First posted October 26, 2018 For additional information, contact: Director Washington Water Science CenterU.S. Geological Survey934 Broadway, Suite 300Tacoma, Washington 98402 The U.S. Geological Survey, in cooperation with the Whatcom Conservation District (WCD), collected groundwater-quality data for roughly 3 years (October 2011–May 2015) from near the water table beneath forage fields receiving regular seasonal applications of liquid dairy manure in Whatcom County, Washington. The work was done as part of an evaluation of WCD’s prototypical Application Risk Management (ARM) decision support system. The ARM system uses a combination of field-specific hydrology, stage of crop-growth, manure management practices, soil conditions, and precipitation forecast to evaluate the timing of manure application via a set of decision support tools (Manure Spreading Advisory, ARM Worksheet, manure application setback distances) in order to reduce the risk of contamination of surface water and groundwater. The ARM system’s effectiveness in reducing leaching of nitrate to groundwater was evaluated by monitoring nitrate concentrations in recently recharged groundwater beneath paired test plots receiving manure application scheduled using either conventional (CON) manure scheduling procedures, which utilize fixed start and end dates for manure application along with projected crop nutrient requirements or ARM manure scheduling procedures using an approach to manure application timing based on projected crop nutrient needs, field conditions, and weather forecast. Water-quality samples from the surface of the water table were collected synoptically from paired test plots (2–5 monitoring wells per test plot) at approximately monthly intervals at three different dairy field sites. Water-quality samples from near the water table were isolated from the underlying aquifer using a combination of an inflatable packer and a fine-grained sand pack encompassing the well-screen interval.Concentrations of nitrate and chloride measured at the water table beneath test plots were highly variable. Concentrations of nitrate ranged from non-detectable to 116 milligrams nitrogen per liter (mg-N/L), and chloride ranged from 1.15 to 153 mg/L. In each test plot, seasonal variations were much greater than spatial variations. Differences in nitrate concentrations in groundwater between the two treatments were inconclusive. Nitrate concentrations in groundwater at paired treatment plots (Mann Whitney, p<0.05) were significantly lower beneath the ARM treatment plot at site B, yet significantly higher beneath the ARM treatment plot at site C. Nitrate concentrations in ground water varied significantly among individual wells at each site (Kruskal-Wallis, p<0.05), indicating that leaching of nitrates from soil following manure application is spatially variable at the field scale tested regardless of manure application strategy. At all three paired test plots, average concentrations of nitrate and chloride at the water table were lowest near the end of the growing season (September) and increased rapidly with the onset of autumn rains (October–December). Under both the conventional (calendar-based) and treatment (ARM-based) manure application scheduling systems, high soil nitrate concentrations in autumn were coincident with rising groundwater levels, suggesting that nitrate and chloride were flushed from soil to groundwater by recharge from the seasonal rains. Under both treatments, concentrations of nitrate in shallow (10–25 feet) groundwater beneath forage fields receiving manure applications were greater than the nitrate drinking water standard of 10 mg-N/L in approximately 85 percent of samples. Yearly mass loading of nitrogen to the groundwater system calculated from nitrate concentrations at the water table and estimates of recharge volume ranged from 86 to 196 pounds-N per acre, which was equivalent to approximately 16–37 percent of the recommended manure application rate for projected forage production yield of 7 dry tons per acre per year. Manure nitrogen applied in the autumn, when crop nutrient needs decrease due to reduced sunlight and cooler temperatures and commensurate with ongoing mineralization of soil organic-nitrogen and increased seasonal precipitation, are more likely to exceed the immediate plant nutritional requirements and hence be flushed to groundwater than manure applications occurring near the peak of the growing season.

First posted March 29, 2016 For additional information, contact: Director, Washington Water Science CenterU.S. Geological Survey934 Broadway, Suite 300Tacoma, Washington 98402http://wa.water.usgs.gov Efforts to assess the likelihood of groundwater contamination from surface-derived compounds have spanned more than three decades. Relatively few of these assessments, however, have involved the use of process-based simulations of contaminant transport and fate in the subsurface, or compared the predictions from such models with measured data—especially over regional to national scales. To address this need, a process-based groundwater vulnerability assessment (P-GWAVA) system was constructed to use transport-and-fate simulations to predict the concentration of any surface-derived compound at a specified depth in the vadose zone anywhere in the conterminous United States. The system was then used to simulate the concentrations of selected agrichemicals in the vadose zone beneath agricultural areas in multiple locations across the conterminous United States. The simulated concentrations were compared with measured concentrations of the compounds detected in shallow groundwater (that is, groundwater drawn from within a depth of 6.3 ± 0.5 meters [mean ± 95 percent confidence interval] below the water table) in more than 1,400 locations across the United States. The results from these comparisons were used to select the simulation approaches that led to the closest agreement between the simulated and the measured concentrations.The P-GWAVA system uses computer simulations that account for a broader range of the hydrologic, physical, biological and chemical phenomena known to control the transport and fate of solutes in the subsurface than has been accounted for by any other vulnerability assessment over regional to national scales. Such phenomena include preferential transport and the influences of temperature, soil properties, and depth on the partitioning, transport, and transformation of pesticides in the subsurface. Published methods and detailed soil property data are used to estimate a wide range of model input parameters for each site, including surface albedo, surface crust permeability, soil water content, Brooks-Corey parameters, saturated hydraulic conductivity, macroporosity and sizes of microbial populations, as well as solute partition coefficients, reaction rates, and meso-micropore diffusion rates. To ensure geographic consistency among the predictions, the only site-specific input data that are used are those that are available for all of the 48 conterminous states.

First posted February 8, 2016 For additional information, contact: Director, Washington Water Science CenterU.S. Geological Survey934 Broadway, Suite 300Tacoma, Washington 98402http://wa.water.usgs.gov Heavy sediment loads in the Sumas River of Whatcom County, Washington, increase seasonal turbidity and cause locally acute sedimentation. Most sediment in the Sumas River is derived from a deep-seated landslide of serpentinite that is located on Sumas Mountain and drained by Swift Creek, a tributary to the Sumas River. This mafic sediment contains high amounts of naturally occurring asbestiform chrysotile. A known human-health hazard, asbestiform chrysotile comprises 0.25–37 percent, by mass, of the total suspended sediment sampled from the Sumas River as part of this study, which included part of water year 2011 and all of water years 2012 and 2013. The suspended-sediment load in the Sumas River at South Pass Road, 0.6 kilometers (km) downstream of the confluence with Swift Creek, was 22,000 tonnes (t) in water year 2012 and 49,000 t in water year 2013. The suspended‑sediment load at Telegraph Road, 18.8 km downstream of the Swift Creek confluence, was 22,000 t in water year 2012 and 27,000 t in water year 2013. Although hydrologic conditions during the study were wetter than normal overall, the 2-year flood peak was only modestly exceeded in water years 2011 and 2013; runoff‑driven geomorphic disturbance to the watershed, which might have involved mass wasting from the landslide, seemed unexceptional. In water year 2012, flood peaks were modest, and the annual streamflow was normal. The fact that suspended-sediment loads in water year 2012 were equivalent at sites 0.6 and 18.8 km downstream of the sediment source indicates that the conservation of suspended‑sediment load can occur under normal hydrologic conditions. The substantial decrease in suspended-sediment load in the downstream direction in water year 2013 was attributed to either sedimentation in the intervening river reach, transfer to bedload as an alternate mode of sediment transport, or both.The sediment in the Sumas River is distinct from sediment in most other river systems because of the large percentage of asbestiform chrysotile in suspension. The suspended sediment carried by the Sumas River consists of three major components: (1) a relatively dense, largely non-flocculated material that settles rapidly out of suspension; (2) a lighter component containing relatively high proportions of flocculated material, much of it composed of asbestiform chrysotile; and (3) individual chrysotile fibers that are too small to flocculate or settle out, and remain in suspension as wash load (these fibers are on the order of microns in length and tenths of microns in diameter). Whereas the bulk density of the first (heaviest) component of suspended sediment was between 1.5 and 1.6 grams per cubic centimeter (g/cm3), the bulk density of the flocculated material was an order of magnitude lower (0.16 g/cm3), even after 24 hours of settling. Soon after immersion in water, the fresh chrysotile fibers derived from the Swift Creek landslide seem to flocculate readily into large bundles, or floccules, that exhibit settling velocities characteristic of coarse silts and fine sands (30 and 250 micrometers). In quiescent water within this river system, the floccules settle out quickly, but still leave between 2.4 and 19.5 million chrysotile fibers per liter in the clear overlying water. Consistent with the results from previous laboratory research, the amounts of asbestiform chrysotile in the water column in Swift Creek, as well as in the Sumas River close to and downstream of its confluence with Swift Creek, were determined to be directly correlated with pH. This observation offers a possible alternative to either turbidity or suspended‑sediment concentration as a surrogate for the concentration of fresh asbestiform chrysotile in suspension.Continued movement and associated erosion of the landslide through mass wasting and runoff will maintain large sediment loads in Swift Creek and in the Sumas River for the foreseeable future. Given the present channel morphology of the river system, aggradation (that is, sediment accumulation) in Swift Creek and the Sumas River are also likely to continue.

BACKGROUND Complex environmental models are frequently extrapolated to overcome data limitations in space and time, but quantifying data worth to such models is rarely attempted. The authors determined which field observations most informed the parameters of agricultural system models applied to field sites in Nebraska (NE) and Maryland (MD), and identified parameters and observations that most influenced prediction uncertainty. RESULTS The standard error of regression of the calibrated models was about the same at both NE (0.59) and MD (0.58), and overall reductions in prediction uncertainties of metolachlor and metolachlor ethane sulfonic acid concentrations were 98.0 and 98.6% respectively. Observation data groups reduced the prediction uncertainty by 55-90% at NE and by 28-96% at MD. Soil hydraulic parameters were well informed by the observed data at both sites, but pesticide and macropore properties had comparatively larger contributions after model calibration. CONCLUSIONS Although the observed data were sparse, they substantially reduced prediction uncertainty in unsampled regions of pesticide breakthrough curves. Nitrate evidently functioned as a surrogate for soil hydraulic data in well-drained loam soils conducive to conservative transport of nitrogen. Pesticide properties and macropore parameters could most benefit from improved characterization further to reduce model misfit and prediction uncertainty.

Tobit regression models were developed to predict the summed concentration of atrazine [6-chloro--ethyl--(1-methylethyl)-1,3,5-triazine-2,4-diamine] and its degradate deethylatrazine [6-chloro--(1-methylethyl)-1,3,5,-triazine-2,4-diamine] (DEA) in shallow groundwater underlying agricultural settings across the conterminous United States. The models were developed from atrazine and DEA concentrations in samples from 1298 wells and explanatory variables that represent the source of atrazine and various aspects of the transport and fate of atrazine and DEA in the subsurface. One advantage of these newly developed models over previous national regression models is that they predict concentrations (rather than detection frequency), which can be compared with water quality benchmarks. Model results indicate that variability in the concentration of atrazine residues (atrazine plus DEA) in groundwater underlying agricultural areas is more strongly controlled by the history of atrazine use in relation to the timing of recharge (groundwater age) than by processes that control the dispersion, adsorption, or degradation of these compounds in the saturated zone. Current (1990s) atrazine use was found to be a weak explanatory variable, perhaps because it does not represent the use of atrazine at the time of recharge of the sampled groundwater and because the likelihood that these compounds will reach the water table is affected by other factors operating within the unsaturated zone, such as soil characteristics, artificial drainage, and water movement. Results show that only about 5% of agricultural areas have greater than a 10% probability of exceeding the USEPA maximum contaminant level of 3.0 μg L. These models are not developed for regulatory purposes but rather can be used to (i) identify areas of potential concern, (ii) provide conservative estimates of the concentrations of atrazine residues in deeper potential drinking water supplies, and (iii) set priorities among areas for future groundwater monitoring.

Pesticide leaching through variably thick soils beneath agricultural fields in Morgan Creek, Maryland was simulated for water years 1995 to 2004 using LEACHM (Leaching Estimation and Chemistry Model). Fifteen individual models were constructed to simulate five depths and three crop rotations with associated pesticide applications. Unsaturated zone thickness averaged 4.7 m but reached a maximum of 18.7 m. Average annual recharge to ground water decreased from 15.9 to 11.1 cm as the unsaturated zone increased in thickness from 1 to 10 m. These point estimates of recharge are at the lower end of previously published values, which used methods that integrate over larger areas capturing focused recharge in the numerous detention ponds in the watershed. The total amount of applied and leached masses for five parent pesticide compounds and seven metabolites were estimated for the 32-km2 Morgan Creek watershed by associating each hectare to the closest one-dimensional model analog of model depth and crop rotation scenario as determined from land-use surveys. LEACHM parameters were set such that branched, serial, first-order decay of pesticides and metabolites was realistically simulated. Leaching is predicted to be greatest for shallow soils and for persistent compounds with low sorptivity. Based on simulation results, percent parent compounds leached within the watershed can be described by a regression model of the form e(-depth) (a ln t1/2-b ln K OC) where t1/2 is the degradation half-life in aerobic soils, K OC is the organic carbon normalized sorption coefficient, and a and b are fitted coefficients (R2 = 0.86, p value = 7 x 10(-9)).

Pesticide transport through the unsaturated zone is a function of chemical and soil characteristics, application, and water recharge rate. The fate and transport of 82 pesticides and degradates were investigated at five different agricultural sites. Atrazine and metolachlor, as well as several of the degradates of atrazine, metolachlor, acetochlor, and alachlor, were frequently detected in soil water during the 2004 growing season, and degradates were generally more abundant than parent compounds. Metolachlor and atrazine were applied at a Nebraska site the same year as sampling, and focused recharge coupled with the short time since application resulted in their movement in the unsaturated zone 9 m below the surface. At other sites where the herbicides were applied 1 to 2 yr before sampling, only degradates were found in soil water. Transformations of herbicides were evident with depth and during the 4-mo sampling time and reflected the faster degradation of metolachlor oxanilic acid and persistence of metolachor ethanesulfonic acid. The fraction of metolachlor ethanesulfonic acid relative to metolachlor and metolachlor oxanilic acid increased from 0.3 to >0.9 at a site in Maryland where the unsaturated zone was 5 m deep and from 0.3 to 0.5 at the shallowest depth. The flux of pesticide degradates from the deepest sites to the shallow ground water was greatest (3.0-4.9 micromol m(-2) yr(-1)) where upland recharge or focused flow moved the most water through the unsaturated zone. Flux estimates based on estimated recharge rates and measured concentrations were in agreement with fluxes estimated using an unsaturated-zone computer model (LEACHM).

This paper is an introduction to the following series of papers that report on in-depth investigations that have been conducted at five agricultural study areas across the United States in order to gain insights into how environmental processes and agricultural practices interact to determine the transport and fate of agricultural chemicals in the environment. These are the first study areas in an ongoing national study. The study areas were selected, based on the combination of cropping patterns and hydrologic setting, as representative of nationally important agricultural settings to form a basis for extrapolation to unstudied areas. The holistic, watershed-scale study design that involves multiple environmental compartments and that employs both field observations and simulation modeling is presented. This paper introduces the overall study design and presents an overview of the hydrology of the five study areas.

Occurrence and fate of 45 pesticides and 40 pesticide degradates were investigated in four contrasting agricultural settings--in Maryland, Nebraska, California, and Washington. Primary crops included corn at all sites, soybeans in Maryland, orchards in California and Washington, and vineyards in Washington. Pesticides and pesticide degradates detected in water samples from all four areas were predominantly from two classes of herbicides--triazines and chloroacetanilides; insecticides and fungicides were not present in the shallow ground water. In most samples, pesticide degradates greatly exceeded the concentrations of parent pesticide. In samples from Nebraska, the parent pesticide atrazine [6-chloro-N-ethyl-N'-(1-methylethyl)-1,3,5-triazine-2,4-diamine] was about the same concentration as the degradate, but in samples from Maryland and California atrazine concentrations were substantially smaller than its degradate. Simazine [6-chloro-N,N'-diethyl-1,3,5-triazine-2,4-diamine], the second most detected triazine, was detected in ground water from Maryland, California, and Washington. Metolachlor [2-chloro-N-(2-ethyl-6-methylphenyl)-N-(2-methoxy-1-methylethyl)acetamide] rarely was detected without its degradates, and when they were detected in the same sample metolachlor always had smaller concentrations. The Root-Zone Water-Quality Model was used to examine the occurrence and fate of metolachlor at the Maryland site. Simulations accurately predicted which metolachlor degradate would be predominant in the unsaturated zone. In analyses of relations among redox indicators and pesticide variance, apparent age, concentrations of dissolved oxygen, and excess nitrogen gas (from denitrification) were important indicators of the presence and concentration of pesticides in these ground water systems.

An unsaturated-zone transport model was used to examine the transport and fate of metolachlor applied to an agricultural site in Maryland, USA. The study site was instrumented to collect data on soil-water content, soil-water potential, ground water levels, major ions, pesticides, and nutrients from the unsaturated zone during 2002-2004. The data set was enhanced with site-specific information describing weather, soils, and agricultural practices. The Root Zone Water Quality Model was used to simulate physical, chemical, and biological processes occurring in the unsaturated zone. Model calibration to bromide tracer concentrations indicated flow occurred through the soil matrix. Simulated recharge rates were within the measured range of values. The pesticide transport model was calibrated to the intensive data collection period (2002-2004), and the calibrated model was then used to simulate the period 1984 through 2004 to examine the impact of sustained agricultural management practices on the concentrations of metolachlor and its degradates at the study site. Simulation results indicated that metolachlor degrades rapidly in the root zone but that the degradates are transported to depth in measurable quantities. Simulations indicated that degradate transport is strongly related to the duration of sustained use of metolachlor and the extent of biodegradation.

Tracer-based ground-water ages, along with the concentrations of pesticides, nitrogen species, and other redox-active constituents, were used to evaluate the trends and transformations of agricultural chemicals along flow paths in diverse hydrogeologic settings. A range of conditions affecting the transformation of nitrate and pesticides (e.g., thickness of unsaturated zone, redox conditions) was examined at study sites in Georgia, North Carolina, Wisconsin, and California. Deethylatrazine (DEA), a transformation product of atrazine, was typically present at concentrations higher than those of atrazine at study sites with thick unsaturated zones but not at sites with thin unsaturated zones. Furthermore, the fraction of atrazine plus DEA that was present as DEA did not increase as a function of ground-water age. These findings suggest that atrazine degradation occurs primarily in the unsaturated zone with little or no degradation in the saturated zone. Similar observations were also made for metolachlor and alachlor. The fraction of the initial nitrate concentration found as excess N2 (N2 derived from denitrification) increased with ground-water age only at the North Carolina site, where oxic conditions were generally limited to the top 5 m of saturated thickness. Historical trends in fluxes to ground water were evaluated by relating the times of recharge of ground-water samples, estimated using chlorofluorocarbon concentrations, with concentrations of the parent compound at the time of recharge, estimated by summing the molar concentrations of the parent compound and its transformation products in the age-dated sample. Using this approach, nitrate concentrations were estimated to have increased markedly from 1960 to the present at all study sites. Trends in concentrations of atrazine, metolachlor, alachlor, and their degradates were related to the timing of introduction and use of these compounds. Degradates, and to a lesser extent parent compounds, were detected in ground water dating back to the time these compounds were introduced.

Abstract : Seven unsaturated-zone solute-transport models were tested with two data sets to select models for use by the Agricultural Chemical Team of the U.S. Geological Survey's National Water-Quality Assessment Program. The data sets were from a bromide tracer test near Merced, California, and an atrazine study in the White River Basin, Indiana. In this study the models are designated either as complex or simple based on the water flux algorithm. The complex models, HYDRUS2D, LEACHP, RZWQM, and VS2DT, use Richards' equation to simulate water flux and are well suited to process understanding. The simple models, CALF, GLEAMS, and PRZM, use a tipping-bucket algorithm and are more amenable to extrapolation because they require fewer input parameters. The purpose of this report is not to endorse a particular model, but to describe useful features, potential capabilities, and possible limitations that emerged from working with the model input data sets. More rigorous assessment of model applicability involves proper calibration, which was beyond the scope of this study.

Chloroform is one of the volatile organic compounds (VOCs) detected most frequently in both ground and surface water. Because it is also one of the four trihalomethanes (THMs) produced in the highest concentrations during the chlorination of drinking water and wastewater, the frequent detection of this compound in ground and surface water of the United States is presumed to be caused primarily by the input of chlorinated water to the hydrologic system. Although anthropogenic sources of the compound are substantial, they are currently estimated to constitute only 10 percent of the total global input to the hydrologic system. Natural sources of the compound include volcanic gases, biomass burning, marine algae, and soil microorganisms. Under most conditions (except in the presence of unusually high bromide concentrations), chloroform is the THM produced in the highest concentrations during chlorination. Furthermore, in most cases where more than one THM is produced from chlorination, the relative concentrations among the different compounds usually decrease with increasing bromination (chloroform > dichlorobromomethane > chlorodibromomethane > bromoform). This phenomenon is presumed to be responsible for the common observation that when more than one THM is detected during investigations of the occurrence of these compounds in the hydrologic system, this same trend is typically observed among their relative concentrations or, for a uniform reporting limit, their relative frequencies of detection. This pattern could provide a valuable means for distinguishing between chlorinated water and other potential sources of chloroform in the environment. Chloroform has been widely detected in national, regional, and local studies of VOCs in ground, surface, source, and drinking waters. Total THM (TTHM) concentrations of the compound, however, were typically less than the Maximum Contaminant Level (MCL) of 80 ?g/L (micrograms per liter) established by the U.S. Environmental Protection Agency (USEPA) for TTHMs. In the studies that compared land-use settings, frequencies of detection of chloroform were higher beneath urban and residential areas than beneath agricultural or undeveloped areas. Because chloroform is a suspected human carcinogen, its presence in drinking water is a potential human health concern. Liver damage, however, is known to occur at chloroform exposures lower than those required to cause cancer, an observation that has been considered by the USEPA as the basis for setting a new, non-zero Maximum Contaminant Level Goal of 70 ?g/L for the compound. As part of its National Water-Quality Assessment Program, the U.S. Geological Survey has been assembling and analyzing data on the occurrence of VOCs (including chloroform) in ground and surface water on a national scale from studies conducted between 1991 and the present. This report presents a summary of current (2004) information on the uses, sources, formation, transport, fate, and occurrence of chloroform, as well as its effects on human health and aquatic organisms.

Information on the concentrations and spatial distributions of pesticides and their transformation products, or degradates, in the hydrologic system is essential for managing pesticide use in both agricultural and nonagricultural settings to protect water resources. This report examines the occurrence of selected herbicides and their degradates in ground water, primarily on the basis of results from two large-scale, multistate investigations by the U.S. Geological Survey—the National Water-Quality Assessment (NAWQA) Program and the Midwest Pesticide Study (MWPS). The NAWQA pesticide data were derived from 2,227 sites (wells and springs) sampled in 20 major hydrologic basins across the United States from 1993 to 1995; the MWPS data were obtained from the sampling of 303 wells in a 12-state area of the northern midcontinent from 1991 to 1994. Data are presented for seven high-use herbicides: five of current interest to the U.S. Environmental Protection Agency for designing Pesticide Management Plans (atrazine, cyanazine, simazine, alachlor and metolachlor), a largely nonagricultural herbicide (prometon), and an agricultural herbicide first registered in 1994 for use in the United States (acetochlor). Six of the herbicides (all except acetochlor) were detected by the U.S. Geological Survey studies in shallow ground water—that is, ground water recharged within the past 10 years—in a variety of agricultural and nonagricultural settings, as well as in several aquifers that are sources of drinking-water supply. Acetochlor was not detected by the MWPS in the summer of 1994, but was detected in shallow ground water during the NAWQA Program by early 1995, and during another U.S. Geological Survey study in Iowa during the summers of 1995 and 1996. The acetochlor observations suggest that, in agreement with results from previous field studies, some pesticides may be detected in shallow ground water within 1 year following their application. In accord with the results from other largescale multistate studies of pesticides in ground water, more than 98 percent of the detections during the NAWQA and MWPS investigations were at concentrations of less than 1 microgram per liter. Consequently, water quality criteria for drinking water—that is, standards established to protect human health—were exceeded at fewer than 0.1 percent of the sites sampled by NAWQA (all of these exceedances involving atrazine alone) and at none of those sampled in 1992 by the MWPS. These criteria, however, may not accurately reflect the overall health risks associated with pesticide detections in water resources because they have been established only for a relatively small number of pesticides and they do not account for the additive or synergistic effects of mixtures, impacts on the health of aquatic ecosystems, or the effects of pesticide degradates. Among the sites sampled during the NAWQA and MWPS investigations, 19.7 and 13.8 percent, respectively, had detections of two or more of the herbicides of interest. Furthermore, for most of the herbicides for which degradates were examined, detection frequencies for major degradates were typically higher than for their respective parent compounds, particularly for the herbicides that are less persistent in aerobic soil. Frequencies of detection at or above 0.01 microgram per liter in shallow ground water beneath agricultural areas during the NAWQA study were significantly correlated with agricultural use in those areas for atrazine, cyanazine, alachlor, and metolachlor (P<0.05; Spearman rank correlations), but not for simazine (P>0.05). In urban areas, overall frequencies of detection of these five herbicides in shallow ground water were positively correlated with their total nonagricultural use nationwide (P=0.026; simple linear correlation). Multivariate statistical analysis indicated that frequencies of detection in shallow ground water beneath agricultural areas were positively correlated with half-lives for transformation in aerobic soil and agricultural use of the compounds (P≤0.0001 for both parameters). Although frequencies of detection were not significantly correlated with their subsurface mobility (Koc; P=0.19) or the median well depths of the sampled networks (P=0.72), the range of Koc values among the five herbicides and the range of well depths were limited.