The world's highest density of tropical glaciers is found in the Cordillera Blanca of the Peruvian Andes. During the dry season, glacial meltwater is a vital fresh water resource in the region as it supplies up to 40% of river discharge. Climate change is driving rapid glacial retreat, causing serious concerns about future availability and quality of fresh water supplies. The purpose of this thesis is to survey water quality in the Rio Santa Watershed (which drains the western side of the Cordillera Blanca), with a particular focus on potentially toxic trace metals released by acid mine drainage and acid rock drainage. In July 2011, major ion samples were collected from 23 sites in the Rio Santa and nine of its tributaries. Samples for trace metal analysis were collected from 11 Rio Santa sites and eight tributaries. pH, temperature, and dissolved oxygen were measured in situ at all sites. Rio Santa discharge was measured directly at 15 locations and calculated by mass balance analysis for another 12 locations. Water in the Rio Santa Watershed is characterized by high concentrations of SO4, Ca, and HCO3. These species are derived from sulfide oxidation (both naturally occurring and enhanced by mining) and carbonate dissolution. The pH is circumneutral at all sites except for two tributaries, Rio Olleros and Rio Quilcay, which both had a pH below 5. Fe-oxyhydroxide coatings cover the streambeds of acid tributaries and some sites along the Rio Santa. As, Cd, Pb, Cu, Mo, Ni, and U concentrations tend to fluctuate above the detection level but below 10 ug/L at most sites in the watershed. Locally elevated concentrations occurred in the acidified Rio Olleros (70 ug/L of Ni) and Rio Quilcay (82 ug/L, 27 ug/L, 34 ug/L and 14 ug/L of Pb, Cd, Ni, and U, respectively.) Elevated concentrations also occurred in circumneutral tributaries Rio Tabla (43 ug/L of Ni) and Rio Llullan (720 ug/L of Mo and 180 ug/L of U.) In the Rio Santa, elevated trace metal concentrations occurred downstream of Ticapampa mine tailings pile (30 ug/L of As) and the city of Huaraz (73 ug/L, 58 ug/L, 53 ug/L, 40 ug/L, and 23 ug/L of Pb, Cd, U, Mo, and Ni, respectively.) Bulk loads of these indicator trace metals were near zero at the headwaters of the Rio Santa, and tended to increase steadily along the river's length in proportion to increasing discharge, with large spikes in loads at Ticapampa and Huaraz. The dramatic decline in loads immediately after these sites is indicative of the non-conservative behavior of dissolved trace elements in the Eh-pH conditions of the Rio Santa. As dry season discharge continues to decline due to glacier recession, current contamination problems may be exacerbated and lower flows will inhibit the capacity of the watershed to buffer against acidic tributary and mining effluent inputs. / La plus grande densité de glaciers tropicaux se trouve dans la Cordillera Blanca des Andes péruviennes. Durant la saison sèche, l'eau issue de la fonte glaciaire est une ressource vitale, car celle-ci alimente jusqu'à 40% du débit des rivières. Les changements climatiques entraînant un retrait rapide des glaciers, l'approvisionnement de la population en eau potable de qualité devient une préoccupation sérieuse. L'objectif de ce mémoire est de faire le relevé de la qualité de l'eau dans le bassin hydrologique du Rio Santa, qui draine la portion occidentale de la Cordillera Blanca, en mettant l'emphase sur les éléments-traces métalliques potentiellement toxiques libérés par le drainage minier acide et par le drainage géologique naturel. En juillet 2011, des échantillons d'ions majeurs ont été récoltés dans 23 sites du Rio Santa et de neuf de ses affluents. Des échantillons d'éléments-traces métalliques ont quant à eux été récoltés dans 11 sites du Rio Santa et de 8 de ses affluents. Le pH, la température et l'oxygène dissout ont été mesurés sur place à tous les sites. Le débit du Rio Santa a été mesuré directement à 15 endroits et calculé par bilan massique pour 12 autres endroits. L'eau du bassin hydrologique du Rio Santa est caractérisée par une forte teneur en SO4, Ca et HCO3, dû à l'oxydation, naturelle et accélérée par les mines, des minéraux sulfureux et de la dissolution du carbonate. Le pH est approximativement neutre pour tous les sites sauf pour les affluents Rio Olleros et Rio Quilcay, qui ont tous deux un pH inférieur à 5. Un revêtement de Fe-oxyhydroxide recouvre le lit de ces affluents acides et quelques autres sites le long du Rio Santa.Les concentrations de As, Cd, Pb, Cu, Mo, Ni et U tendent à fluctuer au-dessus des niveaux de détection mais restent sous 10 ug/L pour la majorité des sites dans les bassins hydrologique. Des concentrations locales élevées apparaissent dans les affluents acides du Rio Olleros (70 ug/L de Ni) et du Rio Quilcay (82 ug/L, 27 ug/L, 34 ug/L and 14 ug/L de Pb, Cd, Ni et U, respectivement.) Des concentrations élevées ont aussi été trouvées dans les affluents dont le pH est approximativement neutre, soit le Rio Tabla (43 ug/L de Ni) et le Rio Llullan (720 ug/L de Mo et 180 ug/L de U.) Dans le Rio Santa, des concentrations élevées d'éléments-traces métalliques ont été détectées en aval des résidus de la mine Ticapampa (30 ug/L de As) et de la ville de Huaraz (73 ug/L, 58 ug/L, 53 ug/L, 40 ug/L et 23 ug/L de Pb, Cd, U, Mo et Ni, respectivement.) Le débit massique de ces éléments-traces métalliques de référence était près de zéro à la source du Rio Santa et tendait à augmenter graduellement au long de la rivière proportionnellement à l'augmentation du débit d'eau avec des grands pics de débits massiques d'éléments-traces métalliques à Ticapampa et Huaraz. La diminution dramatique de ses débits massiques après ces sites indique le comportement non-conservatif d'éléments-traces dissous dans les conditions de pH du Rio Santa. Alors que le débit d'eau de la saison sèche continue de décliner à cause du recul des glaciers, les problèmes de contamination actuels seront potentiellement exacerbés et des flux réduit empêcheront le Rio Santa et ses affluents d'agir comme tampon des affluents acides et des admissions d'eau résiduaires issue des mines.

Low cost, low power wireless sensors (motes) promise to revolutionize environmental data collection, but are they currently refined enough for widespread use by hydrologists? Their viability as a replacement for traditional data collection techniques was investigated in a 7 ha forested watershed in south-western British Columbia. The watershed included 41 instrument clusters measuring air and soil temperature, humidity, throughfall, soil moisture content, overland flow and groundwater head. The foundation of each cluster was a data box containing a MDA300 data acquisition board and a MICA2 processor board from Crossbow Technologies, Inc.™ that allowed for short range wireless data collection. The 41 motes each recorded data every 15 minutes from July, 2006, to April, 2007. In addition to reporting on the reliability of the motes and sensors during the 10 months deployment, the high spatial and temporal resolution data collected by this study gave the opportunity for many analyses of catchment processes. As soil moisture and throughfall are two influential processes in the exchange of water between the earth and the atmosphere, these were the focus of the data analysis. The first analysis was a resampling experiment on seven different events selected from the full data set. Comparing 100 different subsamples each of 5, 10 and 20 points for throughfall and soil moisture showed if increasing the sample size eventually produced diminishing returns in the ability to reproduce the true catchment mean. With significant differences in prediction ability for both soil moisture and throughfall at times of differing hydrologic activity, this analysis provides further support for the theories of changing moisture states of soil moisture and threshold values for throughfall. The second analysis described how the organization of soil moisture and throughfall changed during a range of weather conditions and timescales. Spatial representation of normalized values and Pearson correlation coefficients showed that there were distinct differences between wet and dry periods for soil moisture and between long and short analysis periods for throughfall.

Four artificial tracers were applied to a small headwater catchment in south western British Columbia to study runoff generated from topographically distinct landscape units. The seven hectare catchment is located in the University of British Columbia Malcolm Knapp Research Forest at low elevation (190-280 masl). A weir, multiple tipping bucket rain gauges and several piezometers were used to collect hydrological data. Three separate landscape units were identified based on topography, soil properties and proximity to the stream. The units included an area of shallow slope and deep soil, a riparian area along the intermittent stream channel and an area of very shallow soil with bedrock outcrops on a steep slope. Tracers used included rhodamine-WT, uranine, sodium chloride and potassium bromide. A suite of ion selective and fluorometric probes were used along with automated water sampling to monitor tracer breakthrough. The collected samples were analysed in the lab to validate the field measurements. Tracers were dissolved in solution and applied aerially with a backpack sprayer at the onset of forecasted precipitation events to facilitate rapid infiltration into the soil. The first application took place January 4th, 2006. Measurements were then taken continuously until March 20th, 2006, when a second round of tracers was applied to the landscape units. During the first measurement period, 532 mm of precipitation fell below the forest canopy over 75 days. During the second 78 day measurement period, 290 mm of rain fell. It was found that the overall wetness of the catchment affected travel times significantly. Large storms during the first, significantly wetter, application period exhibited similar lag times from peak event discharge to tracer arrival between the different landscape units. During small precipitation events and under dryer conditions, travel times were greatest in the area of shallow slope and deep soils. These lag times are indicative of longer pathways and perhaps the non-initiation of preferential flow below certain thresholds. In general, it was concluded that delineating catchments into groups of similar landscape units based on physical characteristics may be a promising new approach to explaining catchment runoff response.

<p> Recharge areas for most springs are rarely known because they can be sourced from proximal, shallow, atmospheric sources or long-travelled, deep, regional aquifers and alteration along the flow path is common. Stable isotopic (¹⁸O and ²H) geochemistry of springs water can provide indications of relative flow path distance. Locally sourced springs generally have an isotopic signature similar to the isotopic values of local precipitation for that region and elevation. Springs with a different isotopic composition than local meteoric inputs likely have non-local recharge, representing a regional source. Exceptions to this rule include springs in karst terrain, geothermal locations, or travertine-rich groundwater systems, where other physical or geochemical processes influence the isotopic signature of the water. In this study, we tested local vs. regional sourcing using springs isotopic data from regional studies across Western North America in Arizona, Nevada, and Alberta. These regional study sites included the Basin and Range, Transition Zone, Colorado Plateau, Cordillera, Interior Plains physiographic regions, and lowland to montane aquifers. The combination of location-specific physical data with stable isotopic groundwater data provides an effective method for flow path determination at springs with similar flow and chemistry. Springs from Arizona were found to be a mix of regional and local recharge, those from Nevada were locally sourced, and Alberta's springs are generally sourced from regional flow systems. </p><p> Springs provide a truly unique ecosystem where groundwater first daylights, mixes with surface waters, and both plants and animals find refuge. Variability of springs type (the springs geomorphic properties) can have profound effects on all aspects of a groundwater spring. Water chemistry may exert significant changes on vegetation, which in turn modify the springs ecology. Changes in vegetation composition and density can also change erosion rates and channel morphology, thereby altering geomorphology. Analyses of data from southern Nevada and Alberta, Canada housed in an extensive springs database of western North America were interpreted to determine the interconnectedness of geochemistry, geomorphology, and vegetation cover. Using various statistical techniques, the opposite variables were significant in the two field areas. In the Spring Mountains the highest elevation clusters had the highest plant species diversity and the fewest spheres of discharge. In Alberta the clusters with average elevation, neither highest nor lowest, were the groups with the highest plant species diversity and most variability in geomorphic surface types.</p>

<p> Methods for predicting streamflow in areas with limited or nonexistent measures of hydrologic response commonly rely on regionalization techniques, where knowledge pertaining to gaged watersheds is transferred to ungaged watersheds. Hydrologic response indices have frequently been employed in contemporary regionalization research related to predictions in ungaged basins. In this study, regionalization models were developed using multiple linear regression and regression tree analysis to derive relationships between hydrologic response and watershed physical characteristics for 163 watersheds in the Great Lakes basin. These models provide a means for predicting runoff in ungaged basins at a monthly time step without implementation of any process-based rainfall-runoff model. Major findings from this research study include (1) Monthly runoff in ungaged watersheds was predicted with reasonable skill using regression-based relationships between runoff ratio and watershed physical characteristics; (2) Predictions in ungaged watersheds were highly influenced by the temporal characterization of runoff ratio used to condition the regression models; (3) Watershed classification using regression tree and multiple linear regression techniques resulted in comparable model predictive skill.</p>

<p> This research explores the benefits of conjunctively managed surface and groundwater resources in a volcanic aquifer system to reduce stream temperatures while valuing agricultural deliveries. The example problem involves advancing the understanding of flows, stream temperature, and groundwater dynamics in the Shasta Valley of Northern California. Three levels of interaction are explored from field data, to regional simulation, to regional management optimization. Stream temperature processes are explored using Distributed Temperature Sensing (DTS) data from the Shasta River and recalibrating an existing physically-based flow and temperature model of the Shasta River. DTS technology can collect abundant high resolution river temperature data over space and time to improve development and performance of modeled river temperatures. These data also identify and quantify thermal variability of micro-habitat that temperature modeling and standard temperature sampling do not capture. This helps bracket uncertainty of daily temperature variation in reaches, pools, side channels, and from cool or warm surface or subsurface inflows. The application highlights the influence of air temperature on stream temperatures, and indicates that physically-based numerical temperature models, using a heat balance approach as opposed to statistical models, may under-represent this important stream temperature driver. The utility of DTS to improve model performance and detailed evaluation of hydrologic processes is demonstrated. </p><p> Second, development and calibration of a numerical groundwater model of the Pluto's Cave basalt aquifer and Parks Creek valley area in the eastern portion of Shasta Valley helps quantify and organize the current conceptual model of this Cascade fracture flow dominated aquifer. Model development provides insight on system dynamics, helps identify important and influential components of the system, and highlights additional data needs. The objective of this model development is to reasonably represent regional groundwater flow and to explore the connection between Mount Shasta recharge, pumping, and Big Springs flow. The model organizes and incorporates available data from a wide variety of sources and presents approaches to quantify the major flow paths and fluxes. Major water balance components are estimated for 2008-2011. Sensitivity analysis assesses the degree to which uncertainty in boundary flow affects model results, particularly spring flow. </p><p> Finally, this work uses optimization to explore coordinated hourly surface and groundwater operations to benefit Shasta River stream temperatures upstream of its confluence with Parks Creek. The management strategy coordinates reservoir releases and diversions to irrigated pasture adjacent to the river and it supplements river flows with pumped cool groundwater from a nearby well. A basic problem formulation is presented with results, sensitivity analysis, and insights. The problem is also formulated for the Shasta River application. Optimized results for a week in July suggest daily maximum and minimum stream temperatures can be reduced with strategic operation of the water supply portfolio. These temperature benefits nevertheless have significant costs from reduced irrigation diversions. Increased irrigation efficiency would reduce warm tail water discharges to the river instead of reducing diversions. With increased efficiency, diversions increase and shortage costs decrease. Tradeoffs and sensitivity of model inputs are explored and results discussed.</p>

A study of the Holocene avulsion history of the Brazos alluvial valley revealed that two processes, aggradation and valley tilting, were responsible for generating two styles of avulsion, avulsion-by-progradation and avulsion-by-annexation. As aggradation migrated inland, prograding avulsions tracked the locus of aggradation. Concurrently, a nodal avulsion site developed at 55 km inland, creating avulsions by-annexation. Geomorphic evidence suggests movement along a normal listric fault occurred in proximity to the nodal avulsion site. Within the alluvium above the marine-Oxygen isotope Stage 2 onshore incised valley of the Brazos River, the pattern of stacked channels generated by avulsion was mapped to investigate the aggrading response of the Brazos River to sea level rise. The stacked channels within the valley decrease from eight, at 40 km from the coast, to four, at 65 km from the coast, which reflects the diminishing influence of eustacy inland. As aggradation decreased, while the avulsion frequency remained constant, the younger channels became more isolated, in contradiction to previous stacking models. Those models, however, neglected the influence of antecedent topography during aggradation. Vertically, the eight stacked channels within the lower valley are organized into four stratigraphic units that are attributed to changes in the alluvial valley gradient during aggradation, as calculated from the position of backstepping (retrograding) offshore deltas (paleoshorelines) and their correlative (aggradating) onshore floodplain deposits. The style of avulsion and the channel stacking pattern are both understood with respect to realizable subaerial accommodation. Previous subaerial accommodation models emphasized a proportional upward shift in an equilibrium profile during a sea level rise. Yet, an equilibrium profile must be anchored at both ends. The updip elevation of an alluvial valley is controlled by sediment yield and the cumulative aggradation from all earlier episodes of sea level rise, which should exert a limit on the downdip creation of subaerial accommodation. This study, therefore, quantified the differences in the long-term sediment yield of the Brazos and Trinity rivers of east Texas over past sea level cycles, and concludes that the lower sediment yield of the Trinity River has suppressed its equilibrium profile, thereby limiting the present creation of subaerial accommodation.

In 1993, a tracer solution containing benzene, p-xylene, napthalene, o-dichlorobenzene, and tritium was pulse injected into the subsurface at the Macrodispersion Experiment (MADE) site at Columbus Air Force Base in Mississippi. The hydraulic head, hydraulic conductivity, and contaminant concentration data from this experiment are documented in Boggs et al., 1993a. The Bioplume II biodegradation model was used to simulate this data. All organic tracers were considered as a lumped contaminant requiring 2.731 grams of oxygen to degrade 1.000 gram of total organic contaminants. Natural attenuation was modeled as both first-order decay and instantaneous degradation with first-order decay. The simulations matched the observed tritium data, but did not closely match the organic tracer data.

The development of a method to restrict the land disposal of hazardous waste is an area of great concern for the United States Environmental Protection Agency. A proposed method is to determine allowable concentrations, below cancer limits, of specific chemical contaminants slated for land disposal based on the results of a Monte Carlo type contaminant groundwater transport model. The proposed regulatory model does not account for site specific conditions because the hydrogeologic environment specific data required has not previously been available. A hydrogeologic database is developed and analyzed for specific environments. The behavior of the proposed regulatory model is examined using the hydrogeologic environment specific information as input.

Contaminant sorption to soils has two components: (1) instantaneous sorption, and (2) time dependent sorption. Three coefficients can be used to describe the sorption/desorption of contaminants to soils. (1) The partition coefficient relates the contaminant concentration in the soil to the contaminant concentration in the aqueous phase. (2) The fraction of instantaneous sorption is the ratio of the time independent sorption to the total sorption. (3) The sorption rate constant quantifies the kinetics of the time dependent sorption. This study challenges the long accepted premise that the values of the sorption coefficients are equal to the values of the desorption coefficients and examines the ionic strength and temperature dependence of those coefficients. The sorption/desorption of naphthalene on Lincoln Fine Sand was examined with ionic strengths ranging from 0.001M to 0.5M. The partition coefficients varied as predicted with ionic strength but were about half to a third the predicted magnitude. An apparent minimum partition coefficient was observed at 0.06M. The fraction of equilibrium sorption was a maximum in the range of 0.01M to 0.1M. The sorption rate constants, $\alpha$, decreased with increasing ionic strength. The $\alpha$ for the sorption data was 0.5 hr$\sp{-1}$ at 0.001M decreasing to 0.05 hr$\sp{-1}$ at 0.5M. $\alpha$ for desorption only decreased from 0.13 hr$\sp{-1}$ to 0.09 hr$\sp{-1}$ over the same ionic strength range. Temperature has little effect on the sorption/desorption parameters over the temperature range studied (12 to 30$\sp\circ$C). There is a significant difference between the sorption coefficients and the desorption coefficients. Latent contamination of the soil pore water can be caused by the comparatively slow desorption of naphthalene from the soil. A hitherto uninvestigated mechanism for delayed contaminant release is proposed. Soil organic matter (SOM) may not be immobile on the sediment grain surface. The dynamic nature of the SOM may control the slower desorption.