This study examines differences between rain gauge precipitation estimates and those from the National Weather Service's (NWS's) Multi-sensor Precipitation Estimator (MPE) which blends hourly gauge data with estimates from NWS radars onto a 4×4 km2 grid. Comparisons are made between 1996-2003 on five watersheds within Florida that have a range of sizes and gauge densities. The comparison statistics are found to improve during the period, likely due to improvements in the gauge networks, both the number of gauges and their data quality. MPE hourly totals generally are less than gauge totals, and differences between MPE and gauge amounts (MPE-gauge) often are greater (more negative) for larger gauge values than smaller ones. Results also reveal that the larger the basin, the greater the volume differences and standard deviation of differences. Differences between MPE and gauge-derived precipitation are found to depend greatly on gauge density within the basin which generally increases during the study period. The comparison statistics also are affected by the spatial and temporal variability of rainfall within the basins. Thus, a particular gauge distribution may adequately sample one rainfall scenario and not another. Differences between MPE and gauge-derived precipitation are found to be related to the seasons because of the different types of precipitation generally associated with each. The two data types correlate better during the cold season than the warm season and the standard deviations of differences are greater during the warm season. The tendency for MPE amounts to be less than gauge amounts is more pronounced during the cold season. On the other hand, the trend for MPE amounts to be much less than gauge amounts for the greater gauge totals is more pronounced during the warm season. These seasonal differences are due to several characteristics of the rainfall and its sampling by the two products. The cold season consists of more low top rainfall events than the summer. Radars often overshoot this precipitation, leading to MPE amounts less than gauge amounts. Cold season events also are larger in area and more likely to be sampled adequately by gauges. Conversely, warm season events are more small scale in character and may not be adequately sampled by gauges. Several factors are thought to explain the tendency for smaller MPE than gauge totals with increasing gauge amounts. The standard NWS reflectivity-rainfall (Z-R) relationship produces large underestimates compared to the tropical Z-R relationship. Additionally, the NWS caps reflectivity to avoid hail contamination, and this cap might be set too low for intense precipitation events. Finally, truncation errors in the MPE algorithm prior to 2001 may be a factor in causing the smaller MPE values. Differences between MPE and gauge totals are compared for hourly, 6-hourly, and daily accumulations. The difference statistics improve with increasing periods of accumulation because when hourly observations are added, random errors in the data tend to cancel although systematic errors remain. The results indicate that there are definite differences between MPE and gauge-derived precipitation. However, it is not possible to say which product is superior because there is no "ground truth" at the scale at which Florida precipitation occurs. / A Thesis submitted to the Department of Meteorology in partial fulfillment of the requirements for the degree of Master of Science. / Fall Semester, 2006. / August 9, 2006. / Precipitation Estimation, Multisensor Precipitation Estimator, Rainfall Estimates, MPE / Includes bibliographical references. / Henry E. Fuelberg, Professor Directing Thesis; Paul H. Ruscher, Committee Member; Tiruvalam N. Krishnamurti, Committee Member.

Using a suite of high resolution models, the forecast skills of the superensemble for precipitation and 2-meter temperature over the continental United States and North America are shown. In this study, models and/or gridded fields such as the ETA, WRFARW, MM5, NDFD, several global models, and their ensemble mean are used. The final resolution for the multimodel superensemble is at 32 km at 3-hourly temporal intervals for temperature and ¼ degree daily intervals for precipitation. The forecast length is 60 hours for temperature and 5 days for precipitation. This study utilizes an optimization for the training length (number of days) to arrive at the best results. The period of the study includes March-September 2006. The metrics for the forecast evaluation include the mean absolute error, rms error, bias, and equitable threat scores. The results show a significant improvement of the multimodel superensemble compared to its member models and their ensemble mean. / A Thesis submitted to the Department of Meteorology in partial fulfillment of the requirements for the degree of Master of Science. / Summer Semester, 2007. / June 12, 2007. / NWS, National Digital Forecast Database, MOS / Includes bibliographical references. / T. N. Krishnamurti, Professor Directing Thesis; Robert Hart, Committee Member; Paul Ruscher, Committee Member.

Although spheres and spheroids have been used extensively by researchers as convenient models to approximate "snowflakes" when computing their microwave scattering properties, recent research indicates that the scattering properties of more accurately simulated snowflakes are fundamentally different from the simplified models. To resolve this well-recognized discrepancy, a new snowflake aggregation model is developed in this study and the microwave single-scattering properties of the modeled aggregate snowflakes are characterized for use in radiative transfer modeling and remote sensing algorithm development. Three different aggregate snowflake types (rounded, oblate and prolate) are generated by random aggregation of 6-bullet rosettes constrained by size-density relationships derived from previous field observations. Additionally, they are further constrained to empirically determined aspect ratios (ar) and fractal dimensions (df) of aggregate flakes. Due to random generation, aggregates may have the same size or mass, yet different morphology, allowing for a study into how detailed structure influences an individual flake's scattering properties. Single-scattering properties of the aggregates were investigated using discrete dipole approximation (DDA) at 10 frequencies: 10.65, 13.6, 18.7, 23.8, 35.6, 36.5, 89.0, 94.0, 165.5 and 183.31 GHz. All of these frequencies are currently used in instruments (radar and radiometers) aboard satellites involved in the research of atmospheric ice particles. Results from DDA were compared to those of Mie theory for solid and soft spheres (with a density 10% that of solid ice) and to T-matrix results for solid and soft spheroidal cases with ar values of 0.8 and 0.6 dependent on flake type (rounded, oblate or prolate). Analyzing modeling results, it is found that above size parameter 0.75, neither solid nor soft sphere and spheroidal approximations accurately represented the DDA results for all aggregate types. The asymmetry parameter and the normalized scattering and backscattering cross-sections of the aggregate groups fell between the soft and solid spherical and spheroidal approximations. This implies that evaluating snow scattering properties using realistic shapes, such as the aggregates created in this study, is necessary in radiative transfer modeling and remote sensing studies. When examining the dependence of the single-scattering properties on each aggregate's detailed structure, morphology seemed of secondary importance. Using normalized standard deviation as a measure of relative uncertainty, it is found that the relative uncertainty in backscattering arising from the different morphologies caused by random aggregation is typically ~17%, 13% and 14% for individual particles and ~20%, 30% and 30% when integrated over size distributions for rounded, oblate and prolate flakes respectively. Relative uncertainties for other single-scattering parameters are less. These analyses indicate that a scattering database can be created to approximate the single-scattering properties of realistic aggregate flakes. A database of such aggregate flakes has been created based upon the research detailed herein, and made available for public use. In this work, it is found that flakes with similar size parameters can scatter differently. Ongoing research indicates that this is due to outer layer morphology of the flake (i.e. the arms of a dendritic snowflake) rather than any interior properties. When the interior of an aggregate flake is scrambled, the scattering results are nearly the same as the unscrambled interior whereas if the outer layer is altered, scattering results differ. Another interesting trend noted is that randomly oriented flakes with differing ar values have noticeably differing backscatter cross-sections and could have significant implications for future research. / A Dissertation submitted to the Department of Earth, Ocean, and Atmospheric Science in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Spring Semester 2015. / April 23, 2015. / Includes bibliographical references. / Guosheng Liu, Professor Directing Dissertation; Eric Chicken, University Representative; Mark Bourassa, Committee Member; Robert Ellingson, Committee Member; Vasu Misra, Committee Member.

Wave-driven transport, also known as Stokes drift, is the motion of a particle due to the orbital motion induced by a passing wave. This orbital motion does not form closed loops, leading to a net displacement over a single wave period. Stokes drift has previously been qualitatively shown to be a factor in ocean surface particle transport, with most studies focused exclusively in near-shore regions. However, Stokes drift has never been quantified beyond theoretical studies and case studies limited to small regions. Here, Stokes drift is calculated directly from Wavewatch III model data in the Gulf of Mexico for April-July 2010. Its magnitudes are compared between deep and shelf water areas, and against the magnitudes of surface currents and parameterized wind drift. These comparisons are also made specifically for the time period surrounding the passage of Hurricane Alex through the southwestern Gulf of Mexico. While there is not a major difference between the absolute magnitudes of Stokes drift in shelf vs. deep water areas or when compared to wind drift, Stokes drift is larger in shelf water areas relative to surface currents than in deep water. During Hurricane Alex, Stokes drift magnitudes were much larger in the immediate area of the storm, while in the oil spill area there was little change until after the storm was out of the Gulf, at which time swell had propagated into the region, increasing Stokes drift magnitudes. / A Thesis submitted to the Department of Earth, Ocean, and Atmospheric Science in partial fulfillment of the requirements for the degree of Master of Science. / Summer Semester 2015. / June 30, 2015. / Gulf of Mexico, Oil spill, Stokes drift / Includes bibliographical references. / Mark Bourassa, Professor Co-Directing Thesis; Eric Chassignet, Professor Co-Directing Thesis; Robert Hart, Committee Member.

Uranium contamination is widespread in subsurface sediments at mining and milling sites across North America, South America, and Eastern Europe. In the U.S. alone, the Department of Energy (DOE) is responsible for the remediation of 7,280 km2 of soils and groundwater contaminated due to processes associated with uranium extraction for nuclear weapons production. As a result of waste disposal practices, subsurface sediments at these sites are often co-contaminated with nitric acid, sulfuric acid, and toxic metals. Oxidized uranium, U(VI), is highly soluble and toxic, and thus is a potential contaminant to local drinking water reservoirs. The most promising strategy for in situ uranium bioremediation is immobilization through the biological reduction of U(VI) to insoluble U(IV) by indigenous microbial communities. However, the development of effective U(VI) bioremediation strategies is limited by our current understanding of the composition, metabolic potential and physiological requirements of in situ microbial communities. A polyphasic approach employing microbiological and geochemical techniques was used in this dissertation to link the structure and function of microbial communities in subsurface sediments of the U.S. Department of Energy's Oak Ridge Field Research Center (ORFRC), in Oak Ridge, Tennessee. Subsurface sediments at the ORFRC site are cocontaminated with high levels of U(VI) and nitrate and microbial activity is limited by carbon availability and variable pH. The conditions at the ORFRC site are representative of many radionuclide-contaminated sites; therefore, results from this dissertation will have broader significance for development of bioremediation strategies that can be employed worldwide. To develop effective bioremediation strategies for radionuclide contaminants, the composition and metabolic potential of microbial communities need to be further understood, especially in highly contaminated subsurface sediments for which little cultivation-independent information is available. Thus, the metabolically active and total microbial communities associated with uranium contaminated subsurface sediments were characterized along geochemical gradients (Chapter 1). DNA and RNA were extracted and amplified from four sediment depth intervals representing moderately acidic (pH 3.7) to near neutral (pH 6.7) conditions. Phylotypes related to the Proteobacteria (α−, β−, δ−, and γ−Proteobacteria), Bacteroidetes, Actinobacteria, Firmicutes and Planctomycetes were detected in DNA- and RNA-derived clone libraries. Diversity and numerical dominance of phylotypes were observed to correspond with changes in sediment geochemistry and rates of microbial activity, suggesting geochemical conditions have selected for well-adapted taxa. Sequences closely related to known nitrate-reducing bacteria, comprised 28 and 43% of clones from the total and metabolically active fractions of the microbial community, respectively. Chapter 1 provides the first detailed analysis of total and metabolically active microbial communities in radionuclide contaminated subsurface sediments. The microbial community analysis, in conjunction with rates of microbial activity, points to several groups of nitrate-reducers that appear to be well adapted to environmental conditions common to radionuclide-contaminated sites. To elucidate the potential mechanisms of U(VI) reduction for optimization of bioremediation strategies, the structure-function relationships of microbial communities were investigated in microcosms of subsurface materials cocontaminated with radionuclides and nitrate (Chapter 2). A polyphasic approach was used to assess the functional diversity of microbial populations likely to catalyze electron flow under conditions proposed for in-situ uranium bioremediation. The addition of ethanol and glucose as supplemental electron donors stimulated microbial nitrate and Fe(III) reduction as the predominant terminal electron accepting processes (TEAPs). U(VI), Fe(III), and sulfate reduction overlapped with time in the glucose treatment, whereas U(VI) reduction was concurrent with sulfate reduction but preceded Fe(III) reduction in the ethanol treatments. Phyllosilicate clays were shown to be the major source of Fe(III) for microbial respiration using variable-temperature Mössbauer spectroscopy. Nitrate- and Fe(III)-reducing bacteria (NRB and FeRB) were abundant throughout the shifts in TEAPs observed in biostimulated microcosms and were affiliated with the genera Geobacter, Tolumonas, Clostridium, Arthrobacter, Dechloromonas, and Pseudomonas. Up to two orders of magnitude higher counts of FeRB and enhanced U(VI) removal were observed in ethanol-amended as compared to glucose-amended treatments. Quantification of citrate synthase (gltA) levels demonstrated a stimulation of Geobacteraceae activity during metal reduction in carbon amended microcosms with the highest expression observed in the glucose treatment. Phylogenetic analysis indicated that the active FeRB share high sequence identity with Geobacteraceae members cultivated from contaminated subsurface environments. The results show that the functional diversity of populations capable of U(VI) reduction is dependent upon the choice of electron donor. Metabolic activity and phylogenetic structure of microbial communities mediating U(VI) bioimmobilization were directly linked in microcosms constructed with contaminated subsurface sediments using a stable isotope probing (SIP) approach (Chapter 3). Ethanol and acetate are the electron donors that have been shown in previous bioremediation studies to promote microbially mediated U(VI) reduction. Therefore, microcosms were amended with 13C-labeled ethanol or acetate, as supplemental electron donors, and molybdate was added to select treatments as an inhibitor of sulfate reduction. Activity was assessed by monitoring terminal electron accepting processes (TEAPs) (e.g., nitrate-, sulfate-, Fe(II) and U(VI) concentrations) and electron donor utilization. 13C incorporation into community DNA was examined by density gradient centrifugation along with PCR amplification and terminal restriction fragment length polymorphism (TRFLP) analysis. Incorporation of 13C into microbial DNA was detected by day 3, corresponding with the onset of TEAPs and carbon utilization. Metal reduction commenced only with removal of nitrate and U(VI)-reduction preceded Fe(III)-reduction. Fe(III)-reduction occurred in all treatments, regardless of electron donor or presence of molybdate. Sulfate reduction rates were most rapid in the ethanol-amended treatments, whereas little to no sulfate reduction occurred in the acetate- or molybdate- amended treatments. Microbial community composition was affected by treatment and changed with shifts in TEAPs. The metabolically active denitrifying microorganisms were identified as members of the Betaproteobacteria, whereas, members of the Deltaproteobacteria, Actinobacteria, Firmicutes and Bacteroidetes were active during sulfate- and/or metal-reduction. Our data showed that carbon from supplemental ethanol is coupled to microbial growth either by direct utilization or secondary carbon flow from oxidation end products or dead 13C-biomass. Application of the SIP technique allowed for the definitive identification of metabolically active microbial populations and identification of their role in the separate TEAPs occurring in uranium-contaminated subsurface sediments. A combination of biogeochemical rate measurements and robust microbial community characterization was used to determine the metabolic potential of microbial groups affecting U(VI) reduction in the subsurface. The composition, distribution, and metabolic potential of in situ microbial communities at the ORFRC site varies with subsurface geochemistry and will impact the success of U(VI) bioremediation. Ethanol is an effective electron donor for biostimulation of indigenous microbial communities at the ORFRC site and promotes complete nitrate-reduction and U(VI) removal. U(VI) reduction was observed to overlap with alternative TEAPs, e.g., sulfate- or Fe(III)-reduction and microbial populations involved in these TEAPs included members of the Firmicutes and Deltaproteobacteria. This research provides necessary data for the future optimization of in situ bioremediation practices in uranium-contaminated sediments at the ORFRC. / A Dissertation submitted to the Department of Oceanography in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Summer Semester, 2008. / July 22, 2008. / Iron-Reduction, Bioremediation, Nitrate-Reduction, SSU rRNA, Uranium / Includes bibliographical references. / Joel E. Kostka, Professor Directing Dissertation; Vincent Salters, Outside Committee Member; Markus Huettel, Committee Member; Kirsten Küsel, Committee Member; Thomas DiChristina, Committee Member.

Water residence time is an important parameter in defining how an estuary functions. The focus of this research was on the residence time and its implications on environmental health. Radium isotopes (223Ra, 224Ra, 226Ra, 228Ra) have proven useful for determining the residence time of waters in river dominated estuaries. This approach was applied to St. Andrew Bay near Panama City, FL to estimate the residence time of this non-river dominated estuary. St. Andrew Bay spans 230 km2 and is comprised of four individual estuaries (North Bay, West Bay, St. Andrew Bay and East Bay). The largest freshwater source to this environment is Deer Point Lake which is a man-made lake sourced by natural springs. It is located at the northern tip of North Bay. This study had three main objectives: 1) determine the relationship between two radium isotope sampling methods, 2) estimate the residence time of the northern portion of St. Andrew Bay, and 3) analyze the nutrient concentrations to assess the ecosystem health. The two radium isotope sampling methods compared were a grab sampling method and a stationary mooring method. Both methods utilize manganese dioxide fibers in the field to extract dissolve radium isotopes from the water column. These fibers are processed in the laboratory using a delayed coincidence counter. The measured activities of 223Ra, ex224Ra and ex226Ra were ratioed and the residence time was calculated using the approach described by Moore (2000). Water was collected and frozen for nutrient analysis during sampling. It was later analyzed for the dissolved inorganic nitrogen and phosphorus concentrations using a Lachat QuikChem 8000. The results of this project indicate grab sampling and mooring sampling methods can be combined. However, the inherent difference between the methods needs to be considered when analyzing the data. The grab samples capture the site's radium activities at the particular moment of sampling while the moorings integrate radium isotopes over a tidal cycle. The residence time calculation was compromised because of the multiple local sources which flow into both North and West Bay. While Deer Point Lake is the largest freshwater source into St. Andrew Bay, the flux of radium isotopes is not large enough to be traced throughout the system. The residence time calculated from the radium activity ratios is likely a lower limit estimation of the actual flushing time of the study area because of local sources. This method indicated there is an overall average residence time of approximately 10 days. Comparatively, a water balance was calculated which indicated a residence time from 7 to 20 days assuming the dominant flushing process is the diurnal tidal pulse. The nutrient concentrations measured throughout the study area were low. The DIN concentrations were between 0.49 µM and 4.00 µM and the SRP concentrations range was 0.47-2.41 µM. These concentrations indicate the system is healthy and able to properly incorporate the nutrients entering the system without affecting the water quality. This balance will be maintained as long as the nutrient concentrations and inputs are carefully monitored. / A Thesis submitted to the Department of Earth, Ocean and Atmospheric Sciences in partial fulfillment of the requirements for the degree of Master of Science in Oceanography. / Fall Semester, 2010. / October 18, 2010. / Radium, Mooring, Grab Sampling, St. Andrew Bay, Residence Time, Nutrients / Includes bibliographical references. / Jeffrey Chanton, Professor Co-Directing Thesis; William Burnett, Professor Co-Directing Thesis; Markus Huettel, Committee Member.

This study is focused on the application of 222Rn (radon, t1/2=3.8 d) as a geochemical tracer for evaluation groundwater fluxes in different aquatic systems including submarine springs and lakes. For this purpose improvements of some established methods for detection of 222Rn in natural waters and investigating the possibilities of concurrent measurements of 220Rn (thoron, t1/2=56 s) for detection of groundwater point sources were developed. The 222Rn groundwater tracer technique was applied at study sites in Spring Creek Springs system (Florida Panhandle) and several small lakes in central Florida. The fresh water groundwater fluxes of Spring Creek Springs were evaluating using salinity and 222Rn as geochemical tracers. Two different approaches were applied. The first model is based on time series measurements of either of the tracers (salinity or 222Rn) and data for stream velocity recorded downstream in the spring area. To verify the first approach we developed a simple two-box model that is independent of stream-point measurements and use concurrent salinity data from both upstream and the spring area. Our results from almost two year of monthly based deployments at the study site indicate that the dynamics of the springs' flow is governed most probably by fluctuations of the local water table. The total spring discharge fluctuated between zero to up to ~3.0x106 m3/day (March 2008). The 222Rn approach for assessing groundwater discharge was tested in seven relatively small lakes in North and Central Florida (Lake Newnans, Lake Butler, Clear Lake, Lake Haines, Lake Shipp, Lake Hunter, and Lake Josephine). A mass balance advection-diffusion model a well mixed non-stratified water body showed to be adequate for evaluating the groundwater fluxes in these systems. Comparison of the estimates of some of the lakes with independent seepage meters and water balance studies showed very good agreement. A special investigation on the groundwater end-member for the model evaluations resulted in a decision of using a sediment equilibration approach for determining this parameter. Finally, we used 222Rn-time-series to asses the groundwater discharge in small shallow lakes. We monitored the 222Rn concentration in lake water over time for a period long enough (usually 1-3 days) to observe changes likely caused by variations in atmospheric exchange (primarily a function of wind speed and temperature). We then attempted to reproduce the observed record by accounting for decay and atmospheric losses and by estimating the total 222Rn input flux using an iterative approach. Once a quasi steady-state 222Rn flux was evaluated by balancing the calculated outputs, we divided this flux by the measured or assumed groundwater radon concentration to determine the groundwater discharge / A Dissertation submitted to the Department of Oceanography in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Spring Semester, 2010. / December 2, 2009. / Aquatic Systems, Hydrological Processes, Radon Isotopes / Includes bibliographical references. / William C. Burnett, Professor Directing Dissertation; Xialong Hu, University Representative; Jeffrey P. Chanton, Committee Member; William M. Landing, Committee Member; Michael Wetz, Committee Member.

Numerical models for variable-density flow and solute transport (VDFST) are widely used to simulate seawater intrusion and related problems, including submarine groundwater discharge (SGD). The mathematical model for VDFST is a coupled nonlinear system written in state-space and time form, so the numerical discretization in time and space are usually required to be as fine as possible. As a result, such large space and time transient models are computationally very demanding, which is disadvantageous for state estimation, forward prediction or inverse calculation. The purpose of this research was to develop mathematical and numerical methods to simulate variable-density flow and salt transport via a model reduction technique called "proper orthogonal decomposition" (POD) for both linear and nonlinear models. It was showed that this method can generate representations of data that contain general information about the solution of the original partial differential equations. Data analysis using POD was conducted to extract dominant "model features" (basis functions) through singular value decomposition from experimental data or detailed simulations of high-dimensional systems (snapshots). These basis functions were then used in the Galerkin projection procedure that yielded low-dimensional reduced models. The original full numerical models were presented by the Galerkin finite-element method. The implementation of the POD reduced method was straightforward referring to the complex full model. The developed POD method was applied to solve two classic VDFST problems, the Henry problem and the Elder problem, to investigate the accuracy and efficiency of the POD method. The reduced model can reproduce and predict the full model results very accurately with much less computational time in comparison with the full model. It was showed that the accuracy and efficiency of the POD reduced model is mainly determined by the optimal selection of snapshots and POD bases. / A Dissertation submitted to the Department of Earth, Ocean and Atmospheric Science in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Fall Semester, 2010. / October 19, 2010. / Sigular Value Decomposition, Proper Orthogonal Decomposition, Variable-Density flow, Submarine Groundwater Discharge, Model Reduction, Galerkin Finite Element / Includes bibliographical references. / Bill X. Hu, Professor Directing Dissertation; William C. Burnett, University Representative; Ming Ye, Committee Member; Joseph F. Donoghue, Committee Member; Jeff Chanton, Committee Member.

Submarine groundwater discharge (SGD) is a major pathway for nutrient transport in coastal marine systems. There are indications that SGD may also release large amounts of dissolved organic matter (DOM) to the coastal ocean and thus impact coastal ecosystem functioning. DOM is usually quantified as dissolved organic carbon (DOC), which requires discrete sampling. Therefore, detailed time series on tidally-driven SGD cannot easily be obtained for DOC. The chromophoric component of DOM (CDOM) can be monitored via specific fluorescence in high temporal resolution and in situ. Here we hypothesize that SGD is a significant source of CDOM to the coastal ocean, impacting optical properties and biogeochemical cycles of coastal waters. In this context we also evaluated the possibility of using CDOM as a proxy for DOC in a subterranean estuary. To test our hypothesis, a case study was performed in a shallow bay in the Northern Gulf of Mexico. CDOM was continuously monitored in situ for approximately three weeks in a groundwater well on a beach and 300 m offshore in the adjacent bay. The radon isotope 222Rn was also continuously measured as a conservative tracer for submarine groundwater in the bay. Discrete samples for DOC analysis and associated variables were collected to cover one tidal cycle (15 hours). In a simple SGD model, CDOM concentrations in the bay were predicted by multiplying 222Rn concentrations in the bay (as a measure for SGD) with CDOM concentrations in the well (as the groundwater endmember). This was done for each hour during the entire sampling period. If a lag-time of one hour between groundwater and bay was considered, the predicted CDOM significantly correlated (p<0.01) with the measured CDOM in the bay. Independent statistical tests, including chlorophyll a, salinity and water level data, confirm this finding and demonstrate that CDOM in the bay is mainly driven by freshwater SGD. More detailed analysis of the time series data show that short-term time series of 24 hours or less can lead to erroneous results in estimating SGD. CDOM and DOC significantly correlated for the groundwater endmember, and CDOM could thus be transformed into DOC concentrations. It was estimated that at least 0.6 Mega-mole DOC are delivered to the entire Gulf Coast of Florida in a day via SGD which is similar in order of magnitude as riverine fluxes. / A Thesis submitted to the Department of Oceanography in partial fulfillment of the requirements for the degree of Master of Science. / Summer Semester, 2009. / April 16, 2009. / Subterranean Estuary, CDOM, Submarine Groundwater Discharge, Chromophoric Dissolved Organic Matter / Includes bibliographical references. / Markus Huettel, Professor Directing Thesis; Thorsten Dittmar, Committee Member; William Burnett, Committee Member; Jeffrey Chanton, Committee Member.

A data assimilation method is developed to calibrate a heterogeneous hydraulic conductivity field conditioning on observation of a transient groundwater flow field or transient conservative solute transport field. An ensemble Kalman filter (EnKF) approach is used to update model parameters such as hydraulic conductivity and model variables such as hydraulic head or solute concentration using available data. A synthetic two-dimensional flow case is used to assess the capability of the EnKF method to calibrate a heterogeneous conductivity field by assimilating transient flow data from observation wells under different hydraulic boundary conditions. The study results indicate that the EnKF method will significantly improve the estimation of the hydraulic conductivity field by assimilating hydraulic head measurements and the hydraulic boundary condition will significantly affect the simulation results. The ensemble size should be 300 or larger for the numerical simulation in the study case. The number and the locations of the observation wells will significantly affect the hydraulic conductivity field calibration. Another synthetic case with the mixed Neumann/ Dirichlet boundary conditions is designed to investigate the capacity and effectiveness of a constrained EnKF by assimilating the solute concentration to identify a conductivity distribution. The study results indicate that the constrained EnKF method will significantly improve the estimation of the hydraulic conductivity field by assimilating solute concentration measurements. The larger area for the initial distribution of the solute concentration, the more observed data can be obtained, the better the inversed results. The number of the actual observation wells needed to calibrate the hydraulic conductivity field through the constrained EnKF method via assimilating the solute concentration is very small. The data assimilation method can produce useful results in the first five or seven time step assimilation. The simulated results by the data assimilation method are still very similar with different observation errors. Based on the problems of the filter divergence in the data assimilation application, the localized EnKF method is applied. The covariance inflation and localization schemes are used to the transient state groundwater water flow. The synthetic study case of the transient groundwater flow is the same as the research before, but the assumed real conductivity values are correlated. The simulations by the data assimilation with and without localized EnKF are compared. The hydraulic conductivity field can be updated efficiently by the localized EnKF, while it cannot be updated via just the EnKF. The covariance inflation and localization are found to efficiently solve the problem of the filter divergence. The ensemble size for the localized EnKF method is 100 and less than that only in the EnKF before, which reduce the computer cost. The correlation length is found to affect the simulation by the localized EnKF method much more than the localization radius. Moreover, the updated results of hydraulic conductivity fields produced by the localized EnKF method with the greater correlation length and greater localization radius are a little closer to the real field. Based on the problems of the filter divergence and it is more reasonable to add error perturbations to the forward model because there are so many uncertainties and error sources in the reality, the model error perturbation is added to the EnKF. The synthetic study case and the real hydraulic conductivity field of the transient state groundwater flow are the same as above. The EnKF method by adding the model error perturbation is applied to the transient state groundwater flow to update the hydraulic conductivity through assimilating the observed data of the hydraulic head. After comparing the inverse results obtained via the EnKF by adding model error perturbations with the results produced by the EnKF and the results produced by the covariance inflation scheme via the EnKF method, the problem of the filter divergence is found to be improved to a certain degree by adding the model error to the EnKF method though the updated results at later assimilating time is not good. Even though big error has been added to the forward model, the EnKF method still can efficiently update the hydraulic conductivity field. The EnKF method by adding model error perturbations is more efficient than the EnKF method by the covariance inflation to update the hydraulic conductivity field via assimilating the observed hydraulic head data from the transient groundwater flow. / A Dissertation submitted to the Department of Earth, Ocean, and Atmospheric Sciences in partial fulfillment of the requirements for the degree of Doctor of Philosophy. / Fall Semester, 2010. / October 20, 2010. / Data Assimilation, (Localized) Ensemble Kalman Filter, Hydraulic Conductivity/Head, Transient Groundwater Flow, Boundary Condition, Heterogeneity, Transient Conservative Solute Transport, Solute Concentration, Initial Distribution of the Solute Concentration, Constrained EnKF, Filter Divergence, Model Error Perturbation / Includes bibliographical references. / Bill X. Hu, Professor Directing Dissertation; I. Michael Navon, University Representative; James F. Tull, Committee Member; Yang Wang, Committee Member; Ming Ye, Committee Member.