A spectral wave model (SWAN) coupled with a depth averaged hydrodynamic model (Delft3D) was used to understand the wave and flow dynamics of the Kingston Basin of Lake Ontario during large winter storm events. This model was then used to assess the impact of an offshore wind farm in the Kingston Basin. Results over different model domains with various forcing methods were compared to achieve the highest correlation with wave, current and water level observations from several locations. Storm events were modelled over the complex bathymetry of the basin and results were verified using wave and current profiler data collected during the winters of 2009-10 and 2011-12. Waves were composed of both locally generated wind sea and swell from the main basin of Lake Ontario, while flows throughout the Kingston Basin showed a complex circulation pattern. This circulation is composed of several wind-driven gyres, which are magnified during storm events. The impact of waves on the circulation patterns within the basin is highest in shallow areas where wave breaking drives circulation. To simulate a wind farm, a transmission coefficient was used in the wave model to represent the effects on waves, and an energy loss term was added to the hydrodynamic momentum equations to represent the added drag of the piles on the circulation. The results indicate that the coastal areas in eastern Lake Ontario will be minimally affected. The headlands of Big Sandy Bay, Wolfe Island, could see the largest coastal effects with changes in significant wave height predicted to be < 2%. The majority of impacts to circulation occur in the near-field, with changes in current magnitude of < 0.08 m s-1 (up to 50%). Areas near Wolfe Island exhibit changes of ~ 0.05 m s-1 (30 %), although overall circulation patterns throughout the basin are not affected. The majority of changes to surface waves and wind-driven currents are due to wind farm position with respect to wind direction and the re-direction of flows and waves as they pass through the wind farm. / Thesis (Master, Civil Engineering) -- Queen's University, 2013-09-30 09:30:01.042

Esta tese apresenta um procedimento de identificação experimental de coeficientes hidrodinâmicos de veículos submarinos não tripulados. Apresenta-se o desenvolvimento de uma plataforma experimental para pesquisas em dinâmica, controle e navegação de veículos submarinos. A plataforma experimental inclui: 1) um veículo submarino não tripulado semi-autônomo do tipo estrutura-aberta chamado de LAURS, 2) um sistema multissensorial e multipropulsores para o controle de movimento, e 3) software e arquitetura de controle para a aquisição de dados dos sensores e o controle de movimento. A fim de deduzir as equações do movimento dinâmico do LAURS, apresenta-se a formulação geral das equações hidrodinâmicas não lineares de um veículo submarino com seis graus de liberdade. A partir destas equações gerais acopladas, são deduzidos modelos mais simples para um grau de liberdade (movimento unidirecional) e três graus de liberdade (movimento planar) do veículo. O método de estimação de parâmetros utilizado neste trabalho não requer medidas de aceleração e é baseado na aplicação da técnica dos mínimos quadrados à forma integral das equações dinâmicas do sistema. O procedimento de identificação proposto é baseado na informação dos sensores embarcados. Primeiramente, os coeficientes de arrasto são obtidos a partir de testes de velocidade constante e depois, fixando os coeficientes de arrasto do modelo matemático com estes valores obtidos, são identificados a inércia virtual e os coeficientes de acoplamento a partir de testes de velocidade variável. Nos testes de velocidade variável são aplicadas entradas de força do tipo senoidal. Apresentam-se os valores dos coeficientes hidrodinâmicos para os movimentos de avanço, deriva, arfagem, guinada e caturro do veículo, os quais foram obtidos usando o procedimento de identificação proposto. O desempenho dos modelos dinâmicos identificados é quantitativamente comparado ao movimento do veículo observado experimentalmente. Para o caso dos testes de velocidade constante em avanço e arfagem, foi feita uma comparação dos valores dos coeficientes obtidos usando a abordagem de identificação de sistemas com os dados obtidos a partir dos ensaios de reboque em tanque de provas. Os resultados obtidos validam o procedimento de identificação proposto. Além disso, são apresentados os resultados experimentais obtidos a partir de manobras do tipo zig-zag e é feita uma discussão da identificabilidade de coeficientes de modelos acoplados. Conclui-se que o procedimento de identificação proposto é eficaz na obtenção de valores reais (consistentes com a concepção física do veículo) para os coeficientes hidrodinâmicos de veículos submarinos. A fim de modelar a força hidrodinâmica que atua no veículo em avanço com movimento oscilatório e amplitudes menores do que um comprimento característico, são apresentados os resultados dos ensaios de oscilação forçada usando um mecanismo de movimento planar (MMP). Apresentam-se os valores dos coeficientes de arrasto e de inércia obtidos a partir dos ensaios em avanço no MMP para diferentes números de Keulegan-Carpenter. Os resultados mostram que os coeficientes de arrasto e de inércia para a faixa de velocidades de 0,1 até 0,8m/s dependem fortemente do número de Keulegan-Carpenter e não do número de Reynolds. A partir destes resultados, conclui-se que a amplitude da oscilação do veículo é o principal fator que causa a variação dos coeficientes hidrodinâmicos e não a velocidade do veículo. / In this work, a procedure for experimental identification of hydrodynamic coefficients of unmanned underwater vehicles is presented. At first, the development of a testbed for research on dynamics, control, and navigation of underwater vehicles is presented. This experimental platform includes: 1) a open-frame semi-autonomous unmanned underwater vehicle named LAURS, 2) a multi-sensorial and multi-thruster system for motion control, and 3) software and control architecture for sensor data logging and motion control. In order to derive the LAURS dynamic motion equations, the general formulation of the nonlinear hydrodynamic equations of motion of an underwater vehicle with six degree of freedom is initially presented. From these general coupled equations, simpler formulations with one (unidirectional movement) and three degrees of freedom (planar movement) are derived. The parameter estimation method does not require acceleration measurements and is based on the application of the least squares technique to the integral form of the system dynamic equations. The identification procedure is based on on-board sensor data. First the drag coefficients are obtained from constant velocity tests and afterwards, fixing the drag coefficients in the mathematical model with the obtained values, virtual inertia and coupling coefficients of the vehicle are identified from variable velocity tests. In the tests of variable velocity, sinusoidal force inputs are applied. Values of hydrodynamic coefficients for surge, sway, heave, yaw, and pitch motions are estimated using the proposed identification procedure. Performance of the identified dynamic models is quantitatively compared to the experimentally observed vehicle motion. In the case of constant velocity tests, for the surge and heave motions, comparisons of the hydrodynamic drag coefficient values obtained using the system identification method with data obtained from towing tank tests are presented. Obtained results corroborate for the feasibility of the proposed identification method. Moreover, experimental results obtained from zig-zag maneuvers are presented and the identifiability of coupled dynamic models is discussed. It is possible to conclude that using the proposed method actual hydrodynamic parameters might be estimated. In order to model the hydrodynamic force that acts on the vehicle, in surge motion, with oscillatory movements and with amplitudes that are smaller than or equal to the characteristic length of the LAURS, results of forced oscillation tests in a planar motion mechanism (PMM) are presented. The drag and inertia coefficient values obtained from surge motion tests in the PMM for different Keulegan-Carpenter numbers are presented and discussed. Results illustrates that drag and inertia coefficients, when the vehicle velocity is in the range of 0,1 and 0,8m/s, do not strongly depend on the Reynolds number, however, they are strongly dependent on the Keulegan-Carpenter number. In this context, we can conclude that the oscillation amplitude is the main factor that causes the variation of hydrodynamic coefficients and not the vehicle velocity.

This thesis investigates physical processes in Lake Erie, a large, shallow mid-latitude lake, from two perspectives: climate change impacts on the thermal structure and dissolved oxygen concentration and small-scale eddy dynamics generated by internal Kelvin wave propagation. A three-dimensional hydrodynamic and aquatic ecological coupled model ELCOM-CAEDYM, validated by the field data collected in 2008, is first used to investigate the responses of the thermal structure and dissolved oxygen concentration in Lake Erie to potential changes in air temperature and wind speed. A new method is presented to define spatially and temporally varying regions for the epilimnion, thermocline, and hypolimnion. Four metrics are selected to quantify the characteristics of the thermal structure: mean epilimnion temperature, mean hypolimnion temperature, onset and breakdown of stratification, and thermocline depth. Exploiting the power of the three dimensional model to provide a more authentic characterization of thermal structure in such large lakes, it is shown that patterns inferred from simple isotherm dynamics, as typically done with one dimensional models, are not always accurate. In the dissolved oxygen studies similar analyses are presented. Three factors related to lake hydrodynamics have strong influences on hypolimnetic hypoxia: water temperature, stratification duration, and hypolimnion thickness. The present results show the potential for complicated and interactive effects of climate forcing on important biogeochemical processes in Lake Erie as well as other large mid-latitude lakes. Taking advantage of high performance computing, the generation of eddies when a baroclinic Kelvin wave propagates past a peninsula is studied using the MITgcm. The grid resolution can be refined from 2 km to 200 m in the parallel computing environment. With the finer resolution small-scale processes which cannot be resolved in the coarse resolution applied previously are able to be explored. The eddy dynamics are studied in detail in both an idealized lake and in Lake Erie. This work presents a first attempt at simulating small-scale hydrodynamic processes in large lakes and contributes to our understanding of how energy is moved from large scales (the scale of the basins in Lake Erie) to smaller scales (the scale of the peninsula or Point Pelee).

This thesis investigates physical processes in Lake Erie, a large, shallow mid-latitude lake, from two perspectives: climate change impacts on the thermal structure and dissolved oxygen concentration and small-scale eddy dynamics generated by internal Kelvin wave propagation. A three-dimensional hydrodynamic and aquatic ecological coupled model ELCOM-CAEDYM, validated by the field data collected in 2008, is first used to investigate the responses of the thermal structure and dissolved oxygen concentration in Lake Erie to potential changes in air temperature and wind speed. A new method is presented to define spatially and temporally varying regions for the epilimnion, thermocline, and hypolimnion. Four metrics are selected to quantify the characteristics of the thermal structure: mean epilimnion temperature, mean hypolimnion temperature, onset and breakdown of stratification, and thermocline depth. Exploiting the power of the three dimensional model to provide a more authentic characterization of thermal structure in such large lakes, it is shown that patterns inferred from simple isotherm dynamics, as typically done with one dimensional models, are not always accurate. In the dissolved oxygen studies similar analyses are presented. Three factors related to lake hydrodynamics have strong influences on hypolimnetic hypoxia: water temperature, stratification duration, and hypolimnion thickness. The present results show the potential for complicated and interactive effects of climate forcing on important biogeochemical processes in Lake Erie as well as other large mid-latitude lakes. Taking advantage of high performance computing, the generation of eddies when a baroclinic Kelvin wave propagates past a peninsula is studied using the MITgcm. The grid resolution can be refined from 2 km to 200 m in the parallel computing environment. With the finer resolution small-scale processes which cannot be resolved in the coarse resolution applied previously are able to be explored. The eddy dynamics are studied in detail in both an idealized lake and in Lake Erie. This work presents a first attempt at simulating small-scale hydrodynamic processes in large lakes and contributes to our understanding of how energy is moved from large scales (the scale of the basins in Lake Erie) to smaller scales (the scale of the peninsula or Point Pelee).

Baigiamąjame darbe analizuojama galimybė netradiciniais Lietuvoje metodais (naudojant hidraulinį modelį) įvertinti reguliuoto upelio deformacijų įtaką jo pralaidumui, kadangi dažnai atsitinka taip, kad numatomos nemažos lėšos remontui, kai tuo tarpu upelio pralaidumo charakteristikos dar pakankamai geros. Modelio kalibravimui ir hidrauliniam modeliavimui naudojami realiai išmatuoti ir projektiniai duomenys. Hidrodinaminis modelis sudaromas tuo tikslu, kad būtų galima patikrinti pasirinkto upelio ruožo vagos ir tame ruože esančios pralaidos matmenis kaip pakraštines sąlygas, naudojant iš projekto paimtais hidrologiniais skaičiavimais nustatytus debitus ir įvertinti tėkmės sąlygas esant projektinėms geometrinėms charakteristikoms ir realiai išmatuotoms dabartinėje upelio būklėje. Atliekant natūrinius tyrimus buvo nustatinėjamos šios reguliuoto upelio charakteristikos: debitas, vandens lygiai, vandens paviršiaus nuolydis, šiurkštumo koeficientai; griovyje esančių pralaidų būklė. Darbo tikslas - panaudojant programą HEC-RAS sukurti pasirinkto reguliuoto upelio ruožo hidrodinaminį modelį ir patikrinti griovio ir pralaidos pralaidumo pokyčius lyginant projektinius ir natūroje išmatuotus parametrus. Tyrimo uždaviniai: 1. Literatūros apžvalgoje išnagrinėti atvirų vagų hidraulinio modeliavimo metodus, programinę įrangą ir praktinio taikymo atvejus. 2. Išmokti dirbti su programine įranga HEC-RAS. 3. Sukurti pasirinkto upelio ruožo hidrodinaminį modelį ir atlikti... [toliau žr. visą tekstą] / Final work analyzes the possibility to evaluate the impact of deformations of regulated stream on its capacity by the methods that are non-traditional in Lithuania (using a hydraulic model), as often is the case when substantial funding is provided for repair, while the stream capacity characteristics are still good enough. The actual measured and projected data is used for model calibration and hydraulic modelling. Creation of hydrodynamic model aims to verify the dimensions of bed and culverts in the selected stream section using the discharges based on the project hydrological calculations as peripheral conditions and to evaluate the flow conditions at the projected geometric characteristics and at those measured in the current state of the stream. Field research determined the following characteristics of the stream: yield, water levels, water surface slope, roughness coefficients; condition of the equipment in the ditch: state of culverts. Aim of the work – by modelling the established flow to determine the longitudinal water level profiles for the both cases and to evaluate the differences. Results of the work allow stating that lateral trench deformations have negligible impact on the flow conditions. Objectives of the research: 1. to analyze the hydraulic modelling techniques of open beds, software and cases of practical application; 2. to learn working with the software HEC-RAS; 3. to create hydrodynamic model of the selected stream... [to full text]

Estuaries can be considered as vital natural resources and are unique ecosystems at the interface between terrestrial and marine environments. The increase of population density centred on these coastal features and associated anthropogenic activities such as trade, industry, agriculture and recreation can adversely affect these sensitive environments. The Pumicestone Passage, located in northern Moreton Bay, Australia, is one such estuarine environment where there are concerns about degradation of water quality resulting from rapid land use change. These changes are both immediate to the Passage and within its wider catchment. Of notable concern are the outbreaks of Lyngbya (a toxic blue-green algae) in the Passage itself and near its interface with Deception Bay to the south. Other factors of concern are increased suspended and dissolved loads, and maintenance of ecosystem integrity. In this study, numerical modelling, graphical methods and water surface elevation and current velocity parameter calculations are used to describe hydrological processes in the Pumicestone Passage. A hydrodynamic model is developed using the modelling software SMS and RMA2 as a foundation for future hydrodynamic and water quality modelling. In addition, observed data are used to interpret general hydrodynamic behaviour in the passage, and determine various parameters for use in model development and calibration. Tidal prediction is also discussed and used for model calibration. To support the modelling and for preliminary interpretation of hydrodynamic processes within the Passage, measurements were made in the field of (a) water surface elevation variation at 17 sites; (b) tidal current velocities in four of the tributary creeks and at the northern boundary; (c) volumetric flow rates at two cross-sections within the Passage; and (d) cross-sectional bathymetry at sites where tidal current velocities were measured in the creeks. In general, examination of the observational data reveals a number of important processes in the Pumicestone Passage. Almost all sites within Pumicestone Passage and its tributaries are flood dominant indicating that tidal storage and bottom friction effects are significant. Mesotidal ranges occur at sites close to the southern boundary of the passage, however, bottom friction greatly reduces the tidal response at the remaining sites which results in microtidal ranges. The influence of both the southern and northern tides can be seen in the deformation of tidal waveforms in the central passage. Extensive intertidal areas at and inside the northern inlet to the Passage markedly reduce tidal ranges in the northern estuary and its tributary creeks. Issues involved in hydrodynamic model development and performance are discussed. Overall, model results for the southern estuary have satisfactory correlation with observed data whereas model results for the northern estuary are less satisfactory. In addition, water surface elevation variation model results are generally more accurate than tidal current velocity model results. Reasons for the differences between model and observed values are considered and possible solutions given. Factors discussed relate to boundary condition locations, resolution of bathymetric and geographical data, mesh development methods and parameter assignment.

Esta tese apresenta um procedimento de identificação experimental de coeficientes hidrodinâmicos de veículos submarinos não tripulados. Apresenta-se o desenvolvimento de uma plataforma experimental para pesquisas em dinâmica, controle e navegação de veículos submarinos. A plataforma experimental inclui: 1) um veículo submarino não tripulado semi-autônomo do tipo estrutura-aberta chamado de LAURS, 2) um sistema multissensorial e multipropulsores para o controle de movimento, e 3) software e arquitetura de controle para a aquisição de dados dos sensores e o controle de movimento. A fim de deduzir as equações do movimento dinâmico do LAURS, apresenta-se a formulação geral das equações hidrodinâmicas não lineares de um veículo submarino com seis graus de liberdade. A partir destas equações gerais acopladas, são deduzidos modelos mais simples para um grau de liberdade (movimento unidirecional) e três graus de liberdade (movimento planar) do veículo. O método de estimação de parâmetros utilizado neste trabalho não requer medidas de aceleração e é baseado na aplicação da técnica dos mínimos quadrados à forma integral das equações dinâmicas do sistema. O procedimento de identificação proposto é baseado na informação dos sensores embarcados. Primeiramente, os coeficientes de arrasto são obtidos a partir de testes de velocidade constante e depois, fixando os coeficientes de arrasto do modelo matemático com estes valores obtidos, são identificados a inércia virtual e os coeficientes de acoplamento a partir de testes de velocidade variável. Nos testes de velocidade variável são aplicadas entradas de força do tipo senoidal. Apresentam-se os valores dos coeficientes hidrodinâmicos para os movimentos de avanço, deriva, arfagem, guinada e caturro do veículo, os quais foram obtidos usando o procedimento de identificação proposto. O desempenho dos modelos dinâmicos identificados é quantitativamente comparado ao movimento do veículo observado experimentalmente. Para o caso dos testes de velocidade constante em avanço e arfagem, foi feita uma comparação dos valores dos coeficientes obtidos usando a abordagem de identificação de sistemas com os dados obtidos a partir dos ensaios de reboque em tanque de provas. Os resultados obtidos validam o procedimento de identificação proposto. Além disso, são apresentados os resultados experimentais obtidos a partir de manobras do tipo zig-zag e é feita uma discussão da identificabilidade de coeficientes de modelos acoplados. Conclui-se que o procedimento de identificação proposto é eficaz na obtenção de valores reais (consistentes com a concepção física do veículo) para os coeficientes hidrodinâmicos de veículos submarinos. A fim de modelar a força hidrodinâmica que atua no veículo em avanço com movimento oscilatório e amplitudes menores do que um comprimento característico, são apresentados os resultados dos ensaios de oscilação forçada usando um mecanismo de movimento planar (MMP). Apresentam-se os valores dos coeficientes de arrasto e de inércia obtidos a partir dos ensaios em avanço no MMP para diferentes números de Keulegan-Carpenter. Os resultados mostram que os coeficientes de arrasto e de inércia para a faixa de velocidades de 0,1 até 0,8m/s dependem fortemente do número de Keulegan-Carpenter e não do número de Reynolds. A partir destes resultados, conclui-se que a amplitude da oscilação do veículo é o principal fator que causa a variação dos coeficientes hidrodinâmicos e não a velocidade do veículo. / In this work, a procedure for experimental identification of hydrodynamic coefficients of unmanned underwater vehicles is presented. At first, the development of a testbed for research on dynamics, control, and navigation of underwater vehicles is presented. This experimental platform includes: 1) a open-frame semi-autonomous unmanned underwater vehicle named LAURS, 2) a multi-sensorial and multi-thruster system for motion control, and 3) software and control architecture for sensor data logging and motion control. In order to derive the LAURS dynamic motion equations, the general formulation of the nonlinear hydrodynamic equations of motion of an underwater vehicle with six degree of freedom is initially presented. From these general coupled equations, simpler formulations with one (unidirectional movement) and three degrees of freedom (planar movement) are derived. The parameter estimation method does not require acceleration measurements and is based on the application of the least squares technique to the integral form of the system dynamic equations. The identification procedure is based on on-board sensor data. First the drag coefficients are obtained from constant velocity tests and afterwards, fixing the drag coefficients in the mathematical model with the obtained values, virtual inertia and coupling coefficients of the vehicle are identified from variable velocity tests. In the tests of variable velocity, sinusoidal force inputs are applied. Values of hydrodynamic coefficients for surge, sway, heave, yaw, and pitch motions are estimated using the proposed identification procedure. Performance of the identified dynamic models is quantitatively compared to the experimentally observed vehicle motion. In the case of constant velocity tests, for the surge and heave motions, comparisons of the hydrodynamic drag coefficient values obtained using the system identification method with data obtained from towing tank tests are presented. Obtained results corroborate for the feasibility of the proposed identification method. Moreover, experimental results obtained from zig-zag maneuvers are presented and the identifiability of coupled dynamic models is discussed. It is possible to conclude that using the proposed method actual hydrodynamic parameters might be estimated. In order to model the hydrodynamic force that acts on the vehicle, in surge motion, with oscillatory movements and with amplitudes that are smaller than or equal to the characteristic length of the LAURS, results of forced oscillation tests in a planar motion mechanism (PMM) are presented. The drag and inertia coefficient values obtained from surge motion tests in the PMM for different Keulegan-Carpenter numbers are presented and discussed. Results illustrates that drag and inertia coefficients, when the vehicle velocity is in the range of 0,1 and 0,8m/s, do not strongly depend on the Reynolds number, however, they are strongly dependent on the Keulegan-Carpenter number. In this context, we can conclude that the oscillation amplitude is the main factor that causes the variation of hydrodynamic coefficients and not the vehicle velocity.

The increasing frequency of Cyanobacterial blooms in freshwater bodies raise concerns around the globe. The consequences of this phenomenon impact not only human health but the entire surrounding ecosystem. During the past decades, numerical modelling has been increasingly used to investigate and study aquatic systems. Hydrodynamic and Ecological models are developed in parallel to simulate processes, evaluate potential remedies, and investigate future scenarios. This project aimed at developing a 3D hydrodynamic and water quality (ecological) model to assess the eutrophication conditions of Lake Vomb under current and future scenarios. MIKE 3 FM software was used in the analysis with meteorological, hydrological, and water quality data. The hydrodynamic model performance was satisfactory in terms of water temperature simulation with root-mean-square-error (RMSE) ranging between 0.38-1.2 oC. In the ecological model, Chlorophyll-a (Chl-a) was simulated as a surrogate for Cyanobacteria. The model was adequate in simulating Chl-a concentrations with a Nash-Sutcliffe efficiency (NSE) of 0.94 during calibration and 0.84 after validation. The results showed that Lake Vomb’s nutrient concentrations are highly influenced by external nitrogen loading and internal phosphorus loading. The results also showed that Chl-a levels are correlated with the total phosphorus levels in the lake. Future water quality projections were attempted through two Representative Concentration Pathways (RCPs) for the year 2050. The projections utilized only changes in air temperatures and precipitations and suggested significant increase in Cyanobacteria biomass independence of changes in external nutrient loading.

Per and polyfluoroalkyl substances (PFAS) are found ubiquitously in the environment across the globe.These substances have high persistence due to the strong carbon and fluorine bond. In the aquaticenvironment, due to high persistence, these substances don’t decay easily and are detected on the surfaceas well as in groundwater sources. Human exposure to PFAS has been observed due to ingestion of PFAScontaminated food and water which has an adverse effect on the human health. High concentrations ofPerfluoroalkyl Octanoic Sulphonate (PFOS) and Perfluoroalkyl Hexanoic Sulphonate (PFHxS) wereobserved in Lake Ekoln. The main aim of this study was to analyze the spread of PFOS and PFHxS in thelake. Hydrodynamic modelling of flow and water quality modelling in the lake was performed usingMIKE 3 FM software for the evaluation of the spread pattern of PFOS and PFHxS in the lake. Twoscenarios with the different mass fluxes of PFOS and PFHxS for Fyrisån were assumed for investigatingthe uncertainties and influence of contribution from Fyrisån. Additionally, conductivity from differentinflows was modelled as a passive tracer for understanding the water quality and the circulation in thelake. The results of the simulation showed that Fyrisån and Kungsängsverket are major contributors of PFOSand PFHxS to the lake. Similarly, the analysis of the current spread of PFOS and PFHxS shows that theuncertainty in the model is high and is dependent mainly on the assumption of mass flux from theFyrisån. Due to the lack of sampling data on the concentration of PFOS and PFHxS, it was difficult to geta reliable assumption for the mass flux from the Fyrisån. From the study, it was identified that thesampled concentration of PFOS and PFHxS also had certain variations which might be due to theinfluence of concentration from different sources and processes. Likewise, the simulation result of PFOSand PFHxS was observed to have a similar pattern of spread. Although PFOS is a long-chain PFAS andPFHxS is a short-chain PFAS and they have different physio-chemical properties, the spread patternswere observed to be similar. As only the hydrodynamic processes were influencing the simulation for thespread of PFOS and PFHxS in the lake and other physiochemical processes such as sedimentation andbioaccumulation were not included in the model, the simulated PFAS results were found similar. To conclude, the study shows that the spread of PFOS and PFHxS is mainly influenced by the flow andconcentration in Fyrisån. Also, higher uncertainty in the model performance was observed due to theissue of reliable mass flux estimation from Fyrisån. Similarly, the influence of processes such assedimentation and bio accumulation are necessary to be included in the model for analysis of spread ofPFAS with different physio-chemical properties.

The onset of summer stratification in temperate lakes and reservoirs forces a decoupling of the hypolimnion from the epilimnion that is sustained by strong density gradients in the metalimnion. These strong gradients act as a barrier to the vertical transport of mass and scalars leading to bottom anoxia and subsequent nutrient release from the sediments. The stratification is intermittently overcome by turbulent mixing events that redistribute mass, heat, dissolved parameters and particulates in the vertical. The redistribution of ecological parameters then exerts some control over the ecological response of the lake. This dissertation is focused on the physics of deep vertical mixing that occurs beneath the well-mixed surface layer in stratified lakes and reservoirs. The overall aim is to improve the ability of numerical models to reproduce deep vertical mixing, thus providing better tools for water quality prediction and management. In the first part of this research the framework of a one-dimensional mixed-layer hydrodynamic model was used to construct a pseudo two-dimensional model that computes vertical fluxes generated by deep mixing processes. The parameterisations developed for the model were based on the relationship found between lake-wide vertical buoyancy flux and the first-order internal wave response of the lake to surface wind forcing. The ability of the model to reproduce the observed thermal structure in a range of lakes and reservoirs was greatly improved by incorporating an explicit turbulent benthic boundary layer routine. Although laterally-integrated models reproduce the net effect of turbulent mixing in a vertical sense, they fail to resolve the transient distribution of turbulent mixing events triggered by local flow properties defined at far smaller scales. Importantly, the distribution of events may promote tertiary motions and ecological niches. In the second part of the study a large body of microstructure data collected in Lake Kinneret, Israel, was used to show that the nature of turbulent mixing events varied considerably between the epilimnion, metalimnion, hypolimnion and benthic boundary layer, yet the turbulent scales of the events and the buoyancy flux they produced collapsed into functions of the local gradient Richardson number. It was found that the most intense events in the metalimnion were triggered by high-frequency waves generated near the surface that grew and imparted a strain on the metalimnion density field, which led to secondary instabilities with low gradient Richardson numbers. The microstructure observations suggest that the local gradient Richardson number could be used to parameterise vertical mixing in coarse-grid numerical models of lakes and reservoirs. However, any effort to incorporate such parameterisations becomes meaningless without measures to reduce numerical diffusion, which often dominates over parameterised physical mixing. As a third part of the research, an explicit filtering tool was developed to negate numerical diffusion in a threedimensional hydrodynamic model. The adaptive filter ensured that temperature gradients in the metalimnion remained within bounds of the measured values and so the computation preserved the spectrum of internal wave motions that trigger diapycnal mixing events in the deeper reaches of a lake. The results showed that the ratio of physical to numerical diffusion is dictated by the character of the dominant internal wave motions.