Duration versus date in geological time

[timbrisc]

Issue migrated from trac ticket # 167

component: help | priority: major | keywords: geological time year

2015-03-04 08:09:24: pierocampa created the issue

---

Dear experts,

I am co-chairing the Temporal domain working group at OGC (Open Geospatial Consotrium)(1), and it is our commitment to try to adhere to your standard specs whenever possible when referring to UoMs.

In particular we are starting to standardize 1D coordinate reference systems for time dimensions, which are defined by an epoch date, a direction and a temporal duration (time step). With regards to geological time scales, we found out that an authoritative article ((2), 2009) recommends the use of yr UoM whenever referring to time durations, but that seems to be absent in your spec: a and its multiples only seem to be defined and allowed.

We would like to use yr in our time coordinate reference system when specifying the time step (which is indeed a time duration).

So my question is: have you already debated about this matter? If not, are you going to?

(1) http://external.opengeospatial.org/twiki_public/TemporalDWG/WebHome

(2) http://www.agiweb.org/nacsn/40890_articles_article_file_1641.pdf

[timbrisc]

2015-07-15 23:22:32: gschadow@pragmaticdata.com changed status from new to assigned

[timbrisc]

2015-07-15 23:22:32: gschadow@pragmaticdata.com changed owner from * to gschadow*

[timbrisc]

2015-07-15 23:22:32: gschadow@pragmaticdata.com commented

---

A "yr" referred to by your reference (2) as a julian year.

We have the symbol a_j for that, and we have also later allocated the symbol a as equivalent.

The paper (2) discusses that the symbol "a" conflicts in "Pa" for Peta-year with Pascal. In UCUM we avoided this issue by determining that the year is a "non-metric" unit, in other words, the use with standard SI prefixes is not allowed.

So, what you are bringing up is that we need to revisit the decision on the year as a non-metric unit, which would introduce the conflict, or, indeed, based on this paper, and the preference of the American Institute of Physics (AIP) [Wolfe, H. C., 1962. On uniformity of international usage. Physics Today, 15: 19-20], we could add the unit 'yr', equivalent to 'a' but marked as a metric unit.

This could be done.

Proposal is to add ||year ||time ||yr ||yr ||YR ||yes ||1 ||a_j ||

[timbrisc]

2015-11-04 22:16:08: gschadow@pragmaticdata.com

[timbrisc]

2015-11-04 22:16:08: gschadow@pragmaticdata.com commented

---

Ticket retargeted after milestone closed

[timbrisc]

2015-11-13 19:27:31: gessner@mxdx.de commented

---

With regard to time I have a few comments: There is no "natural" zero for the quantity time, so time measurements are always interval measurements (i.e. "duration"). UCUM is for expressing units of measurement, and this implies that there should not be no differentiation of the quantity measured (i.e. information about the quantity "hidden" in the unit). There are many exceptions to this idealistic rule (e.g. adding units conventionally used in medicine, for the sake of patient safety), but we are against adding more exceptions deliberately. I agree that there is a difference between date, as a qualified quantity, and duration, as any quantity - but as said, this does not affect the units, only the quantity being measured (and therefore, I suggest to cosider it out of scope here).

Points in time (e.g. dates) are also given as time intervals with respect to an arbitrarily chosen "zero". So they could use the same units, although in that case other conventions (or more precisely: notations) are used frequently.

The cited article http://www.agiweb.org/nacsn/40890_articles_article_file_1641.pdf makes it very clear that this discussion is related to discriminating between different quantities, and not between the scales and units to be used.

What remains is the request to enable the use of prefixes with (at least some) year-units. Regarding the irresolvable conflict with Pascal, I am in favor of adopting "yr" as a new metric unit, but without the connotation suggested by the commenter, to differentiate between "duration" and "time before present". Second best would be to use a_j in the same sense.

[timbrisc]

2015-11-13 19:27:31: gessner@mxdx.de

[timbrisc]

2017-11-22 01:15:38: gschadow@pragmaticdata.com

[timbrisc]

2017-11-22 01:15:38: gschadow@pragmaticdata.com commented

---

Ticket retargeted after milestone closed

[timbrisc]

2021-02-16 12:47:06: gschadow@pragmaticdata.com set component to help

[timbrisc]

2021-02-16 12:47:06: gschadow@pragmaticdata.com commented

---

Coming back to this for the board meeting assignment.

Christof and I agree that there isn't any issue with "date" or "point in time", "date", or "timestamp". A point in time is not really a measurement with units.

By the way, I just have to deposit here the broader discussion on the relationship between physical quantities and points in time, intervals, and periodic intervals, which is found here: https://www.hl7.org/documentcenter/public/wg/inm/Acf302.pdf in an earlier but fairly complete version. This here might be a more final specification: http://vico.org/HL7_RIM/infrastructure/datatypes_r2/datatypes_r2.html but it might be more terse on this subject.

Now coming back on actionable proposal, the only thing left is the addition of "yr" unit for year, defining it equal to "a_j" but marking it as "metric" to be able to say "kyr" and "Pyr". Now this may not even be necessary if "a_j" is used, not "a" but "a_j". If we just mark "a_j" as "metric", then you can say "ka_j" and "Pa_j" without a problem, where "Pa_j" doesn't mean a Julian pascal, but according to all syntax rules should be completely unambiguously understood as peta-year.

---

PS: I notice that the original ticket workflow from the unitsofmeasure.org site is not available here (any more?) There were statuses such as "propose reject", etc. that are no longer there. What happened?