Hydrogeology Journal

, Volume 19, Issue 6, pp 1203–1224

Relations of hydrogeologic factors, groundwater reduction-oxidation conditions, and temporal and spatial distributions of nitrate, Central-Eastside San Joaquin Valley, California, USA

Authors

    • US Geological Survey
  • Christopher T. Green
    • US Geological Survey
  • Kenneth Belitz
    • US Geological Survey
  • Michael J. Singleton
    • Environmental Radiochemistry GroupLawrence Livermore National Laboratory
  • Bradley K. Esser
    • Environmental Radiochemistry GroupLawrence Livermore National Laboratory
Report

DOI: 10.1007/s10040-011-0750-1

Cite this article as:
Landon, M.K., Green, C.T., Belitz, K. et al. Hydrogeol J (2011) 19: 1203. doi:10.1007/s10040-011-0750-1
  • 569 Views

Abstract

In a 2,700-km2 area in the eastern San Joaquin Valley, California (USA), data from multiple sources were used to determine interrelations among hydrogeologic factors, reduction-oxidation (redox) conditions, and temporal and spatial distributions of nitrate (NO3), a widely detected groundwater contaminant. Groundwater is predominantly modern, or mixtures of modern water, with detectable NO3 and oxic redox conditions, but some zones have anoxic or mixed redox conditions. Anoxic conditions were associated with long residence times that occurred near the valley trough and in areas of historical groundwater discharge with shallow depth to water. Anoxic conditions also were associated with interactions of shallow, modern groundwater with soils. NO3 concentrations were significantly lower in anoxic than oxic or mixed redox groundwater, primarily because residence times of anoxic waters exceed the duration of increased pumping and fertilizer use associated with modern agriculture. Effects of redox reactions on NO3 concentrations were relatively minor. Dissolved N2 gas data indicated that denitrification has eliminated >5 mg/L NO3–N in about 10% of 39 wells. Increasing NO3 concentrations over time were slightly less prevalent in anoxic than oxic or mixed redox groundwater. Spatial and temporal trends of NO3 are primarily controlled by water and NO3 fluxes of modern land use.

Keywords

Groundwater monitoringHydrochemistryGroundwater protectionNitrateUSA

Relation entre facteurs hydrogéologiques, conditions d’oxydo-réduction de nappe et distribution temporelle et spatiale des nitrates, Centre-Est, de la San Joaquin Valley, Californie, USA

Résumé

Sur une surface de 2,700 km2 à l’Est de la San Joaquin Valley, Californie (USA), des données de sources multiples ont été utilisées pour déterminer les interrelations entre facteurs hydrogéologiques, conditions redox et distribution temporelle et spatiale du nitrate (NO3), un polluant de nappe fréquemment détecté. L’eau de nappe est principalement moderne, ou est un mélange d’eau moderne avec (NO3) détectable en milieu oxydant, mais quelques zones présentent des conditions redox anoxiques ou mixtes. Les conditions anoxiques sont associées à des temps de séjour long qui se rencontrent près de la dépression de la vallée et dans des secteurs de décharge historique de nappe de surface. Les conditions anoxiques sont aussi associées à des interactions entre nappe moderne superficielle et sols. Les concentrations en NO3 sont sensiblement plus faibles dans les eaux anoxiques que dans les eaux oxydantes ou mixtes, principalement parce que les temps de séjour des eaux anoxiques dépassent la durée croissante de pompage et en raison de l’utilisation des fertilisants associés à l’agriculture moderne. Les effets des réactions réductrices sur la concentration en NO3 sont relativement mineurs. Les données sur le gaz N2 dissous indiquent que la dénitrification a éliminé >5 mg/L NO3-N dans environ 10% de 39 puits. L’augmentation des concentrations NO3 dans le temps prévaut légèrement moins en nappe anoxique qu’en nappe à redox oxydant ou mixte. Les tendances spatiales et temporelles de NO3 dans le temps sont principalement contrôlées par l’eau et par les flux NO3 des pratiques culturales modernes.

Relaciones de factores hidrogeológicas, condiciones de oxidación-reducción del agua subterránea, y distribuciones espacial y temporal de nitrato, Valle Centro-Oriental de San Joaquín, California, EEUU

Resumen

Se utilizaron datos de fuentes múltiples en un área de 2,700-km2 en el este del Valle San Joaquin, California (EEUU), para determinar las interrelaciones entre factores hidrogeológicos, condiciones de oxidación – reducción (redox), y las distribuciones especial y temporal de nitrato (NO3), un contaminante del agua subterránea ampliamente detectado. El agua subterránea es predominantemente moderna, o mezclas de agua moderna, con NO3 detectable y condiciones redox óxicas, pero algunas zonas tienen condiciones redox mixtas o anóxicas. Las condiciones anóxicas fueron asociadas con largos tiempos de residencia que ocurrieron cerca del canal del valle y en áreas de descarga histórica de agua subterránea con escasa profundidad del agua. Las condiciones anóxicas también fueron asociadas con interacciones de agua subterránea moderna, somera con los suelos. Las concentraciones de NO3 fueron significativamente menores en agua subterránea anóxica que en agua subterránea óxicas o de redox mixtas, primariamente debido a que los tiempos de residencia de las aguas anóxicas excedieron la duración del bombeo y el uso de fertilizantes asociados a la agricultura moderna. Los efectos de las reacciones redox sobre las concentraciones de NO3 fueron relativamente menores. Los datos del gas N2 disuelto indicaron que la desnitrificación ha eliminado >5 mg/L NO3-N en alrededor del 10% de los 39 pozos. Las concentraciones crecientes de NO3 con el tiempo fueron levemente menos prevalentes en el agua subterránea anóxica que en óxica o de redox mixta. Las tendencias espaciales y temporales de NO3 son primariamente controladas por flujos de agua y NO3 del uso moderno de la tierra.

美国加州东区中部 San Joaquin河谷水文地质条件、地下水氧化还原条件与硝酸盐时空分布的关系

摘要

摘要 : 在位于美国加州的San Joaquin河谷东部一个面积达2700 km 2的区域内, 用多种方法获取数据, 以查明水文地质条件、氧化还原条件与硝酸盐-一种分布较广的地下水污染物-的时空分布之间的关系。地下水主要是由现代水或混入现代水的水源补给, 处于氧化条件下并含有硝酸盐, 但也存在一些缺氧或混合氧化还原环境。还原环境中的地下水一般驻留时间较长, 且位于河谷槽以及浅埋深的古地下水排泄区附近。还原条件还和浅层现代地下水与土壤的相互作用有关。还原环境中地下水的硝酸盐含量明显低于混合有氧化环境的地下水, 主要是因为还原环境中的地下水驻留时间超过了与现代农业相关的开采量增加与施肥活动。氧化还原反应对硝酸盐浓度的影响非常小。溶解 N2数据表明, 39口井的10%经反硝化反应消耗了大于5 mg/L NO3-N 。还原环境中地下水NO3浓度随时间增加的普遍性不如氧化环境或混合氧化还原环境中的地下水。NO3的时空分布主要是由现代土地利用造成的水和NO3通量所控制。

Relações dos factores hidrogeológicos, das condições redução-oxidação das águas subterrâneas e das distribuições espacial e temporal da variável nitrato na área Centro-Este do Vale de SanJoaquin, Califórnia, EUA

Resumo

Numa área de 2,700 km2 localizada a leste do Vale de San Joaquin, na Califórnia (EUA), foram utilizados dados provenientes de múltiplas fontes para determinar as relações entre os factores hidrogeológicos, condições de oxidação-redução (redox) e distribuições espacial e temporal de nitrato (NO3), um contaminante comum nas águas subterrâneas. As águas subterrâneas são predominantemente de origem recente ou resultantes da mistura de águas recentes onde foram detectadas concentrações de NO3 e condições não anóxicas, embora em algumas áreas tenham sido identificadas condições anóxicas ou de características mistas. As condições anóxicas estão associadas a tempos de residência longos que ocorrem perto do vale e em áreas históricas de descarga em aquíferos superficiais. As mesmas condições anóxicas estão também associadas com interacções das águas subterrâneas superficiais recentes com solos. As concentrações de NO3 são significativamente mais baixas em águas com propriedades anóxicas do que em condições não anóxicas ou mistas, porque os seus tempos de residência excedem o tempo de duração de bombagem intensiva e de uso de fertilizantes na agricultura moderna. Os efeitos de reacções redox sobre as concentrações de NO3 são relativamente pouco significativos. A presença de gás N2 dissolvido indica que os processos de desnitrificação eliminaram as concentrações de nitratos acima dos 5 mg/LNO3-N em aproximadamente 10% do total de 39 poços. O aumento das concentrações de NO3 ao longo do tempo foi um pouco menos frequente em condições anóxicas do que em condições não anóxicas ou mistas. As tendências espacial e temporal do NO3 são principalmente controladas pela água e pelos fluxos de NO3 resultante do uso do solo.

Introduction

Reduction-oxidation (redox) conditions influence the transport of many groundwater contaminants derived from both anthropogenic and natural sources (Korom 1992; Smedley and Kinniburgh 2002). Determining the spatial (horizontal and vertical) distribution of redox conditions and the relation to contaminant concentrations is a critical step in understanding the spatial distribution of groundwater quality (Appelo and Postma 1999). Approaches for assessing redox processes based on dissolved concentrations of commonly analyzed constituents using a consistent framework (McMahon and Chapelle 2008; Chapelle et al. 2009; Mendizabal et al. 2010) have been used to relate the spatial distribution of redox conditions to water quality at national (McMahon and Chapelle 2008; Mendizabal et al. 2010), regional (Rose and Long 1988; Rodvang and Simpkins 2001; Gavrieli et al. 2002; Coetsiers and Walraevens 2006; Paschke 2007; Merz et al. 2009), local-flowpath (Tesoriero et al. 2007; McMahon et al. 2008; Green et al. 2008b), and plume scales (Christensen et al. 2000; Wersin et al. 2001). Many of these studies have used detailed geochemical data collected in a limited number of wells to deduce the predominant redox processes that are likely to occur over larger scales. However, there have been relatively few studies that have attempted to map redox conditions in detail using data from a variety of sources on regional scales (Paschke 2007). Given the importance of redox conditions to water quality, it is advantageous to use all available data to determine the spatial distribution of redox characteristics in as much detail as possible in regional-scale investigations of water quality.

Determining the relation of redox conditions to hydrogeologic factors such as depth, position in flow system, lithology, and groundwater age (residence time below the water table) can be useful in predicting hydrological and chemical conditions in areas where they are unknown (Hansen et al. 2008). Redox conditions can vary spatially and temporally in response to changes in the balance of electron donors and acceptors (Chapelle 2001). Redox conditions along groundwater-flow paths commonly proceed along a well-documented sequence (Baedecker and Back 1979; Champ et al. 1979; Berner 1981; Stumm and Morgan 1996) of terminal electron acceptor processes (TEAPs), in which a single TEAP tends to dominate at a particular time and aquifer location (Chapelle et al. 1995; Chapelle 2001). Mixed redox conditions, in which concentrations of dissolved constituents are consistent with multiple TEAPs, have been attributed to either mixing of waters with different redox conditions in well screens or to local heterogeneity of redox processes related to lithologic heterogeneity (McMahon and Chapelle 2008). The limiting factor in TEAP distributions has often been found to be the availability of electron donors, primarily organic carbon or inorganic species such as iron or sulfide (Korom 1992; Zhang et al. 2009). However, detailed information on electron donors is often lacking, particularly on regional scales, necessitating analysis of relations of redox conditions to proxy hydrogeologic factors. McMahon and Chapelle (2008) discussed differences between redox conditions in various principal aquifers across the United States and how they related to geology, climate, and hydrology. Smaller-scale studies have identified cases in which redox conditions are controlled by local groundwater flow dynamics (Promma et al. 2007) or changes in hydrogeologic setting (Veeger and Stone 1996; Gavrieli et al. 2002). In relatively low-permeability glacial tills, the position of a sharp redox boundary between oxidized weathered sediments and reduced unweathered sediments has been found to be related to land-surface elevation (Eidem et al. 1999) and the depth to the water table (Keller et al. 1988; Jørgensen et al. 2004). However, analysis of factors controlling redox at regional scales and in heterogeneous-alluvial-aquifer settings has rarely been documented (Rose and Long 1988; Merz et al. 2009).

Nitrate (NO3) is one of the most widespread groundwater contaminants globally (Spalding and Exner 1993), in the United States (Squillace et al. 2002; US Environmental Protection Agency 2005), and in California (Franco et al. 1994; California State Water Resources Control Board 2002), and increasing NO3 concentrations in groundwater in many regions have been noted (Rosen and Lapham 2008; Rupert 2008). NO3 is regulated as a contaminant in drinking water because it can cause methemoglobinema in infants (Spalding and Exner 1993) and may be associated with some types of cancer (Ward et al 2005). Elevated concentrations of NO3 and increasing concentrations through time are generally attributed to anthropogenic sources including agricultural fertilizers, septic and other wastewater sources, livestock facilities, and atmospheric deposition (Kendall 1998; Böhlke 2002). NO3 transport can be attenuated in aquifers with anoxic conditions, primarily due to denitrification (Korom 1992). Other attenuation mechanisms such as dissimilatory nitrate reduction and assimilation of nitrate into microbial biomass are unlikely to be important in most aquifers (Rivett et al 2008). Although denitrification rates, processes, and effects have been documented along flowpaths at local scales (Tesoriero et al. 2007; Green et al. 2008b; McMahon et al. 2008), there have been few studies describing the relations of redox conditions to NO3 concentrations and changes in concentrations over time in regional aquifers (Otero et al. 2009). In regional aquifers heavily utilized for crop irrigation, large withdrawals from wells and recharge from irrigation applications can substantially increase groundwater velocities and vertical flow components (Burow et al. 2007; Faunt 2009), potentially affecting nitrate transport and degradation rates. However, the factors controlling the distribution of redox conditions and NO3-degradation in heterogeneous-regional-alluvial-aquifer systems with large pumping withdrawals are not well understood.

This study is part of ongoing work by the US Geological Survey (USGS) National Water Quality Assessment (NAWQA) program and the Priority Basin Project (PBP) of the California Groundwater Ambient Monitoring and Assessment (GAMA) program to use groundwater-flow models to simulate changes in NO3 concentrations over time. The purpose of this study is not only to investigate the relations between redox and NO3 concentrations in the Central Eastside San Joaquin Valley of California (Fig. 1) but also to: (1) describe the three-dimensional patterns in redox conditions at a regional scale using existing water chemistry data compiled from multiple sources, (2) analyze relations of redox to hydrogeologic factors, and (3) determine how changes in NO3 concentrations during the last three decades vary in relation to redox zones, groundwater age, and other hydrogeologic factors. This study uses a systematic statistical approach to evaluate hydrogeologic factors as explanatory variables for the distribution of redox and changes in NO3 concentrations over time in a complex regional aquifer system.
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-011-0750-1/MediaObjects/10040_2011_750_Fig1_HTML.gif
Fig. 1

Location of study area in the United States and California with a map of surficial geology and geographic features, and b conceptual cross-section of flow system (modified from Burow et al. 2004; Phillips et al. 2007)

Description of study area and previous investigations

The Central Eastside San Joaquin Valley was selected for study because of the importance of groundwater resources in this area and an abundance of data and detailed previous investigations of water quality (Burow et al. 1998a, b, 2007, 2008b; Wright et al. 2004; Singleton et al. 2007; Green et al. 2008a, b; Jurgens et al. 2008; Landon and Belitz 2008; Landon et al. 2010), hydrogeology (Page and Balding 1973; Burow et al. 2004), and simulation models (Phillips et al. 2007; Burow et al. 2008a; Green et al. 2010). On the basis of these many studies, the Central Eastside San Joaquin Valley is emerging as an important location for better understanding complex hydrogeologic systems and their relation to water quality. For this study, this region serves as a valuable location to evaluate the spatial distribution of redox and its relations to changes in NO3 concentrations over scales of thousands of square kilometers, length scales of tens of kilometers, and age ranges from decades to millennia.

The study area is located within the Central Valley of California (Fig. 1), one of the most valuable agricultural regions in the world, with a cash value of more than $20 billion in 2007 (Great Valley Center 2005; US Department of Agriculture 2007), and a rapidly expanding population dependent on groundwater for drinking water. The southern two-thirds of the Central Valley is the San Joaquin Valley, a 26,000 km2 northwest-trending asymmetrical structural trough containing marine and continental sediments up to 10 km thick (Page 1986). The Sierra Nevada, a mountain range located east of the valley, rises to an elevation of more than 4,200 m (Gronberg et al. 1998).

The 2,700 km2 study area is located on the eastern side of the San Joaquin Valley and includes the City of Modesto (population 206,872; US Census Bureau 2007). The boundaries of the study area (Fig. 1) correspond to those of a numerical groundwater-flow model (Phillips et al. 2007) that is being used in an ongoing study to simulate NO3 transport at a regional scale. However, the areas to the west of the San Joaquin River are excluded from analysis in this study because geochemical conditions on the western side of the valley generally differ from those on the eastern side due to contrasting geologic source materials and climate (Bertoldi et al. 1991; Dubrovsky et al. 1998). The area of primary interest is bounded by the San Joaquin River on the west, the Merced River on the south, the Stanislaus River on the north, and the Sierra Nevada foothills to the east (Fig. 1; Burow et al. 2004).

The study area has a semi-arid climate, with hot and dry summers, and winters that are cool and moist. Average annual rainfall is about 315 mm (Western Regional Climate Center 2007) with approximately 90% of precipitation occurring during November through April. The Stanislaus, Tuolumne, and Merced rivers, as well as their tributaries, are the primary streams in the study area, with most of their flow derived from the Sierra Nevada Mountains to the east. Each of these rivers drains into the San Joaquin River, which flows northwest and empties into the Sacramento–San Joaquin Delta. All rivers in the study area have been significantly modified from their natural state, with multiple reservoirs for irrigation, flood control, and power generation that effectively delay snowmelt runoff.

Land use in the study unit is approximately 65% agricultural (Phillips et al. 2007) with lesser areas of urban and natural land use. The primary crops are almonds, walnuts, peaches, grapes, grain, corn, pasture, and alfalfa (California Department of Water Resources 2009a, b). Fertilizer applications to agricultural crops are the largest source of nitrogen (N) to the study area and county-level average application rates on cropland are estimated to have increased from about 15 kgN/ha/yr in 1950 to about 100 kgN/ha/yr in 2005 (Burow et al. 2008b). The largest urban areas are the cities of Modesto and Turlock with additional areas of urban land use located along the Stanislaus and Tuolumne rivers. The natural land-use areas in the eastern part of the study area are mostly grassland; some riparian woodlands and wetlands are located along streams.

Irrigation began in the study area in the early 1900s (Mendenhall et al. 1916; Davis et al. 1959). Agricultural irrigation supplied by surface water and groundwater currently accounts for about 95% of total water use (Burow et al. 2004). An extensive network of canals is used to deliver water from reservoirs in the Sierra Nevada and foothills for irrigation. Agricultural users also pump groundwater for irrigation, accounting for about 32% of total agricultural water use in 2000 (Burow et al. 2004). In the western part of the study area, numerous shallow drainage wells are pumped to maintain water levels below the root zone, although the amount of drainage pumping has decreased in recent years as groundwater use has increased. Urban water demand is met by both surface-water and groundwater supplies.

The primary aquifers in the study area are a complex sequence of overlapping alluvial fan sediments deposited by three major tributaries to the San Joaquin River that drain the Sierra Nevada (Burow et al. 2004). The main water-bearing units include the unconsolidated alluvial-fan deposits of the Pleistocene-age Riverbank Formation, the deeper unconsolidated Pleistocene-age Turlock Lake and Pliocene-age Laguna Formations, and the semi-consolidated Miocene-Pliocene-age Mehrten Formation (see references in Burow et al. 2004). Holocene flood-basin and dune deposits generally are not saturated except near major rivers. The Mehrten Formation consists of sandstone, conglomerate, siltstone, and claystone derived from fluvial deposits of predominantly andesitic volcanic detritus of the central and northern Sierra and represents a change in lithology and texture from overlying sediments of primarily unconsolidated coarse-grained sediments of arkosic composition. The Mehrten Formation outcrops in the eastern part of the study area but generally is at least 120 m below land surface beneath Modesto.

Groundwater conditions are unconfined, semi-confined, and confined in different zones of the study area. Unconfined conditions are present in unconsolidated deposits above and east of the Corcoran Clay Member of the Turlock Lake Formation, which underlies the southwestern half of the study area (Fig. 1) at depths ranging from 15 to 80 m (Page and Balding 1973; Burow et al. 2004). Confined conditions are present below the Corcoran Clay. Semi-confined conditions are present at depth to the east of the Corcoran Clay, because of many discontinuous clay lenses. The interface between saline and fresh water is generally more than 200 m below land surface (Page and Balding 1973), but may be as shallow as 100 m in parts of the study unit (Burow et al. 2004).

Primary sources of recharge are percolation of irrigation return, seepage from reservoirs and rivers, urban return, and precipitation (Burow et al. 2004; Phillips et al. 2007). Field studies of water-table fluctuations and unsaturated zone tracers in the eastern San Joaquin Valley indicate that recharge during the growing season occurs in response to irrigation and that natural rainfall produces little recharge below fields (Green et al. 2005; Fisher and Healy 2008). In urban areas such as Modesto, however, recharge from precipitation may be more substantial due to thousands of “rock wells”, boreholes backfilled with rock, which route storm runoff into groundwater (Phillips et al. 2007). Recharge from rainfall may also occur near the eastern margin of the valley where valley precipitation increases and runoff from adjacent Sierran foothills occurs. Primary sources of discharge are pumping withdrawals for irrigation and municipal water supply, evapotranspiration from areas with a shallow depth to water table, and discharge to streams (Phillips et al. 2007).

From a regional perspective, groundwater flows laterally from northeast to southwest across the study area driven by topography and the discharge of water to the San Joaquin River and by evapotranspiration in the western part of the flow system (Burow et al. 2004; Phillips et al. 2007; Fig. 1). Because irrigation is a major source of recharge and withdrawals for irrigation are a major source of discharge, there are substantial vertical components of flow (Burow et al. 2008b). Numerical modeling and analysis of atmospheric age tracers in shallow groundwater indicate field-scale vertical velocities of approximately 0.5–2 m/yr in groundwater recharge zones (Phillips et al. 2007; Green et al. 2008a). Pre-agricultural recharge rates were likely much lower (Faunt 2009). Modern vertical flow components enhance movement of water from recharge areas to the perforated intervals of withdrawal wells within shallow to intermediate depths in the system (Fig. 1). These processes occur in both agricultural and urban areas. Groundwater age is vertically stratified, with water less than 50 years old in the upper parts of the system and water that may be tens of thousands of years old at depth (Burow et al. 2008a). In the western part of the study area, the Corcoran Clay may restrict the interaction between the underlying confined and overlying unconfined groundwater. However, well-bores open to the aquifer above and below the Corcoran Clay permit water exchange across the Corcoran (Williamson et al. 1989).

Methods

Groundwater chemistry data included in this study were from three sources, the GAMA PBP, the USGS National Water Information System (NWIS), and the California Department of Public Health (CDPH) database. All NO3 concentrations discussed in this report are in units of milligrams per liter as nitrogen (NO3–N); the US Environmental Protection Agency (USEPA) Maximum Contaminant Level (MCL) for NO3–N in drinking water is 10 mg/L. Data on hydrogeologic factors were assembled from several sources.

Water-quality sample collection and analysis

The GAMA PBP is a statewide assessment of water quality in groundwater used for public-supply in California (Belitz et al. 2003). The GAMA PBP collected samples from 57 wells in the study area during 2005–2006 for a wide range of analytes (Bennett et al. 2006; Landon and Belitz 2008).

PBP and NWIS data were collected in accordance with the protocols established by the USGS NAWQA program (Koterba et al. 1995) and the USGS National Field Manual (US Geological Survey variously dated). These sampling protocols ensure that a representative sample of the groundwater is collected from each well and that samples are collected and handled in a way that minimizes the potential for contamination. All public-supply well samples were collected at discharge points upstream of any chlorination or well-head treatment systems. Detailed descriptions of sample collection procedures, analytical methods, quality-assurance results, and compilations of GAMA PBP data can be found in Bennett et al. (2006) and Landon and Belitz (2008).

Dissolved oxygen (O2) was measured at the well site in groundwater pumped through a flow-through chamber fitted with a multi-probe meter that was calibrated daily. Samples for major ions, nutrients, trace elements, and dissolved organic carbon (DOC) were analyzed at the USGS National Water Quality Laboratory in Denver, Colorado. Carbon-14 (14C) was analyzed at the University of Waterloo and IsoTrace Laboratory (Ontario, Canada) by accelerator mass spectrometry. Samples for the noble gases, helium (He), neon (Ne), argon (Ar), krypton (Kr), and xenon (Xe), were collected in copper tubes (25-cm long, 0.95-cm diameter) using reinforced nylon tubing connected to a hose bib at the wellhead. Copper tubes were flushed with groundwater and clamped at both ends to seal the sample. Tritium (3H) was collected in a 1-L glass bottle. 3H, He, Ne, Ar, Kr, and Xe were measured at Lawrence Livermore National Laboratory (LLNL). 3H was determined by the helium in-growth method (Bayer et al. 1989). He was analyzed using a VG-5400 noble gas mass spectrometer; dissolved Ne, Kr, and Xe were analyzed using a quadrupole mass spectrometer and Ar was measured using a high-sensitivity capacitive manometer (Beyerle et al. 2000; Ekwurzel 2004; Moore et al. 2006). The recharge temperature was estimated from noble gas concentrations (Cey et al. 2008). Tritium-helium piston flow ages were calculated by determining the amount of tritiogenic 3He, after adjusting the measured 3He and 4He for contributions from the atmosphere (equilibrium solubility and excess air) and from subsurface sources (Cook and Solomon 1997; Ekwurzel et al. 1994; Schlosser et al. 1988, 1989). Samples for dissolved gas analysis of N2, Ar, and methane (CH4) were collected in 40-ml baked amber glass sample vials that were purged with three volumes of sample water before bottom-filling to eliminate atmospheric contamination. Dissolved gas concentrations were determined using membrane inlet mass spectrometry (Singleton and Hudson 2005; Singleton et al. 2007). Excess N2 from denitrification was estimated by comparing measured concentrations of Ar and N2 with those expected in water in equilibrium with the atmosphere, using Ne concentrations to constrain estimates of excess air (Singleton et al. 2007).

Water-quality data sources and redox classification

GAMA PBP data included dissolved redox constituents (O2, NO3 plus nitrite as N, nitrite as N, manganese (Mn), iron (Fe), sulfate (SO4), and sulfide (H2S)) at 46 wells, dissolved gas (N2, Ar, CH4), 14C, and DOC at 25 wells, and 3H and noble gas data for age-dating at 49 wells. The PBP data were collected in 2005–2006. Dissolved gas, nitrate isotope, and DOC data from an additional 14 wells immediately to the south of the study area are discussed in this report in the context of estimating denitrification progress (Landon et al. 2010).

Analytical results for dissolved concentrations of O2, NO3–N, Mn, Fe, SO4, H2S, and DOC were retrieved from NWIS along with well locations, depths, perforated intervals, and well type information. These chemistry and ancillary data were compiled from previous and ongoing USGS investigations in the study area (Gilliom 1989; Dubrovsky et al. 1998; Burow et al. 1998a, 2008b; Wright et al. 2004; Jurgens et al. 2008; Green et al. 2008b). Combined, these sources provided data from 226 wells analyzed for concentrations of dissolved constituents used for redox classification between 1979 and 2008 (median year of analysis, 2001).

The CDPH maintains a database of water-quality data collected from public-supply sources for regulatory purposes. Analytical results for total concentrations of NO3, Mn, Fe, and SO4 were retrieved from CDPH along with available information on depths, perforated intervals, and well type. Data for O2 were not available in the CDPH database. Redox data from the CDPH database were available for 402 wells in the study area between 1984 and 2008 (median year of analysis, 2005).

Combining the three data sources, there were redox data available from 4,148 analyses at 674 wells within the study area. A single analytical result for each constituent at each well was selected to characterize the spatial distribution of redox characteristics and avoid biasing the dataset to those wells having many analyses over time. Wells (135) with data for all 5 redox parameters (listed in next paragraph) or 4 redox parameters (433 wells) were included in the redox analysis. At most of these wells, the data from the most recent analysis for all constituents were selected. For 13 wells having O2 data from the USGS and other redox constituent data from the CDPH on different dates, the data from most recent analyses for each parameter were combined into a single composite analysis. For all wells, the most recent values were compared to historical data to evaluate whether the most recent data included anomalous values. Median values of Mn, Fe, and SO4 for the period of record in each well were not significantly different (Wilcoxon test, see the following) from the most recent values; the most recent values of NO3–N were significantly higher (Wilcoxon test statistic Z = 2.11, p = 0.035) than the median values of NO3–N but this result rarely affected the redox classification and may reflect increases in NO3–N over time discussed later in this report. The resulting database contains redox data for wells sampled between 1979 and 2008; the median year of analysis was 2004. The data used in the analysis of redox and relations to explanatory variables are provided in Table 1 in the electronic supplementary material (ESM).

Redox conditions were evaluated on the basis of O2, NO3–N, Mn, Fe, and SO4 concentrations using the classification system of McMahon and Chapelle (2008) (Table 1). An automated workbook program was used to assign the redox classification to each sample (Jurgens et al. 2009). Threshold values of redox indicators were 0.5 mg/L for O2, 0.5 mg/L for NO3–N, 0.05 mg/L for Mn, 0.1 mg/L for Fe, and 0.5 mg/L for SO4 (McMahon and Chapelle 2008). The reporting levels for all constituents in the USGS data sets were well below the redox classification thresholds (Bennett et al. 2006; Landon and Belitz 2008). For the CDPH data the method detection levels (MDL), when reported in the database, were less than or equal to the threshold values used for redox classification, indicating the data had sufficiently low MDLs for use with the redox classification. The highest and most common MDL for Fe in the CDPH dataset was 0.1 mg/L, equal to the Fe classification threshold. For Mn, the most common MDL was 0.02 mg/L; only one sample had an MDL equal to the classification threshold of 0.05 mg/L. H2S was compiled but did not contribute to the classification as it was not detected in any analyses in the study area. The classification scheme includes redox categories (oxic, mixed, anoxic) and specific redox processes (O2-reduction, NO3-reduction, Mn-reduction, Fe-reduction, SO4-reduction, methanogenesis) within the categories. All samples with O2 above the threshold and without other indicators above their thresholds were categorized as oxic. Samples with a mixed redox category contain O2 or NO3–N and another indicator (Mn or Fe) above thresholds. Samples with an anoxic redox category contain O2 below the threshold and one or more other indicators (NO3–N, Mn, Fe) above the thresholds. For samples lacking O2 measurements, samples are classified on the basis of the data available but the category names reflect that O2 could be above or below the threshold (Jurgens et al. 2009). Samples having no O2 value, Fe and Mn below thresholds and NO3–N above threshold are categorized as OxicOrAnoxic. Samples having no O2 value and Fe or Mn above thresholds are classified as Mixed(anoxic)Or(oxic–anoxic) for NO3–N above threshold, and as AnoxicOrMixed(oxic–anoxic) for NO3–N below threshold. In this study, threshold values were not adjusted for total versus dissolved chemical analyses (Wright and Belitz 2010) because no statistically significant differences were observed between dissolved and total constituents for samples collected within 3 years of each other from the same well.
Table 1

Criteria and threshold concentrations for identifying redox processes in groundwater (Jurgens et al. 2009; McMahon and Chapelle 2008) and distribution of wells by redox category and redox process

Redox category

Redox process

Criteria for inferring process

No. of wells in redox category

% of total wells

No. of wells with redox process

% of total wells

O2 (mg/L)

NO3–N (mg/L)

Mn (μg/L)

Fe (μg/L)

SO4 (mg/L)

Water-quality data complete (five redox parameters)

     

Oxic

O2

≥0.5

<0.05

<0.1

99

73.3

99

73.3

Suboxic

Suboxic

<0.5

<0.5

<0.05

<0.1

0

0.0

0

0.0

Mixed(oxic-anoxic)

O2–Mn(IV)

≥0.5

≥0.05

<0.1

12

8.9

6

4.4

Mixed(oxic-anoxic)

O2–Fe(III)/SO4

≥0.5

<0.5

≥0.1

≥0.5

  

5

3.7

Mixed(oxic-anoxic)

O2–CH4gen

≥0.5

<0.5

≥0.1

<0.5

  

1

0.7

Mixed(anoxic)

NO3–Mn(IV)

<0.5

≥0.5

≥0.05

<0.1

10

7.4

9

6.7

Mixed(anoxic)

NO3–Fe(III)/SO4

<0.5

≥0.5

≥0.1

≥0.5

 

 

1

0.7

Anoxic

NO3

<0.5

≥0.5

<0.05

<0.1

14

10.4

10

7.4

Anoxic

Mn(IV)

<0.5

<0.5

≥0.05

<0.1

  

1

0.7

Anoxic

Fe(III)/SO4

<0.5

<0.5

≥0.1

≥0.5

  

1

0.7

Anoxic

CH4gen

<0.5

<0.5

≥0.1

<0.5

 

 

2

1.5

Sum

      

135

100.0

135

100.0

Water-quality data incomplete (four redox parameters)

    

O2 > = 0.5 mg/L

Unknown

≥0.5

<0.05

<0.1

No data

1

0.2

1

0.2

OxicOrAnoxic

O2? or NO3

No data

≥0.5

<0.05

<0.1

≥0.5

323

74.6

323

74.6

OxicOrSuboxic

O2? or suboxic

No data

<0.5

<0.05

<0.1

<0.5

19

4.4

19

4.4

Mixed(anoxic)Or(oxic-anoxic)

Mn(IV)–O2? or NO3

No data

≥0.5

≥0.05

<0.1

52

12.0

11

2.5

Mixed(anoxic)Or(oxic-anoxic)

Fe(III)/SO4–O2? or NO3

No data

≥0.5

≥0.1

≥0.5

  

40

9.2

Mixed(anoxic)Or(oxic-anoxic)

CH4gen–O2? or NO3

No data

≥0.5

≥0.1

<0.5

 

 

1

0.2

AnoxicOrMixed(oxic-anoxic)

Mn(IV)–O2?

No data

<0.5

≥0.05

<0.1

38

8.8

15

3.5

AnoxicOrMixed(oxic-anoxic)

Fe(III)/SO42O2?

No data

<0.5

≥0.1

≥0.5

  

20

4.6

AnoxicOrMixed(oxic-anoxic)

CH4gen–O2?

No data

<0.5

≥0.1

<0.5

 

 

3

0.7

Sum

      

433

100.0

433

100.0

Redox process column: O2 oxygen reduction; NO3 nitrate reduction; Mn(IV) manganese reduction; Fe(III) iron reduction; SO4 sulfate reduction; CH4gen methanogenesis. Chemical species/criteria for inferring process: O2 dissolved oxygen; NO3–N dissolved nitrate as nitrogen; Mn manganese; Fe iron; SO4 sulfate; — criteria do not apply because the species concentration is not affected by the redox process

Hydrogeologic factors and statistical analysis

The purpose of analyzing relations of redox to hydrogeologic factors was to identify mappable variables that serve as surrogates for redox processes at a regional scale. The relations of redox constituents to hydrogeologic factors were evaluated graphically and statistically. The continuous factors investigated included: depth of top of perforations and bottom of well below water table (bwt) and below land surface (bls), depth of water table below land surface, normalized lateral position (defined in the following), sediment texture, and indicators of groundwater travel time below the water table, including 3H, 14C, percent terrigenic helium, and mean piston-flow age based on 3H/3He. Categorical hydrogeologic factors included classification of wells into those that are perforated above, below, or both above and below the Corcoran Clay, and wells that intersect and do not intersect black sands associated with the Mehrten Formation. The variables listed above were those for which mappable data were most readily available and represent direct or proxy indicators of position, flow rates, and conditions along groundwater flow paths.

The various measures of depth were selected as potential explanatory variables because redox conditions have commonly been observed to vary with depth (Appelo and Postma 1999; McMahon et al. 2008) and because depth, particularly below water table, can be a surrogate for groundwater residence time. Well-construction data were determined from driller’s logs, ancillary records of well owners, or NWIS (Landon and Belitz 2008). The elevations of land surface, the water table, and the top and bottom of the Corcoran Clay in 400 × 400-m cells were extracted from groundwater-flow model datasets from Phillips et al. (2007).

Depth to water table was investigated as an explanatory variable because previous studies have noted groundwater tends toward anoxic conditions where depth to water table is shallow (Rose and Long 1988; McMahon and Chapelle 2008). The water-table elevations were simulated values in a steady-state model for the year 2000 calibrated to observed data (Phillips et al. 2007). Depth to water was computed by subtracting the simulated 2000 water-table elevation from land-surface elevations in 400-m cells. The depths to water-table values estimated from the model simulations represent modern average conditions; seasonal fluctuations in water level near the water table are typically < 1 m (Phillips et al. 2007).

Redox conditions were hypothesized to be related to horizontal position in the flow system because groundwater residence time and flowpath length generally increase from the proximal to distal end of the flow system. Normalized lateral position is a proxy for the horizontal position of a well in the regional groundwater-flow system, which primarily flows from the northeastern margin of the valley-fill deposits along the Sierra Nevada mountain front towards the southwestern margin of the flow system, along the San Joaquin River (Fig. 1). Although the local distribution of recharge and stresses can cause local deviations from the regional flow direction, position in the regional system nevertheless serves as a potentially valuable simple proxy for residence time and flowpath length. The normalized lateral position was calculated as the ratio of (1) the distance from the San Joaquin River to the well and (2) the total distance from the San Joaquin River to the east edge of the valley for 30 × 30 cells in the San Joaquin Valley (Faunt 2009). The valley trough, represented by the San Joaquin River, was assigned a value of 0; the east edge of the valley was assigned a value of 1. Lower values of lateral position indicate locations in the downgradient or distal portion and higher values indicate locations in the upgradient or proximal portion of the flow system. Plotting of data, with respect to lateral position, allows for aggregation of areally distributed data into a single, diagrammatic cross section across the study unit.

Relations of sediment texture to redox were hypothesized because groundwater residence time may increase as texture fines and because anoxic conditions commonly occur in fine-textured sediments (McMahon 2004). Faunt et al. (2010) determined the percentage of coarse-grained sediment, defined as the fraction of a vertical interval consisting of sand, gravel, pebbles, boulders, cobbles, or conglomerate, for 15-m depth intervals for 8,500 driller’s logs in the Central Valley. A texture model of the Central Valley was then developed by kriging of the percentage of coarse-grained sediments onto a 1.6-km spatial grid at 15-m depth intervals. The percentage of coarse-sediment values were calculated for wells included in this study by determining the thickness-weighted average value of percent coarse-grained sediments for the perforated interval of the wells (C. C. Faunt, US Geological Survey, personal communication, 14 April 2009).

Indicators of groundwater age were evaluated as explanatory variables for redox based on general observations that groundwater tends to become more reduced as groundwater residence time increases (Appelo and Postma 1999). For the purpose of relating redox characteristics to groundwater age, a groundwater age classification based on threshold values of 3H (1 tritium units, TU), percent terrigenic helium determined from noble gas data (5%), and 14C (90 percent modern carbon, pmc) was used (Landon et al. 2010). 3H activities > 1 TU indicate the presence of groundwater recharged after about 1950 and 14C activities greater than 90 pmc indicate the predominance of groundwater with residence times less than about 1,000 years; together, these two age-dating criteria identify “modern” groundwater (Jurgens et al. 2010). Samples in which > 5% of the total helium is terrigenic likely represent groundwater with a residence time of more than 100 years. Samples with values of 3H and 14C greater than the thresholds and percent terrigenic helium less than the threshold are classified as modern; the inverse of these tracer values relative to thresholds result in a pre-modern classification. If a sample had some tracers suggesting a modern classification and some a pre-modern classification, the sample was classified as being of mixed age.

The Corcoran Clay was hypothesized to have an effect on redox conditions because it serves as a regional confining layer that divides the aquifer system into unconfined and confined zones. Gridded Corcoran Clay surfaces (Phillips et al. 2007) were determined based on analysis of driller’s logs by Burow et al. (2004) and Page (1986).

It was hypothesized that black sands associated with the mafic-rich Mehrten Formation could contain more electron donors than overlying sediments, resulting in development of more reduced groundwater. The Mehrten Formation, whose black sands consist predominantly of andesitic fragments, is generally richer in mafic minerals than overlying alluvial fan deposits (Burow et al. 2004). The first occurrence of black sands in well logs is likely to represent reworked Mehrten sediments rather than the actual top of the Mehrten Formation (Burow et al. 2004). The elevation of the top of these black sands was estimated from interpolation of the depth of black sands from driller’s logs (Burow et al. 2004) to 400 × 400-m cells (C. C. Faunt, US Geological Survey, personal communication, 29 December 2009).

Nonparametric, rank-based methods were used for statistical analysis because these techniques are generally not affected by outliers and do not require that the data follow a normal distribution (Helsel and Hirsch 2002). The significance level (p) used for hypothesis testing for this report was compared to a threshold value (α) of 5% (α = 0.05) to evaluate whether the relation was statistically significant (p < α). Correlations were investigated using Spearman’s method to calculate the rank-order correlation coefficient (ρ, rho) between continuous variables. The Kruskal-Wallis test was used to test differences among more than two groups (Conover 1980). The Wilcoxon rank-sum test was used to evaluate the differences between two groups of data (Helsel and Hirsch 2002). A Pearson’s chi-square (χ2) contingency table test was used to evaluate whether two categorical variables were related. All statistics were evaluated using Tibco Spotfire S + version 8.1 for Windows.

The nonparametric Mann-Kendall test was used to evaluate whether temporal changes in NO3–N concentrations were significant (Helsel and Hirsch 2002) for 90 public-supply wells with CDPH data during the 1980s–2000s. The wells analyzed were those with well construction data. The Sen slope estimator (mg/L/yr) was calculated to estimate trend magnitude (Sen 1968; Hirsch et al. 1991). Trend analysis was performed after calculating annual medians for years with more than two data points to avoid serial correlation and biasing of the data (Helsel and Hirsch 2002).

Land-use statistics for circles with a radius of 500 m around each well (Johnson and Belitz 2009) were used to assess the relation of land use to redox and changes in NO3–N concentrations over time. Johnson and Belitz (2009) demonstrated statistically that land use in a 500-m circle around a well is likely to be predictive of land use in the actual contributing area of a well, which is typically unknown due to uncertainties in stresses and preferential flow paths. Land use was classified using an “enhanced” version of the satellite-derived (30-m pixel resolution), nationwide USGS National Land Cover Dataset (Nakagaki et al. 2007) representing the early 1990s.

Results and discussion

Spatial distribution of redox conditions

Groundwater in the study area is predominantly oxic but has some zones that have anoxic or mixed redox conditions. For the 135 wells with data for the five primary redox parameters (O2, NO3–N, Mn, Fe, SO4), 73% were categorized as oxic, 10% were anoxic, and 16% were mixed redox conditions (Table 1). Among anoxic groundwater, the predominant redox process was NO3–reduction. Among mixed redox groundwater, the most frequent redox process mixtures were NO3–Mn and O2–Mn (Table 1).

For the 432 wells missing O2 but with data for the other four redox parameters, the data suggest possible redox categories. Of these samples, 75% were OxicOrAnoxic, indicating NO3–N above and Mn and Fe below thresholds; 9% were AnoxicOrMixed(oxic–anoxic), indicating NO3–N below and Mn and/or Fe above thresholds; and 12% were Mixed(anoxic)Or(oxic-anoxic), indicating NO3–N above and Mn and/or Fe above thresholds (Table 1). A comparison of redox classifications with and without O2 data was made by removing O2 data for the 135 samples with 5 redox parameters (Table 1) and classifying those samples on the basis of the remaining 4 parameters. Of 107 wells with a redox category of OxicOrAnoxic without O2, 91% were originally categorized as oxic and 9% were originally categorized as anoxic (redox process, NO3-reduction) when O2 was included. This comparison suggests that most OxicOrAnoxic groundwater would be oxic if O2 data existed but some anoxic (NO3-reducing) samples could be included in this category in the absence of O2 data.

With these comparisons of redox classification with and without O2 data in mind, the percentages of samples in oxic and anoxic categories for complete and incomplete redox parameter data were similar (Table 1). The similarity of the percentages between large numbers of samples having complete and incomplete data indicates that the datasets are comparable and that incomplete redox data can be used along with complete data to gain insight into the spatial distribution of redox conditions for this system. Consequently, for mapping purposes, incomplete categorizations of OxicOrAnoxic, Mixed(anoxic)Or(oxic-anoxic), and AnoxicOrMixed(oxic-anoxic)were considered equivalent to oxic, mixed, and anoxic, respectively. While the complete and incomplete data sets are comparable for this system, this would not necessarily be the case in all aquifers, and additional evaluation in a range of settings is needed.

Redox conditions vary areally by normalized lateral position in the valley, with primarily oxic conditions in the eastern four-fifths and primarily anoxic or mixed redox conditions in the western fifth of the study area (Figs. 2 and 3a). These results are generally consistent with previous investigations in the San Joaquin Valley noting that most groundwater is oxic but that anoxic conditions become prevalent toward the center of the valley (Davis et al. 1959; Bertoldi et al 1991; Dubrovsky et al. 1993; Chapelle et al. 1995; and Burow et al. 1998b). O2 is positively correlated, and Mn and Fe are negatively correlated with the normalized lateral position (Table 2). Mn concentrations were particularly strongly related to the normalized lateral position; Mn was generally > 50 ug/L for normalized lateral position < 0.2 and was < 50 ug/L where the normalized lateral position > 0.2 (Fig. 3b).
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-011-0750-1/MediaObjects/10040_2011_750_Fig2_HTML.gif
Fig. 2

Map showing classified redox conditions (using McMahon and Chapelle 2008; Jurgens et al. 2009), study area, normalized lateral position, location of approximate eastern boundary of artesian conditions during pre-development conditions (Mendenhall et al. 1916), eastern extent of Corcoran Clay boundary (Page 1986), and depth to water below land surface determined from difference between land-surface elevation and the simulated 2000 groundwater elevation in 400-m cells in the groundwater model of Phillips et al. (2007)

https://static-content.springer.com/image/art%3A10.1007%2Fs10040-011-0750-1/MediaObjects/10040_2011_750_Fig3_HTML.gif
Fig. 3

Graphs showing normalized lateral position versus well depth with symbol type varying by position of well relative to the Corcoran Clay confining unit and the black sands of the reworked Mehrten Formation (Mehrten) and redox condition or redox indicator concentration: a Redox category (oxic, mixed, anoxic), b Mn, c Fe (Mn and Fe concentration thresholds are in micrograms per liter)

Table 2

Results of non-parametric analysis of correlations of redox indicators and explanatory variables

 

All data

O2

NO3–N

Mn

Fe

ρ

ρ

ρ

ρ

Depth of water table below land surface

0.678(+)

NS

0.313(−)

0.172(−)

Normalized lateral position

0.668(+)

NS

0.377(−)

0.210(−)

Percent coarse-grained sediment

NS

0.106(+)

NS

0.118(−)

Tritium (3H)

0.517(+)

0.575(+)

NS

NS

Percent terrigenic helium

0.586(−)

0.562(−)

NS

NS

Carbon-14

0.506(+)

0.795(+)

NS

NS

Piston-flow model age, 3H/3He, years before present

0.370(−)

0.611(−)

NS

NS

 

Normalized lateral position > 0.2

Depth to top of perforations below water table

0.294(−)

0.402(−)

NS

NS

Well depth below water table

NS

0.343(−)

0.153(−)

NS

 

Normalized lateral position < 0.2

Depth to top of perforations below water table

NS

NS

NS

NS

Well depth below water table

NS

NS

NS

NS

ρ Spearman's rho; values shown for significant positive correlations (+) and significant negative correlations (−) on the basis of significance level (p) less than threshold value (α) of 0.05; NS not significant

Some exceptions to the general pattern of increasingly reduced conditions towards the valley trough occur. First, there are eastward shifts in oxic/anoxic boundaries in the vicinity of major rivers (Fig. 2). Anoxic or AnoxicOrMixed(oxic-anoxic) groundwater occurred further east along the rivers, particularly the Tuolumne River, with lesser inflections along the Stanislaus and Merced rivers, than between the rivers. Second, both within zones that are predominantly anoxic and predominantly oxic, there is a minority of samples with different redox conditions than the prevailing condition. The divergence of some groundwater from prevailing redox characteristics is consistent with the highly heterogeneous character of the alluvial fan sediments comprising the aquifer (Burow et al. 2004). McMahon et al. (2008) identified very low rates of denitrification along a local flowpath in the northeastern part of the Modesto urban area, in spite of prevailing oxic conditions, and hypothesized that denitrification may occur at sand/clay contacts in the aquifer. If so, local textural heterogeneity could produce variability in redox characteristics of water from some wells in comparison with the typical conditions for an aquifer zone.

Relations of redox conditions to depth in the saturated zone vary across the study area (Fig. 3a). The maximum depth below water table of wells classified as oxic is about 170 m at a normalized lateral position of ∼0.5 and decreases to < 40 m towards the valley center (Fig. 3a). After separating the data into portions of the aquifer that were primarily anoxic (normalized lateral position < 0.2) and primarily oxic (normalized lateral position > 0.2) Mn and Fe concentrations were not correlated with depth of the perforations bwt in either part of the system (Table 2). Among wells having a normalized lateral position > 0.2, concentrations of O2 were negatively correlated with depth to the top of perforations bwt (Table 2) but predominantly remained > 0.5 mg/L (see Figure 1 of ESM).

Mixed redox conditions in wells can reflect mixing of waters having different redox characteristics over short, as well as long, vertical intervals. Mixed redox conditions occurred in monitoring or domestic wells with short perforations, possibly indicating mixing of waters from different redox zones that are closely spaced, as well as production wells with long perforation intervals in which mixing of different waters would be expected. For example, of the 22 wells with all 5 redox parameters that were categorized as having mixed redox (Table 1): 15 were from monitoring wells with perforation lengths < 1.52 m within or near the Merced River (Green et al. 2008b; Domagalski et al. 2008) where redox zones are closely spaced; 5 were from monitoring or domestic wells with perforation lengths of 3–6 m above the Corcoran Clay and lateral normalized position < 0.2, a primarily anoxic zone with recent recharge having higher NO3 from land-surface activities; and 2 were production wells with perforation lengths of 24 to 81 m in which mixing of waters having different redox characteristics over large vertical intervals was expected. Among the larger number of sites with 4 redox parameters, wells with mixed redox had significantly longer (Z = 2.41, p = 0.016) perforation lengths than wells with anoxic conditions, suggesting that longer perforation lengths may contribute to detection of mixed rather than anoxic redox conditions in some cases.

In a portion of the aquifer system, redox conditions appear to be affected by the presence of the Corcoran Clay, which acts as a regional confining unit in the distal third of the flow system. Oxic conditions extend further west (distal) above the Corcoran Clay than below; this pattern is most evident for normalized lateral positions between about 0.1 and 0.2 (Fig. 3a). In the eastern portion of the Corcoran Clay (normalized lateral positions of ∼0.2–0.3), groundwater is primarily oxic both above and below the Corcoran clay and redox becomes increasingly mixed or anoxic to the west. Wells perforated both above and below the Corcoran Clay have oxic or mixed redox (never anoxic) and generally have the redox characteristics of the portion of the aquifer above the Corcoran Clay (Fig. 3). These results are consistent with expected downward gradients and fluxes from the unconfined to confined portions of the system under modern developed conditions (Phillips et al. 2007; Faunt 2009).

At normalized lateral positions < 0.1, Mn-reduced conditions occur deeper in the aquifer than Fe-reduced conditions occurring at shallower depths. Some wells with normalized lateral positions < 0.1 and depths greater than 80 m bwt have Mn > 50 ug/L but Fe < 100 ug/L, consistent with Mn-reducing conditions (Fig. 3b–c). In contrast, many shallower wells (well depth bwt < 50 m) in this normalized lateral position range have Fe > 100 ug/L, suggesting Fe-reducing conditions above the Corcoran Clay (Fig. 3c). The occurrence of more highly reduced groundwater at shallow depths than deeper in the aquifer may result from interaction of groundwater with shallow modern Holocene basin flood deposits that are more fine-grained and organic rich than the alluvial fan deposits that compose most of the aquifer on the eastside of the San Joaquin Valley. In the modern developed groundwater flow system, vertical gradients are predominantly downward from the unconfined to the confined portions of the aquifer beneath the Corcoran Clay (Phillips et al. 2007). In the predevelopment system, upward flow occurred at the downgradient end of the regional flow system (Mendenhall et al. 1916; Faunt 2009). It is possible that the occurrence of Mn-reduced conditions underlying Fe-reducing conditions at shallow depths could partially reflect a relict redox zonation that developed during a predevelopment flow system with upward regional flow. It is also possible that there are modern occurrences of upward flow near the valley trough; water-level data are relatively scarce near the river and the groundwater flow simulations of modern conditions in this area are relatively poorly constrained (Phillips et al. 2007).

Relations of redox conditions to hydrogeologic factors

The distribution of redox conditions with respect to normalized lateral position, depth, and position relative to regional confining layers discussed in the previous section is the result of several hydrogeochemical processes. These processes can be deduced from relations of redox to hydrogeologic factors, including depth to water, groundwater discharge zones, and groundwater age.

Anoxic conditions, which primarily occur in areas with shallow depths to water table in the study area, are mostly associated with two hydrogeochemical processes: (1) recharge of young water containing high concentrations of organic carbon, which serves as an electron donor for redox reactions, resulting from interactions of the shallow water table with soils, and (2) discharge of old groundwater that has become reduced due to long-term exposure to electron donors in the aquifer sediments. Specific evidence for these processes is described in the following.

Relation of water-table depth to redox

Redox conditions are strongly correlated with depth to water table bls (unsaturated zone thickness). O2 is positively correlated with depth to water table, indicating decreased O2 when depth to water table is shallow (Fig. 4a; Table 2). Mn, Fe, and DOC (ρ = −0.444, p < 0.001) are all negatively correlated with depth to water table, indicating higher values where the depth to water table is shallow (Figs. 4b–d; Table 2). These results are consistent with those of McMahon and Chapelle (2008), i.e. that O2 was positively correlated and DOC negatively correlated with depth to water table in data from principal aquifers across the United States. Other regional studies have noted relations between shallow depth to water table and anoxic conditions (Rose and Long 1988; Merz et al. 2009).
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-011-0750-1/MediaObjects/10040_2011_750_Fig4_HTML.gif
Fig. 4

Graphs showing concentrations of a O2, b Fe, c Mn, and d DOC with depth to water table below land surface (unsaturated zone thickness) by groundwater-age class

Areas where the depth to water table is within about 2 m of land surface generally correspond with areas where anoxic or mixed redox conditions predominate (Fig. 2). Depths to water table bls of < 2 m commonly occur in the western parts of the study area (normalized lateral positions < 0.2) and along the Stanislaus and Tuolumne River valleys (Fig. 2).

McMahon and Chapelle (2008) hypothesized that climate may influence redox through its control of vegetation, soil development, and recharge and that soil DOC leached to shallow groundwater and microbial mineralization could decrease O2 of shallow recharge where depth to water table is shallow. DOC leached from natural organic matter in soils is more likely to reach the water table where the depth to water is small (Starr and Gillham 1993; Pabich et al. 2001). Under shallow water-table conditions, the shallowest groundwater flow paths can interact with newly generated DOC from soil processes before the microbially labile component has been fully consumed, permitting more DOC to reach groundwater and serve as electron donors for redox reactions (Starr and Gillham 1993; Eberts et al. 2005). Within the study area, Green et al. (2008b) determined using electron balance calculations that 1–3 mg/L DOC in recharge was insufficient to explain the extent of denitrification and other redox reactions along a monitoring well transect terminating at the Merced River. Data from the current study show similarly low DOC for most wells. Elevated concentrations in some shallow samples, suggest local influence of recharging DOC (see section Hydrogeochemical processes related to anoxic conditions). Age tracer analyses indicate, however, that these local DOC “hot spots” are likely not the main factor controlling the relative abundance of anoxic conditions in shallow groundwater in this study area. The majority (86%) of shallow groundwater samples with O2 <1 mg/L were pre-modern or mixed (Fig. 4a), which suggests that the systematic progression of redox reactions with increasing residence time may account for anoxia in shallow (yet old) groundwaters.

Relation of groundwater discharge to redox

Anoxic conditions generally occur within the area of predevelopment artesian conditions mapped by Mendenhall et al. (1916) (Fig. 2), representing a zone of historic groundwater discharge. The boundaries of the anoxic zone and predevelopment artesian area are close to a normalized lateral position of ∼0.2 and are in areas having present day depth to water table of ∼2–5 m (Fig. 2).

The occurrence of regional groundwater discharge in the western portion of the study area was expected to contribute to reducing conditions in the groundwater along and between streams. Along streams, particularly the Stanislaus, Tuolumne, Merced, and San Joaquin rivers, some upwelling groundwater near rivers can have long residence times of hundreds to thousands of years in the aquifer (Green et al. 2010) and would have had a long time to interact with electron donors in the sediments, resulting in increasingly reducing conditions along regional flow paths away from recharge areas. Large volumes of groundwater recharged hundreds to thousands of years before present are present within the aquifer system (Burow et al. 2008b) and these pre-modern groundwaters are discharged to relatively deep pumping wells (Landon et al. 2010) and/or to streams (Green et al. 2010). Under development conditions, areas between streams are primarily recharge areas, with vertically downward flow to pumping wells (Phillips et al. 2007).

Between streams, predevelopment artesian conditions mapped by Mendenhall et al. (1916) in the western portion of the study area imply that groundwater levels were at or above land surface, head gradients were upward, and groundwater discharge occurred in this zone. Net declines in water levels of a few meters occurred from the early 1900s to about 2000 (Burow et al. 2004; Phillips et al. 2007). However, even under development conditions, depths to water table are shallow enough in the western part of the study area that groundwater discharge by evapotranspiration represents about 6% of simulated groundwater outflow (Phillips et al. 2007). Groundwater discharge would have been higher under the predevelopment period because of higher water levels.

Hydrogeochemical processes related to anoxic conditions

Age and chemistry data are consistent with anoxic conditions in the western distal portion of the study area, developing as a result of both discharge of groundwater with long travel times to streams or in the valley trough and recharge of high DOC in shallow water-table settings. Classified ages of groundwater samples with anoxic conditions and shallow depth to water table ranged from modern (last 50 years) to pre-modern (recharged prior to last 50 years). Among samples from wells with depth to water table < 5 m and classified ages, low O2 and high iron were predominantly associated with pre-modern groundwater; specifically, five of the six wells having O2 < 0.5 mg/L had pre-modern or mixed ages (Fig.4a), and a well having iron > 100 ug/L had a pre-modern age (Fig. 4b). Positive correlations of O2 with 3H and 14C and negative correlations with percent terrigenic helium and apparent 3H/3He piston flow age (Table 2) are also consistent with low O2 being associated with older groundwater. Mn > 50 ug/L was not clearly related to classified age, with one sample each having modern, mixed, and pre-modern ages (Fig. 4c). In contrast, high DOC in shallow depth to water-table settings was associated with modern groundwater. Specifically, two of three samples with DOC > 2 mg/L had modern ages and one sample had a mixed age (Fig. 4d). These samples also had O2 < 0.5 mg/L, as would be expected in a sample with high DOC. Wells having DOC > 3 mg/L and depth to water < 3 m (Fig. 4d) were located away from streams (> 4 km); these wells all had NO3- to Mn-reducing redox processes that are likely to reflect the effects of modern recharge with high DOC due to proximity of the water table to soils.

The relative importance of recharge and discharge processes to anoxic conditions may be influenced by modern as well as legacy hydrologic conditions. Areas of groundwater recharge with depth to water less than about 3 m include the western ∼20% of the study area (Fig. 2). Although this area was a regional discharge zone prior to development, local recharge would have occurred at some locations and times during the predevelopment period and is likely to be more prevalent under modern conditions of greater recharge and vertically downward gradients. Recharge of high DOC in shallow water-table areas could be an important mechanism for development of reducing conditions in shallow groundwater. Under pre-development conditions, shallow anoxic groundwater resulting from recharge with high DOC most likely would have remained at shallow depths since the area was a regional-groundwater-discharge zone. Under modern development conditions, vertical-head gradients are downward in most of the study area (Phillips et al. 2007) and anoxic shallow groundwater may have moved downward into the aquifer system during recent decades. Under modern development conditions, about 2% of groundwater discharge from the study area is to streams (Phillips et al. 2007). Substantially larger discharge to wells (70%) than streams under development conditions (Phillips et al. 2007) suggest that discharge to streams would have been higher in the pre-development period than the current period. Substantial quantities of old groundwater that followed long relic flowpaths towards rivers still reside in the system and may explain why groundwater near modern streams tends to be more anoxic than groundwaters at equivalent normalized lateral positions betweens streams, even though groundwater discharge to streams is a small component of the modern regional water budget.

Hydrogeologic factors not related to redox

Potential explanatory variables investigated that were not related to redox are sediment texture and the presence of black sands (representing mafic-rich sediments of the Mehrten Formation). The average percent coarse sediment (sand and gravel) was not correlated with O2 or Mn but was correlated with NO3–N (positively) and Fe (negatively; Table 2). However, the absolute values of the Spearman’s ρ values were the smallest of any significant relations, indicating that these relations were weaker than the other correlations identified, and probably reflect relations of percent coarse sediment with depth. Within the study area, the percent coarse sediment decreased with increasing well depth bwt (ρ = 0.208, p < 0.001), consistent with regional patterns in textural characteristics of aquifer sediments previously noted by Burow et al. (2004) and Faunt et al. (2010). The percent coarse sediment was not correlated with normalized lateral position.

Samples collected from wells intersecting black sands, representing reworked Mehrten Formation sediments, were no more likely to be reducing than samples from other wells for equivalent normalized lateral positions and depths. Within the predominantly oxic portion of the system (normalized lateral position > 0.2, Corcoran Clay absent), the percentage of wells categorized as having anoxic or mixed redox did not differ significantly (χ2 contingency table test, p > 0.05) between wells that intersected black sands (12%, 7 of 56 wells) and those that did not (7.5%, 8 of 107 wells).

Nitrate concentrations and relations to redox

NO3–N concentrations are significantly lower in anoxic groundwater than in oxic or mixed redox groundwater (Kruskal-Wallis χ2 = 64.07, p < 0.001; Fig. 5). This result was expected since the redox classification includes NO3–N concentrations as a parameter. However, NO3–N concentrations are not significantly different between mixed redox and oxic groundwater (Wilcoxon Z = −0.32, p = 0.748). High NO3–N concentrations occur in some anoxic groundwater (Fig. 5). Samples can be classified as anoxic based on O2, Mn, and Fe values and yet have NO3–N > 0.5 mg/L (Table 1). The effects of denitrification on NO3–N concentrations are most likely to be detected in samples with high NO3–N under anoxic conditions. Many samples with oxic or mixed redox conditions have NO3–N concentrations near or above 10 mg/L (Fig. 5).
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-011-0750-1/MediaObjects/10040_2011_750_Fig5_HTML.gif
Fig. 5

Boxplot showing nitrate concentrations by redox and depth categories

Significantly lower NO3–N concentrations in anoxic than oxic or mixed redox groundwater primarily reflect that lower NO3–N and O2 are correlated with increasing depth and older groundwater age and can potentially serve as indicators of these hydrogeological attributes. Of samples analyzed for dissolved N2 gas, current NO3–N and intial NO3–N (NO3–N + N2 from denitrification) concentrations were > 2.8 mg/L in modern groundwater (13 samples). Conversely, most pre-modern groundwater samples (9 of 11) had NO3–N < 2.8 mg/L. In oxic groundwater, NO3–N concentrations are significantly higher in shallow groundwater (depths < 60 m bwt) than in deeper groundwater (depths > 60 m bwt; Wilcoxon Z = 4.53, p < 0.001, Fig. 5). In the data set as a whole, NO3–N concentrations were negatively correlated with depth and with groundwater age (higher NO3–N with higher 3H and 14C, lower percent terrigenic helium and apparent 3H/3He piston flow age; Table 2). Previous investigations have indicated that concentrations of NO3–N are higher in the shallow part of the aquifer system than in the deeper part of the aquifer system where public-supply wells are screened and that groundwater age generally increases with depth (Burow et al. 2007, 2008b; Green et al. 2008b; Jurgens et al. 2008; Landon et al. 2010). However, among anoxic and mixed redox groundwater, NO3–N concentrations did not significantly decrease with depth.

Although NO3–N concentrations are lower under anoxic than oxic conditions, the infrequent occurrence of anoxic conditions (9.5% of wells, Table 1) indicates that NO3–N concentrations in most of the study area are not primarily controlled by redox conditions. Because the anoxic zones are generally located at the downgradient end of the flow system in the valley trough or along lateral rivers (Stanislaus, Tuolumne, Merced), denitrification has the potential to partially mitigate NO3 transport from the regional groundwater system to rivers. Of the 70 wells in the data set with NO3–N < 0.5 mg/L, 59% were located within 1 km of the Stanislaus, Tuolumne, Merced, or San Joaquin rivers. However, of the 149 wells located within 1 km of these rivers, 27% had NO3–N < 0.5 mg/L, and another 26% had NO3–N of 0.5–2.5 mg/L. Similarly, of 55 wells having normalized lateral position < 0.2 but located > 1 km from streams, 22% had NO3–N < 0.5 mg/L, and 5% had NO3–N of 0.5–2.5 mg/L. These results suggest that anoxic limitations on NO3–N concentrations are widespread near streams and near the valley trough within the study area but are not universal.

Regional importance of denitrification

Estimates of the extent of denitrification based on available dissolved gas data in or adjacent to the study area indicate that decreases in NO3–N concentrations as a result of denitrification are generally small and do not currently protect many supply wells from NO3 contamination. Of the 39 wells with data for excess N2 from denitrification in the saturated zone (Landon et al. 2010; Landon and Belitz 2008), there is only one well with measured NO3–N <10 mg/L which would have had NO3–N > 10 mg/L if the N2 from denitrification were added to the measured value (Fig. 6). Thus, there is only one well out of 39 which would change in regulatory status (NO3–N increasing above 10 mg/L) if denitrification were absent. It is possible that denitrification could have a greater impact on NO3–N concentrations in supply wells in the future as increasing NO3–N concentrations in shallow groundwater continue to move deeper (Burow et al. 2008b) into more reactive zones of the aquifer.
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-011-0750-1/MediaObjects/10040_2011_750_Fig6_HTML.gif
Fig. 6

Graph showing measured and initial nitrate nitrogen concentrations for redox and age categories

Minor amounts of denitrification may occur in groundwater deeper in the system. Most of the groundwater with more than 1 mg/L excess N2–N had pre-modern (recharge prior to 1950) or mixed groundwater age (mixture of pre-1950 and post-1950 recharge; Fig. 6). Most samples with initial (sum of residual measured NO3–N and excess N2–N) NO3–N > 10 mg/L and plotting near the 1:1 line are oxic groundwaters with modern ages from relatively shallow monitoring wells. Samples with initial NO3–N < 10 mg/L are mostly production wells or deeper monitoring wells and often plot along or above a curve indicating 1 mg/L excess N2. All four samples having > 5 mg/L excess N2 had anoxic or mixed redox conditions. One of these samples had a measured NO3–N of ∼50 mg/L, excess N2 of ∼6 mg/L, a modern age, and anoxic conditions. Three other samples had nearly complete denitrification of initial NO3–N of 8–10 mg/L, pre-modern ages, and anoxic or mixed redox conditions. Apparent denitrification progress—excess N2/(excess N2 + measured NO3–N)—ranged from 8 to 100% (mean 62%) among 10 samples with anoxic or mixed redox conditions. Samples with excess N2 data also exhibited enriched values of both δ15 N and δ18O of NO3 (Landon et al. 2010), supporting the interpretation that these samples are affected by denitrification.

Because denitrification may occur when O2 > 0.5 mg/L, use of that threshold for defining anoxic conditions may cause the spatial extent of denitrification in the aquifer to be underestimated. In laboratory studies, the O2 concentration threshold required for the onset of denitrification has generally been observed to be < 0.3 mg/L (Tiedje 1988; Seitzinger et al. 2006; Coyne 2008). However, in complex geological settings, some denitrification can occur in reduced microenvironments such as along lithologic contacts, within predominantly oxic environments. Significant excess N2 from denitrification is commonly measured in samples with O2 of up to 2 mg/L (Böhlke et al. 2002, 2007; Beller et al. 2004; McMahon et al. 2004, 2008; Green et al. 2008b). Simulations by Green et al. (2010) confirm that mixtures of travel times in samples can account for the appearance of denitrification in the presence of relatively high concentrations of O2, even in short-screened (0.5 m) monitoring wells. However, the denitrification rate along predominantly oxic flowpaths may be small; for example, in the Modesto urban area, approximately 95% of the NO3 originally present persisted along the sampled flowpath (McMahon et al. 2008). Excess N2 in oxic groundwater (Landon et al. 2010) ranged from <0.4 to 2.9 mg/L with a median of 1.4 mg/L, indicating that denitrification was limited within predominantly oxic zones.

Hydrogeologic factors controlling nitrate concentrations

The relations of NO3 concentrations to redox discussed in the previous are consistent with those of previous studies in the San Joaquin Valley in that groundwater NO3 concentrations are primarily controlled by the land use in recent decades, particularly irrigated agriculture (Burow et al. 2007, 2008a, b). Concentrations of NO3–N in groundwater at concentrations of concern, approaching or greater than the USEPA MCL of 10 mg/L, have been described in the eastern San Joaquin Valley and have primarily been attributed to the effects of agricultural land use (Nightingale and Bianchi 1974; Schmidt 1987; Dubrovsky et al. 1998; California State Water Resources Control Board 2002; Burow et al. 2008b). In a 4,390 km2 study area in the eastern San Joaquin Valley encompassing the area of the present study, a recent study found that NO3–N concentrations were larger than the MCL in 2.1% and between half of the MCL and the MCL in 14.6% of the portion of the aquifer typically used for public supply (Landon et al. 2010). Although the spatial distribution of NO3 concentrations is more strongly related to land-use effects rather than to redox, it was hypothesized that redox conditions could affect changes in NO3 concentrations over time, as is discussed in the following section.

Relations of changes in nitrate concentrations over time to hydrogeologic factors

Analyses of changes in NO3–N concentrations in 90 public supply wells with CDPH data spanning the 1980s–2000s indicated NO3–N concentrations increased in about 33% of the wells, decreased in about 11%, and had no detected change in about 56%, including about 2% of wells where NO3–N was generally not detected—see Table 2 in the electronic supplementary material (ESM; Fig. 7). The median record spanned 21 years with a median of 21 data points (see Table 2 in ESM). Slopes of change in NO3–N concentrations over time ranged from a maximum of 0.64 mg/L/yr increase to a minimum of −0.31 mg/L/yr decrease, with a median of no change (see Figure 2 in ESM)
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-011-0750-1/MediaObjects/10040_2011_750_Fig7_HTML.gif
Fig. 7

Bar chart showing percentage of wells with changes in NO3–N concentrations over time that are increasing, show no detected change, decreasing, or in which NO3–N was not detected for all wells and wells in different redox categories

Changes in NO3 concentrations over time are influenced but are not primarily controlled by redox. It was hypothesized that changes in NO3 concentrations over time would be smaller under anoxic or mixed redox conditions than under oxic conditions. For example, denitrification under anoxic or mixed redox conditions could limit increases in NO3 concentrations over time compared to portions of the aquifer without denitrification. This hypothesis is not supported by the observation that slopes of NO3–N change over time were not significantly different between groundwater with anoxic, mixed, and oxic redox conditions (Kruskal-Wallis χ2 = 3.39, p = 0.335). The similarity of NO3–N change slopes for groundwater having different redox conditions suggests that factors other than redox primarily control rates of change in NO3–N concentrations over time. However, the prevalence of increasing NO3–N concentrations over time in anoxic groundwater may be less than in oxic or mixed redox groundwater. The percentage of wells having decreasing concentrations over time or no detection of NO3–N was higher for anoxic than for mixed or oxic redox conditions (Fig. 7); conversely the percentage of wells having increasing concentrations over time was less for anoxic than for oxic or mixed redox conditions. Overall, the percentages of wells with increasing, decreasing, and no change in NO3–N concentrations over time in anoxic groundwater were significantly different (χ2 = 15.87, p < 0.001) than oxic or mixed redox groundwater.

Changes in NO3–N concentrations over time are primarily related to land use, stratigraphy, and depth in the aquifer system. Most of the wells (7 of 10) having decreasing NO3–N concentrations are located on the southwestern (downgradient side) of Modesto in or adjacent to urban land-use areas (Fig. 8). Historical land-use maps indicate these areas have been in urban land use for decades, whereas agricultural land use predominates in the surrounding areas. Reconstructions of nitrogen fertilizer applications based on county-level data (Alexander and Smith 1990; Ruddy et al. 2006) and nitrate concentrations in recharge for the Modesto area by Burow et al. (2008b) indicate 4- to 5-fold increases during 1950–2000 (“modern” period). Shallow groundwater beneath urban areas in Modesto has lower NO3–N concentrations (< 5 mg/L) than shallow groundwater beneath agricultural areas (> 10 mg/L; Jurgens et al. 2008; Burow et al. 2008b). Other CDPH wells in urban areas analyzed for changes in NO3–N concentrations over time located to the northeast in Modesto have increasing or unchanging concentrations of NO3–N (Fig. 8). These wells may be more strongly influenced by southwestward flowing groundwater with higher NO3–N impacted by agricultural land-use areas to the northeast of Modesto than wells in southwestern Modesto. The presence of thousands of “rock wells”, boreholes backfilled with rock, which route storm runoff into groundwater (Phillips et al. 2007), may also contribute to lower groundwater NO3–N concentrations in the Modesto urban area by increasing the amount of recharge from precipitation, and diluting concentrations from upgradient land use.
https://static-content.springer.com/image/art%3A10.1007%2Fs10040-011-0750-1/MediaObjects/10040_2011_750_Fig8_HTML.gif
Fig. 8

Map showing changes in NO3 concentrations over time

The presence of the Corcoran Clay may be a contributing factor to decreasing NO3–N concentrations over time in southwestern Modesto. The Corcoran Clay may serve to isolate relatively low NO3 urban recharge above the clay from higher NO3 agricultural recharge that may be deeper in the flow system and below the clay. Wells screened entirely above the Corcoran Clay also are generally shallower (bottom of well < 60 m bwt) than wells where the Corcoran Clay is absent (bottom of well typically > 60 m bwt, mean of 90 m). Six of the seven wells in southwest Modesto with decreasing NO3–N concentrations are screened entirely above the Corcoran Clay; the remaining well is screened both above and below. Not all wells in southwest Modesto screened above the Corcoran Clay had decreasing NO3–N concentrations, reflecting that temporal changes in concentrations are influenced by many factors. However, a larger percentage of wells perforated above the Corcoran Clay in southwest Modesto had decreases in concentrations over time (33%, 6 of 18 wells) and a smaller percentage of wells had increases over time (2 of 18, 11%) than was observed for the entire study area (see Table 2 in ESM; southwest Modesto wells above the Corcoran Clay had significantly different (χ2 = 13.33, p < 0.001) percentages of wells with increasing, decreasing, and no change in NO3–N concentrations over time compared to other parts of the study area.

Although slopes of changes in groundwater NO3–N concentrations over time were not significantly different between areas with land use that was predominantly (>50%) agricultural, predominantly urban, or mixed (no land-use category > 50%), a larger percentage of agricultural wells (48%) had increasing NO3–N concentrations than urban wells (30%); similarly a smaller percentage of agricultural wells had decreasing trends (4%) than urban wells (13%). Slopes of change over time were not significantly different for wells with depth bwt <60 m, 60–120 m, and >120 m. However, slopes of changes over time in NO3–N concentrations were more variable at shallower depths, with slopes ranging from −0.14 to 0.64 mg/L/yr for wells < 60 m bwt, –0.31 to 0.19 for wells 60–120 m bwt, and −0.05 to 0.07 for wells >120 m bwt. These results suggest that changes in NO3–N concentrations over time are more variable, in response to land-use effects at shallow depths and tend to be dampened with increasing depth. Because only 17 wells for which analysis of changes in NO3–N over time was done also had age tracer data, and 13 of these wells were categorized as having mixed age, there was insufficient data to directly evaluate the relation of NO3–N change slopes to age category. However, because groundwater age class and depth are strongly related in the study area (Landon et al. 2010), it is likely that dampened changes in NO3–N concentrations over time with increasing depth at least partially reflect that groundwater at greater depths is older and is less likely to be affected by modern changes in NO3–N loading over time at land surface.

Increasing NO3–N concentrations over time were correlated with high NO3–N concentrations. Slopes of change in NO3–N concentrations over time were positively correlated (ρ = 0.21, p = 0.044) with median NO3–N concentration for the period of record. In turn, high NO3–N concentrations in the eastern San Joaquin Valley are associated with substantial increases in nitrogen applications in agricultural areas in recent decades (Burow et al. 2008b). As estimated fertilizer applications through the early 2000s were at or near maximum historical values, relatively widespread high and increasing NO3–N concentrations may occur in the near future.

The analyses of changes in NO3–N concentrations in the same public-supply wells over a 2–3 decade time period in a portion of this regional study area indicate that the hydrogeologic factors explaining changes are complex, but that increases in concentrations are more prevalent than decreases. These results are generally consistent with a study by Burow et al. (2008b) of the entire eastern San Joaquin Valley using NO3 data compiled from multiple sources that indicated that concentrations of NO3–N have increased since the 1950s in the shallow and deep parts of the aquifer system. This study identifies additional complexities in the hydrogeologic factors affecting changes in NO3–N concentrations in the portion of the aquifer used for public supply, including decreasing concentrations related to urban land use and effects of aquifer stratigraphy, depth, and redox. In particular, the results of this study suggest that NO3–N concentrations in shallow groundwater follow loading patterns in areas where the water table is deep (> 2–5 m bls) but NO3–N concentrations may be partially attenuated by denitrification under anoxic conditions where the depth to water table is shallow (< 2–5 m bls).

Conclusions

Although the processes controlling NO3 attenuation have been well characterized along flowpaths at local scales, the hydrogeologic factors controlling the distribution of redox conditions and NO3-degradation in heterogeneous alluvial aquifer systems with large pumping withdrawals at regional scales are less well understood. In a 2,700-km2 study area in the eastern San Joaquin Valley near Modesto, California, redox conditions vary by normalized lateral position in the valley. Reducing conditions were observed along the valley trough and, to some extent, in the vicinity of upgradient major rivers. Within zones that are either predominantly anoxic or predominantly oxic, there are a minority of wells with different redox conditions than the prevailing condition. Redox conditions are generally not related to depth as strongly as lateral position. Including wells with incomplete redox data provided insight into the spatial distribution of redox conditions as well as larger data sets for exploration of explanatory variables. A comparison of redox classifications with and without O2 data indicated that wells with incomplete data were reliably categorized in most cases.

The spatial distribution of redox conditions is related to depth to the water table, historical regional groundwater discharge patterns, and groundwater age. These conclusions are supported by the correlation of depth to water table bls and redox indicators, the spatial correspondence of anoxic conditions with small depths to water table in the western portions of the study area and alluvial valleys along rivers, and the correspondence of anoxic groundwater in the western part of the study area with the location of predevelopment artesian conditions. Anoxic conditions in areas with shallow depths to water table (small unsaturated zone thickness), occur in either: (1) modern recharge water containing an abundance of organic carbon as an electron donor, resulting from interactions of the shallow water table with soils, or (2) discharge of old groundwater that has become reduced due to long-term exposure to electron donors in the aquifer sediments.

While local studies have shown that denitrification of NO3 can be important along some local flowpaths within the study area, this investigation suggests that the effects of denitrification on NO3–N concentrations are small under current conditions at regional scales. Although NO3–N concentrations are significantly less in wells with anoxic conditions than those with oxic or mixed redox conditions, a relatively small proportion of wells (9.5%) in the study area had anoxic conditions. Dissolved N2 and Ar gas data from 39 wells indicate that minor amounts of denitrification occur in anoxic portions of the aquifer, with 4 wells having excess N2 > 5 mg/L and apparent denitrification progress among 10 samples with anoxic or mixed redox conditions of 8–100% (mean 62%). However, denitrification caused NO3–N concentration to decrease from above the drinking water regulatory threshold to below the threshold in only 1 well; this result indicates that denitrification does not currently protect many wells from NO3 contamination. These results are consistent with those of previous studies in the eastern San Joaquin Valley in indicating that groundwater NO3–N concentrations in the regional aquifer are primarily controlled by conservative transport processes and loading from irrigated agriculture over the past few decades. Because the anoxic zones are generally located at the downgradient end of the flow system in the valley trough or along lateral rivers, denitrification has the potential to partially mitigate NO3 transport from the regional groundwater system to rivers.

Although redox conditions do not currently have a substantial effect on NO3–N concentrations in recharge areas, it was hypothesized that increases in NO3–N concentrations over time might be lower in wells having anoxic or mixed redox conditions compared to wells having oxic conditions. Analysis of changes in NO3–N concentrations during the 1980s–2000s in 90 public-supply wells indicates that changes over time are influenced but are not primarily controlled by redox. There was a slightly decreased percentage of wells having increasing NO3–N concentrations over time in anoxic compared to oxic groundwater. Changes in NO3–N concentrations over time were primarily related to land use, with decreasing NO3–N concentrations primarily occurring beneath urban land-use areas, and increasing NO3–N concentrations occurring more frequently in agricultural land-use areas than areas of urban or mixed land use. Increasing NO3–N concentrations over time were also correlated with higher NO3–N concentrations; this observation, coupled with corresponding increases in nitrogen applications on the landscape in recent decades, implies that more widespread occurrence of increasing NO3 concentrations may occur in the future.

Although NO3 concentrations and changes through time in the study area are primarily controlled by nitrate loading from agricultural land use rather than redox conditions, mapping of the regional three-dimensional redox patterns and relations to hydrogeologic factors provide a framework for regional-scale simulation of NO3 fate and transport and for future water-quality monitoring.

Acknowledgements

This study was funded by the US Geological Survey National Water Quality Assessment (NAWQA) Program study of groundwater trends, and by the California Groundwater Ambient Monitoring and Assessment Program. We thank the large number of people involved in collecting the data for these programs as well as the California Department of Public Health for providing access to data utilized in this study. We thank the NAWQA trends team for ideas and suggestions and Barbara Dawson and Claudia Faunt for data and analysis that assisted with this study. This manuscript benefited from reviews by Frank Chapelle, Steve Phillips, and two anonymous reviewers. Any use of trade, product, or firm names is for descriptive purposes only and does not imply endorsement by the US Government.

Supplementary material

10040_2011_750_MOESM1_ESM.pdf (24 kb)
ESM Fig. 1(PDF 24 kb)
10040_2011_750_MOESM2_ESM.pdf (22 kb)
ESM Fig. 2(PDF 22 kb)
10040_2011_750_MOESM3_ESM.xlsx (204 kb)
ESM Table 1(XLSX 204 kb)
10040_2011_750_MOESM4_ESM.pdf (17 kb)
ESM Table 2(PDF 16 kb)

Copyright information

© Springer-Verlag (outside the USA) 2011