General Circulation Models are important tools in the study of the earth's climate system. The terrestrial surface forms the lower boundary to such models over continents and a well-defined lower boundary is crucial for reliable climate simulations because the Earth interacts with the atmosphere via this boundary. The primary motivation for this research is to improve the parameterization of these interactions in General Circulation Models using field data and calibration techniques. For this purpose, a recent version of Biosphere-Atmosphere Transfer Scheme was selected, studied, and then calibrated for five different vegetation types using multi-criteria calibration techniques. The associated parameter sets were then tested in a ten-year climate integration with Version 3 of the Community Climate Model. The present study explored the methodology needed to use the growing number of relevant field data sets effectively and efficiently better to parameterize the land surface in a GCM. It showed that such field data can, indeed, be used in this way, not only to improve simulations but also to understand models' capabilities and deficiencies. Calibrating the land surface parameterization significantly improved simulations relative to the original default parameterization but several physically based land surface models studied, once calibrated, were found to give equally good simulations of the land surface processes. The primary results are that it is possible to obtain a single preferred parameter sets for different vegetation types using multi-criteria calibration, and that using calibrated parameter sets in climate models can improve the representation of surface exchanges and the modeled climate given by a GCM.

We analyze the movement of DNAPL in a three-dimensional randomly heterogeneous porous medium, saturated with water, that is initially pooled above a water table or flows at a constant flux. We consider the front to form a sharp boundary at which the capillary pressure head, assumed equal to the entry pressure head of DNAPL, is prescribed either deterministically or randomly, treat log conductivity as a statistically homogeneous random field with given mean, variance and covariance; cast the corresponding boundary-value problem in the form of an integro-differential equation, in which the parameters and domain of integration are random; expand this equation in a Taylor series about the mean position of the front; and take ensemble mean. To quantify the predictive uncertainty associated with this mean solution, we develop a set of integro-differential equations for the corresponding second ensemble moments. We solve the resulting moment equations analytically and numerically in one and two dimensions to second order in the standard deviation of log conductivity. A comparison of our one-dimensional solutions with the results of Monte Carlo simulations verifies its accuracy. We also show that a probabilistic analysis of wetting front instability due to Chen and Neuman (1996) applies to a DNAPL front.

This work investigates the conservative behavior of boron isotopes and their potential use as an artificial hydrologic tracer. The usefulness of boron isotopes as intrinsic hydrologic tracers and the assumption that boron isotopes are conservative in many natural environments led to the hypothesis that they can be easily adapted as artificial hydrologic tracers. The results are demonstrated in three field studies and a laboratory study carried out under the Soil Aquifer Treatment project. Non-conservative behavior of intrinsic boron isotope ratios was found in vadose zone and groundwater samples. Although this behavior was recognized, its overall importance is relatively small when compared to the range of values found in the vicinity of the field site that ranged from 0 to 39‰. Fractionation of boron isotopes from its source value, domestic effluent with a range of 0 to 6‰; and an average of 3‰, is on the order of 2‰ and is confirmed by a laboratory study. The fractionation is in part related to the humic acid found in organic matter collected from recharge basin surface soils. Boron isotopes, specifically boric acid enriched in 10B, were successfully applied as an artificial tracer. Xenon and oxygen/hydrogen isotopes, and sulfate support interpretation of boron tracer results. However, non-conservative behavior of boron isotopes is also identified in this artificial tracer study. The non-conservative behavior is demonstrated by the late breakthrough of the boron isotope tracer, two days later, relative to that of xenon isotopes added by researchers from Lawrence Livermore Laboratory.

The level of spatial and vertical detail of important hydrologic processes within a watershed that needs to be represented by a conceptual rainfall-runoff (CRR) model in order to accurately simulate the streamflow is not well understood. The paucity of high-resolution hydrologic information in the past guided the direction of model development to more accurately represent processes directly related to the vertical movement of moisture within the watershed rather than the spatial variability of these processes. As a result, many of the CRR models currently available are so complex (vertically), that expert knowledge of the model and watershed system is required to successfully estimate values for model parameters using manual methods. Automatic parameter estimation procedures, developed to reduce the time and effort required with manual methods, do not provide parameter estimates and hydrograph simulations that are considered acceptable by the hydrologists responsible for operational forecasting. Newly available, high-resolution hydrologic information may provide insight to the spatial variability of important rainfall-runoff processes. However, effective and efficient methods to incorporate the data into the current modeling strategies need to be developed. This work describes a new hybrid multicriteria calibration approach that combines the strength of automatic and manual calibration methods. The new approach was used to investigate the benefits of different levels of spatial and vertical representation of important watershed hydrologic variables with conceptual rainfall runoff models.

Spreadsheet programming is used to model the water budget and water management systems of the Upper San Pedro River Basin, Arizona, to examine how hydrologic, demographic, climatic, and economic conditions affect this budget; and how water management policies may affect the budget over the next twenty years. The model represents relationships among urban and rural population and median incomes; urban and rural domestic water consumption; agricultural and industrial consumption; domestic, agricultural, and industrial return flows; phreatophyte and other evapotranspiration losses; natural surface and ground-water recharges and discharges; head-dependent flow between a floodplain aquifer and the river; head-dependent flow between a regional aquifer and the floodplain aquifer; and changes in aquifer storage. Applications indicate that stringent conservation measures and reductions in irrigation are necessary to significantly reduce ongoing losses to storage in both aquifers, which also threaten river discharges and the viability of riparian habitat in the basin.

To improve the understanding of the movement of copper plumes in natural ground water systems, the partitioning and transport of Cu(II) in the presence of natural humic substances were investigated in the lab. The humic substances were isolated from high organic content ground water in Orange County, CA (5mgCL⁻¹). Batch and titration experiments produced one-site Langmuir isotherms for the sorption of Cu(II) (0.1-5.0 mgL⁻¹) to am-SiO₂ (K(L) = 1.46Lmg-1, Q = 5.92μgg⁻¹), α-Al₂O₃ (K(L) = 100L mg⁻¹,Q = 130μgg⁻¹), and aqueous humics (pK₁ = -5.5, pL₁ = 5.1). Results of the column experiments suggest that mobile humics facilitate the transport of Cu(II) by lowering the free Cu(II) concentration. Mineral-bound humics retarded the transport of Cu(II) by increasing the concentration of immobile binding sites and by binding considerably more Cu(II) per mass carbon than aqueous humics. The measured Langmuir isotherms accurately predicted retention times and explained much of the tailing of the desorption breakthrough curves.

Uncertainty in the hydrologic and soil erosion predictions of the WEPP Watershed model due to errors in model parameter estimation was identified through a sensitivity analysis based on the Monte-Carlo method. Identification of parameter sensitivities provides guidance in the collection of parameter data in places where the model is intended to simulate soil erosion. Changes in model predictions caused by changes in model parameters were quantified for model applications in semi-arid rangeland watersheds. The magnitude of the changes in model parameters was defined by the spatial variability of parameters in a watershed. Model sensitivities in predicting overland flow and soil erosion on hillslopes and channels are presented considering rainfall characteristics. The results show that WEPP predictions are very sensitive to attributes that define a storm event (amount, duration, and ip). Model sensitivity to soil erosion parameters also depends of the type of storm event.

The water resources of the Southwestern United States are under significant stress. The historical record of the Colorado River indicates that the commitment allocations (7.5 million acre-feet to both the Upper and Lower Colorado basin states, and 1.5 maf for Mexico) have overestimated the average available streamflow. Compounding the supply problem, the Bureau of Reclamation has projected an average decrease of 9% in the Colorado River streamflow between the years 2011-2060. Improving forecasts of climate and streamflow, at nearly all time scales, is imperative to most effectively manage these strained water resources. Given the challenges confronting the Southwest, three research studies are presented that could be used to assist water managers. The first study targets the lack of skill seen in seasonal forecasts of precipitation across the US issued by the Climate Prediction Center (CPC). An objective and concise methodology is shown to improve overall seasonal forecast skill as an alternative to forecasts made by the CPC. This methodology uses a combined linear and nearest neighbor model to make forecasts, with the NINO3.4 index as the only predictor. The second study shows skillful forecasts of decadal Colorado streamflow using the Atlantic Multidecadal Oscillation (AMO) and Pacific Decadal Oscillation (PDO) indices as predictors. However, even though the instrumental record showed statistically significant skillful forecasts, the reconstructed records of AMO, PDO and streamflow appear to challenge these results. Lastly, the third study investigates the effects of climate change in the 21st century on the Salt, Verde and Rio Grande river basins. Two dynamically downscaled General Circulation Models (GCMs) are first bias-corrected. Then, the output of these models is used as the climatic forcings for the Variable Infiltration Capacity (VIC) hydrologic model. Results suggest that future streamflows are projected to decrease by 22% and 37%, for the respective GCMs, averaged across the basins.

In a water-stressed region, such as the southwestern United States, it is essential to improve current seasonal hydroclimatic predictions. Typically, seasonal hydroclimatic predictions have been conditioned by standard climate indices, e.g., NINO3 and PDO. In this work, a methodology called basin-specific climate prediction (BSCP) is proposed to improve hydroclimatic predictions. The method analyzes the statistically unique relationships between sea surface temperatures (SSTs) and a basin's hydroclimate. The oceanic region which maximizes the correlation is subsequently used as a predictor for hydroclimate. BSCP is used to perform hindcasts for the hydroclimate in the Little Colorado River basin and the results are compared to those using standard climate indices as predictors.

Large rivers in semiarid regions provide valuable resources for municipalities, agriculture, and adjacent ecosystems, yet the limited nature of the resource requires a concerted effort to understand the controls on water quality. To address this issue I continued and adapted biannual geochemical synoptic sampling of the Rio Grande from Del Norte, CO to Elephant Butte Reservoir between August 2003 and 2006 which had started in 2001. During average to drought conditions, river discharge was derived primarily from snowmelt in the headwaters region, total dissolved solutes increased downstream, and wastewater treatment plants were the largest source of nitrogen to the river. Surprisingly, return flows of river water diverted for agriculture had lower average nitrogen concentrations than the original river water, indicating that the agricultural system is a sink for nitrogen. When summer climatic conditions changed to a persistent monsoon regime, both the water and solute sources changed as the river was reconnected to its uplands and floodplain. Stable isotope data indicate that monsoon precipitation represented 10-50% of surface flow and that discharge increases were entirely due to ephemeral flows. An analysis of solute concentrations suggested that 80-100% of the increases in sulfate, chloride, DOC and nitrate were due to ephemeral flows. Ephemeral flows replaced wastewater treatment plants as the largest source of nitrogen to the river.Using data from the synoptic sampling, I developed both a simple chloride mixing model and a dynamic simulation model of nitrate to evaluate the controls on nutrient cycling within the Rio Grande. Results from the chloride mixing model indicate that both abiotic hydrologic processes and biotic processes provide important controls on nutrient concentrations. River characteristics that increase surface water/groundwater exchange are important for determining nutrient retention, a result commonly identified in smaller streams, but never before quantified in a large river. Dynamic simulation modeling indicates that both plant uptake and denitrification remove nitrate within the Middle Rio Grande with denitrification accounting for 55-100% of the nitrate removal downstream of Albuquerque. The results of this dissertation provide a conceptual model for the hydrologic and biologic controls on nutrient concentrations in a heavily managed large semiarid river.