Hi John,
Yes, I guess there may be a difference in the way obs and model data are
handled. In model data (and perhaps many remotely-sensed datasets), the
precise geolocation data ought to be known, including the datum, and
this should be transmitted. But where there is geolocation error (in
in-situ obs for instance) specification of the datum may be less
important than conveying the accuracy of the coordinate data in some
way.
So I agree, it's probably not helpful to insist that a datum is
mandatory. But I think that guidelines and assumptions about how to
interpret different kinds of geolocation error could be valuable.
Jon
-----Original Message-----
From: cf-metadata-bounces at cgd.ucar.edu
[mailto:cf-metadata-bounces at cgd.ucar.edu] On Behalf Of John Graybeal
Sent: 19 April 2010 22:31
To: ngalbraith at whoi.edu
Cc: cf-metadata at cgd.ucar.edu
Subject: Re: [CF-metadata] Default datum for latitude and longitude?
Just because I say my parameter was measured relative to a particular
datum, says nothing about the accuracy or the measurement itself. I
don't think selection of a datum, whether as the de facto default or
explicitly, can be construed in any way to imply accuracy of the
coordinate data.
Since potential error will be introduced by the assumption of *any*
datum -- and the lack of assumption of a datum does nothing to get rid
of the potential error -- I'm not sure there is value in avoiding a
default value for datum. Well, other than saying in a loud voice
"Don't do that", which people will probably do anyway. (Hmm, maybe
there is value in this approach....)
There is a pretty good analogy to the use of accuracy metadata with
data. If this is truly important, than we accept no data without datum
(in the analogy case, we accept no data without accuracy metadata --
note how rare this practice is in the real world). But rather than
saying "data should not be accepted", do we want to say "data can be
accepted at the option of the receiving process, and should be assumed
in the default datum"?
John
On Apr 19, 2010, at 08:48, Nan Galbraith wrote:
> Just want to point out that many of our older datasets, which are
> still available - and actually still used - predate the routine use of
> GPS on ships. These datasets are from buoys that were set using
> dead reckoning.
>
> And, although we now have pretty nearly exact anchor positions,
> the buoys on our moorings deployed in about 5 km of water
> typically have 5 km watch circle diameters.
> So, although the datum specification should be encouraged, there's
> a definite drawback to assuming a default. And, the position's
> accuracy
> is probably more significant than the datum spec for a lot of in
> situ data-
> so we should really encourage the use of QC terms for all coordinates.
>
> Thanks - Nan
n Apr 19, 2010, at 08:51, Bentley, Philip wrote:
> Hi Jon,
>
> Indeed, the UM global climate model, for example, assumes a spherical
> earth.
>
> In the absence of specific datum information I think I would probably
> caution against assuming any particular datum is applicable. I can
> understand the temptation to use WGS 1984 as a default because it is
> widely known and used. However, this implies a degree of accuracy in
> the
> coordinate data which in many instances may well not be justified -
> indeed its use might even compound any relative location errors.
>
> As you rightly suggest, if the end-use/application demands a
> particular
> spatial accuracy then any datasets which demonstrably do not not meet
> those requirements should probably not be included in the analysis -
> or
> else should be with appropriate caveats, provisos and disclaimers!
>
> Regards,
> Phil
>
> Mike Grant wrote:
>> On 19/04/10 15:43, Jonathan Blower wrote:
>>
>>> specification, which datum should be assumed? Spherical Earth?
>>> WGS84?
>>>
>>
>> If you're picking one at random, I'd go for WGS84 - that's a pretty
>> safe
>> bet for a lot of remote sensed and GPS related data.
>>
>> Cheers,
>>
>> Mike.
>>
>>
>>
>>> Hi all,
>>>
>>> The CF conventions define a means to specify the datum used in
>>> certain
>>> map projections
>>>
(
http://cf-pcmdi.llnl.gov/documents/cf-conventions/1.4/cf-conventions.ht
>>> ml#grid-mappings-and-projections). In the (common) case of a data
>>> file
>>> containing latitude and longitude axes, but no explicit datum
>>> specification, which datum should be assumed? Spherical Earth?
>>> WGS84?
>>>
>>> The horizontal error resulting from use of an incorrect datum can be
>>> significant for high-resolution and/or local dataset (hundreds of
>>> metres).
>>>
>>> Cheers, Jon
>
> --
> *******************************************************
> * Nan Galbraith (508) 289-2444 *
> * Upper Ocean Processes Group Mail Stop 29 *
> * Woods Hole Oceanographic Institution *
> * Woods Hole, MA 02543 *
> *******************************************************
>
>
>
> _______________________________________________
> CF-metadata mailing list
> CF-metadata at cgd.ucar.edu
> http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
_______________________________________________
CF-metadata mailing list
CF-metadata at cgd.ucar.edu
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
Received on Tue Apr 20 2010 - 03:24:19 BST
