search

PMassicotte / eemR

Utilities for pre-processing emission-excitation-matrix (EEM).
19 stars 10 forks source link

New wl to calculate FI? #62

Open PMassicotte opened 4 years ago

PMassicotte commented 4 years ago

From Researchgate:

Hi Matthias, Congrats on having made staRdom open! That people can PARAFAC on R is a great idea, and I'm also looking forward to trying it. Although not the main point of your work, I have one suggestion for you about the calculation of the fluorescence index. The FI calculated in staRdom is the old version, which must be applied for EEMs uncorrected for the instrument-specific bias (F450/F500, McKight et al., 2001), and not the new version which must be used for instrument bias-corrected EEMs (F470/F520, Cory and McKnight 2005). Maybe many people are not aware of this difference in usage, partly (or mostly?) because the original authors did not mention at all the reasons why they changed the wavelengths used for FI calculation in 2005 paper. Surprisingly it was first mentioned in 2010 paper by themselves (Cory et al., 2010)...But Nagamitsu Maie, who knew the original authors back then, mentioned the reason in his 2006 paper by citing Cory's PhD thesis. Korak et al. (2014) and Kida et al. (2018) also described the reason briefly, hoping that the correct usage will be spread. If the wrong version of FI is used for a wrong dataset, then too high/low FI values are obtained and FI doesn't work properly (Kida et al., 2018). The FI values for allochthonous vs autochthonous end-members were also modified from the original 1.4 vs 1.9 to 1.21-1.28 vs 1.45-1.52 (in case of corrected EEMs) which is an overall much narrower range (Cory et al., 2010) (Yes, old value for terrestrial endmember of 1.4 is very close to new values for autochthonous endmember of 1.45-1.52). So, the default calculation of FI should be the new version F470/F520, since almost all current EEMs are corrected for instrument-specific biases thanks to the effort of standardization, and I think it's a good idea to prepare an option of the old version for those who do not conduct the correction. In staRdom, I think the calculation of absorbance and fluorescence indices are done with eemR and so maybe it's not appropriate to ask you for this modification. Nonetheless, I'm asking now before increasing people will use your staRdom, coz there is already a quite mess regarding the usage and interpretation of FI in literature... Sorry for the long comment! Cheers, Maru References McKnight et al (2001) Spectrofluorometric characterization of dissolved organic matter for indication of precursor organic material and aromaticity Cory and McKnight (2005) Fluorescence spectroscopy reveals ubiquitous presence of oxidized and reduced quinones in dissolved organic matter. Maie et al (2006) Chemical characteristics of dissolved organic nitrogen in an oligotrophic subtropical coastal ecosystem Cory et al (2010) Effect of instrument‐specific response on the analysis of fulvic acid fluorescence spectra Korak et al (2014) Critical analysis of commonly used fluorescence metrics to characterize dissolved organic matter Kida et al (2018) Contribution of humic substances to dissolved organic matter optical properties and iron mobilization

MorimaruKida commented 4 years ago

Hi @PMassicotte Thank you for generously picking up my concerns:)

PMassicotte commented 4 years ago
MorimaruKida commented 4 years ago

I think the new calculation should be default because it's up-to-date and a more realistic choice. Nowadays EEMs are mostly instrument-bias corrected. Can you please give the old one as a second choice with some explanations and the reference (Cory et al. 2010)?

PMassicotte commented 4 years ago

This is what I have done. Any thoughts @MatthiasPucher ?

MatthiasPucher commented 4 years ago

I was not aware of that before @MorimaruKida pointed it out. From what I read so far, I consider this being the best way.

PMassicotte commented 4 years ago

Add unit tests to verify the results of the new FI.

ETaylor21 commented 3 years ago

Hi Phillippe, I was wondering when this fix might be available/ released to the eemR package? I tried to just copy the corrected code from your commit and to have the function written into my R script but it's not working for what I'm sure is a very obvious reason but troubleshooting that is a little outside my current working knowledge. I also tried downloading the development version of this package to see if the fix was included in that which it appears not to be but please let me know if I'm missing something!

PMassicotte commented 3 years ago

Hi @ETaylor21, I am not sure what is your problem exactly. Have you been able to install the dev version?

ETaylor21 commented 3 years ago

Thanks for getting back to me and sorry for not being clearer! I don't necessarily have a problem, what I'm wondering is when the eem_fluorescence_index function will be updated in the main branch of this package to allow users to choose which way their FI is calculated as mentioned above. I had mentioned that I had successfully installed the dev version of this package but as far as I could tell the FI function in the dev version also did not have the option to choose how the FI is calculated yet. I would like to use the 470/520 option as I am working with corrected eem data.

PMassicotte commented 3 years ago

Can you install the branch that allows it? https://github.com/PMassicotte/eemR/tree/fi