Premium
comPARAFAC: a library and tools for rapid and quantitative comparison of dissolved organic matter components resolved by Parallel Factor Analysis
Author(s) -
Parr Thomas B.,
Ohno Tsutomu,
Cronan Christopher S.,
Simon Kevin S.
Publication year - 2014
Publication title -
limnology and oceanography: methods
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.898
H-Index - 72
ISSN - 1541-5856
DOI - 10.4319/lom.2014.12.114
Subject(s) - dissolved organic carbon , matching (statistics) , equivalence (formal languages) , computer science , principal component analysis , component (thermodynamics) , data mining , quantitative analysis (chemistry) , biological system , statistics , artificial intelligence , mathematics , chemistry , chromatography , environmental chemistry , physics , discrete mathematics , biology , thermodynamics
Parallel Factor Analysis (PARAFAC) is a well‐established method for characterizing dissolved organic matter (DOM). While methods for sample processing and PARAFAC analysis are well defined and robust, subsequent classification of DOM fluorescence components and comparisons of components among studies remain highly qualitative. Because these comparisons often guide the interpretation of subsequent data, it is important that quantitatively accurate comparisons be made. We developed a statistical tool, comPARAFAC, using a modified Tucker's Congruence Coefficient (mTCC), an established method of factor comparison, to provide a quantitative basis for comparing models. To develop and test this tool we used mTCC to compare factors from 35 DOM fluorescence studies using Parallel Factor Analysis (PARAFAC) in marine and freshwater environments. We compared mTCC‐guided component matching with qualitative comparisons made in the literature to describe the current perceptions of component equivalence. Based on our analysis, 21% of the direct comparisons made using the qualitative approach are potentially erroneous, whereas possible matches are missed 14% of the time using that same approach. The procedure and accompanying PARAFAC model library for performing quantitative mTCC‐guided comparisons are available as an R package (see Web Appendix A). This method simplifies and standardizes the process by which researchers identify and compare fluorescent DOM components across studies.