Shale reference compositions and McLennan 2001 crustal compositions

[kaarelmand]

Some new, mostly shale-based reference compositions for pyrolite.

Description

This PR adds some composition .csv files to the refcomp folder, mainly those concerning shales and other sediments.

• Phanerozoic cratonic shale compilation by Condie (1993) -- PHS_Condie1993.csv
• Post-Archean Australian Shale (PAAS) by Taylor & McLennan (1985), updates by McLennan (2001) -- PAAS_TaylorMcLennan1985.csv
• North American Composite Shale (NASC) by Gromet et al. (1984), updates by Taylor & McLennan (1985), Condie, 1993, McLennan, 2001 -- NASC_Gromet1984.csv
• Upper continental crust, lower continental crust and bulk continental crust composition estimates by McLennan (2001) -- UCC_McLennan2001.csv, LCC_McLennan2001.csv, BCC_McLennan2001.csv
• Proterozoic and Archean cratonic shales by Condie (1993) -- PRS_Condie1993.csv, ARS_Condie1993.csv
• "PAAS," being a smoother REY dataset from a different set of Australian shales by Pourmand et al. (2012) -- PAAS_Pourmand2012.csv
• European Shale (EUS) by Bau et al. (2018) -- EUS_Bau2018.csv
• MUd from Queensland (MUQ) by Kamber et al. (2005), being a semi-popular proxy for recent average detritus -- MUQ_Kamber2005.csv

Related Issue

Addresses point 2 of issue #35.

Motivation and Context

Pyrolite was previously missing widely-cited shale reference compositions, most notably PAAS. This PR adds those to pyrolite's already extensive reference composition database.

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)

Checklist:

• My code follows the code style of this project.
• My change requires a change to the documentation.
• I have updated the documentation accordingly.
• I have added tests to cover my changes.
• All new and existing tests passed.

[kaarelmand]

While I did add all these files to the refcomp folder, I am not sure if I have to run any scripts to re-populate the refdb.json file.

Nor am I sure how to update the documentation -- it seems to me that the page on the reference compositions is automatically populated?

[coveralls]

Pull Request Test Coverage Report for Build 1053

• 0 of 0 changed or added relevant lines in 0 files are covered.
• 35 unchanged lines in 5 files lost coverage.
• Overall coverage increased (+0.08%) to 86.632%

---

Files with Coverage Reduction	New Missed Lines	%
pyrolite/geochem/transform.py	2	95.6%
pyrolite/util/math.py	3	90.04%
pyrolite/util/synthetic.py	5	94.79%
pyrolite/plot/spider.py	8	92.31%
pyrolite/util/plot/axes.py	17	72.41%

Totals
Change from base Build 1013:	0.08%
Covered Lines:	4303
Relevant Lines:	4967

---

💛 - Coveralls

[morganjwilliams]

Thanks @kaarelmand, looks great!

I'll have a quick look at the CSVs, but I suspect this should be able to be merged without any issues. If you wanted to check they all import nicely, you could run the pyrolite.geochem tests locally (there's one included for the database update) - although note that the Travis build for the PR passed, and I can confirm that tests pass on my machine too!

To round off the PR, could you add a commit after running the update_database() function locally to add you new files to the refdb.json file? This is where pyrolite will look for the reference compositions.

[morganjwilliams]

On a slight tangent, I have a feeling that soon it will be worth writing up a tutorial about the variety of reference compositions, and their general applicability. Let me know if you're happy to contribute some info for a shale section!

[morganjwilliams]

The documentation example does indeed have an auto-documented list of the reference compositions, so there's no need to update this.

Given we now have 31 compositions, this will be something to break out into a new page soon. I've added #38 to this end, and we can address that separately.

[morganjwilliams]

Looks good! These should be able to be directly merged into develop, and will be released with v0.2.6.

[kaarelmand]

I'm in the process of fixing the .csv files I uploaded earlier. I also added a To Do list in the description.

[kaarelmand]

Some musings, so I don't forget:

I'm thinking I should probably hold off on adding the conglomerates and sandstones from Condie, 1993 -- I don't know/think that anyone would actually need those compositions. And if someone does, it's very easy to add them in. Seems a bit silly in hindsight.

For the European shale, Bau et al. only report REY values, but there's some older data floating around for other elements on the same sample set. Nope, was all just REY values...

Sometime in the future, it might be worth thinking of a function that would take a time-series of element concentration data over Earth history, and normalize it to the changing crust in the Archean--Proterozoic. Condie, 1993, actually has 7 different composition brackets through Earth history, even though I only included his eon-averages.

[kaarelmand]

All right, that should be it for this PR! Not sure if I still have to run any tests or whatnot?

[morganjwilliams]

There's no specific tests for the data files, but I'll pull down the fork and have a look locally.

[morganjwilliams]

@kaarelmand if you could fix the two reference cells for the NASC and PAAS data files I mentioned above and remove the extra text file I think this will then be ready to merge. I'll update the refdb.json after this - have an issue on my side which is likely due to a new version of TinyDB, and I'll update develop to fix this a the same time.

[kaarelmand]

Whoops, these issues should be fixed now!

[morganjwilliams]

Thanks for this one @kaarelmand! I had to re-update the reference database to pass the tests but everything else looks good. You can see some of the new compositions pop up in the normalization example.