Hook Script Reviews

[jaxinewolfe]

Revisit all hook scripts in the Data Library. Make sure the scripts run properly, perform QAQC on the data, and update citations (see issue #64). Please note any problems that arise for particular hook scripts in this issue thread!

Standard quality checks for each hook script:

• testTableCols(): test that column names in each table match CCRCN guidelines
• testTableVars(): test that variable names in each table match CCRCN guidelines
• testRequired(): checks for the presence of mandatory and conditionally mandatory attributes
• test_unique_cores(): test that all core IDs are unique within the study
• test_unique_coords(): test that all core positions are unique (Soon to be replaced by checkDuplicates())
• test_core_relationships(): test that core IDs in the cores table also exist in the depthseries table (and vice versa)
• fraction_not_percent(): test that attributes are expressed as fractions and not percents (where applicable)

• test_numeric_vars(): summarize values in numeric columns for the given table (usu. depthseries)

  Coming soon to a qaqc script near you:

• checkDuplicates(): check if duplicate cores exist in the database
• checkConditional(): check if conditional variables and attributes exist for applicable attributes

Please check off the primary studies you revisit so we can keep track of progress and avoid duplication of effort:

• [x] Abbott_2019
• [x] Arriola_2017
• [x] Baustian_et_al_2021
• [x] Belshe_2019
• [x] Boyd_2019
• [x] Breithaupt_2014
• [x] Breithaupt_2020
• [x] Buffington_et_al_2020
• [x] Callaway_2019
• [x] Chambers_et_al_2019
• [x] Craft_2007
• [ ] Craft_unsorted - no hook script
• [ ] Deegan_2012 - no derivative folder
• [x] Doughty_2019
• [x] Drexler_2009
• [x] Drexler_2013
• [x] Drexler_2019
• [x] Elsey-Quirk_et_al_2011 - no hook script
• [x] Ensign_et_al_2020
• [x] Fourqurean_2012
• [x] Gerlach_2017
• [x] Giblin_2018
• [x] Gonneea_2018
• [x] Holmquist_2018
• [x] Johnson_2007
• [ ] Jones_2017 - no derivative folder
• [x] Kauffman_et_al_2020
• [x] Kemp_2020
• [x] Keshta_et_al_2020
• [x] Krauss_2018
• [x] Lagomasino_et_al_2020
• [x] Luk_2020
• [x] McTigue_2020
• [x] Messerschmidt_2020
• [x] Nahlik_Fennessy_2016
• [x] Noe_2016
• [x] Nolte_2020
• [x] Osland_2016
• [x] Peck_2020
• [x] Piazza_2020
• [x] Poppe_2019
• [x] Sanborn_Coxson_2020
• [x] Sanderman_2018
• [x] Schile-Beers_2017
• [x] Smith_2015
• [x] StLaurent_et_al_2020
• [x] Thom_2019
• [x] Thorne_2015_a
• [x] Trettin_2017
• [x] Vaughn_et_al_2020
• [x] Watson_Byrne_2013
• [x] Weis_et_al_2001
• [ ] Yando_2016 - data download script?

[jaxinewolfe]

General Notes:

• Commenting out code that involves pulling data from a URL and depositing it in the repo so we just load the local data tables (but retain a record of the original download).
• There will still be a site_id in cases where there is no site table for a study
• Is there a reason that some studies lack a derivative folder?
• Salinity_class variables palustrine and fresh have the same definition in our database (<0.5 ppt). Should these be differentiated?
• See resolve_uncontrolled.csv to resolve uncontrolled attributes and variables
• Studies with no methods: Trettin_2017, Osland_et_al_2016, Gonnea_2018, Giblin_2018, Schile-Beers_2017, Sanderman_2018
• Studies that are considered syntheses: Poppe_2019, Schile-Beers_and _Megonigal_2017, Sanderman, Holmquist_2018, Forquean, Boyd et al 2019
• Studies with missing required attributes
• Add a QA function to check if a bibliography ID will be unique to the synthesis

Species Table Notes:

• There are species that are not described/categorized in the species fixes table. This prevents the assignment of habitat and code_type (ex. StLaurent et al 2020). Should we turn this table into a dynamic taxa database?
• Descriptive species codes are assigned to specific habitats in the species fixes table (i.e. Mix is assigned to marsh). Should this be resolved at the hook script level instead?
• Should there be a function to resolve taxa names (i.e. correct misspellings)?

Hook Script Notes:

• Thorne_2015_a: Added a missing citation for the associated article.
• Thom_2019: Workflow wasn't outputting the cores table for some reason. Core_id PB1 exists in the depthseries table but not the core table (Jim says this is OK).
• Smith_et_al_2015: Created citation for the PhD thesis.
• Schile-Beers_2017: elevation_datum of WGS84. Vegetation_notes indicate habitat, should we translate this to a habitat col?
• Sanderman_2018: Methods are created in the script but not output to the derivate folder. Are they incomplete?
• Holmquist_2018: No site_id in the depthseries table. Uncontrolled variables in the methods, cores, and impacts table that need review. Define habitat for species Nuphar advena.
• Piazza et al 2020: Still need to resolve the issue of core_id's that exist in the depthseries but absent from the cores table. These cores are currently available through the Holmquist synthesis.
• Fourqurean_2012: one core from Copertino_unpublished does not appear in any other table (and has no data in the cores table aside from its core_id). Drop it?
• Krauss et al 2018: site table is made from the cores table, is this redundant or should we keep it in?
• Ariolla and Cable 2017: Labeled as a combined dataset and manuscript pub type.
• Gerlach et al 2017: Is there a primary dataset?
• Gonnea et al 2018: what is the core length flag for Hamblin Pond?
• Nahlik and Fennessey 2016: Added two citations => one for the published data and another for the tech report
• Osland et al 2016: 'species_codes' with undefined habitats: Crassostrea virginica, Litter, Wrack. position_method only provided for one entry in the cores table.
• StLaurent et al 2020: species_codes with undefined code_types and habitats: Scirpus spp, Atriplex patula, Pontederia cordata, Schoenoplectus pungens
• Weis et al 2001: Is there a primary dataset?
• Boyd et al 2019: Position method for some studies is missing but the position notes indicate handheld GPS or better.
• Peck et al 2020: Methods need to be split (based on excess pb210 modeling)

[jaxinewolfe]

Studies with duplicate core_ids: Sanborn_and_Coxson_2020 and Baustian_et_al_2020 share core_ids 398 and 399.

[jaxinewolfe]

Methods ID Assignment

Up to this point, when different methods were encountered within a study, we assigned different study IDs (ex. Keshta_et_al_2020_a, Keshta_et_al_2020_b) to the associated cores. Moving forward, we will be assigning a methods_id to distinguish unique methods used in studies. In this way, information collected using mixed methods will still be associated with a single study ID. This methods_id will be added to the database guidance for the following tables: methods, cores, and depthseries. The following studies have been identified as having mixed methods:

• [x] Breithaupt et al 2014 (2 methods)
• [x] Luk 2020 (2 methods; split core study)
• [x] Krauss et al 2018 (3 methods)
• [x] Keshta et al 2020 (2 methods)
• [x] Chambers et al 2019 (3 methods)
• [ ] Noe et al 2016 (2 different coring methods)
• [ ] Peck et al 2020 (2 methods...maybe)

[HolmquistJ]

I think that methods_id should only be indexed in the methods and depth interval tables since one core can have more than one set of methods associated with it. For instance, in the Luk case where they split the core and did different analyses on the different halves or that other case where they had different LOI/BD sample volumes for higher and lower depth increments.

[jaxinewolfe]

We will be reinstating the testConditional() fxn in hook script QAQC. I've performed an initial sweep of the synthesis to compile a table which lists missing conditional attributes for each study. These hook scripts will be revisited for revisions (if possible/necessary).

[jaxinewolfe]

@BettsH @cheneyr Just putting this issue on your radar! Solidifying our standards, QAQC tests, and introducing a new database version last year led to revisiting all the hook scripts. This is where we've been tracking those revisions and documenting things to address in the future that will make this synthesis even more solid. For example, there are some studies that don't have methods tables yet, which is probably one of the top things I'd like to address (esp. since we want to create a workflow eventually which derives carbon stock etc. reliably from these data). I'm thinking we can do a hackathon in the near future to work through a bunch of these.