redefine texture fractions

[bkempen]

The various SandSizeFractions, SiltSizeFractions and ClaySizeFractions (https://vocab.isric.org/glosis_cl/en/) are not really usable since fraction size is not specified (e.g. https://vocab.isric.org/glosis_cl/en/page/sandPropertyCode-SSF1).

The particle size distribution of each fraction should ideally be defined explicitly (for instance based on two/three most commonly used particle size distributions used in soil science). E.g. SandFraction_0.25-0.5mm, SandFraction_0.5-2mm, SiltFraction_0.002-0.02mm, SiltFraction_0.02-0.05mm, SiltFraction_0.02-0.063mm, etc. This would make these entries much more usable. Fractions of various classifications can be considered (e.g. ISO, USDA)

Consider removing current definitions (e.g. SandFraction01...SandFraction09) from the vocab. These come from a previous soil data standardization project at ISRIC defined in a way that the user can specify the particle size for each fraction. As such these are not useful to incorporate in a general vocabulary. In that case, particle size distribution should be explicitly defined.

(relates to #113)

[ldesousa]

These properties were directly derived from the Tier 1 / Tier 2 document submitted by ISRIC to INSII/GSP. I am happy to remove them, but keep in mind they are also present in the GloSIS domain model (UML models and report produced for FAO/GSP in 2019).

There is a code-list for texture Procedures that is referenced in the observations linked to the properties glosis_cl:physioChemicalPropertyCode-Textclay, glosis_cl:physioChemicalPropertyCode-Textsilt, glosis_cl:physioChemicalPropertyCode-Textsand. This code-list does not feature yet in the ISRIC vocab.

[ldesousa]

@bkempen In addition I will also remove these Properties:

• SiltWaterDispersibleFraction01
• SiltWaterDispersibleFraction02
• SandWaterDispersible01
• SandWaterDispersible02
• SandWaterDispersible03
• SandWaterDispersible04
• SandWaterDispersible05

[bkempen]

@ldesousa I would wait with removing until alternatives are added. These should be texture fractions with indicated particle size range. Starting, for instance, with texture fractions as defined by ISO 14688-1:2017 and the USDA and WRB classifications (see tables above). I cannot oversee the consequences for the GLOSIS domain model, perhaps that one should be updated accordingly??

[ldesousa]

@ldesousa I would wait with removing until alternatives are added. These should be texture fractions with indicated particle size range. Starting, for instance, with texture fractions as defined by ISO 14688-1:2017 and the USDA and WRB classifications (see tables above). I cannot oversee the consequences for the GLOSIS domain model, perhaps that one should be updated accordingly??

I see what you mean now. These are not properties, but rather values for the observations glosis_lh:Sand, glosis_lh:Silt and glosis_lh_Clay. However those are supposed to have numerical results instead. This is going to take some work to disentangle.

Envisioned tasks:

1. Confirm the existence of just three properties (Sand, Silt and Clay) and add them to the Physio-Chemical concept scheme.
2. Correct current observations, limiting result to numerical.
3. Create new observations:
   a. SandClass
   b. SiltClass
   c. ClayClass
4. Create new values code-lists:
   a. SandClassValueCode
   b. SiltClassValueCode
   c. ClayClassValueCode

[ldesousa]

The numerical properties are already in the Physio-Chemical concept scheme.

PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
PREFIX glosis_cl: <http://w3id.org/glosis/model/codelists/>

SELECT DISTINCT ?prop ?label
WHERE {
    ?prop a glosis_cl:PhysioChemicalPropertyCode ;
               skos:prefLabel ?label .
    FILTER REGEX(?label, "fraction")
}

prop	label
http://w3id.org/glosis/model/codelists/physioChemicalPropertyCode-Textclay	"Clay texture fraction"@en
http://w3id.org/glosis/model/codelists/physioChemicalPropertyCode-Textsand	"Sand texture fraction"@en
http://w3id.org/glosis/model/codelists/physioChemicalPropertyCode-Textsilt	"Silt texture fraction"@en
http://w3id.org/glosis/model/codelists/physioChemicalPropertyCode-Calfra	"Calcium carbonate equivalent - fraction"@en

[bkempen]

Sand-silt boundary should be 63 instead of 64 in the procedures list:

https://glosis-ld.github.io/glosis/glosis_procedure/index-en.html#/TextureProcedure

[ldesousa]

@bkempen Can you please have a look at the table below? It gathers all the Texture procedures from the AfSP data. Two questions:

• Do they all apply to the fractions identified in the top of this issue?
• Are the disp and nodisp items really necessary? Or can be removed?

Label	Description
adj100	fractions adjusted to sum up to 100%
disp	with appropriate dispersion
disp-beaker	with appropriate dispersion-beaker
disp-hydrometer	with appropriate dispersion-hydrometer
disp-hydrometer-bouy	with appropriate dispersion-hydrometer-Bouyoucos
disp-laser	with appropriate dispersion-laser
disp-pipette	with appropriate dispersion-pipette
disp-spec	with appropriate dispersion-spectrally measured and convertd to sand, silt or clay
fld	field estimate
nodisp	no dispersion
nodisp-hydrometer	no dispersion-hydrometer
nodisp-hydrometer-bouy	dispersion-hydrometer-Bouyoucos
nodisp-laser	no dispersion-laser
nodisp-pipett	no dispersion-pipette
nodisp-spec	no dispersion-spectrally measured and converted to sand, silt or clay

[bkempen]

@ldesousa In my view 'adj100', 'disp', and 'nodisp' can be removed from the list. Furthermore, if I have understood 'hydrometer' is the same as 'bouyoucos'. I therefore suggest to remove 'disp-hydrometer-bouy' and 'nodisp-hydrometer-bouy' as well, and update the descriptions of the hydrometer methods to 'hydrometer (bouyoucos)'.

The main analytical methods are then: beaker, hydrometer (bouyoucos), pipette, laser diffraction, field estimate.

I believe the information about application of a dispersion step is required. I see three options here: 'dispersion', 'no disperson', 'disperson unknown'. Each analytical method (except the 'field estimate') needs to be associated to each of these three dispersion options.

Then there are particle size measurements determined by spectral modelling. Here it is important to have information about the lab method used to determine the particle size measurement of the dataset used to train the spectral model. So I believe the spectral method must always be connected to a wet chemistry method. Having only 'disp-spec' and 'nodisp-spec' is too limited and will hamper data standardization and exchange efforts.

So I foresee a set of procedures that has the analytical method, a dispersion indicator, a spectral indicator (except for the field method). I will share a list with proposed procedures for consideration.

[bkempen]

@ldesousa attached is a table with a proposed list of sand, silt and clay fractions, including definitions, following the USDA, ISSS, WRB and ISO14688 standards. (note the list also contains definitions of the various Aluminium observations that we have also discussed; including a new observation 'Aluminium extractable' that I was missing in the ontology).

The 'procedures' sheet contains a proposed list with procedures (label, code and definition) following the reasoning in my previous comment.

NewGloSIS_PhysioChemical_BK.ods

[bkempen]

Updated table:
NewGloSIS_PhysioChemical_BK.ods

[ldesousa]

@bkempen I corrected the labels:
vcSa_WRB
cSa_WRB
mSa_WRB
fSa_WRB
vfSa_WRB

to:
vcSa_ISSS
cSa_ISSS
mSa_ISSS
fSa_ISSS
vfSa_ISSS

[bkempen]

@ldesousa I do not understand this correction. These fraction definitions relate to the definitions according to the WRB classification, not the ISSS classification.

[ldesousa]

Good that I asked, will restore to the original. Can you point me to the document/resource for the ISSS definitions?

[bkempen]

I got the ISSS definitions from this one (it also has the USDA definitions):
https://www.sciencedirect.com/science/article/pii/S0016706122000908

[ldesousa]

Thanks @bkempen. I now have all the info to complete this issue.

[ldesousa]

Closed with release v1.4.0.