Fifteen years of PARAFAC application to organic matter fluorescence - progress, problems and possibilities
Abstract
The study of dissolved organic matter in aquatic milieu frequently involves measuring and interpreting fluorescence excitation emission matrices (EEMs) as a proxy for studying the total organic matter pool. Parallel Factor Analysis (PARAFAC) is used widely to identify and track independent organic matter fractions. This approach assumes that each EEM reflects the combined fluorescence signal from a limited number of unique, non-interacting chemical components, which are determined via a fitting algorithm. During the past fifteen years, considerable progress in understanding dissolved organic matter fluorescence has been achieved with the aid of PARAFAC; however, very few identical or ubiquitous fluorescence spectra have been independently identified. We studied the influence of wavelength selection on PARAFAC models and found this factor to have a decisive impact on PARAFAC spectra despite receiving little attention in most studies. Because large, chemically-diverse datasets may be too complex to analyse with PARAFAC, we are exploring novel methods for increasing variability in small datasets in order to reduce biases and increase interpretability. Our results suggest that spectral variability in PARAFAC models between studies are in many cases due to artefacts that could be minimised by careful experimental and modelling approaches.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2017
- Bibcode:
- 2017AGUFM.B41H2061M
- Keywords:
-
- 0428 Carbon cycling;
- BIOGEOSCIENCES;
- 1050 Marine geochemistry;
- GEOCHEMISTRY;
- 1055 Organic and biogenic geochemistry;
- GEOCHEMISTRY;
- 1806 Chemistry of fresh water;
- HYDROLOGY