David P. Hamilton

Prediction of water quality in lakes and reservoirs. Part I — Model description

Abstract A one-dimensional water quality model (DYRESM Water Quality) is described which combines a process based hydrodynamic model (DYRESM) with numerical descriptions of phytoplankton production, nutrient cycling, the oxygen budget and particle dynamics. The hydrodynamic component is free from calibration, which ensures that it is readily transferable to other lakes and reservoirs. This improves water quality predictions derived for different hydrodynamic forcing events. It also allows for identification of the specific hydrodynamic processes that influence water quality. The water quality component consists of 13 state variables which may include up to three algal groups, BOD, dissolved oxygen and four components of the dissolved oxygen budget (inflows, biochemical processes, surface aeration and oxygen present in the reservoir at the start of a simulation), nutrients (PO 4 P, NO 3 N, NH 4 N, TP and TN) and inorganic particles. The particle model simulates settling and flocculation/deflocculation of up to seven different size classes of particles. The hydrodynamic, water quality and particle models interact on a sub-daily time step. Forcing data for the model are entered as daily-averaged values. The ecological component requires calibration for each new application through adjustment of several different biological and chemical parameters. Literature ranges for these parameters are wide, but provided the process description is correct, many of the parameters can be validated with measured data.

Prediction of water quality in lakes and reservoirs: Part II - Model calibration, sensitivity analysis and application

A range was assigned to each of the parameters used in the ecological component of the DYRESM Water Quality model based on values found in the literature. The sensitivity of the model to changes in these parameters was determined by individually adjusting parameters to the maximum or minimum of their assigned ranges whilst keeping all other parameters at their assigned means. The effects of these changes were quantified through the mean, the vertical distribution and the temporal variation in chlorophyll a and dissolved oxygen concentrations over a 200 day period, using data for Prospect Reservoir. Parameters that influenced kinetics of phosphorus by phytoplankton, the minimum internal concentration, the half saturation constant and the maximum uptake rate, were amongst the most important determinants of all three measures of chlorophyll a. These parameters were also important determinants of the mean concentration and the vertical distribution of dissolved oxygen, through effects on photosynthetic oxygen production. Sediment oxygen demand had a significant effect on the mean concentration and vertical distribution of dissolved oxygen, phytoplankton density altered the vertical distribution of chlorophyll a, and rates of phytoplankton growth, respiration and mortality influenced the mean concentration of chlorophyll a. The model parameters were calibrated for the 200 day period and the model was validated over an additional 306 days. The mean errors between simulated and measured temperatures, and dissolved oxygen and chlorophyll a concentrations, as percentages of the measured values, were 5.2, 9.9 and 24.1% respectively. Simulated nutrient concentrations (PO4P, NO3N, NH4N, total phosphorus and total nitrogen) all reflected the general temporal and spatial trends observed in the measured data from Prospect Reservoir.

Simulation of vertical position of buoyancy regulating Microcystis aeruginosa in a shallow eutrophic lake

Abstract: Vertical distributions of the cyanobacterium Microcystis aeruginosa are examined in a shallow lake in relation to mixing and thermal stratification over three days. A model of buoyancy regulation by Microcystis aeruginosa, applicable for turbulent environments, is coupled with a one-dimensional hydrodynamic model. The coupled model is applied to Thomsons Lake in Western Australia to examine the relationship between buoyancy regulation and the daily stratification/destratification cycle. The vertical distribution of Microcystis aeruginosa in Thomsons Lake depends on the carbohydrate ballast dynamics and the colony size. When thermal stratification occurs, all the simulations show a similar general pattern of diurnal vertical migration of the Microcystis aeruginosa colonies. The colonies accumulate at the surface during the night and in the morning the colonies lose buoyancy, which leads to a reduction by

Wind induced sediment resuspension: a lake-wide model

\sim

Modelling the impact of zooplankton grazing on phytoplankton biomass during a dinoflagellate bloom in the Swan River Estuary, Western Australia

50% in colony concentration in the top 0.2—0.3 m of the water column. Afternoon winds redistribute the population over the entire water column. When the lake is fully mixed, the vertical migration pattern of the Microcystis aeruginosa colonies may be affected, depending on the colony size and the intensity of the mixing.

Impacts of hydrological changes on phytoplankton succession in the Swan River, Western Australia

Abstract A lake-wide suspended sediment model has been developed and subsequently coupled with a horizontal circulation model to predict depth-averaged suspended sediment concentrations throughout a lake. Resuspension is induced by shear stresses at the sediment-water interface which are considered to be a result of wave action. These shear stresses are calculated using linear wave theory. The model allows for winds from any direction to generate the waves and associated stresses. Sediment deposition is characterised by a formulation which describes the change in suspended sediment concentration with time rather than relying on one or more representative settling velocities. Sediment resuspension is modelled as being independent of deposition. The processes of advection and diffusion in redistributing resuspended sediments horizontally are accounted for in the hydrodynamic model. The result is a general, process based model for predicting depth-averaged suspended sediment concentrations throughout a shallow lake. The model was tested against data collected during 1993 and 1994 from Thomsons Lake, Western Australia.

Flushing of dense, hypoxic water from a cavity of the Swan River estuary, Western Australia

Abstract Ingestion rates of zooplankton were measured in the Swan River Estuary and in the laboratory. These data were used together with data from the literature on phytoplankton and zooplankton physiological parameters, to provide input data to a model of phytoplankton and zooplankton dynamics in the Swan River Estuary. The model also used measured environmental data (nutrient concentrations, light, water temperature and salinity) to simulate phytoplankton biomass (differentiated as four groups) and zooplankton biomass (differentiated as three size classes) at a site in the Swan River Estuary. Zooplankton grazing was shown to be highly important in attenuating a dinoflagellate bloom that occurred over a 3-week model simulation period.

Comparison of two 2-dimensional, laterally averaged hydrodynamic model applications to the Swan River Estuary

The Swan River estuary, Western Australia, has undergone substantial hydrological modifications since pre-European settlement. Land clearing has increased discharge from some major tributaries roughly 5-fold, while weirs and reservoirs for water supply have mitigated this increase and reduced the duration of discharge to the estuary. Nutrient loads have increased disproportionately with flow and are now approximately 20-times higher than pre-European levels. We explore the individual and collective impacts of these hydrological changes on the Swan River estuary using a coupled hydrodynamic-ecological numerical model. The simulation results indicate that despite increased hydraulic flushing and reduced residence times, increases in nutrient loads are the dominant perturbation producing increases in the incidence and peak biomass of blooms of both estuarine and freshwater phytoplankton. Changes in salinity associated with altered seasonal freshwater discharge have a limited impact on phytoplankton dynamics.

Controlling the indirect effects of flow diversions on water quality in an Australian Reservoir

Flushing of dense water from cavities of the upper reaches of the Swan River estuary in Western Australia was investigated using measured salinity and dissolved oxygen profiles and a two-dimensional, laterally averaged hydrodynamic model (TISAT). Seasonal flushing of dense, hypoxic bottom waters from a relatively deep site took place over ∼3 days at the onset of winter in 1994. Model simulations of the purging of this dense water did not correspond closely with changes in the densimetric Froude number. Purging, expressed as depth of the halocline as a fraction of the total cavity depth, occurred when the simulated mean horizontal velocity at 2 m depth (top of cavity) changed from negative to strongly positive, indicating arrest of upstream flow and continuous downstream flow. This corresponded to freshwater discharge of about 50 m3 s−1. Oxygen depletion of bottom waters was closely related to stratification. Oxygen dynamics at the onset of winter river flow was analysed using an exponential decay model, assuning that there was no net inflow or outflow across the halocline and thus no vertical transport of oxygen during a period of strong stratification. The rate constant for oxygen decay at Ron Courtney Island (RCI) was estimated to be 0.232 d−1 for this period. Bottom waters at RCI declined to less than 1 mg 1−1 prior to complete flushing through increased river flows. This study provided in sights to how freshwater flows may be allocated to maintain suitable oxygen levels in the bottom waters of estuarine cavities.

Effects of zinc toxicity on biochemical composition of muscle and liver of murrel (Channa punctatus)

Two laterally averaged, two-dimensional models; TISAT and CE-QUAL-W2, have been applied to the Swan River Estuary. Both models use a finite difference scheme. The time step in the TISAT model is restricted by Courant–Friedrichs–Lewy (CFL) criterion, but this limitation is overcome in CE-QUAL-W2 by using a semi-implicit numerical scheme. In CE-QUAL-W2 numerical diffusion has been reduced relative to TISAT by implementing a third-order QUICKEST horizontal/vertical transport scheme and time-weighted, implicit vertical advection. Predictions of salinity distributions from both models are compared with field observations for the Swan River Estuary in 1994. Both models simulated measured data reasonably well when there were accurate input data. During the period of intense stratification, predictions from CE-QUAL-W2 compared better to measured data than those from TISAT due to the improved transport scheme used in CE-QUAL-W2. Both models gave poor comparisons with measured data at the start of winter, when we hypothesised that the volumetric contribution of ungauged urban drains was particularly significant. In general, CE-QUAL-W2 offers significant advantages over TISAT in simulations of seasonal density stratification in the Swan River Estuary, particularly when there is intense stratification.