This thesis presents a novel methodology for predicting flooding in complex urban environments. MIKE-FLOOD was used to dynamically model the interactions between overland flow, storm sewer drainage, and tail water by linking a modeled storm sewer system and outfall with a detailed 2D hydrodynamic overland flow model to generate animations displaying the location, depth, and duration of flooding. This research project focused on the Texas Medical Center (TMC), located in the Harris Gully watershed, which endured over 3 billion dollars in flood damage from Tropical Storm Allison in 2001. The system of integrated models was calibrated versus TS Allison and matched commendably with an absolute average error of less than 3 inches. This new approach to predicting flooding in complex urban systems has shown to be easier to set up and more dynamic and powerful than other more conventional approaches to urban inundation mapping and flood prediction.

A number of dissolved halogenated volatile organic compounds (VOCs) undergo abiotic transformation reactions under environmental conditions. These compounds also sorb to natural sediments despite their relative hydrophilicity. To date, research has not been performed that considers how sorption and transformation processes combine to influence the fate of these contaminants. 1,1,2,2 tetrachloroethane (TeCA) has been shown by others to undergo base-mediated dehydrohalogenation, producing trichloroethene (TCE) in solution. In this research, reaction rates published by Cooper et al. (1987) were confirmed and were not affected by the presence of sulfide in buffered aqueous solution. In this work, adsorption of TeCA was observed to be linear and consistent with the Freundlich isotherm as well as predictive models based on the octanol-water partition coefficient. Desorption was shown to be reversible for a small number of desorption steps then became irreversible. The results compared favorably with a model previously published by Kan et al. (1998) using semivolatile organic compounds. In a column study with TeCA sorbed to activated carbon, the sorbed reaction occurred with a rate constant approximately half the dissolved reaction rate constant. TeCA in the irreversible compartment of Dickinson Bayou sediment did not react to form TCE. The result is consistent with pH effects, the existence of an energetically stable irreversible compartment, and/or size-related phenomena such as incorporation of the TeCA molecule into the hydrophobic interior of humic acid. TeCA sorbed in the irreversible compartment is apparently not accessible to the hydrophilic hydroxide ion. In a batch reactor, the reaction rate was evaluated by comparing experimental TeCA concentrations with a model based on both the expected reaction rate in solution and the expected impact of sorption on reaction rate as derived from the activated carbon study. The observed reaction rate was approximately two times greater than expected. Although the activated carbon experiment indicated that sorption hinders reactivity, this observation is consistent with heterogeneous catalysis, possibly due to metals incorporated into the humic/fulvic acid structure and natural organic matter.

Failure to rigorously accommodate physical parameter uncertainty in groundwater transport models casts serious doubts on our ability to accurately delineate the contaminant plume at a given site. This failure could also considerably reduce the possibility of success of the remediation scheme intended to clean up a plume within the specified time. In this research, the probability that a contaminant leaking from a waste source will exceed some predetermined target level at a downgradient well is estimated, along with the sensitivity of this probability to the basic uncertainty in input parameters. The relevant parameters are assumed random with prescribed probability distributions. In this work, we present a probabilistic modeling tool based on first- and second-order reliability methods (FORM and SORM) to account for parameter uncertainty in groundwater contaminant transport and remediation. The methodology is applied to analytical groundwater models to provide simple screening tool for the assessment of contamination and remediation. In addition, numerical-based reliability models are developed to account for aquifer spatial heterogeneity and correlation structure in a more complex system. In the analytical phase, the program PROBAN was used for the probabilistic analysis. Results indicate that the greatest impact on the probabilistic event is due to basic uncertainty in seepage velocity. However, chemical-related and source-related parameter uncertainty were also found to be very important factors to consider. In the numerical phase, the finite-element code FLOTRAN was linked to the reliability shell CALREL. Hydraulic conductivity was treated as a spatial random field. Considerable saving in computational time was achieved by using a coarse random variables mesh with a finer numerical mesh. Series system reliability was used to analyze failure at several wells. Probabilistic assessment of plume containment was also provided. FORM and SORM are powerful tools for the probabilistic analysis of groundwater contaminant transport and remediation. Examples given in this work are only samples of the variety of applications that FORM and SORM can address. Their use in areas such as probabilistic risk assessment should be of great potential and interest to regulatory agencies and groundwater modelers alike.

Carboxylate alumoxanes were cast onto alpha-alumina supports by slip-coating to develop asymmetric ceramic membranes. Carboxylate ferroxanes were also analyzed as possible precursor for ceramic membranes. Different support materials were studied. Membrane layers as thin as 1 or 2 mum were achieved. The effect of sintering conditions was investigated. The molecular weight cut off was evaluated for these membranes and compared to pore size data obtained by nitrogen adsorption in an effort to correlate pore size and actual performance of the membrane. Membranes were sintered at temperatures between 600°C and 1100°C. The pore sizes increased with higher sintering temperature, from 7 nm at 600°C to 10 nm at 1000°C, followed by a sharp increase due to the transformation to alpha alumina at higher firing temperatures. The molecular weight cut off showed no considerable variation up to 1000°C. These results are in good agreement with equations reported by several authors that correlate molecular weight with size.

The reactivity of partially reduced metabolites of 2,4,6-trinitrotoluene (TNT), namely arylhydroxylamines and nitrosoarenes, was evaluated with a simple biological system and with components of soil natural organic matter (NOM). This study was carried out to address the long-standing problem of irreversible binding to soil NOM and biomass, commonly observed during the reductive transformation of polynitroaromatic contamination. The study focused on partially reduced metabolites rather than the completely reduced arylamine metabolites that have already been extensively investigated for their role in binding to soil NOM. In the simple bioreduction system of Clostridium acetobutylicum cell-free extract/molecular hydrogen (electron donor), 10% of the initial 14C was found bound to solid proteinaceous material following sequential anaerobic/aerobic treatment. A review of the nitroso and hydroxylamino functional group chemistry revealed that the nitroso-thiol reaction was most likely responsible for the reaction with proteins. The introduction of a model thiol, 1-thioglycerol, into an anaerobic mixture of 4-hydroxylamino-2,6-dinitrotoluene (4HADNT) and 2,4-dihydroxylamino-6-nitrotoluene (DHA6NT) resulted in the formation of a new product, only when the reaction mixture was exposed to air. The results from the model reaction confirmed that thiols could act as competing nucleophiles for nitroso compounds, which are readily formed from hydroxylamino compounds upon exposure to air. The reactivity of arylhydroxylamines and nitrosoarenes with standard humic acids was investigated using 4HADNT and nitrosobenzene as model compounds, respectively. Contrary to results reported by others, 4HADNT was found to be nonreactive towards humic acid at humic acid concentrations in excess of dissolved organic matter concentrations found in nature. Conversely, nitrosobenzene reacted rapidly with humic acids, with the extent of reaction being highest for humic acids that had a high protein content. Humic acids that were pretreated with a thiol derivatizing agent showed diminished capacity for reaction with nitrosobenzene. Since nitroso intermediates from TNT reduction are difficult to synthesize and are rarely observed in nature due to their high instability, their electrophilic characteristics were evaluated using a molecular modeling approach. Molecular models of potential TNT nitroso intermediates were compared with those of the strongly electrophilic nitrosobenzene. The comparison revealed that 2-nitroso-4-hydroxylamino-6-nitrotoluene was more likely to react similarly to nitrosobenzene than 4-nitroso-2,6-dinitrotoluene.

The lead-time and accuracy of a flood alert system designed for a quickly responding urban watershed have been improved by incorporating a variety of new tools and methodologies. These include: the use of computer-mediated voice and data communication systems such as the Internet, high-quality hydrologic data including radar rainfall, real-time hydrologic models, assessment and use of a Quantitative Precipitation Forecast (QPF) algorithm, and the development of improved flood notification levels---providing earlier and more accurate warnings to critical institutions and emergency personnel in flood-prone areas throughout the watershed. While the research focused on one watershed in an urban setting (Brays Bayou in Houston, Texas), the results found are applicable across a broad spectrum of watershed types, provided that the need for more timely and accurate flood forecasts exists. System lead-time improvements were accomplished through the implementation and evaluation of a QPF algorithm increasingly used for short-term weather prediction across the United States. The Brays Bayou watershed provided an excellent test-bed for the collection and evaluation of QPF data. Algorithm accuracy and effectiveness were evaluated at various forecast times and basin sizes commonly found in urban watersheds. Generalized results of these analyses are presented. System accuracy improvements were accomplished with improved radar-rainfall data input and the development of real-time hydrologic models. A real-time interface for the industry standard HEC-1 hydrologic model was created, allowing the hydrologic predictions developed by this model to take greater advantage of the spatial and temporal distribution of real-time radar-rainfall data. The successful implementation of this real-time hydrologic model at the scale of Brays Bayou also provided significant lead-time improvements by providing estimates to when peak flows would actually occur. A successful validation and operational test of the entire system occurred during the November 17th, 2003 storm event. This storm event is utilized as a case study, with results illustrating wide-ranging improvements.

Recharge is one of the most difficult input parameters to evaluate accurately in ground water modeling; this research focuses on analyzing an existing numerical MODFLOW model's accuracy by using more advanced technology for calculating aquifer recharge. The MODFLOW model is of the Northern Tampa Bay (NTB) area (1500 km2) of Florida and monitors the effects of ground water pumping. Ground water overpumping is a serious problem, producing such drastic negative impacts as land subsidence, sinkholes, lowered surface water depths, and seawater intrusion. Improved ground water models will lead to better planning and management of ground water resources. Recharge is a function of several factors, the most important of which is precipitation. This research focused on evaluating the spatial and temporal variability of recharge measurements by incorporating the use of radar rainfall. NEXRAD radar is typically used in hydrologic projects for real-time monitoring and flood prediction. NEXRAD can accurately measure and depict the spatial and temporal patterns of rainfall, leading to spatial improvements in the rainfall measurement since a 4 x 4 km cell-based grid will provide more specific and accurate spatial distribution of rain data than individual rain gages several miles apart. Bias calculations for radar data on a daily basis were analyzed to determine the best method for evaluation of daily data (historical use of NEXRAD calculates bias on storm events). Using a GIS-integrated spatial model with NEXRAD input data, it was determined that NEXRAD could successfully be used for evaluating rainfall in a hydrologic model of the Floridan aquifer.

Ground-level ozone is of a growing concern in many areas of the United States. Ozone is a significant health concern, particularly for people with asthma and other respiratory diseases. Ozone is rarely emitted directly into the air but is formed by the reaction of volatile organic compounds (VOCs) and nitrogen oxides (NOx) in the presence of sunlight. VOCs are emitted from a variety of sources, including motor vehicles, chemical plants, refineries, factories, consumer and commercial products, other industrial sources, and biogenic sources. NOx is emitted from motor vehicles, power plants, and other combustion sources. Ozone and ozone precursors also can be transported into an area from pollution sources found hundreds of miles away. In accordance with the 1990 Clean Air Act Amendments, EPA has required more extensive monitoring of ozone and its precursors in areas with persistently high ozone levels. In these areas, the States have established ambient air monitoring networks consisting of CAMS (continuous air monitoring system) sites, which collect and report detailed data for volatile organic compounds, nitrogen oxides, ozone and meteorological parameters. Analyses of these data help the regulatory agencies to better understand the underlying causes of ozone pollution, to devise effective remedies and to measure air quality trends. This thesis focuses on how to integrate these measurements of VOCs with the receptor modeling techniques in order to identify the sources of VOCs and to attribute ambient VOC concentrations to their original sources. The measurements taken from three CAMS stations in Houston, TX serve as the basis of this research. After presenting the source attribution of volatile organic compounds, where the contribution from different sources to ambient VOC levels are determined, the methods to identify the source regions associated with elevated VOC levels are described. The quantitatively reconstructed emissions from a recently prepared VOC emissions inventory are compared with the receptor model calculations of ambient VOC measurements. Finally, a separate growing concern in the US, the particulate matter pollution, is addressed. The impacts of regional wild fires in Texas on the secondary particulate matter formation are examined. The results of the laboratory investigations on the formation of the secondary sulfate particles through heterogeneous surface reactions are presented.

The principal objective of this research has been the development of advanced data reduction techniques and their application to real-time detection and precise concentration measurement of trace gases for use in spacecraft habitats. This is especially important for human life support technology in long duration space missions such as the International Space Station and future missions to Mars. Trace gas detection was performed using compact infrared diode laser-based absorption gas sensors. These sensors have proven to be very sensitive, selective and rugged to permit use in spacecraft environments. These same sensors can play an important role in unmanned exploratory programs of planetary surfaces or atmospheric research conducted by high altitude NASA aircraft. The infrared laser sources take advantage of recent significant developments of new infrared nonlinear materials, progress in diode and solid state lasers, and fiber optics technology. The techniques developed in this thesis include realtime Voigt fitting, simultaneous multi-species detection and Kalman filtering. These techniques were applied in several field campaigns including formaldehyde monitoring in downtown Houston, Deer Park and Channelview, Texas, as well as ammonia and carbon dioxide monitoring in a Biological Wastewater Processor at the Johnson Space Center in Houston, Texas.

Use of a standard three-dimensional flow model to simulate the saltwater interface in coastal and island aquifers as a no flow boundary is introduced. The method was used to simulate flow in a generic circular island aquifer. The results of this model compared favorably with those generated by an analytical solution and a numerical simulation. The method was used to investigate the freshwater resources of Guemes Island, Skagit County, Washington. The model showed that the northern region of the island is most vulnerable to saltwater intrusion. However, it was determined from the modeling exercise that with careful planning Guemes Island can sustain substantial development of its aquifer system without significant saltwater intrusion. The use of a standard flow model can not simulate changes in well discharge concentration but can be a valuable tool in managing groundwater development in island and coastal aquifers.