GLIMS Minimum Requirements list

[J. Graham Cogley]

     When I collect things, I find it helpful to have both a unique ID
and an "accession number" (code, actually, since it doesn't have to be 
numeric).  The accession code simply reflects the order in which I collect
the things, and the unique ID is - well, unique.

     Perhaps these two concepts are redundant, but I think there is some
redundancy already: the unique-ID concept seems to be getting entwined
with the location concept. If we are allowed two concepts, I think the
two best would be accession-code and location; if only one, I'm not sure.
Two would definitely be preferable. The accession code could have a 
prefix identifying the collector, perhaps.

     Pathological cases: of "our" two glaciers, one (Thompson) is
advancing and is displacing the other (White) which is retreating.  But
surely it should be pretty safe to choose the location somewhere in
the accumulation zone?

     However, that reminds me of another problem which I haven't seen
discussed yet - glaciers which cease to exist.  What do we do about them?
Mauri Pelto, for instance, reports the disappearance of 2 or 3 of his
North Cascade glaciers in one of his recent papers.

     The Minimum Requirements list looks fine to me, though.

     Graham Cogley.


-----Original Message-----
From: owner-glims_database at flagmail.wr.usgs.gov
[mailto:owner-glims_database at flagmail.wr.usgs.gov]On Behalf Of Bruce
Raup
Sent: 10 July 2001 6:00 pm
To: GLIMS Database mailing list
Subject: Re: GLIMS Minimum Requirements list


Jeff,

Yes, we need written guidelines for basic processing steps, such as
generation of the glacier ID, formats for data transfer (that's a big one,
and yes, Luke, we're working on it here), error/uncertainty reporting,
handedness of polygons, etc.  I first wanted to get everyone's nod of
approval (at least through their silence) on this list of minimum
requirements.

Regarding ID generation:  IDs will be produced by different people
simulateously.  Thus, a scheme is necessary so that IDs are still unique,
even without central authority over the process.  This could be done in
different ways, but the one "decided on" was to encode the location into
the ID.  This has the side benefits of being human-interpretable, and
computer searchable (that is, you could do geographic searches just using
IDs if you wanted to).  A potential drawback is that small glaciers are so
close together (less than 100 m) that it's difficult to create unique IDs
for them.  I don't think this should be a problem, though.

[Another way to do the IDs is to have each RC generate a series of numbers
starting with 000001, and prefix that number with the RC ID.  If compelling
reasons are brought up to do something like this, then it's not too late,
if the change is done quickly.]

Assuming we stick with the lon/lat method, Jeff, your thoughts match mine:
that the ID should be roughly on the middle of the centerline.  A simple
average of the lon/lat coordinates doesn't work, since for a curving
glacier, this average position could be in the middle of a different
glacier.

> Are there technical reasons that would cause a hiccup in the
> database if the ID number evolves over time to a place no longer
> on the centerline or even outside the glacier?

Once the ID is set, it won't change, even when the glacier does.  If, in
the future, a glacier recedes past its ID position, there shouldn't be a
problem with retaining the ID that is now over dirt, since it's unlikely
that a different glacier will advance to this position.  Anyone know of
pathological cases where this might happen?

> A point of possible contradiction: "Centerline" is listed as a
> desirable entry; the glacier ID is mandatory but must be on
> the centerline, which suggests that the centerline or at least
> one point of it is mandatory.

I guess I used the word "centerline" in two different ways, one being a
rough idea of where the middle of the glacier is, and the other being a
digitized trace of the central flowline.  I should clarify this in the
min. req. list.

> uneducated opinion: I think that the mandatory glacier ID ought
> to be the lat/lon just somewhere near the center of the glacier
> (roughly the center estimated by eye: roughly along the
> centerline, and roughly half way from summit or top of
> accumulation zone to the toe of the glacier).  Are there any
> science/application reasons or technical reasons that would
> prevent this definition (lat/lon of approximate
> glacier center) of glacier ID?

If there are, they should be identified soon!

-- 
Bruce Raup
National Snow and Ice Data Center                     Phone:  303-492-8814
University of Colorado, 449 UCB                       Fax:    303-492-2468
Boulder, CO  80309-0449                            Bruce.Raup at colorado.edu