H54D-01
Modeling Land Application of Food-Processing Wastewater in the Central Valley, California
California's Central Valley contains over 640 food-processing plants, serving a multi-billion dollar agricultural industry. These processors consume approximately 7.9 x 107 m3 of water per year. Approximately 80% of these processors discharge the resulting wastewater, which is typically high in organic matter, nitrogen, and salts, to land, and many of these use land application as a treatment method. Initial investigations revealed elevated salinity levels to be the most common form of groundwater degradation near land application sites, followed by concentrations of nitrogen compounds, namely ammonia and nitrate. Enforcement actions have been taken against multiple food processors, and the regulatory boards have begun to re-examine the land disposal permitting process. This paper summarizes a study that was commissioned in support of these actions. The study has multiple components which will be reviewed briefly, including: (1) characterization of the food-processing related waste stream; (2) fate and transport of the effluent waste stream in the unsaturated zone at the land application sites; (3) fate and transport of the effluent waste stream at the regional scale; (4) predictive uncertainty due to spatial variability and data scarcity at the land application sites and at the regional scale; (5) problem mitigation through off-site and in-situ actions; (6) long-term solutions. The emphasis of the talk will be placed on presenting and demonstrating a stochastic framework for modeling the transport and attenuation of these wastes in the vadose zone and in the saturated zone, and the related site characterization needs, as affected by site conditions, water table depth, waste water application rate, and waste constituent concentrations. http://www.hilmarsep.com
H54D-02
Probabilistic Modeling for Risk Assessment of California Ground Water Contamination by Pesticides
The California Department of Pesticide Regulation (DPR) is responsible for the registration of pesticides in California. DPR's Environmental Monitoring Branch evaluates the potential for pesticide active ingredients to move to ground water under legal agricultural use conditions. Previous evaluations were primarily based on threshold values for specific persistence and mobility properties of pesticides as prescribed in the California Pesticide Contamination Prevention Act of 1985. Two limitations identified with that process were the univariate nature where interactions of the properties were not accounted for, and the inability to accommodate multiple values of a physical-chemical property. We addressed these limitations by developing a probabilistic modeling method based on prediction of potential well water concentrations. A mechanistic pesticide transport model, LEACHM, is used to simulate sorption, degradation and transport of a candidate pesticide through the root zone. A second empirical model component then simulates pesticide degradation and transport through the vadose zone to a receiving ground water aquifer. Finally, degradation during transport in the aquifer to the well screen is included in calculating final potential well concentrations. Using Monte Carlo techniques, numerous LEACHM simulations are conducted using random samples of the organic carbon normalized soil adsorption coefficients (Koc) and soil dissipation half-life values derived from terrestrial field dissipation (TFD) studies. Koc and TFD values are obtained from gamma distributions fitted to pooled data from agricultural-use pesticides detected in California ground water: atrazine, simazine, diuron, bromacil, hexazinone, and norflurazon. The distribution of predicted well water concentrations for these pesticides is in good agreement with concentrations measured in domestic wells in coarse, leaching vulnerable soils of Fresno and Tulure Counties. The leaching potential of a new pesticide is evaluated by substituting it's sorption and persistence data into the model. Because such Koc and TFD data are often sparse, model inputs are typically derived from sampling of a fitted simple triangular distribution. A new product is considered to be a potential ground water contaminant if the 95th percentile of predicted well water concentrations is greater than 0.05 mg/L.
H54D-03
Water Quality Significance of Wetlands Receiving Agricultural Drainage
The San Joaquin Valley is one of the most productive agricultural regions in the world and this productivity is heavily dependent on irrigated agricultural. An inevitable consequence of irrigated agricultural is the generation of return-flows conveyed down-gradient in agricultural drains that eventually discharge to surface waters. Agricultural drainage often has poor water quality characteristics, but demand for water in California is high and agricultural drainage is often diverted for secondary use, including the maintenance of ponds and wetlands. Additionally, agricultural drainage often discharges into riparian wetlands, rather than into the open river channel. In this study we tested the hypothesis that wetlands were mitigating or buffering the impact of agricultural drainage and that discharge of agricultural drainage into wetland buffer zones would provide water quality benefits. Water samples were collected at wetland, agricultural, and mixed drainages in the San Joaquin River basin and analyzed for a broad array of physical and chemical water quality parameters, including nutrients and organic carbon. At selected wetlands, input-output studies were conducted to determine wetland specific water quality effects. The water quality of drainages influenced by wetlands was compared to drainages that were predominantly influenced by other types of land-use. Wetland influenced drainages are more likely to have higher DOC concentrations that other drainages, including agricultural and mixed urban-agricultural drains. Wetland dominated drainages had lower nitrates than agricultural drainages and studies of individual wetlands demonstrated that wetlands remove soluble phosphate and nitrate, but produce DOC and biochemical oxygen demand (BOD). Overall land use in a drainage was a less significant determinant of water quality than soil type and the presence or absence of wetlands. The specific trihalomethane formation potential (THMFP) of the DOC from wetland dominated drainage waters was not significantly different than the specific THMFP for agriculturally derived DOC. The biodegradation potential for the THMFP was also similar between the different drainage sources, suggesting that there is not a biologically recalcitrant DOC fraction contributing disproportionately to specific THMFP in wetland drainage. Results to date suggest that the THMFP of wetland carbon can be mitigated by biological degradation as would occur in natural ecosystems. Nutrient enrichment of the San Joaquin River is an significant regional water quality problem and diversion of agricultural drainage through wetlands before discharge reduces nutrient loads to surface waters. However, potential waster quality gains from nutrient removal are offset by increases in DOC and associated BOD. The overall benefit of diverting agricultural drainage through wetlands is being investigated in the context of competing water quality priorities and the desire to expand riparian habitat in the San Joaquin Valley.
H54D-04
Nitrate and Dissolved Organic Carbon Concentrations in Riparian-Zone Ground Water of the Lower San Joaquin River, California
Previous studies have estimated ground water inputs to the SJR on the order of 0.1 cms per river km, which could contribute up to 15 percent of downstream flow during the summer. However, there is a paucity of information concerning the chemical composition of ground water accretions in the lower SJR. The objective of this study was to quantify the amount of ground water accretions to the lower SJR and its nitrate and DOC contributions to the river. The study area is a 106 km reach of the SJR from the confluence with Salt Slough to Vernalis. Sampling of nested monitoring wells (3-30 m) on both banks at three sites and at nested (1-5 m) in-stream wells at six sites was initiated to obtain temporal data (monthly) for the modeling of ground water inputs and their associated nitrate and DOC loads. A synoptic study making measurements at 30 sites using temporary drive-points (0.3 and 0.9 m depths) was conducted by boat to provide enhanced spatial coverage. The sampling at each site included measurements of hydraulic and temperature gradients, general water quality characterization and nitrate and DOC concentrations. Based on the first year of monitoring the permanent wells (since September 2006), the specific conductance ranged from 0.3 to 8.3 dS per m (median = 2.6 dS per m) and pH values were near neutral (range = 6.4-7.6; median = 7.2). Nitrate was only present in detectable concentrations (greater than 0.01 mg N per L) in 8 of the 26 wells. Nitrate concentrations ranged from 0.8 to 13.1 mg N per L (median = 2.0 mg N per L) for those wells containing nitrate. Ground water collected from the wells with no detectable nitrate was anoxic with the majority of these wells displaying the presence of sulfide. This suggests that denitrification is a prevalent process leading to the loss of nitrate from many of the ground water sources. Some of the anoxic wells displayed high ammonium concentrations (greater than 2 mg N per L) that could be oxidized to nitrate upon entering the river. DOC concentrations ranged from 0.1 to 3.6 mg per L with a median value of 1.5 mg per L. Preliminary data suggest that many of the in-stream wells (0.3 and 0.9 m depth) are also anoxic. This suggests that nitrate concentrations from in-stream wells are also attenuated by denitrification within the riparian zone and under the river bed. Future aspects of the study include comparison of geochemical, isotopic, and optical characteristics of the ground water with various end-members in an attempt to identify the sources of nitrate and DOC in ground water accretions. Hydraulic gradient and temperature data will be used to model rates of ground water flow into the river. These flux rates will be combined with water quality data to estimate the amount of nitrate and organic carbon contributed to the river from ground water.
H54D-05
Winter Cover Crop Effects on Nitrate Leaching in Subsurface Drainage as Simulated by RZWQM-DSSAT
Planting winter cover crops such as winter rye (Secale cereale L.) after corn and soybean harvest is one of the more promising practices to reduce nitrate loss to streams from tile drainage systems without negatively affecting production. Because availability of replicated tile-drained field data is limited and because use of cover crops to reduce nitrate loss has only been tested over a few years with limited environmental and management conditions, estimating the impacts of cover crops under the range of expected conditions is difficult. If properly tested against observed data, models can objectively estimate the relative effects of different weather conditions and agronomic practices (e.g., various N fertilizer application rates in conjunction with winter cover crops). In this study, an optimized winter wheat cover crop growth component was integrated into the calibrated RZWQM-DSSAT hybrid model and then we compare the observed and simulated effects of a winter cover crop on nitrate leaching losses in subsurface drainage water for a corn-soybean rotation with N fertilizer application rates over 225 kg N ha-1 in corn years. Annual observed and simulated flow-weighted average nitrate concentration (FWANC) in drainage from 2002 to 2005 for the cover crop treatments (CC) were 8.7 and 9.3 mg L-1 compared to 21.3 and 18.2 mg L-1 for no cover crop (CON). The resulting observed and simulated FWANC reductions due to CC were 59% and 49%. Simulations with the optimized model at various N fertilizer rates resulted in average annual drainage N loss differences between CC and CON to increase exponentially from 12 to 34 kg N ha-1 for rates of 11 to 261 kg N ha-1. The results suggest that RZWQM-DSSAT is a promising tool to estimate the relative effects of a winter crop under different conditions on nitrate loss in tile drains and that a winter cover crop can effectively reduce nitrate losses over a range of N fertilizer levels.
H54D-06
Water Quality Signal of Animal Agriculture at USGS Monitoring Stations is Related to Animal Confinement and/or Farm Size
US animal agriculture has undergone major structural changes over the past two decades, with the total number of livestock producers declining dramatically and the average size of the remaining operations increasing substantially. The result has been a pronounced trend towards greater spatial concentration and confinement of livestock. The change raises important questions about the water quality effects of animal agriculture in regions where livestock waste production has become more intensive but recovery, handling, and application of animal wastes to cropland more systematized. In previous research, we developed three separate national-level SPARROW models of surface water contaminants (total nitrogen, total phosphorus, and fecal coliform bacteria). Based on USGS monitoring and ancillary data from more than 400 US stream and river basins, the models include point and nonpoint sources of contaminants, land-to-water transport factors, and in-stream loss processes; parameter estimation is by non-linear regression. In this study we report on a pattern in the statistical results for the three models: The source coefficients (quantity of contaminant delivered to streams per unit of contaminant input) for unconfined animals are consistently larger and more statistically significant than those for confined animals. The implicit meaning is that something associated with waste management on large farms and/or animal confinement (e.g. retention period, recovery of manure for application to crops and subsequent crop uptake, and/or better waste treatment) reduces the average water quality signal of this scale of animal agriculture (per unit of manure input) to barely detectable at downstream monitoring stations, while the water quality signal from unconfined animal agriculture is more clear. The county-level data for confined and unconfined manure inputs (defined primarily by farm size) are from the USDA, and are spatially distributed in the model GIS by 1-km land use data. To date, our analytical probing of the results has not identified a viable alternative to the implicit conclusion. http://water.usgs.gov/nawqa/sparrow/
H54D-07
Uncertainties in modelling sediment and phosphorus transfers from intensive, temperate grassland plots
Erosion of sediment and associated phosphorus from intensively managed grassland is a threat to surface water quality~--–~less "spectacular" perhaps than erosion from arable land but nevertheless significant. Modelling erosion and phosphorus transfer in grassland systems poses both old and new challenges. In particular, our ability to both measure relevant parameters and variables in the field and define effectively the information content in our observations limits the development and testing of models. This is coupled to a shift in focus from surface to sub-surface mobilisation and transport of sediment and phosphorus, brought about by the dense vegetation cover in temperate grasslands which restricts surface erosion processes. While fundamental knowledge is emerging on sub-surface sediment dynamics at the soil core scale, there is little evidence on how to utilise this understanding for models at the plot or catchment scale. In this paper, we explore the uncertainties in model structure, parameters and observed data and their impact on understanding sediment and phosphorus processes at the plot scale. We assess competing model hypotheses for hydrology, sediment and phosphorus dynamics within a model learning framework, i.e. with the aim of rejecting specific models (parameters and structures) where possible. We will discuss the difficulties of inferring process mechanisms robustly from the available data. Furthermore, we shall comment on how this impacts on data collection and model development as we move from small spatial scales (1ha plots) with dense temporal sampling schemes to a larger catchment (40ha) with less data relative to the process complexity operating at this scale.
H54D-08
Aquifer Denitrification: Is it a Zero-Order or First-Order Reaction?
Results from a network of 16 in situ mesocosms (ISMs) used to study aquifer denitrification at 5 sites in North Dakota and 4 sites in Minnesota (with 2 more installations planned for Iowa) are considered. At the Elk Valley aquifer (EVA) site in northeastern North Dakota, denitrification rates from six denitrification experiments were all better modeled as zero-order (0.16 +/- 0.05 mg nitrate-N/L/day), as determined by squared values of the linear correlation coefficient. Denitrification experiments at the other sites showed that denitrification was either below detection (< 0.01 mg nitrate-N/L/day) or was better modeled as a first-order reaction (0.00021/day to 0.0020/day), although squared values of the linear correlation coefficients for both rate models were nearly equal for some of the experiments. Not only were denitrification rates at the EVA site highest compared to the other sites in the ISM network, but sediment concentrations of electron donors at the EVA site were also greatest [ferrous iron about 0.3%, inorganic S (as pyrite) about 0.4%, organic C about 0.4%, weight basis]. These observations support the Michaelis- Menten model for reaction rates, which indicates that reaction rates will be zero-order when the substrate (electron donor) is abundant and first-order when the substrate availability is limited.