Introduction


The Clackamas River in northwestern Oregon (fig. 1) is a valued resource to the region, supporting runs of wild steelhead and salmon and providing drinking water for nearly 400,000 people. From its headwaters near Olallie Butte south of Mount Hood, the Cascade River descends from the High Cascades flowing northwest for 82 mi to reach its confluence with the Willamette River southeast of Portland. Although 72 percent of the 940-mi² watershed is contained within the Mount Hood National Forest (Metro Regional Services, 1997), the lower third of the watershed drains private forests and agricultural, urban, and light industrial land that variously contribute sediments, nutrients, pesticides, and other organic compounds to the Clackamas River (Carpenter, 2003; Carpenter and others, 2008; Carpenter and McGhee, 2009). Although the river, for the most part, is exceptionally clear, it sometimes becomes turbid with sediment and organic matter from storm runoff that degrades the quality of source water at the drinking-water intakes in the lower river. 


Two of the four drinking-water treatment plants (DWTPs) in the lower river—the Clackamas River Water (CRW) and the City of Lake Oswego (LO) DWTPs—use direct filtration as a means to clarify raw source water (fig. 2). Together, these two plants serve about 100,000 people. Both water utilities use chlorination simultaneously with coagulation as part of the water-treatment process. The use of chlorine as a disinfectant, although essential for pathogen control, leads to the halogenation of organic matter present in source water (Croué and others, 1999). Halogenated (chlorine- and bromine-containing) compounds form from dissolved and particulate organic carbon during water treatment and are collectively referred to as disinfection by-products (DBPs). Although only a small fraction of the organic carbon present in source water reacts to form DBPs, several DBPs have been identified as mutagenic and carcinogenic (Krasner and others, 2006; Richardson and others, 2007). For this reason, the USEPA currently regulates two classes of DBPs commonly found in drinking water—trihalomethanes (THMs) and haloacetic acids (HAAs) (U.S. Environmental Protection Agency, 2009). In addition to being a source of DBPs, organic carbon contributes to biofouling, increases chlorine demand, and can affect aesthetic qualities of water such as taste, odor, and color (Cooke and Kennedy, 2001). 


Water managers are concerned about DBPs in drinking water and are interested in identifying the types of organic carbon that contribute DBP precursors in source water to better understand the potential for future deterioration in river quality resulting from a wide array of possible sources (fig. 3). Understanding the timing, sources, and composition of organic matter entering drinking-water intakes will help drinking-water utilities develop source-water-protection programs, facilitate successful and cost-effective treatment strategies, and help plan for future upgrades to treatment plants (Kraus and others, 2010). 


Although much of the carbon in the watershed is contained in its forests and soils, the Clackamas River, along much of its length, is gravel bedded and provides appreciable habitat for benthic algae (periphyton) to attach and grow (see photographs 1a-e). This material contains organic carbon that may decompose to yield DBP precursors (Jack and others, 2002; Huang and others, 2009; Kraus and others, 2011). Periphyton growths occur in the Clackamas River during periods when nutrients, light, and flow conditions are favorable. At times, periphyton biomass reaches nuisance levels along river margins and causes supersaturated concentrations of dissolved oxygen (DO), high (alkaline) pH, and large daily fluctuations in pH and DO. The pH in the lower river is particularly high, and regularly exceeds the State of Oregon water-quality standard during parts of the growing season, particularly in spring (U.S. Geological Survey, 2012). Periphyton biomass is typically higher in the lower river downstream from Estacada, but nuisance biomass levels also occur in the upper river, upstream from North Fork Reservoir (fig. 1). A previous U.S. Geological Survey (USGS) study (Carpenter, 2003) found nuisance levels of benthic algae in the main-stem Clackamas River during the summer, and continuous monitoring of DO and pH—definitive algal-photosynthesis indicators—shows this problem has continued on and off for at least the past decade.


When periphyton detaches from the riverbed and becomes entrained in the flow during algal “sloughing” events, the cells and stalks of diatoms, filaments of green algae, and colonies of blue-green algae (Cyanobacteria) in varying states of decomposition enrich the river with organic carbon. This algae has the capacity to clog water intakes and negatively affect drinking-water quality through the production of tastes and odors and algal toxins (Graham and others, 2010). Algae also contribute carbon that contains DBP precursors (Graham and others, 1998; Jack and others, 2002; Kraus and others, 2011). Phytoplankton (floating algae) also occasionally form blooms during summer in the two primary reservoirs, Timothy Lake in the headwaters of the Oak Grove Fork and North Fork Reservoir on the mainstem (see photographs 2a-c), which can also contribute DBP precursors (Kraus and others, 2011).


Decomposition products of terrestrial plant material, including leaves from deciduous trees, conifer needles, and other plant material contained in soils are certainly a source of organic carbon in the Clackamas River. High rainfall leads to saturated conditions and flow of water through organic-rich surface soils. High rainfall combined with steep topography in much of the basin causes erosion, and landslides are particularly common during large storms. The epic February 1996 rain-on-snow event, for example, produced over 200 landslides in the Fish Creek Basin alone (DeRoo and others, 1998), and much of that material deposited in the main­stem Clackamas River near the upstream end of North Fork Reservoir. Decomposing vegetation buried in that debris could be another source of DBP precursors, along with wastewater from three municipal treatment plants, septic-tank effluents, and other potential sources.


Disinfection By-Products and Historical Trends in Disinfection By-Product Concentrations


DBPs are known to be carcinogenic and are indicated to cause increased risks of reproductive and developmental problems in humans (Richardson and others, 2007). For these reasons, the USEPA regulates the total concentration of four THMs (THM4) and five HAAs (HAA5) in finished (treated) drinking water (U.S. Environmental Protection Agency, 2006). In 2004–05, as part of the USGS National Water-Quality Assessment Program, source and finished water from the CRW DWTP were sampled for a variety of organic compounds, including THMs. Compounds including pesticides and gasoline hydrocarbons were commonly detected in finished water; relative to regulatory standards, however, THMs, were the compound class of most concern (Carpenter and McGhee, 2009). Even though THM4 concentrations were always below current regulatory thresholds, THM4 concentrations may be higher when organic carbon is elevated in source water such as during periods of active rainfall runoff, algal blooms, or periphyton sloughing events—which were not targeted for sampling during that study.


Compliance monitoring data from the CRW and LO DWTPs show THMs in finished drinking water have increased to some degree over the past 20 years (fig. 4). The cause of this increase is not known, and data to evaluate potential causes, such as trends in historic total organic carbon (TOC) concentrations, are not available. Historically, the two highest THM4 concentrations were measured in mid-to-late June 1997 and 2000. It is possible these high THM4 concentrations were caused by sloughed benthic algae because this is a time of year when sloughing events have occurred (as in 2004 and 2005) (U.S. Geological Survey, 2012). 


To better understand this increase in DBP concentrations, basin-specific information is needed regarding the sources of carbon that form the DBPs and factors that contribute to carbon losses from watersheds so that management strategies can be developed and successfully targeted. Information on the type of carbon present can provide insights into the chemical reactions that take place during chlorination that might shape water-treatment strategies to control DBPs. This could be especially important if the types of carbon that contribute to DBPs increase in source water or if the USEPA regulations become more restrictive.


Sources of Organic Matter and Disinfection By-Product Precursors


The DBP precursor pool is a subset of the bulk organic‑matter pool present in source water. Possible sources of organic matter in the Clackamas River basin are shown in figure 3. Previous studies have shown terrestrial plants and soils (allochthonous sources) and algae and macrophytes (autochthonous sources) contribute organic matter and DBP precursors to surface waters (Aiken and Cotsaris, 1995; Reckhow and others, 2004). As described previously, algae are a possible source of organic carbon in the Clackamas River. While less is known about the propensity for periphyton to form DPBs, phytoplankton is a well-known source of dissolved organic carbon (DOC) and DBP precursors, especially HAAs (Jack and others, 2002; Nguyen and others, 2005; Kraus and others, 2011).


The amount and reactivity of organic matter entering a DWTP is a function of the amount and composition of material entering the water throughout the watershed, as well as environmental processes such as biodegradation, photodegradation, sedimentation, and sorption that may take place during transport through the river system. The types of DBPs that form are controlled by the physiochemical properties of the carbon molecules and the complex reactions that occur with disinfectants such as chlorine, along with coagulation treatment, pH, temperature, bromide concentration, and other factors (Crepeau and others, 2004). 


The amount of organic matter in a water sample is typically determined by measuring carbon concentration, assuming that half the organic matter pool is made up of carbon. TOC is commonly characterized by laboratory measurements of whole (unfiltered) water; however, this method has a tendency to under-report carbon concentrations (Aiken and others, 2002). In this report, TOC was derived by summing laboratory measurements of DOC and total particulate carbon (TPC). In addition to these concentration-based constituents, the composition of dissolved organic matter (DOM) was characterized using absorbance and fluorescence spectrophotometry, and continuous in-situ fluorescence was used as an indicator of DOC concentration.


Use of Optical Properties to Characterize Dissolved Organic Matter


Spectral optical property measurements such as absorbance and fluorescence can be used to determine the amount of DOC in water and to broadly characterize dissolved organic matter (DOM) composition (Hudson and others, 2007; Fellman and others, 2010; Matilainen and others, 2011). Shifts in the spectral response of optical properties can help identify sources of carbon within watersheds and inform watershed management (Kraus and others, 2010; Beggs and others, 2011; Bridgeman and others, 2011). Absorbance measures the amount of light absorbed by material in a water sample at specified excitation (ex) wavelengths, and fluorescence measures the light that is re-emitted (emission (em) wavelengths). Depending on the type of material present, the spectral properties can be diagnostic of certain types of organic matter. If material derived from different sources provides a unique “signature”, important sources of carbon within a watershed can be identified. Studies have demonstrated that continuous, in-situ fluorescence measurements can be used as a reliable surrogate for DOC concentration, and advances are ongoing to improve understanding of how changes in carbon composition affect freshwater systems using this approach (Bergamaschi and others, 2005, 2012; Spencer and others, 2007; Saraceno and others, 2009; Pellerin and others, 2012).


Absorbance and Fluorescence Spectroscopy 


The measurement of ultraviolet absorbance at 254 nm (UVA₂₅₄) has been used by the drinking-water industry as a proxy for DOC concentration for several decades (Edzwald and others, 1985; Rathbun, 1996; Korshin and others, 1997; Sadiq and Rodriguez, 2004). In addition to providing information about DOC concentration, absorbance data can provide insight into the chemical make-up of the DOM pool (table 1). For example, UVA₂₅₄ normalized by DOC concentration, also known as “specific” UVA (SUVA, reported in units of liters per milligram-meters, L/mg-m), has been correlated with DOM aromatic content (Weishaar and others, 2003). Similarly, the spectral slope of the absorbance curve has been shown to relate to aromatic content and molecular weight. For example, decreasing spectral slope between 275 and 295 nm is associated with higher aromatic content and increasing molecular weight. Spectral slope has also been shown to change upon irradiation (Helms and others, 2008; Spencer and others, 2009). 


As with absorbance, the fluorescence response and intensity at a single ex/em wavelength pair can be related to DOC concentration; the presence of different peaks, peak slopes, changes in the ratios of ex/em pairs, carbon normalized values, or shifts in peak maxima have been shown to provide information about DOM character and origin (Coble, 2007; Hudson and others, 2007; Stedmon and Bro, 2008). The fluorescence index (FI), calculated as the ratio of em 470 to 520 nm and ex 370 nm, has been widely used to indicate relative contributions of terrestrial- and algal-derived DOM (table 1). FI values obtained in the laboratory typically range from about 1.3 to 1.9; lower FI values are associated with terrestrial soil and plant organic matter—highly processed material having greater aromatic content and higher molecular weight—while higher FI values are associated with lower molecular weights and lower aromatic content indicative of algal and microbial sources (McKnight and others, 2001; Cory and others, 2010). Qualitative information can also be derived from the identification of fluorescence regions that have been linked to different DOM pools such as humic and fulvic acids, protein like substances, and phytoplankton-derived material (Stedmon and others, 2003; Coble, 2007; Hudson and others, 2007). 


In-Situ Fluorometers as Proxies for Dissolved Organic Carbon and Disinfection By-Product Precursor Concentrations


The use of fluorescence around ex 370/em 460 nm has been shown to have similar, possibly better, predictive ability for DOC concentration compared to absorbance measurements (Nakajima and others, 2002; Coble, 2007; Kraus and others, 2010). The continuous measurement of fluorescing DOM (FDOM) with in-situ fluorometers has been successfully used to provide a high-resolution proxy for DOC concentration (Downing and others, 2009; Saraceno and others, 2009; Pellerin and others, 2012). However, because only a subset of the DOM pool fluoresces and this fluorescing pool is a function of DOM composition, the relation between DOC concentration and FDOM needs to be validated for each watershed over the complete range of riverine conditions. The effects of optical density, DOC concentration, temperature, and turbidity on FDOM also need to be evaluated and accounted for (Lakowicz, 2006; Downing and others, 2012). 


Because DBP precursors are a sub-set of the bulk carbon pool, FDOM may also serve as a good proxy for THM and HAA precursor concentrations and, thus, for finished-water THM and HAA concentrations. However, the composition of the DOM pool can affect this relation because it affects the fraction of the DOC pool that reacts to form DBPs and the fraction of the DOM pool that fluoresces. Currently, the information regarding the relation between FDOM and DBP formation is limited (but see Nakajima and others, 2002; Hua and others, 2007, 2010; Beggs and others, 2009; Marhaba and others, 2009; Kraus and others, 2010). Given that the optically-active aromatic fraction of DOM typically dominates the DBP precursor pool in terrestrial environments, there is good reason to believe that in most cases there is a strong relation between these constituents. Furthermore, data from the McKenzie River in Oregon, a similar Cascade Mountain drainage, showed FDOM was a better predictor of THMFP and HAAFP than DOC concentration, suggesting a strong overlap between DOM moieties that are fluorescent and react with chlorine to form DBPs (Kraus and others, 2010).


To date, commercially available fluorometers intended to assess DOM dynamics are centered near ex 370/em 460 nm, referred to as “Peak C,” a humic region of the excitation‑emission (fluorescence) matrix (EEM) commonly identified in surface waters. While measurement of a single ex/em pair provides information about DOC concentration, information about the composition of the DOM pool is currently only available using bench-top fluorometers; sample scans using these instruments produce nearly 2,300 ex/em pairs, which are depicted in the EEM diagrams (fig. 5). 


Ongoing developments in light-emitting diode (LED) manufacturing technology have resulted in the availability of light sources with lower excitation wavelengths into the deep ultraviolet spectrum. Pairing the nearly monochromatic output of these LEDs with a wide array of optical filters makes novel development of miniaturized in-situ fluorometers with different excitation/emission pairs possible. Instruments designed using narrow band-pass filters allow for more focused emission spectra and, thus, also a more specific emission–fluorescence signal around a narrower excitation/emission region. Sensors designed to measure different regions of EEMs can provide signal ratios that may indicate carbon composition changes due to varying proportions of fluorescence associated with different pools of organic matter (humic peaks compared to amino-acid-like peaks, for example). When deployed continuously and in real-time, these sensors present new opportunities to gain insights about how rivers function and identify what factors affect water quality and source-water supplies, especially when a multitude of constituents are measured simultaneously. 


Recent work in an agricultural watershed demonstrated the in-situ FDOM sensor accurately predicted DOC concentrations throughout a precipitation and watershed runoff event, where DOC concentration cycled from a baseline of 2 mg/L to a peak value of 10 mg/L and back again to base‑flow levels (Saraceno and others, 2009). Several recent studies from a range of environments including wetlands, tidal marshes, and forested watersheds produced predictive relations between in-situ FDOM values from a WET Labs™ colored DOM (CDOM) fluorometer and laboratory‑determined DOC concentrations (Downing and others, 2008, 2009; Bergamaschi and others, 2012; Pellerin and others, 2012). This Clackamas study is, to our knowledge, the first to deploy a continuously-operated multi‑channel in-situ FDOM sensor within a drinking‑water treatment plant intake with simultaneous periodic measurement of DBPs in finished water. 


First posted February 11, 2013