The Falling Creek Reservoir (FCR) in Roanoke is monitored for water quality and other key measurements to distribute clean and safe water to the community. Forecasting these measurements is critical for management of the FCR. However, current techniques are limited by inherent Gaussian linearity assumptions. Since the dynamics of the ecosystem may be non-linear, we propose neural network-based schemes for forecasting. We create the LatentGAN architecture by extending the recurrent neural network-based ProbCast and autoencoder forecasting architectures to produce multiple forecasts for a single time series. Suites of forecasts allow for calculation of confidence intervals for long-term prediction. This work analyzes and compares LatentGAN's accuracy for two case studies with state-of-the-art neural network forecasting methods. LatentGAN performs similarly with these methods and exhibits promising recursive results. / Master of Science / The Falling Creek Reservoir (FCR) is monitored for water quality and other key measurements to ensure distribution of clean and safe water to the community. Forecasting these measurements is critical for management of the FCR and can serve as indicators of significant ecological events that can greatly reduce water quality. Current predictive techniques are limited due to inherent linear assumptions. Thus, this work introduces LatentGAN, a data-driven, generative, predictive neural network. For a particular sequence of data, LatentGAN is able to generate a suite of possible predictions at the next time step. This work compares LatentGAN's predictive capabilities with existing neural network predictive models. LatentGAN performs similarly with these methods and exhibits promising recursive results.

This study addressed the geochemical drivers of Mn removal, including pH, alkalinity and the presence of mineral particles. We conducted laboratory experiments and field monitoring at two drinking water reservoirs in southwestern Virginia – Falling Creek Reservoir (FCR) and Carvins Cove Reservoir (CCR). In laboratory experiments in pH and alkalinity-adjusted nanopure water solutions, we observed substantial Mn removal within 14 days only under high pH conditions (pH≥10). In experiments with high pH and moderate to high alkalinity (> 80 mg/L CaCO3), near-total Mn removal occurred within 2 hours, at a rate of 0.25 mg/L-1 hr-1. Mn removal occurred alongside precipitation of microscopic (<5 μm diameter) and macroscopic (>100 μm diameter) particles. Elemental analysis of particles with energy-dispersive X-ray spectroscopy (EDS) supports their identification as Mn(IV) oxides (MnOx), which suggests Mn removal driven by oxidation. Elevated alkalinity in high pH solutions promotes Mn oxidation by maintaining high pH through buffering, which sustains conditions favorable for Mn oxidation. Our results also suggest sorption of Mn and mineral-catalyzed Mn oxidation by Mn oxides formed through oxidation by dissolved oxygen. In experiments using filtered and unfiltered water from the two reservoirs, we observed significant Mn removal in experiments with unfiltered water, suggesting that particles may remove Mn by catalyzing oxidation or nucleating Mn oxide precipitation. Mn removal occurred at 0.05 d-1 in unfiltered FCR water and 0.002 d-1 in unfiltered CCR water. We observed no Mn removal in filtered water from either reservoir. Scanning electron microscope (SEM) and EDS of visible particles from reservoir water experiments suggests that quartz and clay minerals present in the water column may nucleate or catalyze Mn oxide formation. Overall, this research shows that Mn removal under HOx operation is influenced by a variety of factors, including pH, alkalinity and suspended particles. / Master of Science / Elevated concentrations of manganese (Mn), a naturally occurring contaminant, can impair drinking water quality in several ways – by introducing poor taste and smell, staining pipes and appliances, and potentially harming the health of young children. Hypolimnetic oxygenation (HOx) is a novel water treatment method deployed in lakes and reservoirs to control water column contamination of metals and nutrients, including Mn. By pumping oxygen into lakes and reservoirs, HOx systems create conditions favorable for Mn removal from the water column. Previous work in two southwestern Virginia drinking water reservoirs documented differences between sites in how effectively HOx systems are able to remove Mn. These reservoirs have significant differences in their chemical profiles – most notably in pH and alkalinity, which suggests a role for background water chemistry in influencing removal rates in lakes and reservoirs with HOx systems. We used laboratory experiments to simulate the effects of pH and alkalinity on Mn removal rates in oxygenated lakes and reservoirs. We observed substantial Mn removal within 14 days under high pH conditions (pH 10-11) and negligible removal in solutions at or under pH 8. In experiments with pH 10-11 and alkalinity over 80 mg/L, near-total Mn removal occurred within 24 hours. During the 24 hour removal window, we observed yellow-brown discoloration of our experimental solutions within 12 hours, followed by formation of loosely aggregated brown to black particles. Microscopy and elemental analyses indicate that initial discoloration occurs due to formation of 1-2 μm wide manganese oxides with needle-like crystals. The visible aggregates are also manganese oxides. Based on mineral characterization and the time series of Mn removal observed in our experiments, we believe that initial formation of Mn oxides creates a positive feedback loop in solutions of pH 10-11 and alkalinity over 80 mg/L. Mn oxides promote further Mn oxide formation by facilitating conversion of Mn in solution into forms that easily settle from water. Observations of particulate formation and solution chemistry in filtered vs. filtered reservoir water from FCR and CCR supports a pivotal role for particles in facilitating Mn removal. Our research addresses the impacts of water chemistry Mn removal in drinking water, and improves understanding of Mn cycling in natural freshwaters.