<p>Urbanization causes myriad changes in watershed processes, ultimately disrupting the structure and function of stream ecosystems. Urban development introduces contaminants (human waste, pesticides, industrial chemicals). Impervious surfaces and artificial drainage systems speed the delivery of contaminants to streams, while bypassing soil filtration and local riparian processes that can mitigate the impacts of these contaminants, and disrupting the timing and volume of hydrologic patterns. Aquatic habitats where biota live are degraded by sedimentation, channel incision, floodplain disconnection, substrate alteration and elimination of reach diversity. These compounding changes ultimately lead to alteration of invertebrate community structure and function. Because the effects of urbanization on stream ecosystems are complex, multilayered, and interacting, modeling these effects presents many unique challenges, including: addressing and quantifying processes at multiple scales, representing major interrelated simultaneously acting dynamics at the system level, incorporating uncertainty resulting from imperfect knowledge, imperfect data, and environmental variability, and integrating multiple sources of available information about the system into the modeling construct. These challenges can be addressed by using a Bayesian modeling approach. Specifically, the use of multilevel hierarchical models and Bayesian network models allows the modeler to harness the hierarchical nature of the U.S. Geological Survey (USGS) Effect of Urbanization on Stream Ecosystems (EUSE) dataset to predict invertebrate response at both basin and regional levels, concisely represent and parameterize this system of complicated cause and effect relationships and uncertainties, calculate the full probabilistic function of all variables efficiently as the product of more manageable conditional probabilities, and includes both expert knowledge and data. Utilizing this Bayesian framework, this dissertation develops a series of statistically rigorous and ecologically interpretable models predicting the effect of urbanization on invertebrates, as well as a unique, systematic methodology that creates an informed expert prior and then updates this prior with available data using conjugate Dirichlet-multinomial distribution forms. The resulting models elucidate differences between regional responses to urbanization (particularly due to background agriculture and precipitation) and address the influences of multiple urban induced stressors acting simultaneously from a new system-level perspective. These Bayesian modeling approaches quantify previously unexplained regional differences in biotic response to urbanization, capture multiple interacting environmental and ecological processes affected by urbanization, and ultimately link urbanization effects on stream biota to a management context such that these models describe and quantify how changes in drivers lead to changes in regulatory endpoint (the Biological Condition Gradient; BCG).</p> / Dissertation

Water quality is of great importance for ecosystems and society. This thesis characterized and modeled the variation in several key constituents of Swedish surface waters, with particular consideration given to intra-annual variability and sensitivity to climate change. Cyanobacterial data from 29 lakes and basins as well as total organic carbon (TOC) from 215 watercourses were used. Extensive data on catchment characteristics, morphometry, discharge, temperature and other water chemistry data were also analyzed. Models characterizing the seasonality in cyanobacterial concentration and relative cyanobacterial abundance were developed with common lake variables. Concentrations of TOC, iron and absorbance were simulated using discharge, seasonality and long-term trend terms in the Fluxmaster modeling system. Spatial patterns in these model terms were investigated, and the sensitivity of cyanobacteria and TOC to future climate was explored. Nutrients were the major control on cyanobacterial concentration seasonality, while temperature was more important for relative cyanobacterial abundance. No cyanobacterial blooms occurred below a total phosphorus threshold of 20 µg l-1. Discharge and seasonality explained much of the intra-annual variability in TOC, but catchment characteristics could only explain a limited amount of the spatial patterns in the sensitivity to these influences. North of Limes Norrlandicus the discharge term had a larger impact on the TOC concentration in large catchments than in small catchments, while south of Limes Norrlandicus the seasonality had a larger impact in small catchments than in larger catchments. According to the climate change scenarios, both TOC and cyanobacterial concentrations will be higher in the future. The cyanobacterial dominance will start earlier and persist longer. The spring TOC concentration peak will come earlier. The changes in TOC loads are more uncertain due to predicted declines in discharge. Parsimonious statistical regression models could explain observed variability in cyanobacteria and TOC. For predictions, these models assume that future aquatic ecosystems will exhibit the same sensitivity to major drivers as in the past. If this proves not to be the case, the modeling can serve as a sentinel for changing catchment function as indicated by degradation in model performance when calibrations on older data are used to model later observations.

Water scarcity has become a global problem due to diminishing water resource and pollution of the remaining resources. The problems arising fromwater scarcity are exacerbated rapid urbanisation and industrialisation. Water quality management systems are introduced. Numerous water management methods exist some of which, if applied e ectively, can remedy these problems. In South Africa, water management systems are urgently needed to start addressing issues around the longterm sustainability of our limited water resource. Water quality modelling is one of the tools employed to assist in validating decisions made during the planning phase of a water quality management system. It also provides a means of exploring viable options to be considered when these decisions are to be made. A range of management options exist and implementing all of them may prove costly, therefore optimisation techniques are utilised to narrow down options to the most e ective and least costly among the available choices. Commonly, water quality models are used to predict concentrations in the river from which constraint equations are generated. The constraint equations are used in optimisation models to generate feasible solutions by either maximising or minimising the objective function. In this case the objective function is wastewater treatment cost. Constraints equations are based on the set in-stream water quality standard at selected theoretical measuring stations (checkpoints) in the stream and a feasible solution is one that suggests a treatment method that will ensure water quality standards are met at the lowest regional treatment cost. This study focused on the Upper Olifants river catchment near Witbank in Mpumalanga province. This catchment is subjected to extensive wastewater e uents from various mining operations and wastewater treatment plants. The aim here was to develop a water quality model for predicting dissolved oxygen (DO) concentration in the river, and to use a modelling approach to generate constraint equations for the system. A Streeter-Phelps stream simulation model was employed to predict DO concentration in the river. A mixed-integer programming technique was then used to evaluate the impact of nine wastewater treatment facilities discharging e uent into the river. Treatment levels were varied to test model reliability. The coupled stream simulation and optimisation model produced feasible solutions under 2 minutes, with each solution suggesting a range of treatment levels which ensured that the critical DO concentration was above 5 mg/L and the most stringent DO concentration the system could manage without violations anywhere else in the stream was obtained to be 7mg/L. / Dissertation (MEng)--University of Pretoria, 2013. / gm2014 / Chemical Engineering / unrestricted

An important aspect of modeling water quality in water distribution system (WDS) is to predict the temporal and spatial distribution of disinfectant, and the formation of disinfection byproducts (DBP). Consistent efforts have been made to investigate the kinetics of chlorine decay and trihalomethanes (THM) formation in WDS, which are caused by the reaction of chlorine with natural organic matter (NOM). NOM is a heterogeneous mixture of complex compounds. Each specific compound shows individual reactivity with chlorine. Therefore, to better understand and predict the kinetics of chlorine decay and THM formation, the core assumption of this study was established. That is, the variable reactivity of NOM should be involved into the second order kinetics model. Specifically, each single reactive site provided by NOM shows its individual reactivity towards chlorine decay and THM formation, which can be expressed by its individual reaction rate constant, while the mixture compounds of NOM shows the overall reactivity with respect to chlorine decay and THM formation, which should be expressed by an overall reaction rate coefficient. With the reaction progress, the overall reaction rate coefficient was assumed to be continuously decreasing with the reaction time due to the decreased concentration and reactivity of NOM. The decreased overall reaction rate coefficient was referred as a variable reaction rate coefficient (VRRC) in this dissertation. The VRRC was calculated as an exponential function with limited model parameters, which was only related to the characteristics of NOM but independent of chlorine concentration. By introducing the VRRC, the required model parameters were reduced, calibration was simplified and therefore the models showed abilities for a wider application. Consequently, a systematic work has been carried out to develop VRRC models of chlorine decay and THM formation based on the above mentioned assumption, and further extend and validate the models under different chlorination conditions. The following specific topics were addressed. • A VRRC model of chlorine decay was developed and validated under different conditions, including different initial chlorine dosages, different temperature, rechlorination and water mixing conditions. • Based on the identical assumption applied in the chlorine decay model, a VRRC model of THM formation was also developed and also validated under different chlorination conditions, such as, different initial chlorine dosages, changeable temperature condition and rechlorination. • The model application was extended from bulk reaction to wall reaction by considering the presence of pipe deposits in the WDS. Both the chlroine decay and THM formation models were advanced and validated when pipe deposits were present in water. • To further validate the core assumption proposed in this dissertation and also validate the proposed models have a wide application, the residual chlorine and THM concentrations in chlorinated swimming pools were predicted. Both the model accuracy and model adequacy were evaluated through statistical analysis. The results showed that the proposed models were well suited for application in water quality modeling for distribution systems.

I analyzed and compared impacts from global climate change (GCC) and land use change to Deer Creek (United States) a temperate reservoir and Aguamilpa (Mexico), a tropical reservoir by using calibrated CE-QUAL-W2 (W2) water quality and hydrodynamic models based on field data over an extended time period. I evaluated and compared the sensitivity to predicted GCC and land use changes. I individually evaluated changes to air temperature (TAIR), inflow rates (Q), and nutrient loads (PO4-P and NO3-NO2-N) followed by analysis of worst case scenarios. I developed analysis methods using indexes to represent the total reservoir change calculated using the total parameter mass (i.e., algae, dissolved oxygen, total dissolved solids) normalized by the reservoir volume to eliminate apparent mass changes due to volume changes. These indexes have units of average concentrations, but are better thought of as a global reservoir index or normalized concentration. These indexes allow analysis of the total reservoir and not just specific zones. Total normalized algal concentrations were impacted more by changes in nutrient inflows (land use) in both reservoirs than to changes in TAIR and Q. For Deer Creek, PO4-P changes significantly increased normalized algal concentrations in the reservoir and in dam releases when PO4-P inflow was increased by 50%. Aguamilpa was more sensitive to NO3-NO2-N changes, exhibiting significant increases in normalized algal concentration for the +50% NO3-NO2-N simulation. Both reservoirs showed small changes to normalized algal concentration for the +3ºC TAIR simulation with the largest changes occurring during warm seasons. However, Deer Creek exhibited decreased total algal levels when TAIR was increased by 3ºC while Aguamilpa showed increased total algal levels with the 3ºC increase in TAIR. These contrasting trends, a decrease in Deer Creek and an increase in Aguamilpa, were produced by algae succession processes. Changes in Q affected normalized algal concentration in both reservoirs in different ways. In Aguamilpa, total algal levels increased under dry conditions while Deer Creek showed little general change associated with flow changes. Worst case scenario simulations, which included changing more than one parameter, showed that GCC changes can cause large impacts if they occur simultaneously with high nutrient loadings. These results begin to show how GCC could impact reservoirs and how these impacts compare to potential impacts from land use change. The results show that both temperate and tropical reservoirs are impacted by GCC but are more sensitive to nutrients. The methods, plots, and tools developed in this study can assist water managers in evaluating and studying GCC and land use changes effects in reservoirs worldwide.

Municipal separate storm sewer system (MS4) permittees face costly obligations to reduce pollutant loadings needed to achieve waste load allocations (WLAs) and meet total maximum daily loads (TMDLs). Street sweeping is potentially an effective BMP since streets exist throughout urban watersheds, often are directly connected to the storm sewer, and are found to contain an abundance of contaminants. Although pollutant removal from street sweeping has been evaluated for decades, an understanding of the impact on water quality in receiving streams is elusive. Due to numerous variables, the large number of samples necessary to measure impact in receiving streams may never be obtained. In response, modeled pollutant removal efficiencies based on frequency of sweeping have been recommended to the Chesapeake Bay Program, but these results are suspect. Alternatively, the amount of swept material has emerged as a method to quantify reductions. A sampling study was conducted to measure pollutants in swept material. The study identified the fraction of material susceptible to transport in runoff based on timing of sweeping in relation to runoff events. Based on observed pollutant concentration associations with particle size, the study results in estimates of pollutant concentrations for the fraction of material susceptible to downstream transport, dependent on duration since the last rainfall and type of surface swept, whether the area is a streets or a parking lot. Pollutant loadings and required reductions to achieve the Chesapeake Bay WLAs for various land use sample areas are computed for an average year. Modeled removal efficiencies and results from the sampling study were employed to assess impacts from street sweeping. Modeled efficiencies predict significantly lower impact than measurements of pollutants susceptible to runoff in swept material. Modeled loadings are inconsistent with measurements of swept materials and the rigorous sweeping frequency required for modeled removal efficiency credit appears to be unnecessary. / Doctor of Philosophy / Many localities, state agencies and other public entities that own storm sewer systems are increasingly required to reduce pollutants discharged from their systems to surface waters as a result of programs stemming from the Clean Water Act. Traditional stormwater management practices, such as retention ponds, appear limited towards providing the total pollutant reductions necessary due to physical constraints, opportunity and cost. Street sweeping is potentially an effective alternative practice since streets exist throughout urban watersheds, often are directly connected to the storm sewer, are found to contain an abundance of contaminants and can be cost effective. Although pollutant removal from street sweeping has been evaluated for decades, an understanding of the pollutants removed from stormwater is elusive. Past studies suggest the large number of samples necessary to measure impact from sweeping in receiving streams may never be obtained. In response, pollutant removal estimates have been made using computer models, but modeled results are suspect since they cannot be calibrated. Alternatively, a measure of swept material has emerged as a method to quantify pollutant reductions. A sampling study was conducted to measure pollutants in swept material. Results identify the fraction of swept material washed from the swept surface dependent on timing of sweeping in relation to the duration since the last rainfall. Based on observed pollutant concentration associations with particle size, the study results in estimates of concentrations for the fraction of material susceptible to downstream transport, dependent on duration since the last rainfall and type of surface swept, whether the area is a streets or a parking lot. Application of the results are compared to modeled removal efficiencies towards achieving regulatory compliance within various land use sample areas. Modeled efficiencies predict significantly lower impact than measurements of pollutants susceptible to runoff in swept material. Rigorous sweeping frequency required for modeled removal efficiency credit appears to be unnecessary.

No description available.

When properly designed, hypolimnetic aeration and oxygenation systems can replenish dissolved oxygen in water bodies while preserving stratification. A comprehensive literature review of design methods for the three primary devices was completed. Using fundamental principles, a discrete-bubble model was first developed to predict plume dynamics and gas transfer for a circular bubble-plume diffuser. This approach has subsequently been validated in a large vertical tank and applied successfully at full-scale to an airlift aerator as well as to both circular and linear bubble-plume diffusers. The unified suite of models, all based on simple discrete-bubble dynamics, represents the current state-of-the-art for designing systems to add oxygen to stratified lakes and reservoirs. An existing linear bubble plume model was improved, and data collected from a full-scale diffuser installed in Spring Hollow Reservoir, Virginia (U.S.A.) were used to validate the model. The depth of maximum plume rise was simulated well for two of the three diffuser tests. Temperature predictions deviated from measured profiles near the maximum plume rise height, but predicted dissolved oxygen profiles compared very well to observations. Oxygen transfer within the hypolimnion was independent of all parameters except initial bubble radius. The results of this work suggest that plume dynamics and oxygen transfer can successfully be predicted for linear bubble plumes using the discrete-bubble approach. To model the complex interaction between a bubble plume used for hypolimnetic oxygenation and the ambient water body, a model for a linear bubble plume was coupled to two reservoir models, CE-QUAL-W2 (W2) and Si3D. In simulations with a rectangular basin, predicted oxygen addition was directly proportional to the update frequency of the plume model. W2 calculated less oxygen input to the basin than Si3D and significantly less mixing within the hypolimnion. The coupled models were then applied to a simplified test of a full-scale linear diffuser. Both the W2 and Si3D coupled models predicted bulk hypolimnetic DO concentrations well. Warming within the hypolimnion was overestimated by both models, but more so by W2. The lower vertical resolution of the reservoir grid in W2 caused the plume rise height to be over-predicted, enhancing erosion of the thermocline. / Ph. D.

Urbanization alters land cover by increases in impervious areas, resulting in large increases in runoff, sediment, and nutrient loadings downstream. These changes cause flooding, eutrophication, and harmful algal blooms. Stormwater control measures (SCMs) are used to address these concerns and are designed based on inflow loads. Thus, estimating nutrient and sediment loads from developed watersheds is vitally important for meeting the impacts of urbanization. Today, stormwater events are characterized mainly by watershed models using little, if any, actual field monitoring data. The simple event mean concentration (EMC) wash-off approach by land use is a common practice used by practitioners for estimating loads. Pollutants accumulate on surfaces during dry periods, making EMC a function of antecedent dry period (ADP). An EMC results from wash-off of accumulated pollutants from catchment surfaces during runoff events. However, it assumes concentration is constant across events from a particular land use and several studies found little to no correlation between constituent concentrations in stormwater and ADP. Build-up/wash-off equations were developed to account for variation of concentrations between events; however, the required parameters are difficult to estimate. This study applied machine learning approaches with a national dataset along with monitoring and modeling studies at watershed scales to improve predictions of stormwater quantity and quality. First, we obtained stormwater quality data from the National Stormwater Quality Database (NSQD), which is the largest data repository of stormwater quality data in the U.S., and used Bayesian Network Structure Learner (BNSL), a machine learning approach, to discover which climatological or catchment characteristics most significantly affect stormwater quality. Second, we developed and applied Random Forest (RF), a data-driven method, to predict nutrients and sediment EMCs in urban runoff. Third, we applied the Storm Water Management Model (SWMM), a widely used urban watershed model, to an urban watershed and assessed the best fit estimates of SWMM parameters and hydrological response of the watershed during dry and wet hydroclimatic conditions. Last, we conducted a monitoring and modeling study at a catchment scale and assessed the role of land use on stormwater quantity and quality to optimize and investigate the build-up/wash-off parameters for multiple urban land uses for nutrients and sediment. The results presented in this dissertation can help stakeholders, urban planners, and SCM designers improve estimates of nutrients and sediment loads and thus achieve more effective treatment of stormwater, better attain water quality goals, and protect downstream water bodies. / Doctor of Philosophy / Urban development results in increased hardscapes (impervious surfaces), which increases runoff and subsequent pollution from nutrients and sediment carried off land surfaces. This negatively impairs the health of receiving streams, lakes, rivers, and estuaries. A variety of management practices are available for reducing these impacts. Practice size is based on the water quantity and quality it will receive. Thus, estimating the quantity of nutrients and sediment from developed areas is crucial to meet water quality goals. However, designs of stormwater management practices typically use historical data based on land use; rather than conducting new monitoring studies to determine actual pollution loads. Event mean concentration (EMC) is a common method used to estimate wash-off of pollutants from the land. Pollutants accumulate on surfaces during dry periods, making EMC a function of antecedent dry period (ADP) which is the time between storm events. An EMC results from wash-off of accumulated pollutants from urban areas during a storm event. However, EMC assumes pollutant concentration is constant across any storm event from a particular land use. Several studies found little to no correlation between nutrients and sediment concentrations in stormwater and ADP. Build-up/wash-off equations were developed to account for variability of concentrations between storm events; however, there are several parameters that are difficult to estimate. This study applied machine learning approaches to a national stormwater quality dataset and conducted monitoring and modeling studies at progressively smaller scales to improve the predictions of stormwater quantity and quality. First, we obtained stormwater quality data from the National Stormwater Quality Database (NSQD), which is the largest data repository of its type in the U.S., and used Bayesian Network Structure Learner (BNSL), a machine learning method, to discover which climatological or catchment characteristics most significantly affect stormwater quality. Second, we developed and applied Random Forest (RF), also a machine learning method, to predict nutrients and sediment EMCs in stormwater. Third, we applied the Storm Water Management Model (SWMM), which is the most widely used rainfall/runoff model, to an urban area and assessed the best fit estimates of SWMM parameters during dry and wet years. Last, we conducted a monitoring and modeling study at smaller scales and assessed the role of land use on stormwater quantity and quality and estimated build-up/wash-off parameters for multiple urban land uses for nutrients and sediment using optimization. The results presented in this dissertation can help stakeholders, urban planners, and stormwater practice designers improve estimates of the quantity of nutrients and sediment in stormwater, achieve more effective treatment of stormwater, attain water quality improvement goals, and protect the health of receiving streams and downstream water bodies.

Parallel computing has recently appeared has an alternative approach to increase computing performance. In the world of engineering and scientific computing the efficient use of parallel computers is dependent on the availability of methodologies capable of exploiting the new computing environment. The research presented here focused on a modeling approach, known as cellular automata (CA), which is characterized by a high degree of parallelism and is thus well suited to implementation on parallel processors. The inherent degree of parallelism also exhibited by the random-walk particle method provided a suitable basis for the development of a CA water quality model. The random-walk particle method was successfully represented using an approach based on CA. The CA approach requires the definition of transition rules, with each rule representing a water quality process. The basic water quality processes of interest in this research were advection, dispersion, and first-order decay. Due to the discrete nature of CA, the rule for advection introduces considerable numerical dispersion. However, the magnitude of this numerical dispersion can be minimized by proper selection of model parameters, namely the size of the cells and the time step. Similarly, the rule for dispersion is also affected by numerical dispersion. But, contrary to advection, a procedure was developed that eliminates significant numerical dispersion associated with the dispersion rule. For first-order decay a rule was derived which describes the decay process without the limitations of a similar approach previously reported in the literature. The rules developed for advection, dispersion, and decay, due to their independence, are well suited to implementation using a time-splitting approach. Through validation of the CA methodology as an integrated water quality model, the methodology was shown to adequately simulate one and two-dimensional, single and multiple constituent, steady state and transient, and spatially invariant and variant systems. The CA results show a good agreement with corresponding results for differential equation based models. The CA model was found to be simpler to understand and implement than the traditional numerical models. The CA model was easily implemented on a MIMD distributed memory parallel computer (Intel Paragon). However, poor performance was obtained. / Ph. D.