Geology topics

Erik A. Smith

Publications and source records attributed to Erik A. Smith.

18 recordsLinked to original sources

USGS2026

Machine learning generated streamflow drought forecasts for the conterminous United States (CONUS): developing and evaluating an operational tool to enhance sub-seasonal to seasonal streamflow drought early warning for gaged locations

Forecasts of streamflow drought, when streamflow declines below typical levels, are notably less available than for floods or meteorological drought, despite widespread impacts. We apply machine learning (ML) models to forecast streamflow drought 1–13 weeks ahead at 3,219 streamgages across the conterminous United States. We applied two ML methods (Long short-term memory neural networks; Light Gradient-Boosting Machine) and two benchmark models (persistence; Autoregressive Integrated Moving Average) to predict weekly streamflow percentiles with independent models for each forecast horizon. ML models outperformed benchmarks in predicting continuous streamflow percentiles below 30%. ML models generally performed worse than persistence models for discrete classification (moderate, severe, extreme) but exceeded the benchmark models for drought onset/termination. Performance was better for less intense droughts and shorter horizons, with predictive power for 1–4 weeks for severe droughts (10% threshold). This work highlights challenges and opportunities to advance hydrological drought forecasting and supports a new experimental forecasting tool.

conterminous United States↗

USGS2023

Modeling the water-quality effects to the Klamath River from recirculation in drains and canals, Oregon and California, 2006–15

The potential recirculation of Klamath Strait Drain (hereafter called by its local name, “Klamath Straits Drain”) water into Ady Canal to reduce the drain discharge of high nutrient loads into the Klamath River was assessed by the U.S. Geological Survey for the Bureau of Reclamation. To study the feasibility of recirculation, this investigation evaluated three recirculation scenarios over a 10-year period from 2006 to 2015, as a series of 1-year model simulations. A combination of two existing hydrodynamic, water-temperature, and water-quality models (CE-QUAL-W2) were used, including (1) the Link-Keno reach of the Klamath River, using Klamath Straits Drain as a tributary and for calendar years 2006–11, and (2) the same Link-Keno model used for calendar years 2012–15 in combination with an independent Klamath Straits Drain model from 2012 to 2015. Model simulations using the water-quality models were configured for the base case conditions and three different sets of recirculation scenarios: the maximum year-round recirculation without limits (scenario 1), limited year-round recirculation fixed by the current pipe flow configuration (scenario 2), and limited seasonal recirculation (May–September) also fixed by the current pipe flow configuration (scenario 3). In the base case, estimates of annual average daily total nitrogen loads and daily total phosphorus loads exported to the Klamath River from the Klamath Straits Drain were as much as 3,060 and 457 pounds per day (lbs/day), respectively. Currently (2023), the Total Maximum Daily Loads allocations for the Klamath Straits Drain are 21 and 268 lbs/day for total phosphorus and total nitrogen, respectively, so these maximum estimates exceed the current Total Maximum Daily Loads by greater than an order of magnitude. With scenario 1, load reductions occurred year-round for all constituents evaluated (total nitrogen, total phosphorus, 5-day biochemical oxygen demand [BOD5], 5-day carbonaceous biochemical oxygen demand) for the Klamath Straits Drain discharging to the Klamath River. Scenario 2 also had large reductions in total nitrogen, total phosphorus, and BOD5 loads. Substantial reductions did occur for scenario 3 but were constrained to only the active recirculation period from May through September. Despite the restricted period, the average reductions in the annual average daily load for total phosphorus and total nitrogen were 32.1 percent and 26.5 percent, respectively. The Ady Canal diverts high nutrient loads from the Klamath River, so the loading tradeoffs to the Klamath River between no recirculation and the recirculation scenarios were calculated. On an annual basis, the overall net balance between the Klamath Straits Drain and Ady Canal resulted in more total nitrogen and total phosphorus load reductions to the Klamath River for the three recirculation scenarios than the base case, for most years. In contrast, the net balance for BOD5 loads was higher to the Klamath River for the three recirculation scenarios than the base case, for most years. With the recirculation scenarios, the optimal recirculation periods to benefit Ady Canal, Klamath River, and Klamath Straits Drain did not always coincide. Recirculation would be most effective at reducing loads toward the Klamath Straits Drain Total Maximum Daily Load allocations in the spring (March–May) of each year. However, recirculation during these months would also increase salinity in the Ady Canal. In summer, recirculation would reduce Klamath Straits Drain loads toward the Total Maximum Daily Load allocations, though recirculation could decrease Klamath River water quality mostly because of decreased withdrawals of Klamath River water by the Ady Canal. Scenario 3 avoided recirculation into Ady Canal in the early spring months when salinity concerns would be the highest, while still decreasing nutrient loads exported from the Klamath Straits Drain to the Klamath River in the summer months.

California, Oregon↗

USGS2022

Evaluation of the Bushy Park Reservoir three-dimensional hydrodynamic and water-quality model, South Carolina, 2012–15

The Bushy Park Reservoir is a relatively shallow impoundment in southeastern South Carolina. The reservoir, located under a semi-tropical climate, is the principal water supply for the city of Charleston, South Carolina, and the surrounding areas including the Bushy Park Industrial Complex. Although there was an adequate supply of freshwater in the reservoir in 2022, water-quality concerns are present over taste-and-odor and saltwater-intrusion issues. From 2013 to 2015, the U.S. Geological Survey (USGS), in cooperation with the Charleston Water System, engaged in a multi-year study of the hydrology and hydrodynamics of Bushy Park Reservoir to better understand factors affecting water-quality conditions in the reservoir. As part of this study, Charleston Water System worked with Tetra Tech, Inc., a consulting and engineering firm, to develop a Bushy Park Reservoir hydrodynamic and water-quality modeling framework, built upon earlier efforts by both Tetra Tech and the U.S. Army Corps of Engineers. At the completion of the new modeling framework, the USGS was requested to evaluate the calibrated hydrodynamic and water-quality model. The Bushy Park Reservoir Environmental Fluid Dynamics Code (EFDC) model was calibrated for the time period from January 1, 2012, to December 31, 2015. The general modeling approach for the newly revised modeling framework, as briefly detailed in this report, was developed with EFDC. The EFDC is a grid-based modeling package that can simulate three-dimensional flow, transport, and water quality in surface-water systems. This report evaluated the capacity of Tetra Tech’s Bushy Park EFDC model to simulate water discharge, water circulation, surface elevations, temperature, salinity, and other water-quality parameters. The USGS model review focused specifically on the following criteria: (1) determine if the model, with additional effort, could be developed into an adequate planning tool for Bushy Park Reservoir; (2) assess the capacity of the model to specifically address water-quality issues in the reservoir related to taste-and-odor and saltwater intrusions; and, (3) evaluate three preliminary water-management scenarios related to reduced water withdrawals in the reservoir and the effect on saltwater intrusion. Overall, the model was able to simulate discharge, flow velocity, and water-surface elevations with generally good agreement between the simulated and measured values. Specifically, the model was able to demonstrate good agreement for discharge at two USGS continuous discharge locations (USGS station 02172002; USGS station 02172040), with Wilmott index of agreements of 0.86 and 0.75, respectively. A total of seven USGS streamgages, located on the West Branch of the Cooper River, Durham Canal, and the Cooper River, were available for water-surface elevations, with index of agreements ranging from 0.74 to 0.99. However, model-simulated water-surface elevation ranges were appreciably high (compared to measured ranges) for two locations near Pinopolis Dam, farthest upstream on the West Branch of the Cooper River. This result may indicate that too much simulated tidal energy propagated through the model domain. For water temperature, 16 calibration stations were available for at least part of the 4-year simulation. The index of agreement range for temperature comparisons was from 0.95 to 1.00, indicating excellent agreement between the measured and simulated results. One of the primary future applications for the Bushy Park Reservoir EFDC model is to determine the extent of saltwater intrusions. A wide range in the salinity prediction quality was simulated with the model. The prediction quality ranged from an index of agreement of 0.15 at Cooper River approximately 2.75 miles southeast of the Tee, South Carolina, to 0.92 at West Branch Cooper River near Moncks Corner, South Carolina. Although the model did not accurately simulate some of the larger salinity deviations resulting during individual hydrologic events, the seasonal salinity trends were adequately simulated with the model during the study period (2012–15). Therefore, it may be difficult to simulate extreme hydrologic events, such as during large storms, where high salinity water is exchanged with Bushy Park Reservoir. There was agreement in model simulation with the measured data either on the quantitative index of agreement values or qualitative agreement in the seasonal salinity data trends. For water quality, the index of agreement values were generally low for total nitrogen, ammonia, nitrate, total Kjeldahl nitrogen, total phosphorus, and orthophosphate. Although general trends were adequately simulated at specific stations, particularly for Bushy Park Reservoir, the model-simulated fit was low across all the constituents described above with index of agreements usually below 0.50. A limitation for simulating nutrient concentrations across the model domain was the lack of characterization for the constituents directly entering Bushy Park Reservoir, or the lack of data directly attributed to the boundary condition (for example, the Cooper River). The other two calibrated water-quality constituents (besides the nutrients mentioned above) were dissolved oxygen and chlorophyll a . Dissolved oxygen varied from index of agreement values from 0.58 to 0.94 for 11 stations, generally indicating agreement with the available measured data. Chlorophyll a , calibrated for seven stations, had a wider range from 0.11 to 0.74 for the index of agreement. With the current modeling framework, taste-and-odor events, related to cyanobacterial blooms, cannot be directly simulated. However, indirect estimates of cyanobacteria concentrations may be obtained by using the chlorophyll a model outputs, which represent total phytoplankton biomass, and the phytoplankton biovolume data by group (diatoms, green algae, cyanobacteria and others) collected from 2012 to 2015. For the Bushy Park Reservoir modeling framework to be used directly for taste-and-odor issues, cyanobacteria must be simulated and calibrated based on observations of cyanobacteria biomass concentrations. In addition to the cyanobacteria sampling conducted within the reservoir between 2012 and 2015, the new model calibration would also require new algae biomass data-collection efforts to characterize the external sources of cyanobacteria entering the Bushy Park Reservoir from tributaries, as well as the internal cycling, production, and decay of cyanobacteria in the hydrologic system. Further improvements to the EFDC model would include expanding the collection of boundary condition datasets, such as water-quality monitoring to determine improved nutrient loads into the model domain. Along with improved water-quality monitoring for the major boundary conditions, continuous discharge, for both Foster Creek and the Back River, would further constrain the flow balance and the loads into Bushy Park Reservoir. In addition to better boundary-condition characterization, it is important to better characterize possible shortcomings specifically to the model domain, such as the grid resolution, bathymetry, and numerical hydrodynamic errors. Further consideration of the model may involve a sensitivity analysis to determine if errors in the simulation outputs, such as discharge, water-surface elevations, and salinity, were more likely caused by poor boundary condition characterization or, specifically, the model setup. Three model scenarios were run with the revised Bushy Park Reservoir model: (1) reduced withdrawals from one of the large intake-discharge locations for Bushy Park Reservoir, the Williams Station; (2) elevated (above background levels) ocean water level causing saltwater intrusion from the ocean through Durham Canal into Bushy Park Reservoir; and (3) overtopping of the Back River Dam at the southernmost end of Bushy Park Reservoir. For the reduced withdrawals scenarios, the largest shift in flow resulted near the Williams Station intake, with the next largest flow change at the southern end of Bushy Park Reservoir, and a net increase in flow out of the Bushy Park Reservoir to the Cooper River by way of the Durham Canal. The effect resulting from scenario 3 on water quality and salinity was small, with larger increases for dissolved oxygen than other constituents at several monitoring stations. For the two scenarios related to saltwater intrusion (including dam overtopping), the changes in salinity generally were found to dissipate in the following 2 weeks and generally back to baseline salinity conditions within 3 months. This result did vary depending on the severity of the storm or length of the dam overtopping event.

South Carolina↗

USGS2022

U.S. Geological Survey (USGS) Water-Use Data and Research (WUDR) program overview and status as of March 31, 2022

The USGS Water-Use Data and Research Program (WUDR) is an appropriated program and is authorized under the SECURE Water Act (Sec. 9508 (c)). WUDR provides financial assistance through cooperative agreements to State water resource agencies. The WUDR Program has two main goals: • To improve the availability, quality, compatibility, and delivery of water-use data that are collected and/or estimated by States to support national water-use assessments; and • To integrate the water-use data into USGS databases in electronic or machine-readable formats.

Newsletter↗

USGS2020

Hydrodynamic modeling results showing the effects of the Luce Bayou interbasin transfer on salinity in Lake Houston, TX

An overreliance on groundwater resources in the Houston (Texas) metropolitan area led to aquifer drawdowns and land subsidence, so regional water suppliers have been turning to surface water resources to meet water demand. Lake Houston, an important water supply reservoir 24 kilometers (15 miles) northeast of downtown Houston, requires new water supply sources to continue to meet water supply demands for the next several decades. The upcoming Luce Bayou Interbasin Transfer Project will divert up to 500 million gallons per day of Trinity River water into Lake Houston. Trinity River water has significantly different water quality than the Lake Houston tributaries. To evaluate the project’s potential effect on water quality, the U.S. Geological Survey used an enhanced version of a previously released Lake Houston hydrodynamic model. With a focus on salinity and water-surface elevations, the model combined data from 2009 to 2017 with simulated flow from the Luce Bayou Interbasin Transfer to evaluate potential outcomes from three hypothetical flow scenarios. Overall, these scenarios found that the Luce Bayou Interbasin Transfer would cause salinities to moderately rise over most of the modeled time (2009–2017), although salinities were buffered under 2011 drought conditions. Large inflow events equalized salinities under baseline conditions as well as the enhanced flow scenarios.

Texas↗

USGS2020

Simulation of discharge, water-surface elevations, and water temperatures for the St. Louis River estuary, Minnesota-Wisconsin, 2016–17

The St. Louis River estuary is a large freshwater estuary, next to Duluth, Minnesota, that encompasses the headwaters of Lake Superior. The St. Louis River estuary is one of the most complex and compromised near-shore systems in the upper Great Lakes with a long history of environmental contamination caused by logging, mining, paper mills, and other heavy industrial activities. Presently (2020), a widely available, science-based assessment tool capable of evaluating ecosystem-level responses to remediation and restoration projects has not existed for the estuary. To address this need, the U.S. Geological Survey (USGS) built a predictive, mechanistic, three-dimensional hydrodynamic model for the estuary using the Environmental Fluid Dynamics Code framework. In the current version, the model can simulate continuous discharge, water-surface elevations, water temperature, and flow velocity, although the modular framework allows for future additions of water-quality modeling. The model was calibrated using data collected from April 2016 through November 2016 and validated with data collected from April 2017 through November 2017. The four types of data used to evaluate model performance were water-surface elevations, discharge, water temperature, and flow velocities. Streamflow and temperature boundary condition data included a mixture of USGS streamgage data, Minnesota Department of Natural Resources gage data, and estimates derived from the gage data. The model was able to simulate the water-surface elevations with generally good agreement between the simulated and measured values for both years at the daily time step. Specifically, the model was able to demonstrate excellent agreement with the measured data with Nash-Sutcliffe efficiency coefficients greater than 0.8 for all three locations; however, the model was unable to produce hourly water-surface elevations with such accuracy for 2016–17. Discharge was more dynamic than the water-surface elevations, both for the measured and simulated data. Generally, most of the discharge ranged from −650 to 1,200 cubic meters per second, but the constantly changing flux exiting the estuary into Lake Superior (positive flows) and entering the estuary from Lake Superior (negative flows) occurred throughout the year. Even upstream at the St. Louis River at Oliver, Wisconsin, gage (USGS station 0402403250), the effect of flows into the estuary from Lake Superior did occur, demonstrating the strong effect of the Lake Superior seiche on flows for the estuary. From a performance standpoint, the model was able to simulate discharge with generally good agreement in both years, although the 2017 validation was better than the 2016 calibration period. For the daily Nash-Sutcliffe efficiency coefficients, the simulated values were 0.98, 0.62, 0.49, and 0.71 for the Oliver gage; the Superior Bay entry channel at Superior, Wisc., (USGS station 464226092005600); the Superior Bay Duluth Ship Canal at Duluth, Minn., (USGS station 464646092052900); and total entries (combination of the Superior entry and Duluth entry), respectively. For the hourly evaluation criteria, the model performed poorly, with Nash-Sutcliffe efficiency coefficients less than 0 for the two entries into Lake Superior; therefore, as a predictor of discharge at the hourly scale, the model performed worse than using the measured data average. Similar to discharge, the model was a good predictor of flow velocity at the daily time scale but had difficulty matching the measured data at the hourly scale. For discharge and flow velocity, matching at subdaily time steps for a system as complicated as the St. Louis River estuary is considered difficult because the match is highly sensitive to coordinating the exact measurement location to the simulated value. The final calibration target was water temperature, calibrated for the Oliver gage and the Duluth entry. For calibration purposes, the Duluth entry was the more important water temperature target because the Oliver gage was more of an internal check on the model. The Nash-Sutcliffe efficiency coefficients for the Duluth entry were high; hourly Nash-Sutcliffe efficiency coefficients at the Duluth entry were either at or greater than 0.7 for both years, and daily values were 0.84 and 0.82 for 2016 and 2017, respectively.

Minnesota, Wisconsin↗

USGS2020

Potential groundwater recharge rates for two subsurface-drained agricultural fields, southeastern Minnesota, 2016–18

Subsurface drainage is used to efficiently drain saturated soils to support productive agriculture in poorly drained terrains. Although subsurface drainage alters the water balance for agricultural fields, its effect on groundwater resources and groundwater recharge is poorly understood. In Minnesota, subsurface drainage has begun to increase in southeastern Minnesota, even though this part of the State is underlain by permeable karstic bedrock aquifers with only a thin layer of glacial sediments separating these aquifers from land surface. To gain a better understanding of groundwater recharge effects from subsurface drainage, the U.S. Geological Survey (USGS), in cooperation with the Legislative-Citizen Commission on Minnesota Resources, led a 2-year hydrologic study to investigate this connection for two agricultural fields in southeastern Minnesota with subsurface drainage. A total of three monitoring plots were used between the two agricultural fields: two monitoring plots that included an actively drained area with peripheral, undrained areas, and a third monitoring plot without any subsurface drainage. Multiple piezometer transects were set up across the three monitoring plots to characterize the unsaturated zone and shallow water-table flow using pressure transducers and soil moisture probes. From these piezometers, groundwater recharge rates were derived using two different methods: the RISE Water-Table Fluctuation (WTF) method and the DRAINMOD model. In addition to these two methods, the USGS Soil-Water-Balance (SWB) model was used to estimate potential recharge rates for three different monitoring plots. In addition to deriving groundwater recharge rates, the hydrologic budget was analyzed to interpret the water-table surface elevation and soil volumetric water content time series. At one of the two drained plots, the transects exhibited varying water-table surface elevation patterns. Frequent backflow from the adjacent ditch caused subsurface drainage flow to slow down or stop drainage through the main collector drain and cause pipe pressurization, so the closest transect appeared to be mostly controlled by the drain pressurization, whereas the farthest transect was more efficiently drained. Both of the drained monitoring plots had an elevation gradient parallel to the pattern tiles, sloping downward towards the collector drain that aggregated the parallel lines into a single drain. Because the transects were set at different gradients in the field, some of the water-table surface elevation differences were also attributed to lateral flow towards the lowest parts of the field. Three methods were used to derive potential groundwater recharge rates: the RISE WTF method, the USGS SWB model, and DRAINMOD-derived deep seepage rates. Potential groundwater recharge rates, using the RISE WTF method, across all piezometers were 1.55 and 1.94 inches per year, respectively, for water years 2017 and 2018. More differentiation of potential recharge rates between different piezometer types occurred for water year 2018. Although the difference was slightly more than 1 inch between the drained and nondrained piezometers for water year 2018, this difference was statistically significant based on a t-test with a p -value of 0.036 ( α =0.05). When looking at recharge based on distance from the drain, the subsurface drain did not affect potential recharge, although other factors such as variability in screen depths, well construction, and specific yield variability cannot be eliminated. The SWB model was also used to estimate potential recharge rates for water years 2017–18, with rates between 2.44 and 5.92 inches per year for the two drained sites, generally higher than the RISE WTF estimates. DRAINMOD-derived potential recharge rates were generally the highest of the three methods, with potential recharge rates varying from 2.07 to 9.49 inches per year. Overall, there was a lack of agreement between the three methods. These results were not remarkable, considering the fundamental differences in the methodology for each method. However, all methods did show a fundamental difference between piezometers within the drained area and piezometers outside the drained area, including the third undrained monitoring plot. The drained areas show a lower overall potential groundwater recharge compared to the nondrained areas for all three estimates. The difference for the 2018 recharge estimates was slightly higher than 1 inch for the RISE WTF method, the difference was almost double for the nine sites for the DRAINMOD model, and the difference between the drain and undrained plots was even more significant for the SWB model.

Minnesota↗

USGS2019

Updates to the Madison Lake (Minnesota) CE–QUAL–W2 water-quality model for assessing algal community dynamics

A previously developed CE–QUAL–W2 model for Madison Lake, Minnesota, simulated the algal community dynamics, water quality, and fish habitat suitability of Madison Lake under recent (2014) meteorological conditions. Additionally, this previously developed model simulated the complex interplay between external nutrient loading, internal nutrient loading from sediment release of phosphorus, and the organic matter decomposition of the algal biomass. However, the partitioning of Cyanophyta within the modeling framework was simplified to one group and did not account for how different Cyanophyta populations are affected by light conditions, use of nitrogen, temperature growth ranges, and differences in settling rates. Properly capturing Cyanophyta dynamics is important given the potential risks posed by potential large algal blooms. For example, when Cyanophyta form large blooms, recreational activities can become restricted in certain areas because of thick algal scums or algal mats, in addition to the possible production of a class of toxins, known as cyanotoxins, capable of threatening human health, domestic animals, and wildlife. Therefore, we updated the model to partition the Cyanophyta into a group that fixed nitrogen and a second, more buoyant Cyanophyta group that did not independently fix nitrogen. The U.S. Geological Survey, in cooperation with the St. Croix Watershed Research Station (Science Museum of Minnesota) with support from the Environmental and Natural Resources Trust Fund of Minnesota (Legislative-Citizen Commission on Minnesota Resources), updated the Madison Lake CE–QUAL–W2 model to address the shortcomings of simulating Cyanophyta in the previously developed model and better characterize Cyanophyta into two groups. In addition to updating the Cyanophyta group differentiation, the part of the model that handles the simulation of algal community dynamics was updated while preserving model predictive capabilities for nutrients, water temperature, and dissolved oxygen. The calibration and validation of the model was done under recent meteorological conditions with large and persistent Cyanophyta blooms (2014 and 2016). Overall, the model simulations predicted the persistently large total phosphorus concentrations in the hypolimnion of Madison Lake and key differences in nutrient concentrations between 2014 and 2016. The Cyanophyta bloom persistence throughout the summer was also simulated by the model in 2014 and 2016, a critical goal of the model update. Finally, monthly total phosphorus budgets were calculated for the updated Madison Lake model for 2014 and 2016.

Minnesota↗

USGS2018

Specific conductance as a tracer of preferential flow in a subsurface-drained field

Specific conductance (SC), soil volumetric water content (VWC), and discharge were monitored on a subsurface agricultural drain for a 2-yr period (2007–2008) to differentiate preferential flow paths from matrix flow paths. A major observation from the 2-yr period was the fast SC decrease after relatively small rainfall events, often <5 mm. A total of 25 paired rainfall–SC events were classified, with an average preferential flow onset time (from the event start) after 1.7 h and maximum preferential flow after 2.4 h. A specific conductance end-member mixing analysis (SC-EMMA) was used to determine the volume of water that infiltrated through preferential flow pathways. The SC-EMMA was used for 20 of the 25 paired rainfall–SC events; of the 20 classified events, the maximum preferential flow ranged from 11 to 75% of the total subsurface drain flow, with a mean maximum preferential flow of 31%. Overall, SC-EMMA illustrated that a significant portion of the subsurface drain discharge can be attributed to preferential flow, mainly through macropores or other largely open preferential flow pathways. The other primary mechanism, antecedent moisture conditions shifts, could only be shown for four of the 25 classified events. Specific conductance as a tracer of preferential flow was shown to be an effective tool for distinguishing preferential flow to subsurface drains. Even during relatively dry periods, the SC had a substantial decrease shortly after a rainfall event contrary to the conventional idea that macropore flow starts only after all the smaller pores are saturated and surface ponding begins to occur.

Iowa↗

USGS2018

Simulation of hydrodynamics, water quality, and lake sturgeon habitat volumes in Lake St. Croix, Wisconsin and Minnesota, 2013

Lake St. Croix is a naturally impounded, riverine lake that makes up the last 40 kilometers of the St. Croix River. Substantial land-use changes during the past 150 years, including increased agriculture and urban development, have reduced Lake St. Croix water-quality and increased nutrient loads delivered to Lake St. Croix. A recent (2012–13) total maximum daily load phosphorus-reduction plan set the goal to reduce total phosphorus loads to Lake St. Croix by 20 percent by 2020 and reduce Lake St. Croix algal bloom frequencies. The U.S. Geological Survey, in cooperation with the National Park Service, developed a two-dimensional, carbon-based, laterally averaged, hydrodynamic and water-quality model, CE–QUAL–W2, that addresses the interaction between nutrient cycling, primary production, and trophic dynamics to predict responses in the distribution of water temperature, oxygen, and chlorophyll a. Distribution is evaluated in the context of habitat for lake sturgeon, including a combination of temperature and dissolved oxygen conditions termed oxy-thermal habitat. The Lake St. Croix CE–QUAL–W2 model successfully reproduced temperature and dissolved oxygen in the lake longitudinally (from upstream to downstream), vertically, and temporally over the seasons. The simulated water temperature profiles closely matched the measured water temperature profiles throughout the year, including the prediction of thermocline transition depths (often within 1 meter), the absolute temperature of the thermocline transitions (often within 1.0 degree Celsius), and profiles without a strong thermocline transition. Simulated dissolved oxygen profiles matched the trajectories of the measured dissolved oxygen concentrations at multiple depths over time, and the simulated concentrations matched the depth and slope of the measured concentrations. Additionally, trends in the measured water-quality data were captured by the model simulation, gaining some potential insights into the underlying mechanisms of critical Lake St. Croix metabolic processes. The CE–QUAL–W2 model tracked nitrate plus nitrite, total nitrogen, and total phosphorus throughout the year. Inflow nutrient contributions (loads), largely dominated by upstream St. Croix River loads, were the most important controls on Lake St. Croix water quality. Close to 60 percent of total phosphorus to the lake was from phosphorus derived from organic matter, and about 89 percent of phosphorus to Lake St. Croix was delivered by St. Croix River inflows. The Lake St. Croix CE–QUAL–W2 model offered potential mechanisms for the effect of external and internal loadings on the biotic response regarding the modeled algal community types of diatoms, green algae, and blue-green algae. The model also suggested the seasonal dominance of blue-green algae in all four pools of the lake. A sensitivity analysis was completed to test the total maximum daily load phosphorus-reduction scenario responses of total phosphorus and chlorophyll a. The modeling indicates that phosphorus reductions would result in similar Lake St. Croix reduced concentrations, although chlorophyll a concentrations did not decrease in the same proportional amounts as the total phosphorus concentrations had decreased. The smaller than expected reduction in algal growth rates highlighted that although inflow phosphorus loads are important, other constituents also can affect the algal response of the lake, such as changes in light penetration and the breakdown of organic matter releasing nutrients. The available habitat suitable for lake sturgeon was evaluated using the modeling results to determine the total volume of good-growth habitat, optimal growth habitat, and lethal temperature habitat. Overall, with the calibrated model, the fish habitat volume in general contained a large proportion of good-growth habitat and a sustained period of optimal growth habitat in the summer. Only brief periods of lethal oxy-thermal habitat were present in Lake St. Croix during the model simulation.

Minnesota, Wisconsin↗

USGS2018

Drain tiles and groundwater resources: Understanding the relations

Executive Summary Drainage for agricultural production over the past 150 years has been an integral component of human-driven change to Minnesota’s rural landscapes. Benefits of drainage Historically, poorly drained soils across much of the State would often remain saturated or flooded after spring snowmelt, preventing timely farm operations such as tilling and planting crops (Arneman, 1963). Installation of agricultural drainage, both surface ditches and subsurface drainage, accelerated transport of water off farm fields and imparted producers higher crop yields (Beauchamp, 1987; Stoner and others, 1993). Agricultural drainage offered many other benefits such as preventing crop drown out, aerating the soil profile for improved plant growth, limiting surface runoff and soil erosion, and allowing farmers better access to croplands (Fausey and others, 1987). Without agricultural drainage on much of Minnesota’s croplands, it would have been difficult to realize high enough crop yields to remain economically viable. Environmental concerns While drainage of Minnesota croplands provided the benefits mentioned above, several environmental concerns result. These include wetland loss, degradation of downstream water quality, and reduced [potential for] recharge. Early agricultural drainage efforts (pre-20th century) led to the disappearance of much of Minnesota’s natural wetlands. Increased focus on preventing or mitigating wetland loss over the last 50 years has helped curtail further losses, even as agricultural drainage proceeds. Prior to establishment of Minnesota statehood, wetlands accounted for more than 10 million acres in Minnesota, including prairie wetlands, peatlands, and forest wetlands that comprised approximately 19 percent of the total land area (Palmer, 1915; King, 1980). In 2018, only half of Minnesota’s pre-settlement wetlands remain, mostly in parts of the State that have not experienced widespread drainage, such as northern Minnesota. Water-quality monitoring has shown that agricultural drainage, in particular the practice of subsurface drainage, provides a direct flow path for nutrient (nitrogen and soluble phosphorus) losses to surface water resources. The negative consequences of agricultural drainage on surface water quality are well documented (for example, Dinnes and others, 2002; Kladivko and others, 2004; Richards and others, 2008; Rozemeijer and others, 2010; Schottler and others, 2013). Agricultural basins with a high percentage of agricultural drainage have been implicated as part of the cause of the Gulf of Mexico hypoxia zone due to excessive nitrogen export (Goolsby and Battaglin, 2001; Randall and Mulla, 2001). The connection of hydrological effects of agricultural subsurface drainage on groundwater recharge and aquifers, on the other hand, has not been well-established. Agricultural subsurface drainage intercepts infiltrating water below croplands and directly discharges the water to nearby surface waters. However, the size of the water balance shift from drained water that would have evapotranspired or run off the land to drained water that would recharge underlying aquifers has been poorly characterized (Schuh, 2008). Drain Tiles and Groundwater Given the poor accounting of subsurface drainage effects on groundwater resources, the Minnesota Ground Water Association (MGWA) deemed it imperative that we document these effects so that groundwater resources in agricultural regions with substantial drainage can be effectively managed. This white paper documents the relations of drain tiles and groundwater resources and discusses the historical significance of agricultural drainage practices, the recognized positive benefits and potential negative consequences of agricultural drainage practices, and the gaps in understanding of the connections between agricultural drainage and groundwater resources. The major messages emerged from the findings of this white paper are: Complex history. Minnesota has a long history of agricultural drainage, spanning over 150 years. Agricultural drainage, and the eventual widespread usage of subsurface drainage, can be separated into at least four distinct periods of time: (a) early drainage to get water off the land, pre-20th century; (b) the boom and bust era (1900-1945); (c) postwar resurgence of subsurface drainage and early conservation efforts (1945-1960); and, (d) emergence of the environmental movement (1960 to present). The State’s regulatory framework that both allowed for drainage and controlled its usage during these periods is complicated and has been governed by a patchwork of both State and Federal statutes. Drainage Provinces. This white paper advances the concept of tile drainage provinces to aid in the discussion of regional differences in subsurface drainage and its overall effect on groundwater resources. Built upon the concept of groundwater provinces (Minnesota Department of Natural Resources, 2001), three distinct tile drainage provinces were conceptualized: (1) the Southeastern Province; (2) the SouthCentral Province; and, (3) the Western Province. The distinct geology and the soils that developed in these regions have implications for each region’s subsurface drainage density and the potential implications for groundwater. Knowledge gaps. Several critical knowledge gaps are identified in this paper, creating opportunities for further research to improve our understanding for better managed water resources: Extent of drainage is unknown. Direct estimates of the extent of subsurface drainage do not exist in Minnesota. However, several indirect methods have been utilized to estimate subsurface drainage, from the field-scale to countylevel through the use of geographic information system (GIS) analysis and aerial photography. Based on a 2012 U.S. Geological Survey estimate of subsurface drainage extent (Nakagaki and Wieczorek, 2016), about 21% of the land area in Minnesota has some density of subsurface drainage. Effect of drainage on underlying aquifers is unknown. A basic understanding of unconfined and confined aquifers and their recharge is necessary to connect any hydrological effects from agricultural drainage to groundwater. The basic goal of subsurface drainage to efficiently drain saturated soils clearly alters the water balance in croplands. However, its overall effect on groundwater resources has been poorly characterized, and is in large part determined by the geology below drained areas and the arrangement of underlying aquifers. Water balance shifts. An improved understanding of historical water balance shifts from pre- to post-drainage periods is necessary to understand long-term implications on net groundwater recharge. Also, more direct field-scale studies and indirect modeling studies are needed to characterize water budgets for fields with subsurface drainage.

Minnesota↗

USGS2017

Streamflow distribution maps for the Cannon River drainage basin, southeast Minnesota, and the St. Louis River drainage basin, northeast Minnesota

Streamflow distribution maps for the Cannon River and St. Louis River drainage basins were developed by the U.S. Geological Survey, in cooperation with the Legislative-Citizen Commission on Minnesota Resources, to illustrate relative and cumulative streamflow distributions. The Cannon River was selected to provide baseline data to assess the effects of potential surficial sand mining, and the St. Louis River was selected to determine the effects of ongoing Mesabi Iron Range mining. Each drainage basin (Cannon, St. Louis) was subdivided into nested drainage basins: the Cannon River was subdivided into 152 nested drainage basins, and the St. Louis River was subdivided into 353 nested drainage basins. For each smaller drainage basin, the estimated volumes of groundwater discharge (as base flow) and surface runoff flowing into all surface-water features were displayed under the following conditions: (1) extreme low-flow conditions, comparable to an exceedance-probability quantile of 0.95; (2) low-flow conditions, comparable to an exceedance-probability quantile of 0.90; (3) a median condition, comparable to an exceedance-probability quantile of 0.50; and (4) a high-flow condition, comparable to an exceedance-probability quantile of 0.02. Streamflow distribution maps were developed using flow-duration curve exceedance-probability quantiles in conjunction with Soil-Water-Balance model outputs; both the flow-duration curve and Soil-Water-Balance models were built upon previously published U.S. Geological Survey reports. The selected streamflow distribution maps provide a proactive water management tool for State cooperators by illustrating flow rates during a range of hydraulic conditions. Furthermore, after the nested drainage basins are highlighted in terms of surface-water flows, the streamflows can be evaluated in the context of meeting specific ecological flows under different flow regimes and potentially assist with decisions regarding groundwater and surface-water appropriations. Presented streamflow distribution maps are foundational work intended to support the development of additional streamflow distribution maps that include statistical constraints on the selected flow conditions.

Minnesota↗

USGS2017

Groundwater discharge to the Mississippi River and groundwater balances for the Interstate 94 Corridor surficial aquifer, Clearwater to Elk River, Minnesota, 2012–14

The Interstate 94 Corridor has been identified as 1 of 16 Minnesota groundwater areas of concern because of its limited available groundwater resources. The U.S. Geological Survey, in cooperation with the Minnesota Department of Natural Resources, completed six seasonal and annual groundwater balances for parts of the Interstate 94 Corridor surficial aquifer to better understand its long-term (next several decades) sustainability. A high-precision Mississippi River groundwater discharge measurement of 5.23 cubic feet per second per mile was completed at low-flow conditions to better inform these groundwater balances. The recharge calculation methods RISE program and Soil-Water-Balance model were used to inform the groundwater balances. For the RISE-derived recharge estimates, the range was from 3.30 to 11.91 inches per year; for the SWB-derived recharge estimates, the range was from 5.23 to 17.06 inches per year. Calculated groundwater discharges ranged from 1.45 to 5.06 cubic feet per second per mile, a ratio of 27.7 to 96.4 percent of the measured groundwater discharge. Ratios of groundwater pumping to total recharge ranged from 8.6 to 97.2 percent, with the longer-term groundwater balances ranging from 12.9 to 19 percent. Overall, this study focused on the surficial aquifer system and its interactions with the Mississippi River. During the study period (October 1, 2012, through November 30, 2014), six synoptic measurements, along with continuous groundwater hydrographs, rainfall records, and a compilation of the pertinent irrigation data, establishes the framework for future groundwater modeling efforts.

Minnesota↗

USGS2017

Water-quality models to assess algal community dynamics, water quality, and fish habitat suitability for two agricultural land-use dominated lakes in Minnesota, 2014

Fish habitat can degrade in many lakes due to summer blue-green algal blooms. Predictive models are needed to better manage and mitigate loss of fish habitat due to these changes. The U.S. Geological Survey (USGS), in cooperation with the Minnesota Department of Natural Resources, developed predictive water-quality models for two agricultural land-use dominated lakes in Minnesota—Madison Lake and Pearl Lake, which are part of Minnesota’s sentinel lakes monitoring program—to assess algal community dynamics, water quality, and fish habitat suitability of these two lakes under recent (2014) meteorological conditions. The interaction of basin processes to these two lakes, through the delivery of nutrient loads, were simulated using CE-QUAL-W2, a carbon-based, laterally averaged, two-dimensional water-quality model that predicts distribution of temperature and oxygen from interactions between nutrient cycling, primary production, and trophic dynamics. The CE-QUAL-W2 models successfully predicted water temperature and dissolved oxygen on the basis of the two metrics of mean absolute error and root mean square error. For Madison Lake, the mean absolute error and root mean square error were 0.53 and 0.68 degree Celsius, respectively, for the vertical temperature profile comparisons; for Pearl Lake, the mean absolute error and root mean square error were 0.71 and 0.95 degree Celsius, respectively, for the vertical temperature profile comparisons. Temperature and dissolved oxygen were key metrics for calibration targets. These calibrated lake models also simulated algal community dynamics and water quality. The model simulations presented potential explanations for persistently large total phosphorus concentrations in Madison Lake, key differences in nutrient concentrations between these lakes, and summer blue-green algal bloom persistence. Fish habitat suitability simulations for cool-water and warm-water fish indicated that, in general, both lakes contained a large proportion of good-growth habitat and a sustained period of optimal growth habitat in the summer, without any periods of lethal oxythermal habitat. For Madison and Pearl Lakes, examples of important cool-water fish, particularly game fish, include northern pike ( Esox lucius ), walleye ( Sander vitreus ), and black crappie ( Pomoxis nigromaculatus ); examples of important warm-water fish include bluegill ( Lepomis macrochirus ), largemouth bass ( Micropterus salmoides ), and smallmouth bass ( Micropterus dolomieu ). Sensitivity analyses were completed to understand lake response effects through the use of controlled departures on certain calibrated model parameters and input nutrient loads. These sensitivity analyses also operated as land-use change scenarios because alterations in agricultural practices, for example, could potentially increase or decrease nutrient loads.

Minnesota↗

USGS2015

Potential groundwater recharge for the State of Minnesota using the Soil-Water-Balance model, 1996-2010

Groundwater recharge is one of the most difficult components of a water budget to ascertain, yet is an important boundary condition necessary for the quantification of water resources. In Minnesota, improved estimates of recharge are necessary because approximately 75 percent of drinking water and 90 percent of agricultural irrigation water in Minnesota are supplied from groundwater. The water that is withdrawn must be supplied by some combination of (1) increased recharge, (2) decreased discharge to streams, lakes, and other surface-water bodies, and (3) removal of water that was stored in the system. Recent pressure on groundwater resources has highlighted the need to provide more accurate recharge estimates for various tools that can assess the sustainability of long-term water use. As part of this effort, the U.S. Geological Survey, in cooperation with the Minnesota Pollution Control Agency, used the Soil-Water-Balance model to calculate gridded estimates of potential groundwater recharge across Minnesota for 1996‒2010 at a 1-kilometer (0.621-mile) resolution. The potential groundwater recharge estimates calculated for Minnesota from the Soil-Water Balance model included gridded values (1-kilometer resolution) of annual mean estimates (that is, the means for individual years from 1996 through 2010) and mean annual estimates (that is, the mean for the 15-year period 1996−2010). The Soil-Water-Balance model uses a modified Thornthwaite-Mather soil-water-balance approach, with components of the soil-water balance calculated on a daily basis. A key advantage of this approach includes the use of commonly available geographic information system data layers that incorporate land cover, soil properties, and daily meteorological data to produce temporally and spatially variable gridded estimates of potential recharge. The Soil-Water-Balance model was calibrated by using a combination of parameter estimation techniques, making manual adjustments of model parameters, and using parameter values from previously published Soil-Water-Balance models. Each calibration simulation compared the potential recharge estimate from the model against base-flow estimates derived from three separate hydrograph separation techniques. A total of 35 Minnesota watersheds were selected for the model calibration. Meteorological data necessary for the model included daily precipitation, minimum daily temperature, and maximum daily temperature. All of the meteorological data were provided by the Daymet dataset, which included daily continuous surfaces of key climatological data. Land-cover data were provided by the 2001 and 2006 National Land Cover Database: the 2001 classification was used from 1994 through 2003, and the 2006 classification was used from 2004 through 2010. Soil data used in the model included hydrologic soils group and the available soil-water capacity. These soil data were obtained from the Natural Resources Conservation Service Soil Survey Geographic (SSURGO) database and the State Soil Geographic (STATSGO) database. The statewide mean annual potential recharge rate from 1996–2010 was 4.9 inches per year. Potential recharge estimates increased from west to east across Minnesota. The mean annual potential recharge estimates across Minnesota at a 1-km resolution for the overall simulation period (1996–2010) ranged from less than 0.1 to 17.8 inches per year. Some of the lowest potential recharge rates for the simulation period were in the Red River of the North Basin of northwestern Minnesota, and generally were between 1.0 and 1.5 inches per year. The highest potential recharge rates were in northeastern Minnesota and the Anoka Sand Plain in central Minnesota. Eighty-eight percent of the potential recharge rates (by grid cell) were between 2 and 8 inches per year from 1996–2010. Only about 3 percent of all the potential recharge estimates (by grid cell) were less than 2 inches per year, and 9 percent of estimates were greater than 8 inches per year. On an annual basis, however, potential recharge rates were as high as 27.2 inches per year. The highest annual mean recharge estimate across the State was for 2010, and the lowest mean recharge estimate was for 2003. Although precipitation variability partially explained the annual differences in potential recharge estimates, precipitation alone did not account for these differences, and other factors such as antecedent moisture conditions likely were important. Also, because precipitation gradients across the State can vary from year to year, the dominant land-cover class and hydrologic soil group combinations for a particular region had a large effect on the resulting potential recharge value. During 1996–2010, April had the greatest monthly mean potential recharge compared to all other months, accounting for a mean of 30 percent of annual potential recharge in this single month.

Minnesota↗

USGS2014

Water quality and algal community dynamics of three deepwater lakes in Minnesota utilizing CE-QUAL-W2 models

Water quality, habitat, and fish in Minnesota lakes will potentially be facing substantial levels of stress in the coming decades primarily because of two stressors: (1) land-use change (urban and agricultural) and (2) climate change. Several regional and statewide lake modeling studies have identified the potential linkages between land-use and climate change on reductions in the volume of suitable lake habitat for coldwater fish populations. In recent years, water-resource scientists have been making the case for focused assessments and monitoring of sentinel systems to address how these stress agents change lakes over the long term. Currently in Minnesota, a large-scale effort called “Sustaining Lakes in a Changing Environment” is underway that includes a focus on monitoring basic watershed, water quality, habitat, and fish indicators of 24 Minnesota sentinel lakes across a gradient of ecoregions, depths, and nutrient levels. As part of this effort, the U.S. Geological Survey, in cooperation with the Minnesota Department of Natural Resources, developed predictive water quality models to assess water quality and habitat dynamics of three select deepwater lakes in Minnesota. The three lakes (Lake Carlos in Douglas County, Elk Lake in Clearwater County, and Trout Lake in Cook County) were assessed under recent (2010–11) meteorological conditions. The three selected lakes contain deep, coldwater habitats that remain viable during the summer months for coldwater fish species. Hydrodynamics and water-quality characteristics for each of the three lakes were simulated using the CE-QUAL-W2 model, which is a carbon-based, laterally averaged, two-dimensional water-quality model. The CE-QUAL-W2 models address the interaction between nutrient cycling, primary production, and trophic dynamics to predict responses in the distribution of temperature and oxygen in lakes. The CE-QUAL-W2 models for all three lakes successfully predicted water temperature, on the basis of the two metrics of absolute mean error and root mean square error, using measured inputs of water temperature and nutrients. One of the main calibration tools for CE-QUAL-W2 model development was the vertical profile temperature data, available for all three lakes. For all three lakes, the absolute mean error and root mean square error were less than 1.0 degree Celsius and 1.2 degrees Celsius, respectively, for the different depth ranges used for vertical profile comparisons. In Lake Carlos, simulated water temperatures compared better to measured water temperatures in the epilimnion than in the hypolimnion. The reverse was true for the other two lakes, Elk Lake and Trout Lake, where the simulated results were slightly better for the hypolimnion than the epilimnion. The model also was used to approximate the location of the thermocline throughout the simulation periods, approximately April to November, in all three lake models. Deviations between the simulated and measured water temperatures in the vertical lake profile commonly were because of an offset in the timing of thermocline shifts rather than the simulated results missing thermocline shifts altogether.

Minnesota↗

USGS2013

Electron donor concentrations in sediments and sediment properties at the agricultural chemicals team research site near New Providence, Iowa, 2006-07

The concentrations of electron donors in aquifer sediments are important to the understanding of the fate and transport of redox-sensitive constituents in groundwater, such as nitrate. For a study by the U.S. Geological Survey National Water-Quality Assessment Program, 50 sediment samples were collected from below the water table from 11 boreholes at the U.S. Geological Survey Agricultural Chemicals Team research site near New Providence, Iowa, during 2006-07. All samples were analyzed for gravel, sand (coarse, medium, and fine), silt, clay, Munsell soil color, inorganic carbon content, and for the following electron donors: organic carbon, ferrous iron, and inorganic sulfide. A subset of 14 sediment samples also was analyzed for organic sulfur, but all of these samples had concentrations less than the method detection limit; therefore, the presence of this potential electron donor was not considered further. X-ray diffraction analyses provided important semi-quantitative information of well-crystallized dominant minerals within the sediments that might be contributing electron donors.

Iowa↗

USGS2007

Anthropogenic organic compounds in ground water and finished water of community water systems in the Greater Twin Cities metropolitan area, Minnesota and Wisconsin, 2004–05

As part of the U.S. Geological Survey’s National Water-Quality Assessment (NAWQA) Program, two Source Water-Quality Assessments (SWQAs) were conducted during 2004–05 in unconfined parts of the glacial aquifer system and in unconfined parts of the Prairie du Chien-Jordan aquifer in the Greater Twin Cities metropolitan area of Minnesota and Wisconsin. SWQAs are two-phased sampling activities in the NAWQA Program. The first phase evaluated the occurrence of 265 (258 are included in this report) anthropogenic organic compounds (AOCs) through monitoring source water in 30 of the largest-producing community water system wells completed in the aquifers underlying the Greater Twin Cities metropolitan area. The AOCs included volatile organic compounds (VOCs), pesticides, and other AOCs. During the second phase of the study, 15 of the original community water system wells, those with the greatest number of AOC detections, were resampled along with associated finished water. Results from the first phase of sampling indicated that 40 AOCs were detected, and 83 percent of the samples had at least one detected AOC. Concentrations of AOCs detected in the source water generally were low (defined in this report as concentrations less than 1.0 microgram per liter). Human-health benchmarks for these compounds (Maximum Contaminant Levels for regulated compounds or Health-Based Screening Levels for unregulated compounds, when they existed) typically were not exceeded. Fifteen VOCs were detected in the source-water samples. However, concentrations were low. Seventeen pesticide compounds were detected generally at concentrations less than concentrations for VOCs. Most of the pesticide compounds detected were triazine- or alachlor-parent compounds or their breakdown products. Eight other AOCs were detected in the source-water samples but generally at low concentrations. Results from the second phase of sampling indicated a total of 13 and 12 VOCs were detected in source-water and in finished-water samples, respectively. Most of the VOCs, except for those associated with disinfection by-products, were detected more frequently in source-water samples than in finished- water samples. Concentrations of most VOCs detected in either source water or finished water were less than human-health benchmarks. Twenty-one pesticide compounds were detected in either source water or finished water. Concentrations of detected pesticides in source-water and finished-water samples were low. The most frequently detected compounds in both the source and finished water were triazine-parent pesticides or their breakdown products and breakdown products of alachlor and metolachlor. In general, pesticides, if detected in source water, also were detected in the corresponding finished water. Concentrations of pesticides detected were less than human-health benchmarks in both source and finished water. A total of nine other AOCs were detected in the source-water or finished-water samples, and about the same number of compounds was detected in each of the sample groups in either source water or finished water. Detected concentrations of other AOCs were low. Water-quality results from source-water samples were compared to characterize differences between aquifers. VOC and other AOC detections were more frequent in water from the Prairie du Chien-Jordan aquifer compared to the glacial aquifer. Pesticides, however, were detected more frequently in the glacial aquifer. On the basis of study results, the hydrogeologic setting, land use, and aquifer productivity are important in explaining the occurrence of AOCs in community water system wells. Results of this study indicate that monitoring for pesticides in source water generally indicates the potential occurrence of pesticides in finished water but that this is not necessarily true of VOCs. Additional monitoring is needed to better understand the occurrence of other AOCs in source and finished waters.

Minnesota, Wisconsin↗