Hydrology is a field fraught with uncertainty. Uncertainty comes from both our inability to perfectly know the true nature of constant system components of hydrologic systems (e.g. hydraulic conductivity, geologic structure, etc.) and our inability to perfectly predict the behavior of variable system components (e.g. future precipitation, future streamflow, etc.). Hydrologic literature has increasingly recognized that within the bounds of uncertainty, many acceptable hydrologic models exist and differ in their predictions. Modeling applications that recognize this uncertainty have become more practical as a result of increasing computing power and improved software. Given a set of model predictions, the applied hydrologist or water resource manager is faced with an important question: in light of this uncertainty, how do I make the best decision? Many decision making criteria are valid for use in water resources, however, decision making criteria are subjective in their nature and require input from the decision maker about their values and outlook. Decision making criteria can range from optimistic to pessimistic, and can be probabilistic or non-probabilistic. This dissertation explores the importance of hydrologic uncertainty and the stance of the decision maker in selecting an appropriate decision making criterion. The dissertation comprises four manuscripts. The first manuscript presents an analysis of uncertainty arising from choice of groundwater sampling method. The study analyzes how three sampling methods compare across a range of analytes and well constructions. The second manuscript presents an analysis of the risk that a wellfield will not be able to meet water demands. A Monte-Carlo model is used to evaluate how uncertainty arising from variable groundwater recharge in an alluvial aquifer translates to total wellfield risk. The third manuscript reviews multi-model methods used to support decision making and makes an argument that non-probabilistic decision making methods deserve a larger role in hydrologic studies. A groundwater recharge example is presented that compares the performance of model selection, model averaging, probabilistic, and non-probabilistic decision making methods when used for decision making. The final manuscript presents the Discrimination Inference to Reduce Expected Cost Technique (DIRECT). DIRECT is a MATLAB® based computer code that optimizes project design under uncertainty using an expected utility decision criterion. Examples are presented for remediation system design and groundwater pumping.

Water quality in the Sierra Vista Ground-Water Basin is of extreme importance due to the basin's unique ecosystem and predicted future population growth. Portions of the Upper San Pedro River, flowing through the Sierra Vista Basin, contain some of the few remaining perennial streamflows in the southwest. Baseflow in the perennial reaches of the river are maintained almost entirely by the regional and floodplain aquifer systems. A population increase is predicted for the Sierra Vista Basin, and an impact on groundwater quality and availability can be expected. Due to the closely linked hydrologic systems within the basin, contamination or depletion of the regional aquifer could have direct implications for the San Pedro River. Water samples were collected within the study area from the regional and floodplain aquifers, the San Pedro River, and a bedrock spring in the Huachuca Mountains. Samples were analyzed for field parameters, major-ions, and stable isotopes to describe the main chemical characteristics of the hydrologic systems within the basin. Analysis of regional aquifer geochemistry indicates a ground-water system strongly controlled by calcite precipitation. Specific conductance, deuterium and oxygen-18 values indicate a mixing of regional-aquifer ground water and San Pedro River surface water within the floodplain aquifer. Estimates of inflow to perennial reaches of the floodplain aquifer from the regional aquifer vary from 50 to 80%, depending on location. Inflow to the San Pedro River at Charleston from the regional aquifer is estimated to be about 50 to 70% of the stream discharge.

A new, robust and versatile method to solve the hydrological inverse problem has been conceived, developed and thoroughly tested. This method obtains transmissivity values in the form required for commonly-used computer models. By minimization of an objective function that is reduced to retain solely the most important components, the approach is to find a transmissivity distribution and slightly modified head and flux fields. The equations resulting from the objective function are then solved iteratively by successive overrelaxation, starting with a guess value and converging toward a solution. / The method is tested by means of a synthetic application. Transmissivity, head and flux distributions are chosen as functions of space so as to meet the steady-state hydrological equation. The transmissivity is then put aside, and another transmissivity distribution is obtained from the head and flux values at grid points using the inverse method. The true and recovered transmissivities are then compared, with percentage of recovery greater than 90%. Varying initial guesses for transmissivity has very little effect. / The method is applied with success to a portion of the Sand-and-Gravel Aquifer of northwest Florida. Comparison of the results of the application with transmissivity values available from pumping tests shows good agreement. Sensitivity analysis reveals only a reasonable amount of amplification of noise error. / The method's ease of use and its format for the microcomputer make it well-suited for the hydrologist without expertise in inverse methods or extensive computer facilities. Its application to both synthetic and real aquifer situations demonstrates its potential as a practical tool in water resource management. / Source: Dissertation Abstracts International, Volume: 47-06, Section: B, page: 2353. / Thesis (Ph.D.)--The Florida State University, 1986.

Well-developed karst aquifers contain high permeability limestone matrix and much higher conductive conduits, this dual porosity system behaves totally different from other kinds of aquifers and becomes a challenging task for modern hydrogeological study. High permeable conduit system provides idea pipes for contaminant transporting in rapid flowing groundwater, this effect may cause wide range pollution in a short time. One of these serious problems is seawater intrusion. Seawater intrusion has been found in many coastal aquifers, produced contaminated fresh groundwater resources and induced ecosystem problems. Seawater intrusion in a well-developed karst aquifer such as Woodville Karst Plain (WKP) is simulated by Dr. Zexuan Xu (Xu and Hu, 2017a), he developed a new model VDFST-CFP (Variable-Density Flow and Solute Transport - Conduit Flow Process) which considers the variable density flow in dual porosity system. VDFST-CFP provides an accurate simulation of seawater intrusion in a coastal karst aquifer with conduit networks. It couples the variable density flow field and the density function of salinity in the porous medium and non-laminar groundwater flow within karst conduits. Currently, the VDFST-CFP model is used to simulate seawater intrusion condition at a synthetic level, the present numerical simulation only considered the idea circumstance that is one conduit in a 2D model, and data analyses mainly focused on the horizontal source. In this study, an improvement of VDFST-CFP will concentrate on the vertical source model in the WKP, the roughness of conduit wall and multiple pipes will be considered. Two improvement are implemented in the new model: (1) multi-conduit networks in the domain; (2) micro- and macro-structures on the conduit wall (conduit wall roughness). The simulation results show that dual-pipe system produced a larger contaminant plumes than single-pipe system. Meanwhile, rougher micro-structures and more macro-structures on conduit wall slow down the velocity of seawater intrusion in conduit system, however, have a limited affect salinity distribution in the matrix. In addition, local sensitivity analysis and global sensitivity analysis of seven parameters (conductivity, diameter, dispersivity, exchange permeability between conduit and matrix, effective porosity, mean roughness height and specific storage) are conducted in this study. Sensitivity results indicate that conductivity, diameter and porosity are more important to head and salinity distribution simulations than other four parameters. Diameter is the most important parameter to the conduit simulation, while matrix simulations is more sensitive to effective porosity. Furthermore, scenarios study about variation of boundary conditions is conducted, the result shows that a decreasing of salinity at submarine spring or a decreasing on sea level moves seawater intrusion backward both in conduit and matrix, while the intrusion in conduit and matrix have different sensitivities to the change of boundary conditions. / A Thesis submitted to the Department of Earth, Ocean and Atmosphere Science in partial fulfillment of the Master of Science. / Summer Semester 2017. / June 1, 2017. / coastal karst aquifer, Seawater intrusion, Sensitivity analysis, VDFST-CFP / Includes bibliographical references. / Xiaolong Hu, Professor Directing Thesis; Yang Wang, Committee Member; Stephen Kish, Committee Member; Ming Ye, Committee Member.

Transport of organic contaminants in a sand and gravel aquifer has been investigated in order to increase our understanding of hydrologic and geochemical processes affecting movement of selected organic solutes in groundwater. An abandoned wood-treatment facility in Pensacola, Florida, was chosen for the investigation. The geometry of the system, including locations of hydrologic sources and sinks, was documented in development of a calibrated three-dimensional model of groundwater flow. These data were then combined with contaminant distribution and groundwater velocity data in simulations of solute transport. / Both numerical and analytical transport simulations were performed, in order to compare and contrast results from each. Even with dispersion equal to zero, advective flow resulted in particle transport well beyond the boundaries of the site. All solutes investigated were shown to require both increased retardation (sorption) and increased decay (biodegradation) relative to a conservative constituent. Naphthalene was best simulated with retardation values between 2 and 4 and a decay constant between 0.001 and 0.002 d$\sp{-1}$. The phenolic compounds were best simulated with retardation values between 1 and 2 and a decay constant between 0.003 and 0.02 d$\sp{-1}$. / The complexities of solute migration in groundwater preclude any study from providing unique solutions. Available laboratory experiments on behavior of organic solutes and comparable field investigations at other sites support similarities among the studies. The existence of large vertical chemical concentration gradients which persisted after over 200 meters of lateral travel distance supported the idea of very small vertical dispersivities at the site. The small value for dispersivity suggests that results of vertically-averaged simulations should be used with caution in evaluation of organic transport in the subsurface. This investigation represented an effort to provide a conceptual framework for further studies, including predictions of changes to the environment, based on increased understanding of hydrologic and geochemical processes in the subsurface. / Source: Dissertation Abstracts International, Volume: 50-02, Section: B, page: 0474. / Major Professor: James Bryant Cowart. / Thesis (Ph.D.)--The Florida State University, 1988.

No description available.

Estuaries are important sites of organic matter (OM) transformation, exchange, and burial but remain one of the least understood regions in the global carbon cycle. The carbon cycle within these regions is complex due to strong gradients in biological and physical processes, and increasing anthropogenic impacts. This is further complicated by the many sources of particulate and dissolved organic matter (OM) in estuaries, including materials derived from terrestrial and anthropogenic sources as well as aquatic and marine primary production. This study combined lipid biomarker analyses with stable and radiocarbon signatures of lipids and source-specific biomarkers to better understand the sources and aging of OM in Delaware River and Bay, a model estuarine system. The lipid composition of particulate organic matter (POM, > 0.7 μm) and ultrafiltered dissolved organic matter (UDOM, 1kDa – 0.1 μm) was investigated along the salinity gradient in the Delaware River and Bay during five separate cruises. Sources of OM associated with POM and UDOM were examined using chlorophyll a, C:N ratios, stable carbon and nitrogen isotopes (δ13C and δ15N), total lipid extracts, and fatty acid (FA) biomarker compounds. Multiple hierarchical models explored which environmental characteristics were the primary drivers of POM and UDOM composition. These models revealed that chlorophyll a, POC, and TSS influenced POM sources and composition along the estuary, while a variety of drivers influence UDOM composition. Stable carbon (δ13C) and radiocarbon (Δ14C) measurements of dissolved inorganic carbon, bulk particulate organic carbon (OC), and neutral and polar lipids from particulate organic matter (TLEPOM) and ultrafiltered dissolved organic matter (TLEUDOM) were measured in order to gain insights about the source and age distribution of lipids along the Delaware River and Bay. Overall, Δ14C values for neutral TLE were more depleted (i.e., had “older” radiocarbon ages) than polar TLE. Radiocarbon ages for neutral TLEPOM were younger than neutral TLEUDOM by approximately 10,000 YBP, while polar TLEPOM and polar TLEUDOM were similar in age. Using a 14C isotope mass balance, changes in contributions of modern and fossil OC were quantified along the estuary for TLEPOM and in TLEUDOM. Complementary to determining the radiocarbon ages of different lipid classes, this study was the first to apply compound specific radiocarbon analyses to fatty acids (FA) associated with estuarine POM. Δ14C values indicate that the ages of terrestrial and algal FA change along the estuary. Terrestrial FA increased in age along the estuary due to downstream sources, while algal FA became “younger” along the estuary due to contributions from autochthonous sources. FA biomarker and radiocarbon analyses revealed changing composition of OM along the Delaware River and Bay: (1) older, terrestrial sources of OM characterized riverine OM, (2) the ETM was a location of shifting sources and introduction of “older” POC, and (3) the bay was dominated by younger, marine sources of OM. Lipid age was not based on within estuary processes but on the delivery of “aged” OM from the watershed and along-estuary mixing of different sources. Overall, this study provided new insights about the sources and ages of OM along the estuarine salinity gradient and the complex processes by which they are controlled.

Watershed modeling is an important tool for quantifying the inputs of fresh water, sediments, and nutrients into receiving estuaries and potential changes in those loads under scenarios including changes in land use and climate. There are a variety of existing watershed loading models available, from simple to complex, but a spectrum of these models have yet to be applied and compared in a low relief, coastal plain setting. This project has been conducted as part of the Defense Coastal/Estuarine Research Program (DCERP), which has focused on the impact of Marine Corp Base Camp Lejeune (MCBCL) and activities in the surrounding watersheds on the New River Estuary (NRE), located in southeastern North Carolina. As part of DCERP, nine sub-watersheds on MCBCL with contrasting land use were monitored to allow computation of freshwater, sediment, and nutrient loads to the NRE. In the current project, these loads were used to assess the performance of two existing watershed models using the Environmental Protection Agency’s (EPA) Better Assessment Science Integrating point and Nonpoint Sources (BASINS) 4.0 modeling suite: the relatively complex, temporally-resolved Hydrologic Simulation Program-Fortran (HSPF), and the relatively simple, annually-resolved Pollutant Loading (PLOAD) model. For both models, the 2001 National Land Cover Data were used for analysis; this dataset was compared to the recently released 2006 NLCD dataset and changes were found to be small. Monthly HSPF model output generally followed precipitation trends, and tended to over-predict freshwater stream flow and under-predict sediment and nutrient loads. PLOAD reproduced annual loads of total nitrogen within measured ranges, under-predicted annual loads of total suspended solids, and was less successful at predicting PO43- loads. Results from HSPF and PLOAD were combined with those from six other modeling approaches applied during DCERP to complete a spectrum of models from simple to complex. Model output from HSPF and PLOAD was scaled up to estimate loads entering the NRE from that portion of its watershed lying on MCBCL. Model estimates suggest that approximately 5-6% of the total nitrogen entering the NRE from external sources originates from the MCBCL watershed, a value on the lower end but within the range of estimates from other models applied to the system. Scenarios were run within HSPF to investigate how the conversion of forested land to impervious surfaces on MCBCL may alter existing loads; the model was relatively insensitive to changes in impervious surfaces. Neither PLOAD nor HSPF predicted nitrogen, phosphorus, and sediment loads better than the other watershed models applied during DCERP. The results of this study combined with development of other models suggest that simpler models, such as PLOAD, are able to estimate loads to the NRE as well as more highly technical models, such as HSPF, and that regardless of model choice a focus on loads at the annual scale is most justifiable.

No description available.

No description available.