[Cube/ImageDM] - Characterisation and Accuracy

[CresitelloDittmar, Mark]

Mireille,

>I would just insure that the Characterisation DM is still in agreement.

Yes.. I hope that a greater consistency between the models is a primary
goal for this effort.
This is one reason that I (for Cube) and Arnold (for STC) have been working
with Gerard to
aid in his efforts to represent them in VO-DML.  With Characterisation also
folded in there,
the task of identifying inconsistencies would become fairly simple.
(Granted, this is a longer
term goal :) ).


As for now.
  CubeDM uses Characterisation-1.13 as is.  So, unlike the Spectral model,
there is no
  chance of being out of sync.

  But I believe you mean between Char and STCMod ?
    - that shouldn't be necessary since Char-1.13 specifically pulls from
STC-1.33
      so that remains unchanged.
    - the Char.Accuracy content would still be summarized from the STCMod
coordinate content.
    - when Char moved to a modified STC, they would come into sync.

Char Error model:
   I have done some diagramming of this part since I wanted to be sure the
STCMod coordinates
   included all of the pieces that Characterisation expected. (see: Char
Model Overview)
   NOTE: the connection from ErrorRefValue and resolution ReferenceValue ->
STC is not diagrammed.

   + char:ErrorRefValType ==> stc:CError substitution group (so is in
STCMod:Error model diagram)
   + char:ErrorBoundsType ==> stc:coordIntervalType  which is unchanged in
STCMod save for the base value type as Quantity.

   The problem with the stc:CError substitution group tree is that it can't
be directly represented
    in UML (I don't think so anyway).  The best I could do is create an
object tree with the role
    as an attribute of the relevant type.

>About the term "Uncertainty", which is new , I think we should use
Accuracy which already exist >instead.
>Same concepts with two different names might lose our standard  readers /
implementers .

Equating objects, Char:Accuracy would be a collection of "Uncertainty"
elements.  Accuracy.SysError and
Accuracy.StatError would both be of type "<Uncertainty>", as would
Char.Resolution.ResolutionRefVal.

I'm flexible about the name of the object. I chose 'Uncertainty' to convey
some idea of the role they are
used for, rather than as a set of shapes.

Mark

In Response To:
>
>Thank you Mark for this careful improvement and follow up of Image/DM.
>
>I haven't check your update in details yet .
>I would just insure that the Characterisation DM is still in agreement.
>That should be the case , because syserror and staterror inside Accuracy
were based on the STC >object you rely on .
>Same for Resolution.
>
>About the term "Uncertainty", which is new , I think we should use
Accuracy which already exist >instead.
>Same concepts with two different names might lose our standard  readers /
implementers .
>
>Is it really a new concept?
>
>best wishes , Mireille
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.ivoa.net/pipermail/dm/attachments/20140502/b208a2ee/attachment.html>