GLIMS definition issues

[Hugh Kieffer (GD.Flagstaff) (520)556-7015]

Bruce:
 I am just back from ~4 weeks travel. 

I think the glacier definition is fine.


Re ice shelves.

 1. I guess exclusion of ice shelves is strictly appropriate for the L
 in GLIMS. However, excluding them will break a tie to historic
 observations and discussions. A place where this
 is particularly important is in the Antarctic Penisula RC, and we
 should get their input.
   If ices shelves are dropped completely, then
 the paragraph is OK as stands, with the possible exception of:
 2.  I thought "grounding line" was a more common term than "grounding
 zone"; how do you define the latter?


Re coordinates:

Johns' discussion is usefull, and points out some of the subtleties
which must be addressed.  A (perhaps unexpressed) point to having both
N/E and L/L is to separate the issues of precision(N/E) and accuracy
(L/L). The challenge is this: when someone pulls polygons for single
glacier at two times, there must not be appearent change in the
boundary location due solely to inaccuracy of georeferenceing the two
source images. If the GLIMS DB had only L/L coordinates, AND the
uncertainty in absolute locations in either/both scene were comparable
to the changes that actually occured, then the user would get an
incorrect measure of change unless they projected the L/L information
for both polygons into some local cartographic system (E.g., a N/E
system) and slid one relative to the other to removed the absolute
offsets.

Keeping the N/E coordinates, with an estimate of the uncertainty of
the absolute locations, forces (we hope) the user to explicitly deal
with these uncertainties in combining scenes or looking at temporal changes.

Re the datum to be used for any specific image. 
  I think the RC's should have significant say here.  The central
  parallel and meridian must be close enough to the scene that
  distortion (sphere to plane) is insignificant. In many cases, the
  largest contribution to absolute location inaccuracy will be the
  "dead-reckon" error of the scene, and perhaps the absence of any
  elevation control, not cartographic distortions.  Thus, the need for
  a local immobile feature to be used in "registering the rocks" for
  temporal ice changes.

The bottom line is that users must be aware of the georeferenceing
precision/accuracy issues in using GLIMS DB polygons.

L/L will probably have to be in double precision, N/E can be single precision.


 
Hugh

-----------------------------------------------------
Hugh H. Kieffer         hkieffer at flagmail.wr.usgs.gov
U.S. Geological Survey  Phone  928-556-7015 <-- Note new area code 
2255 N. Gemini Drive    FAX           -7014
Flagstaff, AZ 86001     Main Office   -7000 
-----------------------------------------------------