Synthesis Version Updates

[jaxinewolfe]

This update is scheduled for March 2024

Please continue working in the develop branch!

Overall Goals:

• Resolve errors in existing hook scripts
• Add any necessary QAQC tests (depth intervals, data viz, taxa resolution)
• Add necessary uncontrolled attributes and variables
• Avoid duplication of cores (esp. international cores from seagrass or mangrove habitat)

Additional goals:

• Remove instances of modeled fraction carbon values
• Make sure that every core has an assigned habitat
• Every study ID has at least one associated citation
• Track down coordinates for every core (site-level resolution is better than nothing)
• Activities data (radioisotope) units are standardized across all studies

[jaxinewolfe]

Resolving errors in existing hook scripts:

• Weis 2000: standardize the "counts_per_hour" activity unit
• Morgan et al 2024 depthseries missing study_id
• Make sure data is output to derivative folders with no row index column
• Poppe et al 2019 has some cores with duplicate impact classes
• Holmquist synth impact classes need site_id (only core_id presently)
• Resolve encoding errors in datasets that have accents (Cifuentes, Eagle, author names, etc)
• Determine whether the following studies just have cores with duplicate intervals or if they contain cores that need to be separated into unique IDs: "Callaway_et_al_2019", "Fourqurean_and_Kendrick_unpublished," "Howard_and_Fourqurean_2020", "StLaurent_et_al_2020"
• Edits to hook scripts with radiocarbon dates/ages? These would be: "Watson_and_Byrne_2013" "Shaw_et_al_2020" "Rodriguez_et_al_2022" "McTigue_et_al_2020" "Marot_et_al_2020" "Lafratta_et_al_2018"
  "Krauss_et_al_2018" "Johnson_et_al_2007" "Gerlach_et_al_2017" "Drexler_et_al_2009" "Costa_et_al_2023" "Belshe_et_al_2019" )

The following cores have a depth interval where the min and max are likely reversed:

study_id	core_id
Thom_1992	PB1
DelVecchia_et_al_2014	M1314
Costa_et_al_2023	PanamaCaribbean-Station Forest (SF)-1
MacKenzie_et_al_2021	Catanauan_216_714
Sharma_et_al_2021	Koh_Kohng_138_384
Sharma_et_al_2021	Koh_Kohng_138_386

The following studies have intervals with NA depths:

"Nahlik_and_Fennessy_2016" "Langston_et_al_2022" "De_Iongh_et_al_1995" "Agawin_et_al_1996" "Townsend_and_Fonseca_1998" "Holmer_et_al_2007" "Van_Engeland_2010" "Boyd_et_al_2017"

Studies with NA Coords:

• 1 Thom_1992
• 2 Schile-Beers_and_Megonigal_2017
• 3 Nsombo_et_al_2016
• 4 Eid_and_Shaltout_2016_Egypt
• 5 Eid_et_al_2016_Saudi_Arabia
• 6 Poppe_et_al_2019
• 7 Marot_et_al_2020
• 8 Drexler_et_al_2013
• 9 Belshe_et_al_2019

Studies with NA Habitat:

• 1 Thom_1992
• 2 Schile-Beers_and_Megonigal_2017 (likely sabkah, which we should add to the database)
• 3 Osland_et_al_2016
• 4 Marot_et_al_2020

Studies with Modeled C:

• "Ceron-Breton_et_al_2011" (likely)

Synthesis and Post-Processing

• How to indicate island municipalities or colonies? (ex. you can't filter for Bonaire)
• should we truncate decimals?
• flag cases where values that should be positive are negative? (ex. DBD for Lagomasino and Morrissette)

[jaxinewolfe]

We need a QAQC function to catch studies that have no associated citation

Some starter code:

no_citations <- ccrcn_synthesis$cores %>%
filter(!(study_id %in% unique(ccrcn_synthesis$study_citations$study_id)))

if (nrow(no_citations) > 0) {
warning("NOTE: The above studies were removed because they did not have citation information present.
Please review the CCN library synthesis to confirm that all synthesis studies have proper study citation information in '/data/CCN_synthesis/CCN_study_citations.csv' ")

unique(no_citations$study_id)
}

[jaxinewolfe]

@cheneyr @BettsH

Here are the QAQC results for our current version of the synthesis. It looks like a lot, but I think it's fairly small stuff that we can knock out! For example, theres a bunch of columns starting "..." which resulted from tables being output using write.csv() without specifying that row.names = F (idk why the default is set to true, its annoying) to prevent it from creating a column with the row number index included. If you spot stuff that is related to datasets you've worked on you can go for those quick fixes, or we can chat about some of the more nuanced things in our meeting (or whenever).

Also, we are well past 10k cores, holy moly! 🎉

index	test	result
1	Core ID uniqueness	Check the following core_id(s) in the core-level data: 2B, 305, 398, 399, AL, B1, B2, B3, B4, G15, G4, G5, G9
2	Valid core ID links in core table	No core ID in depthseries table: WBWA1109_01PU, WBWA1109_02PU, WBWA1109_03PU, WBWA1109_04PU, NSOR1209_01PU, NSOR1209_02PU, NSOR1209_03PU, NSOR1209_04PU, NBOR1409_01PU, NBOR1409_02PU, NBOR1409_03PU, Catlett_1m, Catlett_Transect, Goodwin_1m, Goodwin_Transect, Pamunkey_Transect, SweetHall_1m, SweetHall_Transect, Taskinas_Transect
3	Valid core ID links in depthseries table	No core ID in core table: PB1, RC_U_A, RC_M_A, PR_U_A, PR_M_A, W_U_A, W_M_A, F_U_A, F_M_A
4	Test coordinate uniqueness	1373 sets of coordinates are associated with more than one core. Check 'data/QA/duplicate_coordinates.csv'
5	Validity of column names in depthseries table	Undefined columns: ...33, ...38, th234_activity, th234_activity_se, k40_activity, k40_activity_se, ...56, ...57, date, pb210_crs_age, pb210_crs_age_sd
6	Validity of column names in cores table	Undefined columns: ...29, salinity, ...31, ecological_condition_flag, ...37, ...38, core_date, core_position_method, geomorphic_id, ...42
7	Validity of column names in sites table	Passed
8	Validity of column names in species table	Undefined columns: ...7, ...8
9	Validity of column names in impacts table	Undefined columns: impact_notes, ...6
10	Validity of column names in methods table	Undefined columns: ...30, ...32, ground_or_sieved_flag, ...35, pb210_background_assumption, ...37
11	Validity of column names in study_citations table	Undefined columns: keywords, day, ...20, issue, ...22, issn, abstract, eprint, ...30, ...31, article-number
12	Validity of variable names in depthseries table	Passed
13	Validity of variable names in cores table	Undefined variables: WGS84, riverine, palustrine, deltaic, brackish to fresh, brackish to saline, other, mudflat, plain, submerged subtidal
14	Validity of variable names in sites table	Undefined variables: palustrine
15	Validity of variable names in species table	Passed
16	Validity of variable names in impacts table	Undefined variables: managed, restoring, canalled
17	Validity of variable names in methods table	Undefined variables: PVC tube or thin-walled metal tube, Eijkelkamp peat core sampler, shovel, shovel core, gouge corer, polycarbonate tube, duplicate measurements, duplicate measurements, ground and sieved, not specified, not specified, not specified, selected intervals
18	Validity of variable names in study_citations table	Undefined variables: primary source, article

[jaxinewolfe]

Synthesis QAQC Checks:

Note: the following have now been added to the synthesis report output

Depthseries

• Cases where there are multiple observations per depth increment
• Cases where there is only one interval observation for a core (consider adding a flag for these surface samples?)
• Make sure that depth_max - depth_min is a positive number

Cores

• Check that coordinates exist for each core
• Each core is assigned a country and habitat type

Bib

• Check that there is a bib citation for every study

[jaxinewolfe]

@cheneyr

Thanks for renaming the citation tables! Some are still getting flagged in the synthesis QA and it looks like it's because the study_id was left out. (Though the citation table for Drake 2024 may still be missing from the derivative folder). So one more annoying edit there for the following:

[1] "Stahl_et_al_2024" "Palinkas_and_Engelhardt_2024"
[3] "Palinkas_and_Cornwell_2024" "Drake_et_al_2024"
[5] "Craft_2024"

[jaxinewolfe]

@BettsH Could you take a look at the coordinates for Bukoski et al 2017 in the Sanderman synthesis? They were made fuzzy per the authors request, but there are a few that have ended up in Laos when they should be in Vietnam (so, a bit too fuzzy). Maybe check out the original paper or the supplementary data and/or use google earth engine to see if we can't update those so they a least get assigned the right country.

Cores in question: "M1566" "M1567" "M1568" "M1569" "M1570" "M1571" "M1572" "M1573" "M1574" "M1575" "M1576" "M1577" "M1578"

[jaxinewolfe]

@cheneyr two tasks for you (if you haven't already done them)!

• Could you please remove the modeled fraction carbon values from Snedden_2021 (if you know what equation they used, you can make a note of this in the carbon_profile_notes of the methods table)
• Can you resolve the mismatch in the core_id's for the Drake 2024 hook? (ex. PR_M_A is in the depthseries by not the core table)

Thank you!