Estimates of soil hydraulic parameters are essential for predicting and describing water movement in unsaturated soils. Inverse approaches to estimating soil hydraulic parameters have gained great favor. However, limited effort has been placed on obtaining estimates using observed experimental data. Moreover, little has been advanced in applying these approaches to larger scale, multidimensional systems. In this research, several transient laboratory experiments were conducted using both repacked and intact soil cores, and larger scale 1 and 3-dimensional repacked soil columns. Measurements of soil water matric potential and water content were used to obtain parameter estimates for the closed form van Genuchten soil water relations. For the soil cores, measurements were obtained using pressure outflow, upward infiltration and evaporation procedures. For the 1-dimensional soil columns, these same data were obtained using upward and downward infiltration procedures. For the 3-dimensional soil columns, a point source application of water was used. Optimizations were carried out With HYDRUS-1D and HYDRUS-2D using observed matric potential and water content data to define the objective function. For the repacked and intact soil cores, parameters were also estimated by nonlinear least squares fit to retention data obtained from the pressure outflow and evaporation procedures. Parameter estimates obtained by nonlinear least squares fits to the pressure outflow and evaporation retention data were seen to be similar to those obtained by optimization. However, parameter estimates based on data obtained from the upward infiltration experiments were considerably different than those obtained from the evaporation experiments. These differences are attributable to hysteresis. Parameter estimates obtained for the repacked soil cores were seen to be inadequate in terms of predicting matric potential responses in larger scale soil columns. This was also true when parameter estimates obtained from the larger scale repacked soil columns were used to predict matric responses during 3-dimensional flow in the largest scale soil columns. The results of this work show that inverse methods when used in conjunction with upward infiltration and evaporation procedures can provide accurate estimates of unsaturated hydraulic parameters. When used together, these experimental methods can be used to obtain data and hydraulic parameters describing both the imbibition and drying branches of the soil water retention response. These results further show that estimated parameter values based on measurements made at one scale tend to be inappropriate in terms of adequately describing systems at larger scales.

Various methods are used to postulate, compare and rank alternative conceptual-mathematical models of unsaturated flow and transport during infiltration and tracer experiments at the Maricopa Agricultural Center (MAC) near Phoenix, Arizona. The models include one- and two-dimensional flow and transport in a uniform soil, a soil consisting of uniform layers, and a stratified soil having laterally varying properties. Characterization of the soil as a uniform medium is based on information obtained from public sources. Characterization of the soil as a layered medium is based on site data. Soil hydraulic properties are estimated from pedologic data. Uniform soil hydraulic properties are ascribed to each layer on the basis of soil type using mean values of three generic databases. Variable soil hydraulic properties are ascribed to individual soil samples using pedotransfer models based on site soil data. By treating these variable hydraulic property estimates as measurements, Bayesian updates of their mean values together with the variance of each estimate are obtained. Geostatistical analysis of soil pedologic and hydraulic properties provides support for a layered conceptual model with relatively large-scale lateral variability in each layer. The above conceptual-mathematical models and hydraulic parameter estimates are applied to analyze flow and transport under uncertainty and to compare simulated and observed water contents during one of the infiltration experiments at the MAC. It is shown that in order to reproduce observed behavior, it is necessary to further modify the hydraulic parameter estimates through inverse modeling. Various conceptual-mathematical models and parameter estimates are compared and ranked using likelihood-based model discrimination criteria. The inverse results are confirmed by using those to simulate the flow of an earlier experiment. Finally, the inverse estimates of soil hydraulic parameters based on one infiltration experiment are used for solute transport modeling. An advection-dispersion model with linear mass transfer to and from a zone of immobile water, or anion exclusion, is employed in forward and inverse modes. The study illustrates how a combination of methods and data sets can be used sequentially and in tandem to improve one's understanding of unsaturated flow and transport conditions at a site.

This research has developed a method for estimating hydraulic roughness coefficients for overland flow models in a dynamic approach, to more effectively simulate runoff on natural, agricultural and urban slopes. The hydraulic roughness coefficients are then generated with a series of neural networks. First, a laboratory experiment was designed to explore the effects of soil microtopography, slope and Reynolds number on the magnitude of Darcy-Weisbach, Manning and Chezy roughness coefficients. It was found that three parameters were necessary to describe the soil surface microtopography. Neural networks developed in a preliminary phase were able to reproduce the roughness coefficients obtained in the laboratory experiment by using five predictor variables: bed slope, Reynolds number, and the three parameters used to describe the microtopography. However, these networks failed to generate roughness coefficients for different input variables (generalization). Second, more complex algorithms were developed as combinations of neural networks in parallel. The algorithm output, the sought hydraulic roughness estimate, was estimated with the arithmetic average of the individual network outputs. Results presented in this study demonstrate that combining multiple neural networks reduced the prediction error and improved on the generalization ability of the neural networks. It was also observed that the estimate accuracy was influenced by the characteristics of the dataset, and especially by the relationship between the roughness coefficient and Reynolds numbers. Finally, a field experiment was performed to explore the applicability of the algorithms. A numerical model based on the 1-D diffusion approximation to the Saint Venant equations was constructed, and two surface irrigations were performed to collect data to test the model estimates. The model was used under two scenarios: (1) with constant hydraulic roughness coefficients, and (2) using variable hydraulic roughness predicted with the algorithm. Discharge at the end of the plot and irrigation front advance estimated using both models matched the observations well. However, when using a variable hydraulic roughness, the front was initially delayed until there was a sufficient surface storage to push it forward. The methodology described in this research should be useful for 2-D overland flow models applied to natural slopes with unsteady rainfall.

The results of a study on phytoremediation of nitrate and ammonium contaminated soil and groundwater at the Monument Valley Uranium Mill Tailings Remedial Action (UMTRA) Project site are presented in this dissertation. The following is a summary of the findings: A phytoremediation plot was established to remediate soil nitrate and ammonium contamination at a former uranium ore-processing site. Atriplex canescens were planted and deficit irrigated, preventing recharge into the shallow aquifer. Initial soil concentration of nitrate-N was approximately 180 mg kg⁻¹ but decreased to only 80 mg kg⁻¹ after 41 months. A decrease in nitrate-N concentrations was observed throughout the 4.6 m soil profile, which was a unique observation. Ammonium-N concentrations remained at initial levels of approximately 180 mg kg⁻¹ and did not decline over the study period. Soluble salts decreased only 20% in the soil, which was attributable to nitrate loss. Residual soil nitrogen became enriched in ¹⁵N, indicating biological denitrification. Nitrate-N loss was 1,360 kg ha⁻¹yr⁻¹, which is approximately three times higher than agricultural soils. These findings may provide a low-cost method for soil nitrate remediation. A. canescens and Sarcobatus vermiculatus are native phreatophytic shrubs at the UMTRA Project site that were evaluated for their potential to remove nitrate from the shallow aquifer. Stable isotope signatures from plant stem water were similar to the groundwater isotope signatures, suggesting the plants are rooted in the plume. Currently only 7% of the plume area is vegetated, mainly due to heavy grazing. When protected from grazing, plants increased in cover by over 50% per year during a three-year period. Transplants of A. canescens that were protected from grazing and irrigated during the first summer after planting were rooted in the plume within three years, growing greater than 2-m in height. Based on these results, the nitrate plume could be removed within 13 years if grazing were restricted and vegetation were to be enhanced to 50% cover, whereas almost six decades would be required for remediation under current conditions. This study shows that phytoremediation may be an inexpensive and non-invasive means of nitrate remediation at this and other arid locations.

Tracer studies using Br⁻ and bacteriophage PRD1 in both surface and subsurface flow constructed wetlands were conducted to analyze their hydrodynamic behavior and efficiencies in removing viruses from wastewater. A survival test in situ was also conducted to analyze the persistency of PRD1 in wetland environments. Concurrently, a sampling program for microbial and chemical indicators in the surface flow wetland for a period of 16 months was conducted. The tracer studies revealed a reduction of 99 and 84 percent in the subsurface and surface flow wetland, respectively. Bromide recovery at the outlet of both wetland systems was about 75 percent. The Convective-Dispersion Equation was able to predict the observed PRD1 and Br⁻ breakthrough curves obtained during the tracer study in the surface flow wetland. The monitoring program of pollution indicators showed that biochemical oxygen demand and total suspended solids can be reduced efficiently, reaching the tertiary effluent standard of 10 mg L⁻¹ required by The Arizona Department of Environmental Quality. This sampling program suggested that coliphages may be a better indicator of fecal contamination than total and fecal coliforms in surface flow wetlands.

Eleven series of replicated tests were conducted using 38.1 mm, 15.9 mm, and 9.5 mm gravel to determine the most effective soil surface cover to prevent soil erosion from rainfall. A sediment tray one meter square in size with an integrated rainfall simulator was used to generate data after initial trial runs had established test procedures. Various size gravels and a control with no cover were tested in a laboratory using simulated rainfall to evaluate their effectiveness in preventing erosion. Through thirty-three experiments, signature traits of specific rock sizes were identified. Experiments on 38.1 mm gravel indicated the usefulness of rock mulches in soil erosion prevention. Evaluations with 9.5 mm material indicated that erosion prevention varies inversely with particle size. Experiments with 15.9 mm gravel suggested that this material could increase erosion. This study reflects the ambivalence in the literature and points to the complexity of micro-interactions and erosion potential as influenced by gravel size. Six mechanisms governing rock mulch erosion were proposed.

A number of factors have contributed to the interest in reclaimed wastewater irrigation of vegetables. However safety of water as related to fresh cut vegetables has been a paramount concern of responsible agencies, growers and consumers. The objective of this study was to evaluate the risk mitigation potential of subsurface drip irrigation during reclaimed wastewater irrigation. Virus detection methodologies on produce were first optimized. Beef extract (3%) +0.04M sodium pyrophosphate was found to be the most efficient eluent to recover viruses from soil and plant material. The recovery efficiency of poliovirus type l and adenovirus type 40 from Pima clay loam soil ranged from 12--16%, and recoveries from Brazito sandy loam ranged from 58--81%. The recovery efficiencies of poliovirus type 1, adenovirus type 40, MS2 and PRD1 from lettuce ranged from 45 to 70%. The recovery efficiencies of MS2 and PRD1 from tomato fruit were in the range of 90%. Poliovirus type 1, adenovirus type 40, MS2 and PRD1 survived longer in Pima clay loam than in Brazito sandy loam. All enteric viruses remained stable at 4°C suggesting that little virus inactivation occurs during transportation and marketing. Poliovirus type 1 and adenovirus type 40 lost 1 log₁₀ in 11 and 17 days respectively on lettuce at room temperature. Hepatitis A virus lost 2.5 log₁₀ in 50 days on a lettuce head in a greenhouse during the winter season. Tomato, lettuce and cucumber crops were irrigated with virus-seeded water by subsurface and surface drippers. Subsurface drip irrigation resulted in 99% less viral contamination of vegetable leaves as compared to surface drip irrigation. The greatest risk of infection occurs from the outer leaves of lettuce. The risk of infection from consumption of reclaimed wastewater irrigated tomatoes and cucumber was 32% and 72% less than lettuce. The risk of infection from rotavirus by ingestion of vegetables is greater than poliovirus type 1. The risk of infection from subsurface drip irrigated vegetables did not approach the United States Environmental Protection Agency's acceptable annual risk (1:10,000) until the concentration of viruses in the irrigation water reached 100/l. No internal contamination by viruses of the vegetables was observed during their growth. The coliphages PRD1 survived longer than poliovirus type I, adenovirus type 40, and MS2. It would thus appear to be a good model for studying the persistence of those viruses on produce and in irrigated agricultural systems.

Over the past century, woody plants and shrubs have increased in abundance at the expense of grasslands in many semiarid regions. The availability and concentrations of nutrients influence the relative success of plants, but the effects of fire frequency on soil nutrients is unknown for semiarid grasslands. On the gunnery ranges of Fort Huachuca in southeastern Arizona, study sites were established to examine the effects of fire frequency on soil biogeochemistry, plant biochemistry, and δ¹³C values in soil organic matter (SOM). The sites were on homogeneous granitic alluvium where wildfire frequency history is known from 1973 to present and no cattle grazing has occurred in recent decades. Subplots represent fire frequencies of no burns, 3 fires per decade, and 5 fires per decade. The "no burn" plot has abundant C3 Prosopis velentina (mesquite) trees, whereas the burned plots are open C4-dominated grasslands with scattered mesquite trees. Prosopis trees have altered SOM pools by the concentration of plant nutrients and the addition of isotopically light shrub litter. Frequent fires have altered the basic geochemistry and nutrient availabilities of the soil, and the changes appear to be significant enough to affect plant growth. Soil pH increases with burning frequency, and TOC, total nitrogen, and plant-available phosphorus show significant increases on the infrequently burned plot. Burning is advantageous for preservation or restoration of grasslands, as total living grass biomass is greater on the two burned plots. Root biomass 11 is significantly lower on the "frequently burned" plot. Concentrations of the key nutrients nitrogen and phosphorus are reduced in plants on the burned sites compared to plants on the unburned site. Fires help re-distribute nutrients but evidence of nutrient concentrations and δ¹³C values are retained in SOM for many decades. Estimates of bulk carbon turnover rates range from 112 to 504 years. Evidence for modern C3 shrub expansion is found in the shift of SOM δ¹³C values from values characteristic of C4 grasses to C3 shrubs in surface soil layers. δ¹³CSOM values indicate that the Holocene and Late Pleistocene were dominated by C4 grasslands, and the pre-Late Pleistocene vegetation was a C4-grass savanna with abundant C3 plants.

A numerical first order approach is proposed to conduct stochastic analyses of head and concentration under variably saturated conditions. The approach is based on a first-order Taylor series expansion and an adjoint state method. To implement the approach in different flow and transport regimes, numerical models are adopted to evaluate sensitivities of head and concentration with respect to hydrological parameters. This provides the possibility of conducting stochastic analyses of flow and transport problems with any kind of boundary and initial conditions. As a result, limitations of analytical approaches such as the spectral/perturbation approach can be avoided. In addition, the use of adjoint state method also alleviates the computational burden encountered in Monte Carlo simulation by allowing us to evaluate the sensitivities of head and concentration only at interesting/measurement locations. Several numerical simulations are performed to examine the sensitivities and moments of head and concentration under different flow conditions. The results show that the existence of water tables in the simulation domain can have a significant impact on the moment calculation of head and concentration. The calculated statistical moments are used to estimate log-conductivity by cokriging. The conditioning effect of head, concentration, and arrival time in estimating log-conductivity is investigated under different flow conditions. The results show steady state head is the best secondary information compared to solute concentration and arrival time in estimating conductivity by providing stable and consistent results. Estimates can be error prone when concentration measurements are used to estimate LnKs because of the nonlinear relationship between concentration and LnKs and the large variability in the simulated solute plumes. A sequential estimating technique is shown to be able to overcome some of these inadequacies of using concentration measurements. Arrival time, requiring a large amount of CPU time, does not show any advantage over concentration and head in estimating conductivity.

Contamination of soil and groundwater environments by toxic metals and organic compounds is of major concern because of the potential health hazard posed for humans. Remediation of such sites may require the addition of chemical agents that help in the mobilization of contaminants which are likely to be bound to solid surfaces. Surfactants are one of the agents proposed for addition to enhance the removal of soil-bound contaminants. In this dissertation, a series of laboratory experiments were conducted to investigate constraints to the potential application of a microbially produced surfactant (biosurfactant) for removal of metal contaminants from soil. The first part of the dissertation describes the measurement of stability constants and 13 metals including ten of the metals most frequently found in contaminated sites as well as three of the most common metal cations found in soil, Ca²⁺, Mg ²⁺, and K⁺. The second part of the dissertation describes a series of experiments designed to determine the interaction of the biosurfactant with soil matrix components including clays, metal oxides, and organic matter. The biosurfactant used in this research was monorhamnolipid produced by Pseudomonas aeruginosa ATCC 9027. A mixture of mono- and dirhamnolipid produced by P. aeruginosa UG2 was also used in some cases. Results showed that selectivity of the monorhamnolipid followed the order: Al³⁺ > Cu²⁺ > Pb²⁺ > Cd²⁺ > Zn²⁺ > Fe³⁺ > Hg ²⁺ > Ca²⁺ > Co²⁺ > Ni²⁺ > Mn²⁺ > Mg²⁺ > K⁺. These results suggest that monorhamnolipid binds common metal contaminants in preference over common soil cations such as Ca²⁺, Mg²⁺, and K⁺. Rhamnolipid was shown to bind to some soil constituents very strongly including hematite, illite, kaolinite, and montmorillonite. These results indicate that the effectiveness of rhamnolipid in soils with high amounts of iron oxide or clay may be limited due to extensive sorption. Finally, it was found that monorhamnolipid sorbed more strongly than a rhamnolipid mixture containing both monorhamnolipid and dirhamnolipid. This suggests that the use of a biosurfactant mixture may improve the effectiveness of rhamnolipid in the removal of organic and metal contaminants from soil.