A temperature compensation method for CDOM fluorescence sensors in freshwater

Abstract

296As interest in the aquatic cycle of organic carbon (OC) has increased, the deployment of in situ optical sensors to measure CDOM fluorescence (chromophoric dissolved organic matter) as a proxy for OC concentration has become more common (e.g., Downing et al. 2009;Sandford et al. 2010). CDOM sensors typically use UV light (~350 nm) to excite the emission of blue light (~450 nm) from certain organic fluorophores, allowing investigators to distinguish CDOM from more commonly measured phytoplankton pigments. Given that CDOM may be more labile than previously thought and given that rates of OC mineralization may vary with fluctuating environmental factors, such as temperature and light, these inexpensive sensors could afford a substantial advantage over traditional wet chemistry methods-provided that the artifactual effects of environmental factors on fluorescence efficiency are well constrained (Graneli et al. 1996;Bertilsson and Tranvik 2000;Bastviken et al. 2004;Hanson et al. 2003;Vahatalo 2009).Here, we quantify the effect of temperature on the fluorescence of CDOM from two dystrophic Wisconsin lakes and an aquatic NOM reference material. Based on laboratory experiments over a wide range of OC concentrations, we derive a function that can be used to standardize CDOM measurements to any reference temperature (and, thereby, remove the effect of temperature variation on CDOM fluorescence). Using a reference temperature of 20°C, we then apply the function to field data and show how temperature compensation affects temporal changes in CDOM fluorescence under natural conditions. Methods and proceduresTwo commercial CDOM fluorometers were used: 1) the C3 Submersible Fluorometer from TurnerDesigns, Inc.; and 2) the ), and the subscripts r and m stand for the reference and measured values. (We note that an analogous function is used widely to calculate temperature-specific conductance from the measured conductivity of natural waters.) For the two sensors we tested, the temperature-specific fluorescence coefficients (r) were -0.015 ± 0.001 and -0.008 ± 0.0008 for Wisconsin bog waters at 20°C. When applied to field data, temperature compensation removed the effect of multi-day trends in water temperature, and it also damped the diel CDOM cycle. We conclude that temperature compensation is a necessary and important aspect of CDOM monitoring using in situ fluorescence sensors.

Reply to a comment by Watras et al. (2014) on temperature compensation method for field measurements of CDOM fluorescence

“…In addition, we proposed a temperature compensation method for field measurements of CDOM fluorescence. We also compared the performance of our temperature compensation method to that proposed in Watras et al (2011) and stated that our equation produced a more successful temperature correction. have now defined q in an equation as follows:…”

“…We now realize that we misinterpreted the calculation of the temperature correction factor q as described in Watras et al (2011). Based on equation 1, the two equations are mathematically equivalent.…”

“…

AbstractThe recent comment by clarifies the calculation of the temperature correction coefficient (q) in Watras et al (2011). Based on this clarification, we accept that the equation to compensate for temperature quenching of chromophoric dissolved organic matter (CDOM) fluorescence presented in Ryder et al (2012)

Comment

In our recently published paper Ryder et al (2012), we showed that the degree of temperature quenching of chromophoric dissolved organic matter (CDOM) fluorescence differed between sites and over time.

…”

“…The recent comment by clarifies the calculation of the temperature correction coefficient (q) in Watras et al (2011). Based on this clarification, we accept that the equation to compensate for temperature quenching of chromophoric dissolved organic matter (CDOM) fluorescence presented in Ryder et al (2012) …”

“…From Equation 1 above and as defined in , it is now clear that q is the quotient of the slope divided by the regression equation solved for the reference temperature. Watras et al (2011) then provided a second definition for q within the same paragraph, again as the quotient of the slope/intercept, but for the regression of CDOM(c) on (T m -T r ) where m and r were subscripts for measured and reference respectively and was where c is the concentration between 0 and 1.We regret that there was misinterpretation in our understanding of the text in Watras et al (2011) on the calculation of the q value. We agree this invalidates the comparisons presented in Tables 1 and 3 and Figures 3 and 4 in Ryder et al (2012).…”