Principal

Dissolved Organic Carbon Concentrations and Compositions, and Trihalomethane Formation Potentials in Waters from Agricultural Peat Soils, Sacramento-San Joaquin Delta, California:
Implications for Drinking-Water Quality

U.S. Geological Survey
Water-Resources Investigations Report 98-4147

Prepared in cooperation with the
California Department of Water Resources

Sacramento, California, 1998

ABSTRACT

Soil water was sampled from near-surface, oxidized, well-decomposed peat soil (upper soil zone) and deeper, reduced, fibrous peat soil (lower soil zone) from one agricultural field in the west central Delta over 1 year. Concentrations of dissolved organic carbon in the upper soil zone were highly variable, with median concentrations ranging from 46.4 to 83.2 milligrams per liter. Concentrations of dissolved organic carbon in samples from the lower soil zone were much less variable and generally slightly higher than samples from the upper soil zone, with median concentrations ranging from 49.3 to 82.3 milligrams per liter.

The dissolved organic carbon from the lower soil zone had significantly higher aromaticity (as measured by specific ultraviolet absorbance) and contained significantly greater amounts of aromatic humic substances (as measured by XAD resin fractionation and carbon-13 nuclear magnetic resonance analysis of XAD isolates) than the dissolved organic carbon from the upper soil zone. These results support the conclusion that more aromatic forms of dissolved organic carbon are produced under anaerobic conditions compared to aerobic conditions.

Dissolved organic carbon concentration, trihalomethane formation potential, and ultraviolet absorbance were all highly correlated, showing that trihalomethane precursors increased with increasing dissolved organic carbon and ultraviolet absorbance for whole water samples. Contrary to the generally accepted conceptual model for trihalomethane formation that assumes that aromatic forms of carbon are primary precursors to trihalomethanes, results from this study indicate that dissolved organic carbon aromaticity appears unrelated to trihalomethane formation on a
carbon-normalized basis. Thus, dissolved organic carbon aromaticity alone cannot fully explain or predict trihalomethane precursor content, and further investigation of aromatic and nonaromatic forms of carbon will be needed to better identify trihalomethane precursors.

INTRODUCTION

Figure 1. Location of the study area, Twitchell Island, Sacramento-San Joaquin Delta, California

Drainage water from Delta islands is estimated to contribute from 20 to 50 percent of the DOC contributing to the formation of THMs (THM precursors) in water samples collected at the H.O. Banks pumping plant (Amy and others, 1990; California Department of Water Resources, 1990). The H.O. Banks pumping plant is one of the primary diversion points from the Delta for drinking-water supply. The release of DOC from organic peat soils is believed to be the primary source of the DOC and THM precursors to the island drainage water from Delta islands (Amy and others, 1990; California Department of Water Resources, 1994a), although other carbon sources, such as recent crop residue and microbial biomass, also contribute to DOC releases. Island drainage water is pumped over the levees and into the channel waters of the Delta.

Organic soils in the Delta developed primarily from the accumulation of decaying plant material in this once tidal wetland during the last 10,000 years (Atwater and others, 1977; Atwater, 1980). Decomposition of the organic material by bacteria, fungi, and other organisms has contributed to the formation of the peat soils present on Delta islands, up to 60 feet (ft) deep in some areas (California Department of Water Resources, 1993). By the mid-1800's, settlers began farming the rich organic soils of the Delta, and, by 1869, extensive levee systems were built around Sherman and Twitchell Islands (California Department of Water Resources, 1993) to allow for the expansion of agriculture. Reclamation of Delta islands for agricultural purposes continued into the 1920's, and agriculture accounts for about 70 percent of the current land-use in the Delta (Templin and Cherry, 1997).

Reclamation of Delta islands by the construction of levees and the dewatering of soils for agricultural production has increased the exposure of organic soils to oxygen, resulting in subsidence of the land. Microbial oxidation of the peat soils is the predominant process that contributes to the loss of land-surface elevation in the Delta (Rojstaczer and Deverel, 1993, 1995; Deverel and Rojstaczer, 1996; Deverel and others, 1998); parts of the Delta are more than 20 ft below sea level (California Department of Water Resources, 1993). The resulting large difference between Delta channel water levels and island land-surface elevation increases hydraulic pressures on levees, increasing the probability of levee failure and also requiring the pumping of large volumes of drainage water off the islands to maintain ground-water levels below crop root zones.

Ground-water levels under the Twitchell Island agricultural field studied for this report are maintained at about 3 ft below land surface under nonflooded and nonirrigated conditions, resulting in a vertical oxygen gradient within the soil profile--aerobic at the surface, transitioning to anaerobic below the water table. Under these contrasting redox conditions, microbial decomposition of the soil organic matter (SOM) differs. Under aerobic conditions, microorganisms decompose SOM at much faster rates, with carbon dioxide (CO₂) and water (H₂O) as their metabolic end products. Under anaerobic conditions, bacterial decomposition rates of SOM are much slower and result in reduced compounds such as gaseous nitrogen (N₂), hydrogen sulfide (H₂S), and methane (CH₄) [depending on the availability of oxidized forms of nitrogen (N), sulfur (S), and carbon (C) as terminal electron acceptors] as their metabolic end products (Stevenson, 1985, 1994; Tate, 1987). Under both redox conditions, decomposition of SOM releases carbon that is potentially available to the aqueous phase. These contrasting decomposition pathways result in two soil layers that grade from one to the other. The surface and near-surface soils form an oxidized, well-decomposed, organic soil layer. Below the water table, deeper than about 4 ft below land surface, is a reduced, fibrous peat soil layer that is much less decomposed. The chemical characteristics of the different organic compounds released to the aqueous phase under these different redox conditions are discussed in detail in the Composition and Reactivity of Dissolved Organic Carbon section of this report.

Organic matter derived from different sources (for example, historically accumulated peat and recent agricultural inputs) and affected by different processes (for example, decomposition under varying redox conditions) has distinctive chemical characteristics associated with those materials and processes. The DOC released from SOM is a complex, heterogeneous mixture of a multitude of organic compounds whose chemical characteristics and reactivity are the result of all of the factors (organic matter sources, land-use practices, hydrology, and so forth) and biogeochemical processes (microbial decomposition, redox reactions, sorption to solids, and so forth) affecting DOC within the system (Aiken and Cotsaris, 1995). Throughout the Delta, these factors and processes vary spatially, and the DOC produced varies in composition and reactivity. Thus, not all DOC in drainage water is the same, and not all organic compounds react in the same way or to the same degree. This underlying premise must be considered when assessing the origin of THM precursors in the Delta.

The primary focus of this study was to evaluate the concentration and composition of DOC released from the soil and in drainage water from one agricultural field on Twitchell Island in the Delta (fig. 2) and to relate the DOC composition to its propensity to form THMs.

The purpose of this report is to transfer information and understanding of the results of the DOC/THM drainage-water study to the cooperator and other water-resource managers. The report focuses on data and interpretation of the more salient aspects of the DOC/THM drainage-water study, and many of the details regarding methods and additional data are in the appendices. Appendix A reports sample collection methods and the analytical results from the DWR Bryte Laboratory. The DOC, ultraviolet absorbance (UVA) at 254 nanometers (nm), specific UVA (SUVA or UVA/DOC), and analytical methods for samples analyzed by the U.S. Geological Survey (USGS) are presented in appendix B. Appendix C describes the methods and results of the DOC fractionation and isolation techniques using XAD resins. The methods and data for THM formation potential of selected whole-water samples and DOC isolates (fractionated using XAD resins) analyzed by the USGS are presented in appendix D. Appendix E presents methods and results of liquid-state carbon-13 nuclear magnetic resonance (13C-NMR) analyses of selected XAD fractionation isolates, solid-state 13C-NMR analyses of selected DOC isolates, and preliminary analysis of two soil samples by solid-state 13C-NMR. Appendix F presents DOC, SUVA, and other water-quality data for three wetland-habitat test ponds.

This study was done by the USGS in cooperation with the California Department of Water Resources (DWR) Municipal Water Quality Investigation Program and is part of the USGS National Drinking Water Initiative.

DESCRIPTION OF STUDY SITE AND STUDY DESIGN

Three pairs of stainless steel lysimeters and piezometers were installed at the northwestern end of the field (fig. 2), spaced about 40 to 80 ft apart. The lysimeters were installed from 0.5 to 1.5 ft below land surface to sample soil water influenced by the oxidized, well-decomposed peat soil layer (upper soil zone, USZ) (fig. 3). The piezometers were installed from 4.5 to 6.5 ft below land surface to sample ground water associated with the reduced, fibrous peat soil layer (lower soil zone, LSZ). Placement of the lysimeters and piezometers was designed to sample DOC released from soils influenced mainly by aerobic and anaerobic soil microbial decomposition processes, respectively. The samplers were installed parallel to and about 50 ft south of the ditch that drains the field, and their placement was intended to capture water draining from the field to the ditch.

Four pairs of lysimeters and piezometers were initially installed in the field. During installation of the lysimeter and piezometer at site 3, water was filling the borehole from an opening about 2 ft below land surface. It appeared that this represented a direct connection from the ditch draining the field to the borehole, either a channel in the peat or a rodent hole. Water-quality results from Lysimeter 3 and Piezometer 3 supported this observation. The results indicated that the quality (DOC and specific conductance) of water sampled at this site was considerably different from the other three sites and was similar to the quality of water in the drainage ditch. It was concluded that soil water sampled at site 3 was being significantly influenced by water originating from the drainage ditch and did not represent water that was influenced mainly by surrounding peat. Therefore, sampling from this site was discontinued and results for Lysimeter 3 and Piezometer 3 are not included in this report.

Lack of available water in the unsaturated zone for much of the year greatly hindered using the lysimeters to sample soil water in the USZ. Although lysimeters are designed to sample unsaturated soil, obtaining sufficient water for all the analyses was difficult, except when the field was intentionally flooded in February 1996. Even during the period when the field was irrigated, sample volumes obtained from the lysimeters were small, which limited the types of analyses done. In contrast, piezometers, which were installed below the water table, supplied ample water for analyses.

Hydrologic Framework for Study Design

Prior to flooding, precipitation during late December 1995 and early January 1996 began saturating the soils and leaching accumulated salts and DOC toward the water table and laterally toward the drainage ditch. When intentional flooding began (fig. 4a), most water probably moved downward, saturating the soil and releasing carbon from the soil, thereby increasing interstitial DOC concentrations in the USZ and LSZ with time.

The field was drained in March after 1 month of flooding. During this period, water transported DOC from the field to the ditch (fig. 4b). Following draining, the field was plowed and corn was planted in June. The crop was first irrigated in mid-July, after which irrigation water was applied at various intervals until September 5. After the field was drained and during the cropping/irrigation period (fig. 4c), the near-surface soil went through periods of wetting and drying, resulting in considerable variations in soil moisture. Under conditions of variable soil moisture and relatively high temperatures, it was assumed that microbial activity varied, and reached periods of maximum activity in surface and near-surface soils, which most likely released large amounts of available carbon.

Corn was harvested in mid-October and the field was left fallow (fig. 4d). In the near-surface soils, microbial activity continued to release available carbon from the peat soil, and evaporation increased the levels of soil salinity during this period. Small amounts of precipitation occurred in early December 1996 and large amounts occurred at the end of December and in early January 1997, which caused flooding throughout northern California and the Delta. Final sampling of the field took place in early January. Near-surface soil salinity was high at this time, with specific conductance values as high as 6,780 microsiemens per centimeter (mS/cm) (fig. 5), reflecting dissolution of accumulated salts.

Sampling and Analytical Approach

In August, additional analyses were added to better assess redox conditions and THMFP. Flow-through chamber measurements of dissolved oxygen (DO), pH, and platinum-electrode redox potential (Pt-electrode Eh) were made on the piezometer samples. Ground water was pumped through the airtight, flow-through chamber (fitted with DO, Pt-electrode, and pH probes) from the bottom to exclude any air, thus permitting measurement of these parameters while minimizing the influence of atmospheric gases.

For seven of the samplings, additional water samples were collected to characterize the DOC in more detail (DOC fractionation and analysis of isolated fractions). These samplings included the beginning and end of the intentional flooding period (February 6 and March 11, 1996), after flooding but before irrigation (June 19, 1996), at the beginning and in the middle of the irrigation period (July 17 and August 4, 1996), before the winter rains (November 13, 1996), and during winter rains but before flooding (January 2, 1997). We hypothesized that these periods would provide critical information regarding DOC quality for periods when drainage water production and DOC loads would be greatest and when microbial generation of available carbon in soils would be important.

During previous studies in the Delta, DOC, UVA, and THMFP were measured in an attempt to develop relations to predict or estimate THMFP. SUVA (UVA/DOC) is a spectroscopic measurement that estimates the molecular, aromatic structure of the bulk DOC, normalized to carbon, in a water sample; the aromatic part of DOC is believed to contain the major precursors of THMs (Rook, 1976, 1977; Reckow and others, 1990). However, simple linear correlations among DOC, SUVA, and THMFP (on a molar basis) for drainage and channel waters from throughout the Delta have not resulted in useful predictive relations, even though some correlations were significant for specific regions or units in the Delta (California Department of Water Resources, 1994a). These results indicate that the nature and character of the DOC resulting from drainage in different parts of the Delta vary, influenced by differing conditions throughout the Delta. Sources of carbon (peat vs. recent vegetation) and the relative amounts of organic and mineral soils present in the different regions are important factors that cause the differences in DOC quality and quantity. These factors contribute to the lack of correlation on a regional scale among DOC, SUVA, and THMFP in drainage and channel waters.

The general approach in this study was to examine the chemical variability of the DOC released from the two soil zones and in the drainage ditch over the course of a year for one agricultural field and to relate the chemical character of the DOC to its propensity to form THMs. This approach should help discern the effects of redox conditions and land- and water-management practices on DOC quantity, quality, and reactivity with respect to the formation of THMs. In addition to measuring DOC, UVA, and THMFP of the whole-water samples, DOC for seven selected samples (described previously) were fractionated and isolated using XAD-8 and XAD-4 resins. Amberlite XAD resins are nonionic macroporous copolymers with large surface areas that have been used by many investigators to sorb organic acids such as humic substances (Aiken, 1985). This method divides the DOC into operationally defined organic acid fractions extracted by XAD-8 and XAD-4 resins (fig. 6) (Aiken and others, 1992) and provides valuable information about the types of organic constituents present in the bulk sample.

The part of the DOC extracted by the XAD-8 resin and eluted with base [0.1 molar (M) sodium hydroxide (NaOH)] represents the hydrophobic acid (HPOA) fraction that can contain aliphatic carboxylic acids of five to nine carbons, one- and two-ring aromatic carboxylic acids, one- and two-ring phenols, and other humic substances (Aiken and others, 1992). This fraction contains humic and fulvic acids that contain the more aromatic compounds and are considered the primary reactive component of DOC contributing to the formation of THMs and other disinfection byproducts (DBPs) (Amy and others, 1990; Reckow and others, 1990; Owen and others, 1993). The part of DOC extracted by the XAD-4 resin and eluted with base (0.1 M NaOH) represents the hydrophilic acid (HPIA) fraction that contains polyfunctional organic acids and aliphatic acids with five or fewer carbon atoms (Aiken and others, 1992). This fraction contains fewer aromatic compounds; therefore, it should contain relatively fewer THM and DBP precursors compared to the HPOA fraction.

The DOC was further characterized by examining selected isolates in more detail. The SUVA of HPOA and HPIA isolates provided an initial evaluation of their aromaticity. The THMFPs of the isolates provided a direct assessment of the relative contributions of the HPOA and HPIA fractions to the whole-water THMFP. Liquid- and solid-state ¹³C-NMR analyses provided a quantitative and qualitative assessment of structural and functional group composition of the isolates, and this information--in particular the aromatic carbon content--was used to further assess (on a relative basis) the probable THM precursors contained in each fraction.

Statistical comparisons of differences between lysimeter and piezometer DOC, SUVA, THMFP, specific trihalomethane formation potential (STHMFP), and HPOA and HPIA related parameters used the nonparametric Mann-Whitney test for population medians (Helsel and Hirsch, 1992).

DISSOLVED ORGANIC CARBON CONCENTRTION

Upper Soil Zone

After the field was drained (March 1), lysimeter DOC concentrations increased during March, April, and May, with the median DOC concentration for replicate samples reaching 73.9 mg/L in May (fig. 7). During this period (post-leaching/pre-irrigation) (fig. 4b), the soils became drier and temperatures increased, creating conditions conducive to increased microbial activity and the release of DOC. Results of previous studies of gaseous CO₂ fluxes from peat soils (indicative of microbial activity) on Twitchell Island indicate that fluxes increased with increasing temperature, and the highest fluxes occurred at soil moisture content ranging from about 20 to 30 percent on a volume basis (Deverel and others, 1998). A decrease in soil moisture in June may have resulted in near-surface conditions dry enough (<20 percent by volume) to decrease microbial activity, thereby decreasing available DOC in the USZ.

The first irrigation of the field began on July 13. This field was irrigated using "spud ditches," a common method for irrigation in the Delta. Spud ditches are temporary ditches that are trenched approximately 1-ft wide and 2- to 3-ft deep and run parallel to the length of the field and perpendicular to the ditch draining the field. Water was siphoned from the San Joaquin River into the ditches that convey irrigation water to the southern end of the field where the spud ditches were filled. Water in the spud ditches took about 3 days to reach the drainage ditch at the north end of the field, during which time water flowed laterally from the spud ditches and increased the moisture content of the near-surface soil. The DOC concentrations in the USZ for the July 17 sampling (median DOC of 52.7 mg/L) increased slightly compared to the June sampling (median DOC of 46.4 mg/L) (fig. 7), probably reflecting the initial wetting of the soil and the release of DOC.

Between July 13, when irrigation of the field began, and September 5, when irrigation for the season ended, there were four irrigation periods that resulted in varying degrees of water saturation of the near-surface soils. Decreases in specific conductance of the lysimeter samples during the irrigation period reflect the dilution of soil salinity by the applied irrigation water (fig. 5). The variable moisture content of the near-surface soils caused difficulties in obtaining samples from lysimeters, resulting in the collection of DOC samples from only one lysimeter during both the August 14 and August 21 samplings (fig. 7).

Spatial variability in soil moisture and soil organic matter seems to have contributed to large differences in DOC concentrations between lysimeter samples on each of the samplings in September, October, and November (fig. 7). For each of these sampling dates, only two lysimeters produced enough water for DOC measurements, and the differences between the lysimeter DOC concentrations were large. Lysimeter 1 DOC concentrations increased during this period from 91.7 to a maximum of 128 mg/L, the highest concentration of DOC detected during this study. In contrast, lysimeter 2 yielded DOC concentrations of 42.9 and 41.7 mg/L for the September and October samplings, and lysimeter 4 had a concentration of only 38.1 mg/L in November. These large differences most likely reflect the variability in soil moisture between sites while the field was drying (irrigation was terminated on September 5). Lysimeter 1 is located at the lowest point in the field (the northwest corner), and almost always had the wettest soil conditions. These relatively wet conditions during this period undoubtedly created soil-moisture conditions that favored increased microbial activity and contributed to the much higher DOC concentrations observed for lysimeter 1.

Considerable amounts of precipitation fell during the month of December 1996 and the beginning of January 1997 resulting in flooding throughout the Delta and northern California. Concentrations of DOC in the lysimeter samples (January 2, 1997; fig. 7) were much less variable, most likely reflecting the mixing of near-surface soil water and the torrential rains that occurred just prior to and during the sampling.

Lower Soil Zone

Although median DOC concentrations were similar for many piezometer samples (fig. 10) and lysimeter samples (fig. 7) for the same sampling dates, the nonparametric Mann-Whitney test for population medians (Helsel and Hirsch, 1992) indicated that median DOC concentrations in the LSZ (piezometer samples) (fig. 10) were significantly higher (a=0.05) than those in the USZ (lysimeter samples) (fig. 7). These differences are apparent for periods when the field was intentionally flooded (February) until the field was irrigated (July) and for the winter sampling in January 1997 (figs. 7 and 10). The piezometers were installed to sample mainly ground water, but the ground-water table fluctuated, and the water quality varied in response to water management and flow in the field. Concentrations of DOC in samples from the piezometer during the intentional flooding period indicated a slight increase for piezometer samples from February 6-14 (fig. 10). After the field was drained (March 1), median DOC concentrations in the LSZ increased for the March, April, and May samplings to a high of 82.0 mg/L in May (fig. 10). During this period, the soils above the water table were becoming less saturated, and microbial decomposition of SOM may have contributed additional carbon to the ground water.

After irrigation began (July 13), DOC concentrations in piezometer samples reached their lowest median concentration of 49.3 mg/L on July 17 (fig. 10). Subsurface irrigation through spud ditches apparently allowed water low in DOC (2 mg/L) to rapidly migrate downward through the highly permeable peat soils (J.L. Meyer and A.B. Carlton, University of California, written commun., 1975; Delta organic soil salinity and nutrient status study: Report of laboratory analyses, progress report by the University of California Agricultural Extension to the California Department of Water Resources) and to dilute the ambient ground-water DOC concentrations. The effect of dilution on specific conductance also is apparent in the piezometer samples taken during and after the irrigation period (fig. 11).

For the remainder of the irrigation season (irrigation ceased on September 5) and through the October sampling, DOC concentrations in piezometer samples were variable (fig. 10), reflecting a combination of irrigation-influenced processes. In most cases, median DOC concentrations in lysimeter samples exceeded median DOC concentrations in piezometer samples. In the short term, application of irrigation water with low DOC concentrations tends to decrease DOC in the soil zone, thus influencing the piezometer samples, as discussed previously. During the longer term irrigation period, irrigation cycles cause wetting and drying of soils above the water table thus creating variable conditions for microbial decomposition of SOM and the release and transport of available carbon. These complex hydrologic and microbially related processes both act to produce the high variability in DOC concentrations observed during this period.

In contrast, piezometer samples collected during November and January were much less variable (fig. 10). This decreased variability in DOC following the irrigation season probably reflects the lack of irrigation-induced alternating wetting and drying cycles and the associated effect on DOC concentrations.

Water flux and DOC transport data would have aided interpretation of DOC concentration trends in the USZ and LSZ and allowed estimates of DOC loading to the drainage ditch. This, however, was beyond the scope of this investigation.

Drainage Ditch

COMPOSITION AND REACTIVITY OF DISSOLVED ORGANIC CARBON

The propensity for DOC to form THMs, as measured by the THMFP of the samples, is examined for whole-water samples in relation to the UVA (aromaticity) of the bulk DOC. The THMFPs of selected DOC isolates are compared to isolate properties, as determined by SUVA and ¹³C-NMR, to assess reactivity of HPOA and HPIA fractions in relation to their composition and source and to factors affecting DOC production under the conditions studied.

Specific Ultraviolet Absorbance

Fractionation of Dissolved Organic Carbon into Hydrophobic and Hydrophilic Acids

Complete DOC fractionation data for lysimeter, piezometer, and drainage ditch samples are presented in appendix C, table C1. For purposes of this discussion, averaged data are used when replicate analyses were available. The discussion focuses only on data for the HPOA and HPIA fractions, the DOC extracted by and eluted from the XAD-8 and XAD-4 resins, respectively. These two fractions combined account for the majority of DOC in all samples (58 to 76 percent), probably contain most of the THM precursors, and are the fractions for which other compositional and structural data are available.

In general, the fractionation data (fig. 14) indicate that (1) the sum of the HPOA and HPIA DOC fractions are significantly greater for the piezometer samples compared to the lysimeter samples, indicating that the lysimeter samples contained more DOC that was not sorbed by the resins (probably ultra hydrophilic acids); (2) the quality of the ditch samples was very similar to that of the piezometer samples, reflecting a potentially greater DOC contribution to the ditch from the LSZ; and (3) the HPOA fraction for the piezometers was significantly greater than that for the lysimeters (a=0.05), indicating that greater amounts of humic substances were produced from the LSZ. This result is in agreement with the SUVA data for the whole-water samples that had significantly higher values of SUVA for the piezometer samples compared to the lysimeter samples. Together, these results support the conclusion that more aromatic forms of DOC are produced under anaerobic conditions (piezometer samples) than under aerobic conditions (lysimeter samples) and imply that more THM precursors should be produced by the LSZ under anaerobic conditions.

Trihalomethane Formation Potential

Samples collected from the beginning of the study through September 1996 were analyzed by the dose-based THMFP method only, samples collected in October 1996 were analyzed by the reactivity-based THMFP method only, and 22 samples collected from November 1996 through January 1997 were analyzed using both methods. Dose-based and reactivity-based THMFP results for samples analyzed by both methods were significantly correlated, R²=0.983 (fig. 15), and indicate that the dose-based THMFP results consistently are about 11 percent higher than the
reactivity-based THMFP results. Because results of both methods are highly correlated (indicating no dilution effect problem for the dose-based method for these samples) and the data record for the dose-based THMFP determinations is much longer, the results and discussion below for whole-water sample THMFP utilize the dose-based THMFP data. The exceptions to this are the THMFPs for samples collected on October 23, 1996, which were analyzed by using only the reactivity-based method.

For samples collected from the USZ (lysimeter), from the LSZ (piezometer), and from the drainage ditch, the linear relation between DOC and THMFP (R²=0.864) (fig. 16) is excellent. The variance about the regression line in figure 16 indicates the variability in DOC quality and composition in relation to THM precursors. This high correlation most likely is due to the predominance of peat as the major source of DOC. In other systems that are not as homogeneous (for example, systems that contain more diverse sources of DOC or higher amounts of mineral soil), such a high correlation would not be expected. For example, Owen and others (1993) examined seven different source waters from throughout the United States and found poor correlation between DOC and THMFP (R²<0.50).

Although the correlation is high for samples from this study, use of DOC to predict THMFP could lead to considerable error in estimating THMFP, especially at higher DOC concentrations. For example, the regression equation predicts a THMFP of 67 micromolar (µM) (8,200 µg/L) for a sample containing about 85 mg/L DOC. For this DOC concentration, actual THMFP concentrations range from about 57 µM (7,000 µg/L) to 78 µM (9,500 µg/L) (fig. 16). Thus, use of this relation to predict THM precursor loading, for example, could lead to significant errors in the load prediction. In addition, high correlation between DOC and THMFP is not expected for samples from less homogeneous areas that contain diverse sources of DOC.

The linear relation between UVA and THMFP (fig. 17, R²=0.702) for the lysimeter, the piezometer, and the drainage ditch samples also is good. Because UVA is an indicator of DOC aromaticity and aromatic forms of DOC are considered primary THM precursors (Rook, 1976, 1977; Reckow and others, 1990), the higher correlation between THMFP and DOC (R²=0.864), compared to that for THMFP and UVA, was not expected. The THMFP and UVA data normalized to carbon were not linearly correlated (R²=0.156) (fig. 18). Specific THMFP (THMFP/DOC or STHMFP) provides an indication of the average potential for the carbon in a sample to form THMs, a measure of the potential THM precursor content on a molar basis normalized to carbon. The generally accepted model for THM formation is that the primary THM precursors are aromatic forms of carbon (such as resorcinol), in which case a strong linear relation is expected between STHMFP and SUVA. The lack of correlation between STHMFP and SUVA suggests that a more detailed assessment of aromatic compound species may help to better identify THM precursor compounds, and that forms of DOC other than aromatic compounds also may be significant THM precursors in these samples.

Upper and Lower Soil Zones

The THMFP results for the lysimeter (fig. 19) and the piezometer (fig. 20) samples generally followed trends similar to the lysimeter and the piezometer DOC results (figs. 7 and 10) over the course of this study. This similarity also is reflected by the correlation between DOC and THMFP (R²=0.864) (fig. 16). The same processes affecting the release of DOC from the USZ and the LSZ, as discussed in the Dissolved Organic Carbon Concentrations section of this report, also affect the release of THM precursors.

No significant difference (a=0.05) was observed between the median concentrations of THMFP for the lysimeter (5,950 µg/L) and the piezometer (6,750 µg/L) samples. The STHMFP medians for lysimeters [9.02 micromolars per millimolar (µM/mM)] and piezometers (9.84 µM/mM) also were not significantly different (a=0.05). This result is in contrast to DOC, for which concentrations of piezometer DOC were significantly greater than lysimeter DOC. The smaller number of lysimeter samples (17) analyzed for THMFP, relative to the piezometer samples (42), may not accurately represent the seasonal variability of USZ water. The lysimeter samples analyzed for THMFP were mainly from wetter periods during the year (winter flooding and irrigation) when the USZ is relatively saturated, and results may have been influenced by near-saturated soil moisture conditions that are similar to those in the LSZ.

Trihalomethane Formation Potential of Isolated Fractions

Many investigators have shown that THMFP measurements are highly pH dependent (Rook, 1974; Symons and others, 1975; Rathbun, 1996), THMFP concentration increasing with increasing pH. Differences in temperature also will affect THMFP, with more volatile THM species being produced at the higher temperature. Thus, the effects of differences in these two variables help to explain the differences in THMFP results in figure 21 for the two methods.

In this section of the report, results from the modified reactivity-based THMFP method (pH 7.0, 20°C) for whole-water samples are used for comparison purposes because the DOC-isolate samples were analyzed using only this method. Although results using this modified method differ from results for the other reactivity-based method (pH 8.2, 25°lC) and the dose-based THMFP method, THMFP comparisons between whole-water samples and DOC-isolate samples require using data obtained by the same method.

As described in appendix D, DOC isolates (XAD-8 and XAD-4) were redissolved and the resulting solutions analyzed for THMFP and SUVA. Therefore, STHMFP was used to compare the whole-water and the isolate data. Figure 22 shows that the STHMFP of the HPOA fraction was greater than that for the HPIA fraction for five of the seven samples analyzed. But the median STHMFP value for the HPOA fractions [43.8 micrograms per milligram (mg/mg)] was not significantly greater (a=0.05) than that for the HPIA fractions (41.7 µg/mg). This result was not expected because it is generally thought that the humic fraction (HPOA) contains most of the THM and DBP precursors (Owen and others, 1993). For these samples, almost equivalent amounts of STHMFP were observed for both humic (XAD-8) and nonhumic (XAD-4) fractions. Owen and others (1993) observed similar results for seven source waters from throughout the United States. They found that the reactivity of the nonhumic and humic fractions were similar and state that this result "is somewhat contrary to conventional wisdom ... namely that it is the humic fraction that serves as DBP precursor material."

The SUVA of the HPOA fractions is significantly higher (a=0.05) than that for the HPIA fractions (fig. 23), indicating greater aromaticity of this DOC fraction. The greater aromaticity of the HPOA fraction compared to the HPIA fraction is apparent in figure 24, where the ¹³C-NMR data clearly indicate the greater aromatic composition of the HPOA isolates compared to the HPIA isolates. The ¹³C-NMR data provide solid evidence that supports the use of SUVA as an indicator of DOC aromaticity for samples in this study.

The relatively small differences in STHMFP between the HPOA and HPIA fractions (fig. 22) did not reflect the significantly greater aromaticity of the HPOA fraction over that of the HPIA fraction (fig. 24). It is generally thought that the HPOA fraction contains most of the aromatic forms of carbon, which is supported by data in figure 22, and that aromatic forms of carbon are the primary THM precursors (Rook, 1976, 1977; Reckhow and others, 1990). Even though the HPOA fraction was higher in aromatic composition (fig. 24) and carbon (fig. 14) compared to the HPIA fraction, the THMFP concentration contributed by the HPOA fraction was not significantly greater than that contributed by the HPIA fraction (fig. 22). This result indicates that DOC aromaticity alone cannot fully explain or predict THM precursor content.

SUMMARY AND CONCLUSIONS

The generally accepted conceptual model for THM formation assumes that aromatic forms of carbon (such as resorcinol) are primary precursors to THMs. Natural environments with reducing conditions, such as peat bogs and water-logged soils, tend to produce greater amounts of aromatic DOC compared to well-oxygenated environments. Thus, one of the principal hypotheses of this study was that the reduced peat soils beneath the shallow ground water on Delta islands would have a tendency to release greater amounts of aromatic carbon (THM precursors) relative to near-surface oxidized peat soils. To test this hypothesis, soil water was sampled from near-surface, oxidized, well-decomposed peat soil in the upper soil zone (USZ) and deeper, reduced, fibrous peat soil in the lower soil zone (LSZ) from one agricultural field in the west-central Delta. Soil redox conditions also are influenced by water-management practices--irrigation and intentional flooding of fields and drainage ditch depths that help control ground-water levels. Therefore, results from this study have implications for potential management alternatives to reduce production of THM precursors in Delta drainage water.

The general approach of the study was to examine the chemical variability of DOC samples from the USZ and the LSZ during a 1-year period. The chemical character of the DOC was related to its propensity to form THMs while taking into consideration the effects of redox conditions and land- and water-management practices on the biogeochemical processes affecting the release of DOC from the soils.

The analytical approach to the study involved analysis of whole-water samples for DOC, ultraviolet absorbance (UVA) at 254 nm, trihalomethane formation potential (THMFP), and various inorganic constituents. Considerable focus was given to the aromaticity of carbon in whole-water samples and isolated carbon fractions as indicators of THM precursor content. The quality of the DOC was evaluated by first examining the whole-water specific UVA (SUVA, UVA/DOC) as an indication of DOC aromaticity. Whole-water DOC was isolated and fractionated into hydrophobic acid (HPOA) using XAD-8 resin and hydrophilic acid (HPIA) using XAD-4 resin, where the HPOA fraction contains the more aromatic humic and fulvic acids and the HPIA fraction contains the less aromatic and more aliphatic forms of carbon. The amount of DOC isolated on these columns, the proportion of HPOA and HPIA, as well as analysis of the intrinsic chemical structure of the isolates by carbon-13 nuclear magnetic resonance (¹³C-NMR), all provided insight into the processes controlling the aromaticity of a sample. Selected isolates from XAD fractionations were further analyzed for THMFP and related to SUVA values of the isolates and to the more quantitative ¹³C-NMR measure of aromatic carbon, which also provided further DOC structural and functional group information. Isolate THMFPs were interpreted in terms of isolate properties (SUVA and ¹³C-NMR results) to assess the reactivity of HPOA and HPIA fractions in relation to their composition, source, and factors affecting DOC production under the conditions studied.

Conclusions

• Waters from the anaerobic LSZ contained slightly higher concentrations of DOC and greater amounts of humic material than waters from the aerobic USZ, indicating that carbon released from peat soils under anaerobic conditions is a substantial contributor to DOC and that the quality of the DOC released from the two zones differs.
• DOC in waters from the LSZ contained a higher proportion of aromatic compounds, the putative THM precursors, than that from the USZ.
• DOC in waters from the drainage ditch were compositionally more similar to the DOC in waters from the LSZ than that from the USZ, suggesting the major source of DOC in ditch water was the LSZ.
• DOC aromaticity alone cannot explain fully or predict THM precursor content. This finding indicates that:
  1. UVA is not a suitable tool for predicting THMFP in these waters because this measurement is most sensitive for aromatic carbon.
  2. Processes other than those that control bulk aromatic carbon content control THM precursor concentrations in these waters, which explains the historical observation that on a regional basis, there is a poor correlation between UVA and THMFP on a carbon-normalized basis.
• The highest DOC levels, the highest variability in DOC, and the highest THMFP levels followed summer irrigation and winter flooding periods, which suggests that repetitive cycles of wetting and drying promotes the release of DOC and THM precursors in these soils.

Summary of Supporting Observations

Dissolved Organic Carbon Concentrations

• DOC concentrations in the oxidized USZ were highly variable. For the oxidized USZ, median DOC for lysimeter samples ranged from 46.4 to 83.2 mg/L. Variations in DOC are attributable to effects of water-management practices (flooding and irrigation) and precipitation on microbial soil processes and subsequent release and transport of DOC.
• DOC concentrations in the reduced LSZ were generally slightly higher than that from the USZ. Median DOC for the reduced LSZ (piezometer samples) ranged from 49.3 to 82.3 mg/L and were significantly higher (a=0.05) than DOC from the USZ. In general (1) variations in piezometer DOC, at times, apparently were caused by drainage from soils above; and (2) the lowest and most variable concentrations of DOC are associated with the irrigation period, during which the drop in specific conductance of samples clearly indicated dilution of ground water by irrigation water.
• DOC concentrations in the ditch draining the field were always lower than the median DOC for either USZ or LSZ waters for all sampling dates, ranging from 9.8 to 54.9 mg/L. Lower values are consistent with mixing of waters from several different sources, including lower DOC irrigation waters. Interpretation of ditch DOC is complicated by the multiple sources of water and DOC that contribute to the drainage ditch throughout the year.

Quality and Composition of Dissolved Organic Carbon

• The DOC from the LSZ had significantly higher (a=0.05) aromaticity than the USZ as measured by SUVA. This indicates that the deeper, more reduced fibrous peat releases more aromatic forms of carbon compared to the near-surface, oxidized decomposed peat.
• Isolation of DOC as HPOA (XAD-8) and HPIA (XAD-4) fractions accounted for 58 to 76 percent of the total DOC, demonstrating that our analytical scheme accounted for the majority of DOC in all samples.
• The sum of the HPOA and HPIA DOC fractions were significantly greater (a=0.05) for piezometer samples compared to lysimeter samples, indicating that the lysimeter samples contained more forms of DOC not retained and eluted by the resins (probably ultra-hydrophilic acids).
• Water from the LSZ contained more humic materials than that from the USZ. Piezometer samples had greater HPOA fractions (humic substances) compared to those in lysimeter samples, were richer in the more aromatic HPOA, and were in agreement with SUVA results for whole-water samples. These results support the conclusion that more aromatic forms of DOC are produced under anaerobic conditions compared to aerobic conditions.
• The composition of the drainage ditch samples, as indicated by the HPOA and HPIA distributions, closely resembled that of the piezometer samples, potentially reflecting a greater DOC contribution to the ditch from the LSZ.
• DOC isolated in the HPOA fraction was more aromatic than that isolated in the HPIA fraction. For the seven DOC samples isolated and fractionated, the SUVA of the HPOA isolates was significantly higher (a=0.05) than the SUVA for the HPIA isolates, denoting greater aromaticity of the HPOA fractions relative to the HPIA fractions. This difference is explicitly demonstrated by the 13C-NMR data, which clearly show greater aromatic carbon composition of the HPOA isolates in comparison to that of the HPIA isolates.

Trihalomethane Formation

• DOC concentration, THMFP, and UVA were all highly related. Linear correlations were found between THMFP and DOC (R2=0.864) and THMFP and UVA (R2=0.702) for samples from lysimeters, piezometers, and the ditch, indicating that THM precursors increased with increasing DOC and UVA for whole-water samples.
• In contrast, no significant correlation was found between STHMFP (THMFP/DOC) and SUVA, indicating that no significant relation exists between carbon aromaticity and THMFP on a carbon basis. This result suggests that a more detailed assessment of aromatic compound species may help to better identify THM precursor compounds and that forms of DOC other than aromatic compounds also may be significant THM precursors in these samples.
• There is no consistent difference in the capacity of USZ and LSZ waters to form THMs. Even though median values of THMFP and STHMFP were greater for piezometer samples (6,750 mg/L and 9.84 mM/mM, respectively) compared to those for lysimeter samples (5,950 mg/L and 9.02 mM/mM, respectively), the differences were not significant (a=0.05). A smaller number of lysimeter samples (17) were analyzed relative to piezometer samples (42) because of lack of available water in the USZ during dry parts of the year when the field was not irrigated. Thus, results from the lysimeter samples may not be representative of THM precursor release from the USZ throughout the year.
• The aromaticity of DOC appears to be unrelated to THMFP on a carbon-normalized basis. Although the HPOA fractions were obviously more aromatic than the HPIA fractions, the median STHMFP for the HPOA isolates (43.8 mg/mg) was not significantly greater (a=0.05) than the median for the HPIA isolates (41.7 mg/mg). This result again emphasizes that DOC aromaticity alone cannot explain fully or predict THM precursor content and that further investigation of aromatic and nonaromatic forms of carbon will be needed to better identify THM precursors.

Other Significant Observations

• The anaerobic redox condition of the LSZ was characterized by extremely low dissolved oxygen (<0.38 mg/L) and platinum-electrode redox potentials (<310 mV) indicative of anoxic conditions.
• Twenty-one water samples analyzed for THMFP using the dose-based method and the reactivity-based method were linearly correlated (R²=0.983), with the dose-based results consistently 11 percent greater than the reactivity-based results, indicating no dilution effect problem for the dose-based method for these samples.

Implications of Study Results

The DOC and aromaticity of LSZ samples generally were greater than those of the USZ samples, but no significant differences were found for THMFP results for samples from the two zones over the annual cycle. Data from this study suggest that aromaticity alone is not an accurate indicator of THM precursors. This result implies that THM precursors do not arise from the same source as most of the aromaticity; thus, a more accurate indicator of THMFP than UVA is needed for screening water quality in the Delta.

Another implication is that the processes that concentrate THM precursors are not significantly affected by redox conditions and appear to be unrelated to the processes that concentrate aromatic DOC in the LSZ. This finding explains why there is no observed general regional relation between SUVA and THMFP in Delta waters. This lack of relation is likely related to the type of organic material released by peat soils, and care should be exercised in extrapolating this finding to other locations. Also, this finding is based on concentration measurements and may not be true for the total amount (loads) of precursors released from the soils. However, if studies confirm the finding on a regional basis, it may be possible to directly identify the source of the THM precursors and to seek a remedy for high THM precursor concentrations in Delta waters.

REFERENCES CITED

Aiken, G.R., 1985, Isolation and concentration techniques for aquatic humic substances, in Aiken, G.R., McKnight, D.M., Wershaw, R.L., and MacCarthy, P., (eds.), Humic substances in soil, sediment, and water--Geochemistry, isolation, and characterization: New York, John Wiley & Sons, p. 363-385.

Aiken, G.R., and Cotsaris, E., 1995, Soil and hydrology: Their effects on NOM: Journal of American Water Works Association, January 1995, p. 36-45.

Aiken, G.R., McKnight, D.M., Thorn, K.A., and Thurman, E.M., 1992, Isolation of hydrophilic organic acids from water using nonionic macroporous resins: Organic Geochemistry, v. 18, no. 4, p. 567-573.

Amy, G.L., Thompson, J.M., Tan, L., Davis, M.K., and Drasner, S.W., 1990, Evaluation of THM precursor contributions from agricultural drains: Research and Technology, v. 82, no. 1, p. 57-64.

Atwater, B.F., 1980, Attempts to correlate Late Quaternary climatic records between San Francisco Bay, the Sacramento-San Joaquin Delta, and the Mokelumne River, California: Dover, University of Delaware, Ph.D dissertation (Dissertation Abstract 80-19923), 214 p.

Atwater, B.F., Hedel, C.W., and Helley, E.J., 1977, Late Quaternary depositional theory, Holocene sea-level changes, and vertical crustal movement, southern San Francisco Bay, California: U.S. Geological Survey Professional Paper 1014, 15 p.

Bohn, H.L., McNeal, B.L., and O'Connor, G.A., 1985, Soil chemistry (2nd ed.): New York, John Wiley & Sons, 341 p.

California Department of Water Resources, 1990, Delta island drainage investigation report, water years 1987 and 1988--Interagency Delta Health Aspects Monitoring Program: California Department of Water Resources, 110 p.

----, 1993, Sacramento-San Joaquin Delta atlas: California Department of Water Resources, 121 p.

----, 1994a, Five-year report of the municipal water quality investigations program: Summary and findings during five dry years, January 1987--December 1991: California Department of Water Resources, 119 p.

----, 1994b, California Department of Water Resources Bryte Laboratory's quality assurance and quality control manual: California Department of Water Resources, 63 p.

Cook, R.L., Langford, C.H., Yamdagmi, R., and Preston, C.M., 1996, Modified cross polarization magic angle spinning 13C-NMR procedure for the study of humic materials: Analytical Chemistry, v. 68, p. 3979-3986.

Deverel, S.J., and Rojstaczer, Stuart, 1996, Subsidence of agricultural lands in the Sacramento-San Joaquin Delta, California: Role of aqueous and gaseous carbon fluxes: Water Resources Research, v. 32, no. 8, p. 2359-2367.

Deverel, S.J., Wang, B., and Rojstaczer, Stuart, 1998, Subsidence of organic soils, Sacramento-San Joaquin Delta, California, in Borchers, J.W., (ed.), Land Subsidence Histories and Current Research: Proceedings of the Dr. Joseph F. Poland Symposium, Association of Engineering Geologist Special Publication No. 8: Belmont, Star Publishing Co.

Eaton, A.D., Clesceri, L.S., and Greenberg, A.E., eds., 1995, Standard methods for the examination of water and wastewater (19th ed.): Washington, DC, American Public Health Association, American Water Works Association, Water Environment Federation, various pagination.

HACH, 1992, HACH water analysis handbook (2nd ed.), HACH Company, 831 p.

Helsel, D.R., and Hirsch, R.M., 1992, Statistical methods in water resources: Studies in Environmental Science: Amsterdam, Elsevier, v. 49, 522 p.

Krasner, S.W., and Sclimenti, M.J., 1993, Characterization of natural organic matter: Disinfection byproduct analysis, in Natural Organic Matter Workshop, Chamonix, France, American Water Works Association Research Foundation, p. 9.

Means, E.G., III, and Krasner, S.W., 1993, D-DBP regulations: Issues and ramifications: Journal of American Water Works Association, February 1993, p. 68-73.

Owen, D.M., Amy, G.L., and Chowdhury, Z.K., 1993, Characterization of natural organic matter and its relationship to treatability: Denver, American Water Works Association Research Foundation and American Water Works Association, 250 p.

Pontius, F.W., 1991, Disinfectant-disinfection byproducts rule update: Journal of American Water Works Association, December 1991, p. 24-26, 113-115.

Preston, C.M., 1987, Review of solution NMR of humic substances, in Wershaw, R.L., and Mikita, M.A. (eds.), NMR of humic substances and coal: Lewis Publishers, Inc., Chelsea, Michigan, p. 3-32.

Rathbun, R.E., 1996, Trihalomethane and nonpurgeable total organic-halide formation potentials of the Mississippi River: Archives of Environmental Contamination and Toxicology, v. 30, p. 156-162.

Reckhow, D.A., and Edzwald, J.K., 1989, Development and laboratory evaluation of the haloform formation potential tests: Research Report, Subject Area: Water Quality and Analysis: Denver, American Water Works Research Foundation and American Water Works Association, 176 p.

Reckhow, D.A., Singer, P.C., and Malcolm, R.L., 1990, Chlorination of humic materials: Byproduct formation and chemical interpretations: Environmental Science and Technology, v. 24, p. 1655-1664.

Rojstaczer, Stuart, and Deverel, S.J., 1993, Time dependence in atmospheric carbon inputs from drainage of organic soils: Geophysical Research Letters, v. 20, no. 13, p. 1383-1386.

----, 1995, Land subsidence in drained histosols and highly organic mineral soils of California: Soil Science Society of America Journal, v. 59, p. 1162-1167.

Rook, J.J., 1974, Formation of haloforms during chlorination of natural waters: Water Treatment and Examination, v. 23 (part 2), p. 234.

----, 1976, Haloforms in drinking water: Journal of American Water Works Association, v. 68, p. 168.

----, 1977, Chlorination reactions of fulvic acids in natural waters: Environmental Science and Technology, v. 11, no. 5, p. 478.

Snoeyink, V.L., and Jenkins, D., 1980, Water chemistry: New York, John Wiley & Sons, 463 p.

Steelink, C, Wershaw, R.L., Thorn, K.A., and Wilson, M.A., 1990, Application of liquid state NMR spectroscopy to humic substances, in Hayes, M.H.B., MacCarthy, P. Malcolm, R.L., and Swift, R.S. (eds.) Humic substances II: In search of structure: John Wiley & Sons, New York, p. 281-308.

Stevenson, F.J., 1985, Geochemistry of soil humic substances, in Aiken, G.R., McKnight, D.M., Wershaw, R.L., and MacCarthy, P., (eds.), Humic substances in soil, sediment, and water: New York, John Wiley & Sons, p. 13-52.

----, 1994, Humus Chemistry: Genesis, composition, reactions (2nd ed.): New York, John Wiley, 496 p.

Symons, J.M., Bellar, T.A., Carswell, J.K., DeMarco, Jack, Kropp, K.L., Robeck, G.G., Seeger, D.R., Slocum, C.J., Smith, B.L., and Stevens, A.A., 1975, National organics reconnaissance survey for halogenated organics: Journal of American Water Works Association, v. 67, p. 634-647.

Tate, R.L., III, 1987, Soil organic matter: Biological and ecological effects: New York, John Wiley & Sons, 291 p.

Templin, W.E., and Cherry, D.E., 1997, Drainage-return, surface-water withdrawal, and land-use data for the Sacramento-San Joaquin Delta, with emphasis on Twitchell Island, California: U.S. Geological Survey Open-File Report 97-350, 31 p.

Thurman, E.M., 1985, Organic geochemistry of natural waters: Developments in biogeochemistry: Dordrecht, Martinus Nijhoff/Dr. W. Junk, 497 p.

U.S. Environmental Protection Agency, 1994, Drinking water regulations and health advisories, 822-R-94-001, 10 p.

U.S. Geological Survey, 1980, Groundwater, Chapter 2 of National handbook of recommended methods for water-data acquisition: U.S. Geological Survey, Office of Water-Data Coordination, p. 2-1 to 2-149.

Wershaw, R.L., 1985, Application of nuclear magnetic resonance spectroscopy for determining functionality in humic substances, in Aiken, G.R., McKnight, D.M., Wershaw, R.L., and MacCarthy, P., (eds.), Humic substances in soil, sediment, and water: Geochemistry, isolation, and characterization: New York, John Wiley & Sons, p. 561-582.

Wershaw, R.L., and Makita, M.A., 1987, NMR of humic substances and coal: Lewis Publishers, Inc., Chelsea, Michigan, 236 p.

URL: http://ca.water.usgs.gov/rep/wrir984147/
Contact: webmaster@maildcascr.wr.usgs.gov