Photobleaching of CDOM: A new approach to quantify apparent quantum yield matrices applicable to spectral photochemical models

Xiaohui Zhu, Boston University, Earth & Environment, Boston, MA, United States, William L Miller, University of Georgia, Marine Science, Athens, GA, United States and Cedric G Fichot, Boston University, Dept. of Earth & Environment, Boston, United States
Photobleaching is an important degradation mechanism for chromophoric dissolved organic matter (CDOM) which can regulate the vertical distribution and availability of solar radiation in the surface ocean. However, a rigorous quantitative assessment of the magnitude, variability, impacts and significance of this process in the surface ocean remains lacking, primarily because photobleaching is a spectrally complex process. In order to facilitate its modeling, photobleaching efficiency data must capture not only the wavelength (energy) of the photon absorbed, but also the loss in absorption coefficient that occurs across the entire spectrum. Because CDOM both absorbs radiation and photobleaches polychromatically, mathematical description of its apparent quantum yield (AQY) is complex, making quantitative models of photobleaching rates in natural waters challenging. We have developed a novel experimental approach to determine the photobleaching AQY in natural water samples. Our approach combines results from photochemical experiments carried out under controlled and measured illumination conditions, a partial least square regression, and a machine learning algorithm to produce a three-dimensional AQY matrix for the photobleaching process. Application to several river- and coastal-water samples revealed AQY spectral features consistent with previous observations for photobleaching of the CDOM absorption spectrum, and the approach can reproduce observed changes in spectral CDOM absorption with high fidelity. Results also show that both the magnitude and spectral characteristics of derived AQY depend on exposure time (light history) and temperature, further capturing the complex interactions needed to be considered for modeling application.