Known Issues With LinkedEarth Representation of the PAGES2k Dataset
This page lists the issue identified with the representation of the PAGES2k dataset in LinkedEarth.
Contents
Ghost measured variables
Because the LinkedEarth Ontology requires the representation of the archive type to be made on a measured variable, it was decided to create "ghost" measured variable if only the inferred variables were archived. The following datastes have improperly set the inferred variable to a measured variable with a Property:ProxyObservationType (L) set to "Ghostmeasured".
Instead, these datasets should have an additional entry in the JSON-LD file with the proper Property:ProxyObservationType (L).
Earliest Sample Date/Latest Sample Date
The period covered by the record should be provided at the variable level through the HasMinValue and HasMaxValue properties.
EarliestSampleDate
LastestSampleDate
Calibration models on measured Variable
The newest version of the LinkedEarth Ontology should only allow the calibration model to be attached to an inferred variable through the property calibratedVia. Some PAGES2k datasets have information relating to calibration on the measured variable.
Improper naming of ProxyObservationType
Some types have a number after them probably lifted from the name. Should be removed. This inlude:
- d18O1
- calcification 1,2,3,4,5,6,7
- count (1 thru 18)
- trsgi (1 thru 4)
- temperature (1 thru 3)
In addition, uncertainty_temperature and notes should not be types of proxy observations.
Improper formating for trace element names
The LinkedEarth Ontology uses the proper "/" for trace elements rather than a space as used by PAGES 2k, therefore "Mg Ca" should be "Mg/Ca".
Mg/Ca
Sr/Ca
C/N
SensorSpecies placed under the variable category
In the LinkedEarth Ontology, the proper path is Category:MeasuredVariable (L) -> Property:HasProxySystem (L) -> Category:ProxySystem (L) -> Property: ProxySensorType (L) -> Category:ProxySensor (L) -> Property:SensorSpecies (L).
Note: This may be a problem with the mapping rather than the PAGES2k LiPD files.
Physical Sample
For most datasets, the name of the physical sample doesn't correspond to anything. In some instances, it's just an fancier name for the timeseries. Should be removed.
LiPD version mixture
Some files have shown to have an invalid mixture of LiPD v1.2 and v.1.3 structures. The known issue so far is ensemble tables are an incorrect data type. In v1.2, ensemble tables are a fixed single entry for one table. In v1.3, ensemble tables are an array of multiple possible entries for multiple tables. Files downloaded from the Wiki have ensemble arrays as per v1.3, but have other characteristics of v1.2 like "chronModel", "chronMesurementTable", "chronDataTableName" and other discontinued keys.
