however it matches the lower bound of this range to within an order of magnitude. After an initial drop in Nodularia concentration at the beginning of the "senescence" stage, the simulated bloom again stabilises. This is not consistent with empirical observations. The results show that the model correctly reproduces the germination, growth, and stationary phases of a Nodularia bloom. It does not, however, adequately simulate the final phase, during which the bloom is expected to collapse Figure 3-12. A closer look at the influences of some physico-chemical parameters on predicted growth and decline of Nodularia is warranted. 3.4.4.3.l
Salinity and Temperature
Salinity and temperature affect modelled Nodularia populations by influencing the average life-span of cells: (average lifespan)= (maximum lifespan)
* f(S) * f(T)
Examining temperature and salinity conditions during the four phases of the model run, the following is obtained: Table 3-4 Temperature and salinity effects on Nodularia life span during each phase of the model run. Germination Phase 0
~-
rremperature ( C) Salinity f(S) f(T) f(S)xf(T) life-soan (davs)
21 12 J.81 J.50 0.40 80
n:lxponential Growth ~hase
21.5 14 Kl.83 0.41 Kl.34 67
Stationary Phase
23.5 21 J.93 J.20 Kl.IS 36
_Senescence
24 30 0.92 0.08 Kl.08 15
These figures do not look unreasonable given the expected results. Clearly, however, > -;!
Nodularia growth matches mortality after a brief period of decline in the simulated
'
results during the final "senescence" phase. There are a few possible reasons for this
'~
result:
104
l
i'
i Modelling dynamics ofNodularia spumigena blooms
1)
The effects of salinity and temperature on Nodularia used in this model were
observed by Huber under laboratory conditions. In these conditions, Nodularia continued to grow even at salinities up to 60. In the field, however, Nodularia blooms always collapsed by the time the salinity had risen to 30. This may indicate a) that laboratory trials did not reflect field conditions; b) that, contrary to the conclusions of Huber and others, salinity was not the primary factor triggering bloom collapse; or c) that there were cumulative interrelations between the adverse effects of rising salinity and other factors which have not been identified and are not represented in this model. 2)
That incorporating the effects of salinity as a control on mortality rates rather
than on growth rates is ineffective; or 3)
That other factors, such as light and decay rates, which would also limit
growth rates during this phase are not properly represented in the model. 3.4.4.3.2
Light
Figure 3-10 shows the relationship between light and Nodularia growth rate in the model. This relationship is, again, based on the laboratory results of Huber (l 984a). As it shows, photoinhibition begins to be a limiting factor at a relatively low light intensity, while the growth rate declines to zero when the photon flux density is close to zero. As a result of this and the tendency of Nodularia to form a thick scum at the surface during blooms, the effects of self-shading are very important. The current run was set up as a one-layer model, with Nodularia assumed to be evenly distributed throughout this layer and the PAR used in calculations determined at a depth midway down the water-column. This may be overly simplistic, given that Nodularia can form a scum up to 60cm thick at the surface during blooms (EPA, 1988). Further, Nodularia detritus (unlike living Nodularia) is assumed not to float on the surface. If
it does, it would decrease the light available for growth as the mortality rate increases in the final phase of the bloom. Other possible problems with the representation of light in this model include the assumed absence of other plankton for this run, the simple representation of sediment suspension, and the choice of values for light
105
Modelling dynamics ofNodularia spumigena blooms
absorption and scattering coefficients for all organic components represented in the model. 3.4.4.3.3
Decay
In the model as it currently stands, decay of Nodularia and other detritus is mediated by bacteria, and the rate of decay is relatively high when bacterial populations are high. Upon decay, nutrients in the detritus are released immediately into the water column. Recycling of nutrients from decaying Nodularia is important to the maintenance of the bloom during the stationary phase (Huber, 1984a), however the gradual increase in the proportion of phosphorus which becomes unavailable as the bloom continues may be a factor in triggering bloom decline (Huber, 1984a). If decay and phosphorus release rates in the model are too high, phosphorus may not be limiting growth sufficiently during the final phase of the simulation. Given that no empirical data on bacterial populations or decay in the Estuary is available, it may be appropriate to remove bacteria from the model and include instead a simple, calibrated, decay rate. 3.4.4.4 Forcing senescence in a seasonal Nodularia bloom
The results of an unmodified run of the biochemical component of the model were unsatisfactory as the simulated bloom failed to collapse as expected. This was probably due to the application of laboratory results for response of Nodularia to variations in salinity that may not reflect responses under field
conditions. In the Estuary, bloom collapse always occurred when salinity reached 2830 (Hamel, 1985). The simplest way to reflect this empirical evidence in the model is to force Nodularia growth to stop once salinity reaches this level. Figure 3-23 compares the
results from the trial previously described with results obtained when such a condition is added to the model. Other trials (not shown) in which parametres defining the response of Nodularia to light, phosphorus, or temperature, were not able to reproduce this pattern. This indicates that salinity or some closely associated factor was critical in determining Nodularia bloom collapse in Harvey Estuary.
106
Modelling dynamics of Nodularia spwnigena blooms
Nodularia
25
20
A
,.-.,
· '· - · - · - · ~ ·-~ ~ · --- r J
I
I
n i ': s!
i\
I
s
! '
!\ s
)
\\ '
!_/ 0
20
40
\,
140 120 100 80 lime from start of simulation idays)
60
I
---1-~ 1.80 160
Figure 3-23 A comparison of the results of the trial described in section 0 (A) with results obtained when the Nodu/aria growth rate is set to zero when the salinity exceeds 29 (B).
3.5
Closing comments
Although this model makes many simplifying assumptions, it remains complex and requires considerable calibration within the range of available literature values. Although the flexibility of a model of this complexity may be required for long-term predictive uses (when conditions may vary considerably), a simpler model may be more beneficial for other goals. The following chapter takes a closer look at observational data from Harvey Estuary, keeping in mind the relationships identified in the model above, with the aim of
107
l Modelling dynamics of Nodularia spumigena blooms
identifying the key elements responsible for controlling Nodularia blooms in Harvey Estuary. With this information, a much simpler model capable of replicating the observed behaviour of Nodularia in the system can be produced, and used to test present understanding of the major factors involved.
108
Chapter 4:
Analysis of water quality data
4. 1 Introduction McComb and Lukatelich (1990) presented an examination of inter-relations between some biological and physicochemical parameters observed in Harvey Estuary between 1978 and 1987. The present chapter expands upon this work, including observational data from 1987 to 1993, commenting on a broader range of water quality parameters, and including some brief comparisons with Tamales Bay.
4.2
Relationships between observed variables - Harvey Estuary
Table 4-1 shows Spearman rank correlation coefficients for instantaneous surface values of variables measured at a site in the middle of Harvey Estuary (Figure 2-1). It is immediately apparent that most of these variables are correlated with most others, which is in accordance with the web of interactions shown in the conceptual diagram of Figure 3-2. Low correlations between total alkalinity and other water quality parameters, including salinity, with which a strong linear relationship might be expected, are probably not significant, given the low number of alkalinity observation - only 47 data points are available for the entire 15-year period. Interestingly, the correlation between chlorophyll a concentration and temperature is very low, and significant only at p:0;0.05. In general, phytoplankton growth rates tend to increase with temperature. More specifically, the results of Huber (1984a, 1985), showed that, in laboratory conditions, Nodularia growth rates (Figure 3-7) and germination (Figure 3-8) increase with temperature, up to an optimal value of approximately 28°C. In light of this, a stronger correlation between temperature and chlorophyll a was anticipated. It is likely that several factors contribute to this result. First, phytoplankton growth
tates decline at temperatures above the optimum, and it is likely the optimal temperature changes over the course of the year as well as between different
109
n I'
: I
Analysis of water quality data
phytoplankton groups in the system. Figure 4-1 shows observations of temperature and chlorophyll a. If it is assumed that the optimum temperature for phytoplankton is 21 °C, lines of best fit on either side of this optimal value can be drawn as shown below. The rank correlation between chlorophyll a and temperature, considering only data points for which the observed temperature was below 21°C, is approximately 0.15 (which, with 404 degrees of freedom, is significant at p:S:0.01). The correlation between chlorophyll a and temperature, at temperatures above 21°C, is -0.17 (significant at p:S:0.05 with 180 degrees of freedom).
110
Analys1.\ of '"1 aler qual11y dara
Table 4-1 Rank correlation coefficients for surface obseNations at a site in the middle ofHaNey Estuary, 1977-1993.
Date
'. alk.
·s
T
:no
;pH
............................... ......................1. ................... .................. ..!.. . .................................. ..l.................. ...i. .................
s
-0.22
I.oo
0.13
0.45
·
jSecchi :depth 'light
lo/oSat
jP04
:org P
1
NIL
org N !Si0 4 :chl a !phaeo !N:P ..i........................................J...................:.. ..................l....................!................. 0.11 0.10 -0.20 10.13 !-0.35 :-0.42 10.39 :
l.. . . . . . . . . .L. . . . . . ....J~>.-.t.i~~.. .!.....................L....................i... . . . . ... . . . . . .. . . . .
:-0.60 1-0.13 :o.51
:0.02
... ..........................................•........... •...................• .............•....... -·········..···· ...·-·····-···....:.....................~•••........•.•..•....!.....................! ............-
:-o.51 :-0.19 1-0.35 1-0.22
1
N0 3
.J
....L ..•. __ ,, ..........i ................-...:......................;......._,_,......... ..................... ,_.......................................1 ..............--..~..................... :.....................~---········-.....:
:~:l ~::~ ~:~~ ~:~~ !:~;1 I ~ : ~: 1~:: ~~ i~: ~~ 1~~~~ i~:~~ !~:~~ 1~;~ ~:;~ ~:~~ ~:~~-t-0.05 I~:~! .~~~~ 1~::~~ I ~g.. ~~~.: ·1· -· · · · · · · · · · · · ·~·:~·~··-· · · ~~~~·~· ·-·!1~·:~·~······ · 6~f,}. . . .i.~~~·~· · · · t~·~~··-· -·~~l~·· · ·r~~~3·-· · I~:·!·~· · · · ~~·:~·~· · · · .·~6~~4······~·:~~·· · · · ~~·:l·~· · · ·~·:·l-~· · · ~·:!~·· · · · t6~6~?.. . ..~~·~l·· · · · i~:·:·}· · · · ·~:·~~·· · · I ~~:~~ty · 1
j
:
:
.
6
2
:
:
8
I
:
:
!
,:;
:
0
~~{~Pii)§i : ~~~=·· ~~r r~~r ~ 1~~ =m; /g~::::~ 1~ :~=m: f:i~~-:r~~ - ~;! =~ir ~;~ ·1~rri~ri=t~:~r1r~r- 1 iiihi.e~1inc: "(ill~')
~
o sa~tion 1
· ·«ri3 · ·· · ·-0-:-sf·····r-0:·0"2"····· «f«»s·· · · ·1o:If·· · · 10.49 -0.17 0.21
-0.19 -0.35
P04 (µgL- ) ~~g;tl~. 1 »·(µgL·:1).......... ·0:·07········ :0·:2:2......
0.24 -0.05
0.10 j0.82 -0.30 10.14
: -0:«j3·····-r~rf:f·· rroff·······10:44·········ro5f · · ·o:Kc·····..::oj7·· · · ~o:«»s··· · ·· · · ··t< f h·· · ··lojo·· · · ·ro:Kr····l-0j§··· 1 j-0.48 1-0.12 j0.44 !1 o~ j-0.11 I0.33 -0.27 -0.26 0.31 ·-0.07 I0.50 j0.30 j-0.20 .J
:o.58 j-0.04 j-0.22
j0.33
j0.31
l-0.11
~ 100
j0.17
0.33
0.34
0.18
J0.41
j0.15
10.28
!-0.45
i
-0·.-0K····· 0·:2"0·······10:-io·······T0:·49········Fo:«s6·· ··r::o·:·1·3·····ro:Kc······:o:-33"··· · · ro:·ff.......IT«>o-·-·· ·:0:·20...... :0:0"9"..... 2k.g/m
3
(Sl(atlfled)-->
Figure 4-5 Observed relationship between minimum salinity and spring stratification in Harvey Estuary. The dashed line shows a linear best fit.
It should be noted that density variations in the Estuary are strongly dominated by variations in salinity, rather than tomperature. The correlation between vertical density difference and vertical salinity difference in all observations is very strong (r,
~
0.99, ps0.01).
l 119
Analysis of water quality data
Bottom data for Tamales Bay is not available for the present study, so no comparison can be made on this matter. 4.4.2 Stratification and oxygen depletion When the water column is stratified, vertical Illlxmg is reduced and the rate of replenishment of oxygen in the bottom layer of the water column is therefore also reduced. Hence, if biochemical oxygen demand due to respiration and decay processes in the sediments is sufficient, water near the sediments is likely to become depleted of oxygen. Figure 4-6 shows observed bottom dissolved oxygen concentrations and stratification in Harvey Estuary between 1978 and 1992. The correlation between these two variables is significant
(rs~
-0.45, p:'>0.01). An even stronger correlation
(rs~
0.79,
p:'>0.01) is observed between vertical density difference and oxygen deficit (i.e. difference between surface and bottom oxygen concentration). This is illustrated with a logarithmic best fit in Figure 4- 7 .
.
1
f -!
Il
120
Analysis of water quality data
•
.. '
•
•
•
•
• • • • • •• • • •••
•
• •• • • •• • •• • • "'• ••
.
• •
•
•
••
• • • ••• .,. •• • •• .,.• • • 2
•
4
•'
6
• ••• s
•
•
•. • . ·'
•,
10
• 12
• •14
• 16
Figure 4-6 Observed dissolved oxygen concentration at the bottom of the water column at a site in the middle of Harvey Estuary against vertical density difference (kg m3).
121
Analysis of water quality data
xx x ,,,. x
x
>«
x
x
xx xx
10- 2 '--~~~~~~~~~~~~~~~~~~~-'-'"~~~~~~~ 1 1
10~
1~
10-
10
1~
observed density difference
Figure 4-7 Observed oxygen deficit (mgL-1) against observed difference between surface and bottom density (kg3m- 3).
This relationship, together with surface oxygen concentration, can often be used to provide a reasonable prediction of bottom oxygen concentration, according to:
ob =O, -0.754t.p, -o.666
(4-1)
where Ob is bottom oxygen concentration (in mgL-1), O, is surface oxygen concentration, and L'.pv is the difference between surface and bottom density (in kg 3m·3).
The results are shown in Figure 4-8. '~
.
122
Analysis of water quality data
14 0
0
0
12 0
0
0
0
0 0
"' 0
0
o~ 0
0
2
4 6 8 10 observed bottom dissolved oxygen concentration
12
14
Figure 4-8 Bottom oxygen concentration predicted from surface oxygen concentration and vertical density difference against observed bottom dissolved oxygen concentration (mgL-') at a site in the middle of Harvey Estuary.
The theoretical saturated oxygen concentration (which can be derived from salinity and temperature) is strongly correlated (r, = 0.57, ps0.01) with surface oxygen concentration, and may be used in place of 0, where oxygen records are unavailable. Of more interest in validating our hypothesis, however, is the question of whether the Estuary was vertically stratified for a sufficient proportion of each spring to create oxygen poor conditions at this time of year. Depletion of oxygen in bottom waters in spring should facilitate the release of phosphorus from sediments, which may in turn trigger the germination and exponential growth phases of a Nodularia bloom. Although the available data are sparse in time, Figure 4-9 provides evidence of a clear relationship between the proportion of each spring when the water column is .,
stratified and the proportion of observations of bottom water samples for which dissolved oxygen concentration is low.
123
I
I
Analysis of water quality data
'
'
0.9 I-
0.1
0.1 2
0.8
0.9 (stratified)-->
Figure 4-9 Relationship between observed deoxygenation in Harvey Estuary bottom water and spring stratification in the Estuary. The dashed line shows a linear best fit.
4.4.3 Phosphorus release
1 1
i
I ! l' t
Phosphorus is released when the sediments become hypoxic (i.e., when chemically reducing conditions exist). In the model, this is assumed to occur when water in the bottom layer of the water column becomes anaerobic, i.e., when the oxygen concentration falls below a critical level, assumed in this instance to be approximately 1 mgL- 1. No direct observations of sediment phosphorus release rates in Harvey Estuary are available on a relevant time-scale, as this was not a process that was routinely monitored (although some laboratory measurements were taken, Lukatelich, (1987); McAuliffe (1986)). Degree of stratification (as measured by the vertical density difference calculated from salinity and temperature records) and phosphate concentration at the bottom of the water column are, however, positively correlated (rs"" 0.16, p:'>0.01), which is consistent with the hypothesis.
l
124
·1,.•l
!
Analysis of water quality data
4.4.4 Stratification and chlorophyll a The relationship between observed stratification during spnng and maximum observed chlorophyll a concentration at site A is shown in Figure 4-10. It is evident that higher chlorophyll a concentrations (indicating the occurrence of Nodularia blooms in this estuary) did tend to occur during years for which the water column was stratified for a significant proportion of the preceding spring. Again, this is compatible with the hypotheses outlined in section 1.1.
-'1985
2000
"' 1500
~
\ii
il"' ~
1000
500
o1__~~!l!L~BiL~___J~-'l!J!ljl9Li...~~-"-~----'~~--'-~~'---~_L~___j
0
01
.•
e
*83
.Q
.s:::;
82
* *86
"20
*78 10 87 5
10
79
77 15
20
minimum winter salinity
Figure 5-1 Mass of chlorophyll a at peak of spring bloom of Nodularia plotted against minimum salinity so of Harvey Estuary for preceding winter. The points refer to the 11 years from 1977 to 1987 and are labelled by the year; they are taken from an identical plot by McComb and Humphries (1992). The figure shows the model prediction. The Nodularia measurements are averages over three sites down the axis of estuary.
This chapter considers an explanation of the threshold that is an expansion of a mechanism due originally to Lukatelich ( 1987). The hypothesis is simply that the threshold involves a critical riverflow that creates two of the conditions necessary for phosphorus release from sediments in ~pring. The first condition (due to Lukatelich, 1987) is that the river delivers a sufficient nutrient load to the ·estuary to create a winter phytoplankton bloom whose decaying detritus is large enough to produce a sediment oxygen demand in spring able to de-oxygenate the bottom water. The
127
An analytical model of interannual variability of Nodularia spuniigena bloonis
second condition (which is being proposed by the present chapter) is that the spring salinity difference between the estuary and the ocean is sufficient to create a stratified water column (during tidal advection into Harvey Estuary) able to stop vertical mixing oxygenating the bottom water. This mechanism is illustrated by the data of Figure 5-1, which show oxygen concentrations and salinity at top and bottom of the water for the wet year 1985, and dry year 1987. The figure clearly demonstrates the link between the salinity of the estuary, the vertical salinity difference, and bottom hypoxia. Figure 5-2 presents oxygen and salinity profiles based on data collected approximately weekly from late winter to early spring. Figure 5-3 shows vertical profiles of salinity and oxygen at the southern and northern end of Harvey Estuary on three days in the spring of 1989. The profiles in Figure 5-2 and Figure 5-3 were all taken between 9.30 and 11.30 a.m., and again show bottom hypoxia increasing with salinity stratification.
128
-....
I
Error/ Reference source not found.
oxygen concentration '10
_:[o~-~
August 4
0
August 18
-2 0
August 25
-2 0
September 2
-2 0
September 15
-2 0
September 21
-2
_:[
.:r .:r
0 -1 -2 0
September 29
l_ ~,
~
October 13 .. ·
~l
October 27 November 3
10
30
20 salinity
Figure 5-2 Series of vertical proff/es of salinity (full line) and oxygen (dotted) at a station at centre of Harvey Estuary. The proff/es were taken at approximately weekly intervals from the late winter (August) to late Spring (November) of 1992 (the year proceeding the opening of the Dawesville Channel).
129 !
I
Error! Reference source not found.
southern end 0
I
E-o.5 ::::i
0u ..... 2
-1
~ -1 .5 c:
-
-2
.c
0)
'Qi
.c
-2.5 0
northern end
0
1.
10 Oct 1989
1: 1: 1: 1·
-0.5
--- 19 Oct 1989 ... 24 Oct 1989
\ .
-1
' -1.5
.
I
10 salinity
20
-2.5 0
30
'"
I·
u .....
30
-0 .5
1:
_,
0
10 20 salinity (psu)
0
E-o.5 ::::i
'\
-2
0
I
'
I ·
-1
/
/
Q)
~ -1.5 c:
/
.. ·· / ·
-1.5
,)
(
/
-2
E
/
-2
0)
'Qi .c
-2.5 0
5 10 oxygen (mgtliter)
-2.5 0
15
5 10 oxygen (mgtliter)
15
Figure 5-3 Vertical profiles of salinity (upper boxes) and oxygen (lower boxes) at the southern (left hand boxes) and northern end (right hand boxes) of Harvey Estuary on three days in the spring of 1989. The profiles were all taken between 9.30and11.30 am.
The present chapter attempts to develop a simple analytical detailed model of the Nodularia blooms based on the occurrence of the two conditions, discussed above,
for phosphorus release from sediments and make a quantitative comparison with the data of Figure 5-1. The growth of Nodularia is dependent on the accessibility of the phosphorus released from the sediments. This process is not explicitly considered here, and clearly under conditions of limited mixing, the phosphorus may not mix through the water column. However, the algae are able to sink through the water column at night and take up luxury intracellular phosphorus (Lukatelich, 1987).
5.3
Phosphorus Release Rate
McComb and Lukatelich (1995) reported laboratory measurements on intact sediment/water-column cores removed from Harvey Estuary at least monthly over the period 1987 to 1989. In the laboratory, hypoxic conditions developed easily,
130
An analytical tnodel of interannual variability of Nodularia souniieena bloonis
producing phosphorus release rates, R, which corresponded to a mean of about 3 tonnes per day from the whole basin. This is derived as an average over oxygen concentrations below 30% saturation since the rate of phosphorus release from sediments increases very rapidly with de-oxygenation. For simplicity, the release rate is assumed in the present model to suddenly turn on when the concentration drops below some fraction a of saturation, and then to have the value R. Control experiments on the cores showed that most of the sediment oxygen demand in early spring was due to diatom detritus from the winter bloom. 5.4
Time dependence of the Nodu/aria Bloom
The growth of the Nodularia bloom is illustrated by Figure 5-4, showing Nodularia mass, (converted from cell counts) from the end of the (austral) winter until summer, for two typical wet years. Also shown is an accompanying model result, which will be discussed later. The initial rapid growth phase lasts about one or two months, and is followed by a period during which the bloom size is steady. This is followed by collapse of the bloom in early summer when the salinity of the estuary reaches about 25. Decay of diatom detritus commences in winter and creates hypoxic bottom conditions but Nodularia blooms do not grow until spring. This is due partly to the decrease of nitrate within the water column after cessation of major riverflow. In the absence of significant nitrate, detital decay creates reducing conditions in the sediments, facilitating the release of phosphorus (Lukatelich, 1987). Secondly, water temperatures and light become suitable for germination of Nodularia akinetes in spring. Presumably, the steady state period that follows involves a balance of growth through phosphorus release (or recycling) and natural mortality. This is supported by the results of the model shown in section 3.4.4.4, which found that the bloom collapsed due to natural mortality rates when growth was switched off at high salinities. The reason for the collapse of the bloom at a salinity of about 25 could be due to the inability of Nodularia to fix nitrogen at such high salinities. However, that is contrary to laboratory experiments by Huber ( l 984a) who found that Nodularia continued to grow and fix nitrogen until much higher salinities.
131
""""
I
An analytical model of interannual variability of Nodularia spumigena blooms
5.5
Sediment oxygen demand
All masses in this chapter are expressed in terms of equivalent phosphorus. Hypoxic conditions depend on the sediment oxygen demand in relation to the saturated oxygen content Cs of the water column per unit height of the entire estuary (tonnes of equivalent phosphorus per meter). It is convenient to write the sediment oxygen demand as the oxygen flux expression
I
CsV
(tonnes of equivalent phosphorus per day)
where vis a velocity (metres per day).
1j
If the interannual variability of the sediment oxygen demand is ignored, then v is a
l
constant. The general formulation of this chapter will assume that the v is controlled by the detritus load from the previous winter, which varies inter-annually with the size of winter diatom populations. The simpler assumption of constant sediment oxygen demand corresponds to the special case in which that load is replaced by a constant. The minimum (austral) winter salinity, s0 , occurs near the beginning of August and this is chosen as the start of the model time series t. Akinete germination is also assumed to occur at this time, although in reality, the timing of the germination phase depends on water temperatures and light (see section 3.2.2.1. l ). This assumption simplifies the model, allowing interannual variability of estuarine temperature, which is small in relation to variability of riverflow and minimum salinity, to be neglected. The initial combined load of diatom and zooplankton detritus (tonnes of equivalent phosphorus) at the end-of-winter/start-of-spring, is po. Although, as discussed in Chapter 3: , diatom and zoop!ankton populations depend on a range of factors, including nitrogen and phosphorus concentrations, these factors have been shown to be strongly correlated with winter riverflow and hence, minimum winter salinity in Harvey Estuary.
In this simpler model, therefore, initial detrital load is assumed to
depend simply on minimum salinity,so,
,1
i
132
An analytical n1odel of interannual variability of Nodularia spumigena blooms
(5-1)
where pr is the detritus load in an extreme winter when the estuary becomes completely fresh, and
Smarine
denotes marine salinity. All notation is listed in Table
5-1 to Table 5-4. Table 5-1 Definition of symbols in this chapter. Parameters for which values assumed are given in Table 5-2 to Table 5-4. svmbol a m m* p pO v Yr
q r R* s ~s
so t tgrowth t, c Cb
K, u z w A
meaning phosphorus release coefficient (0
cta3
-
-
-
-
-
---
-
---
.Jan84-
Apr84
.JulB4-
-
C>ct84
164
.
-
-
-
--
.Janas
-
-
-
Apr85
.JulSS
C>cti:
Figure 6-6 Derived inputs for Harvey Estuary: a) Evaporation (mmld) b) Short-wave (solar) radiation c) Long-wave radiation input d) Long-wave radiation output e) Sensible heat flux
Hydrodynamics of shallow mediterranean estuaries
~
(a)
E E
~t·····-··· '.
: • : • ~]
400
(b)
~ '.:~ }:7~V,=, ~-/=/:/. ·=:~ 0
(c)
(d)
~::[~~·~~ ~mt.;:~:. ~···~~~ 300
(e)
·~ _:~~s@}::;~;+~v:~. ; ~~ ?l~~d .Janaa
APraa
.JulSS
Octaa
.Jan89
AprB9
165
.JulS9
C>ct89
.Jan9D
Apr90
Figure 6-7 Derived inputs for Toma/es Bay: a) Evaporation (mmld) b} Short-wave (solar) radiation c) Long-wave radiation input d) Long-wave radiation output e) Sensible heat flux
Hydrodynamics of shallow mediterranean estuaries
Mass fluxes due to evaporation and precipitation were applied at the surface using the same subroutine as for fresh water inflow. Evaporation rates and surface heat fluxes, derived in this way during model runs, are shown in Figure 6-6 and Figure 6-8.
6.4.2 Inclusion of River Inflow As shown in Figure 2-5, a number of drains and rivers flow into the Peel-Harvey system. Draining into the southern end of Harvey Estuary are the Harvey River, the Mayfields, Meelup and South Coolup. Collectively, these are the major source of fresh water for the Estuary: direct precipitation and flow from smaller drains each account for a much smaller volume of water and the contribution of groundwater (less than 1% of total inflow) can be ignored for out purposes. To represent fresh water inflow into the Estuary with POM, one possibility was to leave the southern boundary open, set salinity and temperature at the open boundary to those of river water, and apply a velocity proportional to total fresh water flow across this open boundary. This approach, however, is somewhat inflexible and could, in combination with strict elevation and other condition at the northern boundary, lead to instabilities or non-physical model behaviour. For these reasons, an alternative approach was taken: the southern boundary of the model estuary was closed, and fresh water applied as a source term at each internal time-step. This is handled by the subroutine, river() , in this model. First, the flow rate, totalflow, for the present time-step is set. The range of cells in the x and y dimensions over which this inflow should initially be distributed, the depth to which the water added should immediately be mixed, and whether inflow should be tapered across this range in the x-dimension are specified. Distributing inflow over more cells vertically and horizontally, as well as tapering the inflow, reduces numerical stress on the model by reducing the initial qensity and elevation gradients applied at each time-step. This also allows the effects of the different drains to be combined into one term, which reduces the model's run-time and allows collective inflow estimates to be used as model input without artificially splitting them.
166
Hydrodynamics of shallow rnediterranean estuaries
From this inflow data, river() calculates the amount of fresh water to be added to a column during each time-step, and then updates the data structures accordingly. First, external-mode surface elevation is adjusted:
EL "'
=EL n
+ flowx!:,.t dxx dy
(6-4)
where: ELn is the external-mode surface elevation at time-step n, flow is inflow rate 3
(m s-
1 )
for this column, M is the size of the time-step (in seconds), dx is the
(longitudinal) length of the column (m), and dy is the (lateral) breadth of the column (m).
Internal-mode surface elevation (ET) is adjusted in the same way, and internal- and external- mode depths (DT and D) are corrected accordingly. The overall effect of the addition of fresh water on the salinity and temperature of the water column is easily calculated. For salinity:
(6-5)
where Sn is the average salinity in the water column at time-step n, Vn is the total volume of the water column at time-step n, S, is the salinity of the added fresh water, and V, is the volume of fresh water added during the time-step. Similarly, for temperature:
(6-6)
167
Hydrodynamics of shallow 1nediterranean estuaries
where T,, is the average temperature of the water column at time-step n and T, is the temperature of the fresh water added during the time-step. In practise, the calculation is not quite so straightforward. Under the s1gmacoordinate system, the number of layers (and the proportion, DZ, of the water column taken up by each layer) remains constant as the height of the free surface is adjusted. As a result, the Z-coordinate width of each layer (DZ x D) increases when fresh water is added. Thus, when water is added to the top of the water column, the top of the bottom layer moves upwards. As a result, some of the water that was previously represented as being in the layer above (i.e. layer 2) is absorbed into the bottom layer. In this simple case,
(6-7)
where S(3),, is the salinity of layer 3 at n, S(2)n is the salinity of layer 2 at n, and V( 3 ),, is the volume of layer 3 at n.
Similarly, layer 2 expands and is pushed upward by the expansion of the layer beneath it. Hence, layer 2, after the addition of river water, has lost some water to layer 3 and includes some water that was previously in layer 1, so;
S(2),,..1
S(2),,V(2)n -S(2)n(V(3),,.., -V(3),,)+S(l)n(V(2),,.., -V(2), +V(3),,,J-V(3)n V(2),
(6-8)
The top layer (layer 1) loses some water to the layer beneath it and expands to include the river water that has been added. The salinity in layer 1 after the addition r
of river water may therefore be calculated by:
168
Hydrodynamics of shallow mediterranean estuaries
S(l)n+, = S(l),,V(l),, -S(l),,(V(2),""1 -V(2),, +V(3),""1 -V(3),,)+S,V,: S(l),
2 kg m-3
Figure 7-1 Relationship between observed and modelled stratification at a site in the middle of Harvey Estuary
Figure 7-2 and Figure 7-3 show the problem in more detail. In each, the black line indicates the proportion of observations for which vertical density difference was less than the value indicated, while the blue line indicates the proportion of time in model output for which this was the case. Figure 7-2 shows results for the entire June 1989 to June 1993 period, while Figure 7-3 shows combined results for each spring period (September to November). The water column is more often stratified during spring, and it is at high stratification that the model's predictions are least accurate (Figure 7-4).
219
Applicability of the hy drodynamic model in modelling biogeochemica/ processes
Overall
0.95
0.9 (/)
~0.85 ~
0
c: 0.8 0
~
g-
5.0.75
ill
:;
E 0.7 ::l 0
0.65
0.6
0.55
0
2
4
6
8
10
12
14
16
stratification (density difference between surface and bottom, kg m-3
Figure 7-2 Stratification in Harvey Estuary (September to November, 1989-1993). This shows the proportion of records for which the difference between density of bottom water and density of surface water at a site in the middle of the basin was less than the value given on the X-axis (in kgm~). The black line represents the (approximately weekly) observations, the blue line shows model results (with daily output).
220
Applicability ofthe hydrodynamic model in modelling biogeochemical processes
Spring
0.9
If)
"C
~ 0.8
(J) .....
0 c 0
~8- 0.7 .....
c..
~
~
"5
E 0.6 ::i
0
I
0.5
0.4
0
2
14 12 10 8 6 4 stratification (density difference between surtace and bottom, kg rn-3
16
Figure 7-3 Spring stratification in Harvey Estuary (September to November, 1989-1993). This shows the proportion of records for which the difference between density of bottom water and density of surface water at a site in the middle of the basin was less than the value given on the X-axis (in kgm.J). The black line represents the (approximately weekly) observations, the blue line shows model results (with daily output).
221
Applicability ofthe hydrodynamic model in modelling biogeochemicaf processes
'V
'V
:
··~· ·· · ·· ·· ··· · V·· · ···!· · · ·· ······ ·· ·- .. ···
'
w
w
'V
'b?\'V
5
v'
'V
v:
"~
~ 'V v,"' \7~
~,,
·······-- ...
'V
; ;~
\7
~~- ~'V \7 : \I v .
.... .......... .............
......: " .... .....................
0 --- ----- --- -- --
~... . .
... . ....• ..
..... ·-·· .. . .··-·-·· .
.. ··-
''V -15'--~~~-'-~~~------'-~~~----''----~~~----'-~~~-'-~~~--'
-5
0
5
10 15 modelled density difference
20
25
Figure 7-4 Error in modelled difference between surface and bottom water density.
7.2.1 Possible sources of error
7. 2. 1. 1 Sigma coordinate pressure gradient errors The Princeton Ocean Model (as mentioned in section 6.4) is a sigma-coordinate model. Haney (1991) was the first to report in the oceanographic literature concerns from meteorological modellers (e.g. Janjic, 1977; Mesinger, 1982) regarding errors introduced by sigma coordinate systems when steep gradients exist in the underlying topography. It is suggested that, for "hydrostatic consistency", grids used with sigma-coordinate models should conform to the following condition: !!__ oxH < 1 H &r
(7-1)
where cr is the vertical sigma coordinate, ocr is depth in sigma-coordinates, H is depth (in metres) and OxH is the horizontal change in depth between adjacent grid-cells.
222
d
.I Applicability of the hydrodynamic niodel in niodelling biogeoche1nical processes
!!q
i •1
!1
Mellor et al. (1994 ), however, note that this condition is restrictive, and show that the pressure gradient error decreases as the square of vertical and horizontal grid size. Further, they suggest that initial pressure gradient errors are advectively eliminated
1J
:~
:ii
!1
I
after a long time integration due to small compensatory errors in the density field. These factors should combine to minimise the influence of this error in the present application given the long integration times involved. Mellor et al. (1998) identify an additional error that occurs with three-dimensional flows in sigma coordinate models and which, unlike the first kind of error, does not disappear with long time integrations. They label this vorticity error a sigma error of the second kind (SESK). This kind of error is small and can be reduced by subtracting a horizontally averaged initial density field before computing the baroclinic integrals, or by using a curvilinear grid that follows bathymetric contours (Mellor et al., 1998). POM by default subtracts a horizontally averaged density field before computing baroclinic integrals. Condie and Webster (1999) applied POM to a very shallow (1.2 m) lake, and compared the results with those of a z-coordinate model, DieCAST (Dietrich et al., 1990; Dietrich, 1993; Dietrich and Ko, 1994), and with observations. Their results were consistent with those presented here, in that POM over-estimated vertical diffusion and, hence, underestimated stratification. This did not appear to be an effect of the Mellor-Yamada turbulence closure scheme (Mellor and Yamada, 1982) used by POM, as the z-coordinate model did not exhibit this behaviour, even when the same turbulence closure scheme was used, although DieCAST produced other errors at high wind speeds. The results of Condie and Webster (1999) also suggested that the excessive mixing was not caused by diffusion along sigma levels, but could be related to varied bathymetry and errors in the horizontal pressure gradient term. The problem disappeared when the model was run with flat bathymetry (Condie and Webster, 1999). 7.2.1. l. l
Trials
In attempts to improve prediction of vertical stratification in Harvey Estuary, several additional runs were conducted:
223
1
Applicability of the hydrodynamic model in modelling biogeochemica/ processes
Mellor et aL (1994), however, note that this condition is restrictive, and show that the pressure gradient error decreases as the square of vertical and horizontal grid size. Further, they suggest that initial pressure gradient errors are advectively eliminated after a long time integration due to small compensatory enors in the density field. These factors should combine to minimise the influence of this error in the present application given the long integration times involved. Mellor et aL (1998) identify an additional error that occurs with three-dimensional flows in sigma coordinate models and which, unlike the first kind of enor, does not disappear with long time integrations. They label this vorticity error a sigma error of the second kind (SESK). This kind of error is small and can be reduced by subtracting a horizontally averaged initial density field before computing the baroclinic integrals, or by using a curvilinear grid that follows bathymetric contours (Mellor et al., 1998). POM by default subtracts a horizontally averaged density field before computing baroclinic integrals. Condie and Webster (1999) applied POM to a very shallow (1.2 m) lake, and compared the results with those of a z-coordinate model, DieCAST (Dietrich et aL, 1990; Dietrich, 1993; Dietrich and Ko, 1994), and with observations. Their results were consistent with those presented here, in that POM over-estimated vertical diffusion and, hence, underestimated stratification. This did not appear to be an effect of the Mellor-Yamada turbulence closure scheme (Mellor and Yamada, 1982) used by POM, as the z-coordinate model did not exhibit this behaviour, even when the same turbulence closure scheme was used, although DieCAST produced other errors at high wind speeds. The results of Condie and Webster (1999) also suggested that the excessive mixing was not caused by diffusion along sigma levels, but could be related to varied bathymetry and errors in the horizontal pressure gradient term. The problem disappeared when the model was run with flat bathymetry (Condie and Webster, 1999). 7 .2.1.1.1
Trials
In attempts to improve prediction of vertical stratification in Harvey Estuary, several additional runs were conducted:
223
..•.
,
Applicability of the hydrodyna1nic 1nodeL in niodeLling hiogeocheniical processes
I
a)
With bathymetry smoothed to satisfy condition 7-1, for hydrostatic consistency (shown in blue on Figure 7-5 and Figure 7-6). No significant
l
improvement was found. b)
With a flat basin (i.e. constant depth) to examine whether the problem was eliminated with smooth bathymetry (shown in cyan). Although trial b did show a noticeable difference in vertical stratification in the estuary in spring (Figure 7-6), this run produced incorrect results for seasonal salinity trends, overestimating minimum winter salinities in particular (Figure 7-7). This highlights the significance of bathymetric variations to the hydrodynamics of Harvey Estuary, but indicates that runs with flat bathymetry will not be useful in predicting stratification in the Estuary.
c)
With the vertical diffusivity, KN, set to zero when the vertical buoyancy gradient exceeded 0.1 kg m·3 m· 1. The aim of this run (shown in mauve) was to determine whether vertical mixing was overestimated only when the buoyancy gradient was high. No improvement was found.
d)
With KN set to 70% of the calculated value (green). improvement was found.
No significant
Runs with KN set to lower percentages of the
calculated value were also attempted, but the model became unstable under these conditions, so no useful results were obtained. e)
With the horizontally averaged density field, RMEAN, recalculated after each time step (not shown). (RMEAN is used in POM to reduce errors in density calculations). No improvement in results occurred in this run.
f)
With vertical resolution increased from 7 layers to 21 layers (not shown). No significant improvement in results was found.
224
4 Applicability of the hydrodynamic model in modelling biogeochemical processes
Overall
0.9
~ 0.8
~ ~
0
c:
~ 0.7
e-
a.
·i
0.6
'5
E
::l 0
0.5 observations standard hydrostatic consisten flat basin KN=O when stratified
0.4
KN=0.1 