Carbon Characterization Using Optical Properties


Optical measurements, including UVA₂₅₄ and FDOM (ex 370/em 460 nm), were strongly correlated with concentrations of DOC (table 11) and TOC in the Clackamas River. Although DOC concentration remained relatively low in the mainstem, there were large changes in SUVA values (1.5 to 4.4 L/mg-m) over the study period, indicating a shift in DOM composition (fig. 19). SUVA values were highest during high-flow periods, indicative of contributions of higher aromatic carbon associated with humic-like material. During low-flow periods, SUVA values decreased, suggesting the DOM pool contained a greater proportion of non-aromatic, lower molecular‑weight carbon derived from less-degraded organic material (Weishaar and others, 2003). SUVA increased slightly at the downstream sites in the mainstem. SUVA values were also higher in the tributary sites, particularly during the October 2010 storm event (3.2–4.3 L/g-m). Like SUVA, spectral slope values are also used to indicate changes in DOM composition. As was seen with SUVA, seasonal changes in spectral slope suggest the DOM contains higher molecular‑weight, aromatic material during storm events compared to base-flow periods (appendix G1).


Contour plots of selected EEMs obtained from the fluorescence analysis are shown in fig. 20 to highlight observed changes in DOM amount and composition. The highest intensity EEMs were from Deep, Rock, and Sieben Creeks where DOC concentrations were highest. The first set of EEMs for the upper, middle, and lower mainstem during low streamflow contain a Peak T-like carbon signal near 270–340 ex/em associated with algal-derived organic matter high in protein and (or) freshly leached plant material high in polyphenols (Coble, 2007; Hernes and others, 2009; Beggs and others, 2011). The presence of this peak is best demonstrated in the last EEM from North Fork Reservoir that represents the algal grab sample collected at the surface during the September 2011 phytoplankton bloom. It should be noted that the Peak T signal was only apparent, however, in the contour plots when overall fluorescence intensities were lower, reflecting periods when DOC concentration was low. This likely results because the Peak T region is over-shadowed by the stronger humic-like signal (for example, Peaks A and C) that appeared during high-flow events (fig. 20). In order to more accurately quantify the presence of different pools of organic matter contributing to the overall EEMs spectra, PARAFAC modeling was used; this approach enables the more subtle underlying signals that are not necessarily visible in the EEM color contours to be detected and the relative contribution of each of the underlying fluorophores to be quantified. 


The PARAFAC model developed and validated with 167 samples from the mainstem, tributary, North Fork Reservoir, and source and finished water produced five components: C1–C5 (fig. 21). Model components C2, C3, and C4 are associated with terrestrial humic-like substances commonly associated with fluorescence Peaks A and C (Stedmon and Markager, 2005; Coble, 2007; Murphy and others, 2008; Yamashita and others, 2008). There are, however, subtle differences among these three components. Fluorescence associated with C1 has been associated with several different sources of DOM, including marine and terrestrial humic acids that have been microbially processed, freshly produced (phytoplankton derived, for example) labile material identified as Peak N, and material exported from agricultural and wastewater-impacted catchments identified as Peak M (Stedmon and Markager, 2005; Coble, 2007; Fellman and others, 2010). C2 is typical of terrestrial organic matter composed of high molecular-weight and aromatic compounds (McKnight and others, 2001; Stedmon and others, 2003). C5, located in a region frequently referred to as “protein-like” because tryptophan and tyrosine fluoresce in this region, is associated with less-processed carbon derived from fresh terrestrial plants, algae, and (or) wastewater (Murphy and others, 2008; Hernes and others, 2009; Beggs and others, 2011).


Examination of the relative contributions of the different PARAFAC components shows that C1, C3, and C4 represent the bulk of DOM fluorescence (fig. 22). Component C5 generally represented a smaller and more highly variable fraction. Although there were slight seasonal shifts in the relative proportions of the different PARAFAC components in CRW source water (fig. 22B), the overall trend shows a consistent pattern. Average percentages were 26 percent for C3, 21–22 percent for C1 and C4, and 15–16 percent for C2 and C5, reflecting dominance by terrestrial types of carbon in the lower mainstem.


The fluorescence spectra of Eagle and Clear Creeks, two streams draining predominantly forested basins (table 2), were similar to the main-stem sites (fig. 22), reflecting dominance by terrestrial humic substances. The other three tributaries—Deep, Rock, and Sieben Creeks—have less forested area and are variously influenced by agriculture and urban development (table 2). These streams contained the highest proportion of component C1 (24–39 percent; fig. 22), suggesting the DOM contributed from these tributaries differed in composition compared to the other sites. Rock and Sieben Creeks also showed the lowest proportion of C5.


The fluorescence spectra of samples from North Fork Reservoir had the highest average percentage of component C5 (fig. 22), likely reflecting the presence of recently added DOM from phytoplankton algae that were prevalent in the reservoir during all samplings. Organic matter recently contributed by algae is expected to contain a greater protein‑like signal. 


Although there were only minor changes in the FI values (1.3–1.5 across the watershed samples), seasonal trends suggest DOM in the mainstem is more dominated by microbial-derived carbon between August and October (fig. 19), when more algal contributions would be expected. This trend is in accordance with the SUVA values. FI values were also higher in the tributary sites compared to main‑stem sites for the May 2010 and September 2011 basin-wide samplings. The FI value was highest (2.3) for the algal grab sample collected from the surface of North Fork Reservoir in September 2011 (not shown in fig. 19), which contained a high abundance of blue-green algae (Anabaena sp.). Although the FI did not vary a lot, these results suggest this measure could be indicative of algal-derived carbon despite the relatively low levels of chlorophyll-a observed during the study.


The HIX (table 1) ranged from about 2 to 8; the most notable trends were higher values in some of the tributaries, particularly in September 2011, possibly from highly-processed humified material, and lower values for reservoir samples, indicative of more recently added “fresh” material (appendix G1).


Carbon characterization based on optical properties provided complementary evidence to the carbon concentrations during the two autumn storms in 2010 that offers insights into the different effects of these storms on watershed processes. The second, larger storm (fig. 13), for example, produced higher SUVA and HIX values and lower FI values generally associated with greater aromatic content and higher molecular-weight material. This was accompanied by declines in the percentages of carbon components C1 and C5 and increases in components C2, C3, and C4—all indicative of the carbon quality shifting from a more labile microbial/algal source with lower aromaticity during the initial storm to a higher molecular-weight, aromatic source of terrestrial origin after the larger storm (table 8). These results point to the success of fluorescence technologies in detecting these shifts in DOM quality (and quantity) and provide a means by which the quality of source water can be closely monitored for treatment-plant operations, for gaining a deeper understanding of river conditions and processes, and for evaluating trends over short and long time scales. Through implementation of these instruments in studies like this, new technologies are being developed to advance these capabilities.


First posted February 11, 2013