The `anlys_id` for Kahiltna Glacier does not exist in GLIMS

[ezwelty]

RGI version
RGI v7.0

Describe the issue
Found a case of an anlys_id value in RGI 7.0 not matching any in GLIMS. Have not checked whether this anlys_id used to exist (in which case this would be an issue with GLIMS) or how widespread it is.

Expected outcome
All anlys_id in RGI 7.0 should exist in GLIMS.

Screenshots

Suggested solution

• Check prevalence of problem
• Check whether missing records were deleted from GLIMS (hence a GLIMS issue)
• If not, check RGI 7.0 production scripts for how invalid anlys_id came to be used

[fmaussion]

Uh that's a good one (as in: interesting one). @bruceraup do you have any idea? Otherwise I'll have to dig deep in the RGI scripts.

[bruceraup]

Spot-checking, it seems common in Alaska, but not a problem over Mt. Rainier. That's all I've checked now. In my database redesign, a few IDs did change, but this doesn't appear to be the cause in Alaska, since the correct analysis IDs seem to be lower numbers than what's in RGI7. The other possibility is that in my data move, the outline with the correct analysis IDs got dropped? I hope that's not the case, but I'm just brainstorming. I do still have the old database so can check more thoroughly in coming days.

[ezwelty]

So actually, anlys_id: 396706 is present in GLIMS 20250318 (by which I mean, the pre-packaged north/south database dumps with that date). But it is not found by GLIMS Search (https://www.glims.org/maps/textsearch/). Perhaps this is a problem with the web interfaces? Or was this record changed or altered since 2025-03-18? (I didn't check GLIMS 20250408.)

There are however 123 anlys_id in RGI 7.0 not present in GLIMS 20250318. All are from region 19 (which also, spot checked, are missing from GLIMS Search) except for 3 in region 10 (which are in GLIMS Search).
rgi70-glims20250318-anlys_id.csv

Comparing to GLIMS 20230725 (north: 20230725, south: 20230607), those 3 from region 10 are the only anlys_id missing.
rgi70-glims20230725-anlys_id.csv

[fmaussion]

Sorry, digging this out now as I am doing RGI work - do I understand correctly that it's unlikely to be an issue in the RGI scripts?

RGI scripts used to access GLIMS differently, i.e. not from the global pre-packaged database dumps, but via non-documented API download. Could it be that something happened between the download and the dump?

[ezwelty]

@fmaussion Not sure, I don't have any additional information to provide. @bruceraup Do you have any information on this? I was hoping to use lookup tables to link RGI 7+ ids to GLIMS geometries, but if RGIs derived from GLIMS cannot be guaranteed to be available from GLIMS at a later date, such an effort is compromised.