study area makes it desirable to use management practices that leave this water where it is, rather than bring it to the land surface or allow it to move into parts of the aquifer that may be used for water supply. Low concentrations of selenium in water near the water table imply that, where evaporative concentration is controlled, continued irrigation with lowselenium water will result in a growing volume of this low-selenium ground water. Drainage strategies aimed at removing this low-selenium ground water near the water table may be feasible in some areas, but selenium concentrations still may substantially exceed waterquality criteria.

Selenium in Tile-Drain Water

Drainage systems play a key role in the evaluation of management alternatives because of their effective control of the water table and the large number of systems already in operation. Selenium concentrations vary greatly between drainage systems but tend to be consistent over time in drain water from a particular system. The exception is the first 1 to 5 years, when concentrations tend to be the highest and most variable. The low variability of selenium concentrations in water from existing mature drainage systems underscores the fact that drainage systems withdraw ground water, which tends to be of a generally constant chemical character in a particular place.

Selenium from the San Joaquin River probably is not having a measurable effect on water quality in...San Francisco Bay at present.

drained fields shows that local water-table history, geologic conditions, irrigation history, and drainage-system design, all factors that vary greatly between fields, can markedly affect the type of water that is removed by the drains.

Mobility of Soil Selenium

Soils that contain selenium and are irrigated are a source of selenium in shallow ground water. Readily soluble forms of selenium in present-day soils are only a small fraction of the total selenium, but the quantities of soluble selenium can be substantially different between soils in different fields and at different depths. For example, in one field that had been irrigated for decades and tile-drained for 15 years, the soils were highly leached throughout the unsaturated zone. In two fields irrigated for just as long, but drained for less than half as long, saline soils with substantial quantities of soluble selenium were at the 3-foot depth, even though the near-surface soils were highly leached. Even after 40 years of irrigation, water percolating through soils of these latter two fields still contains high concentrations of selenium.

Even in the most highly leached soils, dissolved-selenium concentrations in solution ranged from a median of 13 μg/L at the 1-foot depth to 20 μg/L at the 3-foot depth. These selenium concentrations probably are the minimum that occur in actual recharge water passing through this soil to the ground water. These concentrations probably result from the slow dissolution of slightly soluble forms of selenium, which may be a long-term source at such levels of leaching. Thus, even after the most soluble forms of selenium have been leached from the soil by earlier irrigation, continued irrigation can result in recharge to ground water that contains lower, but still undesirably high, concentrations of selenium.

The high variability in selenium concentrations between existing drainage systems reflects the high variability of selenium in shallow ground water, variable hydrologic conditions at individual fields, and the age of the drainage systems. Detailed study of individual

Selenium in the San Joaquin River

At the present time, drain water from about 77,000 acres of tile-drained farmland eventually flows to the San Joaquin River, mainly from two tributaries, Salt and Mud Sloughs. Water from

individual drainage systems that discharge to waterways that eventually reach these sloughs or smaller tributaries contains from less than 10 to 4,000 μg/L of selenium, and mixtures of these waters contain from 20 to 100 μg/L. Concentrations of selenium in the San Joaquin River depend primarily on the magnitude of the selenium load from the sloughs and dilution by low-selenium water from all other sources of streamflow to the river. There are no substantial gains or losses of selenium between the sloughs and the delta.

During the study period, selenium concentrations at several sites on the river at times exceeded proposed waterquality criteria. Available data indicate that selenium from the San Joaquin River probably is not having a measurable effect on water quality in the delta and San Francisco Bay at present. Selenium concentrations in the San Joaquin River can be reduced by limiting tiledrain water discharges to Salt and Mud Sloughs and may be partly controllable by dilution with water from low-selenium sources in the basin.

Role of Sediment Chemistry in Water Quality

By Arthur J. Horowitz

Recent USGS research in sediment chemistry has concentrated on two major types of investigation: those intended to provide usable techniques for collecting, concentrating, and analyzing both suspended and bed sediments and those designed to explain the processes-the concentrating and releasing mechanismsthat control sediment-chemical concentrations.

One of the major problems facing investigators working with suspended sediments has been to obtain adequate quantities of material for study and analysis. The standard method of collecting and concentrating suspended sediment using membrane filtration is inadequate for large-scale national programs. At

Significant shifts in suspended-sediment concentrations...have been detected...over distances of as little as 16 to 33 feet.

best, current filtration techniques permit the processing of only 1 to 2 quarts of water per hour. Use of a centrifuge in the laboratory permits the processing of 3 to 5 quarts of water per hour. More rapid techniques have been investigated that permit the processing of water samples at the rate of 1 to 4 quarts per minute for the separation of suspended sediments for trace-element analyses. A similar study is now underway to see if these techniques can be used for organic analyses.

To adequately assess how chemicals are transported in sediments, scientists must have representative samples from different areas and at different time intervals; thus they must know how the distribution of suspended sediment varies across a section of river. Even when the most accurate measuring methods currently available are used, significant shifts in suspended-sediment concentrations and their associated chemical components have been detected in river cross sections in as little as 20 to 30 minutes and over distances of as little as 16 to 33 feet. The development of models designed to predict sediment-associated trace-element transport requires a twopronged approach: (1) A thorough understanding of the relation among riverine flow and bed, bank, and overbank deposits and the conditions that induce sediment movement or deposition and (2) a thorough understanding of the interaction between sediments (suspended and bed) and water. A further issue is the accurate modeling of suspended-sediment transport. At present, most sediment-transport models deal with sand-sized (>63 micrometers) particles; however, the highest traceelement concentrations are associated with the silt- and clay-sized (<63 micrometers) fractions. The USGS began a 2-year study in 1988 to investigate and model fine-sediment transport. Studies

are also underway to help clarify sediment-water interactions.

Research to develop an understanding of the geochemical processes that influence the accumulation of trace elements in a variety of benthic organisms from different segments of the food chain is continuing in both the natural environment and the laboratory. These studies include assessments of traceelement partitioning in oxidized sediments and of the physiological processes that influence the accumulation of trace elements and their effects on organisms. These studies also will define the relation between trace-element concentrations in sediments and concentrations of those trace elements in biota.

Reliable Remediation of Contaminated Aquifers

By Steven M. Gorelick

In the United States alone, more than 29,500 uncontrolled hazardouswaste sites have been inventoried by the U.S. Environmental Protection Agency. If the trend during the past 10 years continues, the number will exceed 33,000 by early 1990. It is estimated that about 80 percent of these sites have the potential to cause contamination of ground water. Once ground water is contaminated, it can cost millions of dollars and can take decades to restore an aquifer to a condition where the water is again usable. In some industrialized areas, many sites and many aquifers have been contaminated. For example, in California's Santa Clara Valley ("Silicon Valley") where ground water furnishes half of the water supply, there are more than 50 significantly contaminated sites, and private industry alone has spent more than $200 million on remediation since 1982.

Given this nationwide problem, scientists at the U.S. Geological Survey have developed computer models to analyze and to serve as an aid in the design of site-specific, reliable ground-water restoration schemes. At the forefront of this

Once ground water is contaminated, it can cost millions of dollars and can take decades to restore an aquifer.

new methodology are techniques that use combined simulation-management models to identify the ground-water withdrawal strategies that will most effectively remediate or restore contaminated aquifers. As the name implies, the models join computer simulation techniques, which predict the pattern of subsurfacecontaminant migration, with advanced mathematical and statistical methods, which determine the most economical design for a remediation scheme.

These simulation-management models determine the best location and pumping rates for wells that could be used to reliably contain and remove contaminated water. The major component of the models is computer simulation to describe contaminant migration. Unfortunately, the physical, chemical, and biological mechanisms controlling transport of underground contaminants are highly complicated.

Ignoring the chemical and biological mechanisms, even the simplest soluble compound will migrate and spread because of highly variable transmission and storage properties in aquifers. The transmission property, known as hydraulic conductivity, varies from place to place, both laterally and vertically. This variation can be more than a factor of 10 within the distance of a few meters. Much of the uncertainty in predicting contaminant migration stems from the variability of aquifer properties. The simulation models are based on the physics of subsurface flow and require knowledge of local aquifer properties, yet it is virtually impossible to measure the tremendous spatial variability over a substantial portion of an aquifer. Therefore, the variation must be characterized statistically. Statistical information on the spatial variability of aquifer properties is specified in the simulation models, which

then produce statistical information on the chances that contamination will occur at a particular time and place.

Finally, the simulation-management model determines the best remediation strategy using ground-water withdrawals so that water-quality standards are met at a desired reliability, perhaps 90 percent, at water-supply wells. This promising tool for aquifer management requires substantial computer effort. The assessment and design of reliable remediation schemes must be analyzed for each particular hazardous-waste site; thousands of simulations are needed to quantify the uncertainty of subsurface-contaminant migration.

Microbial Reduction of Iron in Sedimentary Environments

By Derek R. Lovley

It is becoming increasingly apparent that microorganisms catalyze many of the important geochemical reactions that influence the quality of water supplies. A new group of microorganisms, the dissimilatory iron-reducing bacteria that catalyze most of the reduction of ferric iron to ferrous iron in sedimentary environments, has been discovered. Before this discovery, iron reduction was thought to be a trivial process in the metabolism of bacteria and much of the iron reduction in sedimentary environments was thought to be abiological. Now it is clear that direct microbial activity is required for the reduction of iron in most environments and that models of iron geochemistry must consider the physiological limitations and capabilities of the iron-reducing microorganisms.

Microbial iron reduction converts insoluble ferric iron oxides to more soluble ferrous iron forms, which affect water quality in several ways. High concentrations of dissolved iron are one of the most prevalent ground-water quality problems. Furthermore, the extent of microbial iron reduction also controls the fate of toxic trace metals in aquatic envi

ronments. Iron oxides bind trace metals and concentrate them in the sediments. When microorganisms reduce iron, the trace metals are released into the surrounding water. In a similar manner, microbial iron reduction releases the phosphate held by iron oxides in the bottom sediments of lakes and estuaries. This released phosphate is often the nutrient that controls the development of nuisance algal blooms. Excessive growth of algae, for example, is thought to produce the low levels of oxygen in waters of some areas of the Potomac River Estuary and Chesapeake Bay.

Studies conducted with freshwater sediments from the Potomac River Estuary in Maryland resulted in the isolation of the first microorganism known to break down organic compounds to carbon dioxide with the reduction of iron or manganese. The organism does not fit any previously described genera and has temporarily been given the strain designation GS-15. Organisms with a similar metabolism appear to be distributed in a wide variety of environments, as such organisms were subsequently found in brackish-water estuarine sediments and in a deep aquifer in which products of iron reduction were degrading the water quality.

Studies on sediments and various iron-reducing isolates demonstrated that most naturally occurring organic compounds may be decomposed by ironreducing bacteria. However, each particular type of iron-reducing bacteria can only metabolize a restricted number of compounds. Therefore, food chains of several different types of iron-reducing bacteria are required to completely oxidize sediment organic matter.

Iron-reducing microorganisms were also found to be potentially important in the removal of organic contaminants from water supplies. Aromatic compounds such as benzene are common organic contaminants of ground water. Geochemical evidence suggests that iron reduction is an important mechanism for oxidizing these compounds and, thereby, mitigating their adverse effects. An ironreducing organism was isolated that could oxidize aromatic compounds by using the iron oxides naturally present in the sediments of aquifers. The metabolism of this organism demonstrates a

likely mechanism for the oxidation of aromatic contaminants with iron reduction and provides a model organism (GS-15) with which to begin studying the factors controlling the rate of contaminant removal by this process.

Iron-reducing bacteria were found to generate copious quantities of ultra-finegrained magnetite as a product of iron reduction, a novel mechanism for the magnetization of sediments that may have important implications for fields of study such as paleomagnetism. This metabolism also provides a likely mechanism for the development of the massive accumulations of magnetite in ancient iron ore deposits and for the magnetic anomalies associated with hydrocarbon deposits.

Studies on the metabolism of ironreducing bacteria, and microorganisms involved in other anaerobic (oxygen-free) geochemical cycles such as methane production and sulfate reduction, have led to the development of mathematical models of these processes. These models, which are based upon the physiological characteristics of the sediment microorganisms, have been found to accurately predict the distribution of important organic and inorganic constituents in the waters of a wide variety of anaerobic environments. As more is learned about the microorganisms responsible for catalyzing geochemical reactions, these models will increase in predictive power.

National Water-Quality
Assessment Program

By William M. Alley and
William G. Wilber

During the past two decades, Federal, State, and local governments and industry have made significant commitments to the protection of water quality. Estimates by the Bureau of Economic Analysis show that about $184 billion was spent from 1972 to 1982 for water pollution abatement and control. Future expenditures for pollution abatement and control through the year 2000 have

been projected to be as much as $600 billion. Given the large financial investments in water-quality management and protection already made, the potential for even larger investments in the future, and the concerns about solving myriad problems, there is a need for reliable and nationally consistent information on the status of and trends in the quality of the Nation's water resources, as well as for scientifically valid explanations of these conditions and trends.

Critical to understanding water quality is the ability to make comparisons among different locations and over time. Consistent information is necessary to make valid comparisons. Although this need has been well recognized for a long time in conducting individual waterquality studies, it is only recently that the need for nationally consistent information to make valid regional and national comparisons about current water-quality conditions and about changes in these conditions has been well recognized.

Because of this stated need for nationally consistent water-quality information, beginning in 1986, the Congress has annually appropriated funds for the U.S. Geological Survey to test and refine concepts for a National Water-Quality Assessment Program. Such a full-scale program would provide a nationally