Meeting Notes 2004.05.18

Standard Naming Exercise

MeetingNotes20040518 (meeting 2)

What do we want to do today?

• fill in blanks

• standardize domain column items

• • this is useful regardless of how the results will be used (e.g., ocean/river/lake vs water)

  • could construct a class hierarchy (but how useful is that?)

  • will get gnarly when we get to current

• discuss how units column should be used

• standardize most basic variable names (large percentage of file contents)

• • how to indicate quality flag, e.g., _QF?

• identify references

=== Review of Element Issues ====

• how is domain applied?

• • use the most generic term for it

• should Domain always be a part of name?

• • appears to be a user- and use-specific issue

  • keeping domain and parameter names separate is very useful (sayeth the DB pros)

• distinction between bottles and in-situ?

• • not at this level, one may be treated as the other

• columns 'instrument' and 'sensor' confusing (which is which?)

• • could improve definitions in the spreadsheet

  • how primary is this information?

• units has been subject to a lot of work

• • don't store as knots, since it isn't metric (or maybe just be sure to convert for science use)

  • have precise terminology, so we know degC = degrees C = centigrade

Discussion of Preferred External Standards

McCann: [http://sweet.jpl.nasa.gov/sweet/esmf.xls esmf.xls-ESMF/CF Taxonomy] [http://sweet.jpl.nasa.gov/sweet/gcmd.xls gcmd.xls-GCMD Keywords Taxonomy]

• extended presentation of variables vs Realm, Phenomena, Property, Substance, Space, Numerics, Time, Biosphere, Services, Data

• lots of measurements exist, across a wide range of domains

• includes domain in some of the names

• this ontology is used in the registration of data sets (via form) at GCMD

• form exists to enter new terms for describing their data

• • indication is NASA is quick to respond to new requests

  • Gene Major is implementing the GCMD, process for adding names is not rigid

  • we could use these for comparison as we're filling out our variables

  • how mature is this? can we use it as a sole basis?

  • • interesting that Earth Realm doesn't include atmosphere in several places that it might

    • using this model, "first submission wins", so it may not be totally systematic

  • what distinguishes between CTD profile and measurement at given depth?

  • • this may be different kind of information, not captured in this information space

• we're looking at automating processes, rather than providing human-searchable data

Kolber: [http:// ARGO Data Management User Manual] (.pdf)

• contains a very well described structure with detailed info on lots of variables

• variable names are short and do not include domain, which is expressed in a long name

• how to handle different domains in short name?

• • short name tends to be meaningless, since long name is what appears on plots and contains domains

  • interoperability affected by repeated use of same names

• ARGO is dealing with limited data types -- this simplifies the issues (e.g., they are not dealing with water temperature and air temperature)

• might be interesting to take some of these common names (lat, long, jday, depth) and map the information they provide to our spreadsheet

• • ARGO vs GCMD? neither one necessarily what *we* wanted to do

McCann: [http://sweet.jpl.nasa.gov SWEET Ontology (with Protoge)]

• download Protege from Stanford site, and install Ontology Web Language (OWL) plugin

• can also view on web (but not with Safari)

• tool can be used to create new ontologies or additions to existing ontologies

• large user and developer community for this tool

• submissions can be submitted to web library

• could take SWEET ontologies (which are based on GCMD) and add our own information

• putting cart before horse?

• • take spreadsheet and group thing together

Others For Future Discussion

• Schramm: EPIC Key File Variables

• • contains a long list of variables and associated information

  • variables are organized by numeric indices

• Graybeal: SensorML

• • an ontology for what data describes sensors (their location, etc.)

  • recent release just came out

• MarineXML

• • complicated histories here: European and Australian versions

  • BODC reconciled European version with their own? Australian based on BODC originally?

  • BODC has old data sets that have to be compatible

  • very hard to keep up with this work or know where it's going

• starting to id discrepancies with other references could start feeding this back to those references

Completing the Parameter Column of the Compiled Spreadsheets

• lots of classes (for an ontology) becoming apparent

• starting to normalize some names

• time could be challenging to represent

• almost through this column

The Plan

Get together next Tuesday at 8:30 AM to continue this process.

• Anyone who wants to fill out the easy Parameter entries for the rest of their data, send results to John

• Next steps include:

• • Finish the Parameter column

  • Do the same for Domain as we did for Parameter

  • Go down the Name In Use column and propose better names, given what we now know