Experimental and modeling considerations for working with CDOM photochemistry

Miller [32] has provided a concise consideration of issues when designing photochemical experiments using CDOM, and the careful measurements that must be made. Here, we will provide a general design concept for an experimental exposure of CDOM to a polychromatic light source (e.g., solar radiation) and

Initial Dark Light

Initial Dark Light

Initial Dark Light

Treatment

Figure 5. Changes in bog water DOM after exposure to solar radiation. (A) Loss of CDOM absorbance at 320 nm, (B) decrease in molecular weight (indicated by increase in ratio), (C) decrease in DOC concentration, and (D) increase in ¿13C of lyophilized DOM.

After Osburn [158].

then discuss the methods for constructing a predictive model based on the results. We use as an example the process of photobleaching and the measurement of dissolved absorbance. While not comprehensive, this section should provide for new readers a general approach for designing and interpreting photochemical experiments with CDOM.

6.6.1 Experimental methods

The experimental design for measuring changes in bulk parameters of CDOM such as dissolved absorbance is straightforward. First, water is collected and then filtered to remove particles and bacteria, usually through a pre-cleaned (baked at 450 °C then rinsed with ultrapure deionized and distilled water) glass fiber filter with a pore size of ~0.7 fim. Next, the filtrate is mechanically sterilized by passage through a 0.22 fim filter and carefully transferred to clean (acid-washed and baked) quartz vessels, usually test tubes or round-bottom flasks, reserving an aliquot of the sample for analysis of initial parameters. The quartz should be at least 99% pure to ensure transmission of all environmentally relevant UVR through the walls of the quartz vessel. In some cases, a bactericide may be used to suppress microbial growth, though many microbial inhibitors (NaN3, HgCl2) absorb UVR near the UV-B range, near 275 nm [132], The vessels are then placed in a water bath and exposed to solar or artificial radiation. The water bath buffers the temperature of the CDOM solution that can vary by tens of degrees Celsius from early morning to evening for solar radiation exposures. Similarly, artificial light sources can also generate high temperatures around the samples. The exposure time may vary depending on the goals of the experiment. Some samples may be collected periodically during the exposure to generate a time-series of photochemical changes, which is useful for determining rate constants.

6.6.2 Reportingphotobleaching results

Several methods exist to report the optical changes in CDOM after the photo-bleaching experiment. Loss of absorbance and fluorescence are the two most common parameters to report. The loss of absorbance is determined by subtraction of final minus initial absorbance spectra measured with a UV-Vis spectrophotometer. A sample spectrum of absorbance loss per nm for photobleached CDOM is shown in Figure 6. This figure illustrates that absorbance exhibits wide variation across the UV spectrum and between different sources of CDOM. Often, loss of absorbance is reported at specific wavelengths, with wavelengths in the UV-B or UV-A region being most common (e.g., 320 or 350 nm). It has been

280 300 320 340 360 380 400 420 440 460 480 500

Wavelength (nm)

Figure 6. Loss of absorbance per nm in several different types of CDOM.

280 300 320 340 360 380 400 420 440 460 480 500

Wavelength (nm)

Figure 6. Loss of absorbance per nm in several different types of CDOM.

popular to report absorbance loss at 250 nm, which is an indicator of aromatic content. Similarly, loss of DOM fluorescence may be measured using excitation at one wavelength and emission at another wavelength. Recently, published reports of excitation-emission matrices have provided a three-dimensional picture of CDOM photochemical changes [41,122,133-135]. CDOM fluorescence is particularly useful because the results may be normalized to a standard, such as quinine sulfate. Currently, no accepted standard for CDOM absorbance exists and reporting loss of absorbance as absolute values makes for difficult comparisons among CDOM sources that vary in CDOM quantity and quality. Moreover, dissolved absorbance may be affected by dissolved species other than DOM. Thus, some workers have reported loss of absorbance normalized to the DOC concentration of the water sample, which approximates the molar absorptivity used in classic photochemistry [8,15,116,124,136-139].

Osburn et al. [132] have used the integrated photobleaching (loss of absorbance over the range of 280 to 500 nm) divided by the total absorbed energy (also from 280 to 500 nm) to calculate the photoreactivity of various CDOM sources. This parameter is descriptive of the capacity for the CDOM to lose absorbance, and normalizes different CDOM types to their absorbed energy, which is a function of CDOM quality and quantity. This is similar to the apparent quantum yield calculated for CDOM by Whitehead et al. [140]. Similar integrated photobleaching calculations have been used for CDOM photobleaching studies from estuarine [122] and lake [123] sources.

6.6.3 Modeling photochemical changes in CDOM

One goal of the study of CDOM photochemistry is to predict the effects of enhanced UV influx, from stratospheric ozone depletion, on CDOM. Because photon energy is inversely related to its wavelength (Planck's relationship, section 6.2), the energy per photon increases with decreasing wavelength (Figure 7). This is analogous to the absorption spectrum of CDOM, and means that wavelengths are variable in their ability to bring about chemical change. We might expect that higher-energy UV-B wavelengths can cause more photobleaching than lower-energy UV-A wavelengths, but we must also consider the quantity (or fiuence) of photons that reach the aquatic environment. Figure 7 also shows that the fluence for UV-B wavelengths is orders of magnitude lower than the fluence of UV-A wavelengths. Thus, to model photobleaching, we need to account for the effectiveness of photons at each wavelength in the solar spectrum, as well as their fluence.

Mathematically, it is possible to deconvolute the effectiveness of each wavelength and assign each wavelength a weight; in effect, generating a spectral weighting function (SWF) for photobleaching. Osburn et al. [132] have described in detail the methodology for computing SWFs using photobleaching data obtained from multiple optical cutoff filters based on the Rundel method [141] and comments by Cullen and Neale [142]. The cutoff filters successively manipulate solar spectra by successively removing more UVR and creating an array of

3.0E+06

8.0E-19

8.0E-19

0.0E+00

280 300 320 340 360 380 400 420 440 460 480 500 Wavelength (nm)

Figure 7. A comparison between a typical solar spectrum generated for temperate regions (Washington, DC, USA) and the energy per photon estimated by the Planck equation.

0.0E+00

280 300 320 340 360 380 400 420 440 460 480 500 Wavelength (nm)

Figure 7. A comparison between a typical solar spectrum generated for temperate regions (Washington, DC, USA) and the energy per photon estimated by the Planck equation.

spectral treatments. A simple exponential was assumed for the shape of the SWF (based on the dissolved absorbance of CDOM and equation 1) and an iterative nonlinear regression of the iterative photobleaching in each spectral treatment versus the cumulative absorbed energy in each treatment was run to optimize the fit of the SWF to the observed data. The equation for the SWF was:

where fV300 is the weight at 300 nm, and Sw is the slope of the exponential. The seed value for oo was estimated from a regression of the differential photo-bleaching between adjacent optical cutoff filters versus the differential absorbed energy in adjacent treatments. The seed value for Sw was estimated from the slope of In «cdom(^) versus wavelength (5, the spectral slope). The values for an average SWF (based on multiple experiments of surface water CDOM from several lakes) were: Wm = - 0.0103 ± 0.006 and Sw = 4.34 ± 1.79 x 10~6).

This function can then be used to predict the amount of photobleaching that has occurred by multiplying the SWF by a measured irradiance spectrum. For example, we used the summary SWF of Osburn et al. [132] to predict daily changes in dissolved absorbance for humic Lake Lacawac, northeastern Pennsylvania, USA. First, daily measurements of ground level incident energy at four wavelengths (305, 320, 340, and 380 nm) were used to construct daily solar spectra for northeastern Pennsylvania, USA, during the summer of 1998. The measurements were recorded on a Biospherical Instruments GUV-521 radiometer and the total incident energy on the surface of Lake Lacawac was modeled to generate daily solar spectra at 1 nm intervals from 280 to 500 nm.

Modeled attenuation coefficients for the lake were used to modify the surface solar spectrum at 10 cm interval of depth throughout the mixed layer. Thus, they computed spectral energy at depth and could then predict the photobleaching at depth using the SWF. The daily spectra were multiplied by the spectral weights from the SWF to calculate daily photobleaching at the 10 cm intervals in the mixed layer of Lake Lacawac. The sum of the predicted photobleaching at depth was subtracted from an initial daily dissolved absorbance value, which equaled a daily change in dissolved absorbance for the mixed layer of Lake Lacawac. The model was run successively for that was repeated for each day of the study period. Thus, the model predicted daily changes in integrated dissolved absorbance caused solely by photobleaching. Figure 8 shows the measured changes in dissolved absorbance and the predicted changes using the SWF. The close match between modeled and measured dissolved absorbance suggest that CDOM photobleaching by solar UVR is the process that controls dissolved absorbance, and thus transparency, in lakes. Deviations of the model from observed change in the dissolved absorbance of the lake are attributed to recharge of fresh CDOM from various sources (precipitation, runolf, and advection).

6.6.4 Use of polychromatic vs. monochromatic radiation

The use of polychromatic radiation (e. g., solar radiation) more closely resembles

Figure 8. Predicted change in dissolved absorbance for the epilimnion of Lake Lacawac during 1998. Changes in dissolved absorbance were computed using the average SWF of Osburn [158]. Solid line is the model run from Day 1 to Day 104. Other symbols represent the model run for intervals in between dissolved absorbance measurements made during the modeling period (open squares).

Date

Figure 8. Predicted change in dissolved absorbance for the epilimnion of Lake Lacawac during 1998. Changes in dissolved absorbance were computed using the average SWF of Osburn [158]. Solid line is the model run from Day 1 to Day 104. Other symbols represent the model run for intervals in between dissolved absorbance measurements made during the modeling period (open squares).

the natural environment, and is different from classic photochemistry, which often uses monochromatic radiation. In this experimental setting, a mono-chromator is used to separate wavelengths generated from a light source. Thus, in an experiment, a sample is exposed to only one wavelength at a time and the effect of that one wavelength is recorded. This approach allows the researcher to easily calculate an effect per wavelength (i.e., an action spectrum). If the radiant energy at each wavelength is known, the researcher can calculate a quantum yield for the photoreaction. However, the monochromatic approach implicitly assumes that the measured effect is only due to energy at the wavelength of irradiation. This makes application of action spectra to the natural environment difficult.

With polychromatic radiation, the effect is less obvious and necessitates the manipulation of the energy spectrum (e.g., with optical cutoff filters) to measure the effect and then to deconvolute the weighted effect at each wavelength. While the use of optical cutoff filters to modify polychromatic radiation more closely simulating the natural environment, their use introduces error to the calculations and may reduce the sensitivity of the analysis [141]. However, several lines of evidence suggest that with polychromatic solar radiation, multiple wavelength reactions contribute to the photodegradation of CDOM. Both Osburn et al. [132] and Whitehead et al. [140] have measured photobleaching at wavelengths that were excluded by optical filters. They suggest the interactive effect of photons from multiple wavelengths caused photobleaching at any one wavelength. The mechanism that drives this phenomenon may be that a chromo-phore absorbs over a range of wavelengths or a change in the relative abundance of chromophores (absorbing and different and multiple wavelengths) in the bulk DOM. This effect would further complicate the application of action spectra for CDOM photobleaching to measurements of polychromatic solar spectra.

0 0

Post a comment