Hillshade
IfSAR DTM (~2010), 5 m product.
Terrain base for context; native ~5×9.8 m ground pixels.
Glacier outline
Manually digitized from Landsat; areas from UTM 5N geometry.
2005 is a byte-identical copy of 2000 — 5 independent epochs, not 6 (duplicate dropped).
Branches
North/middle/south branch polygons (2000 extent).
Stakes sit on the south branch, ~10–30% snowier than the glacier mean — basis of the β_w bias term.
Ice thickness
Ice-thickness model — farinotti track.
Farinotti vs IceBoost volumes disagree ~38% (6.90 vs 9.07 km³); no local measurement arbitrates — both carried.
Snowlines
End-of-summer snowlines; elevations re-derived per branch against the IfSAR DTM.
SAR-timing trust filter keeps 5 of 22 scenes; the middle branch (nunatak horseshoe) is excluded.
Stakes
Spring-2024 GNSS survey (UTM 6N, reprojected).
Per-observation σ from the field debrief; 5 estimated 2025 entries flagged, the ACC-2025 estimates excluded from the likelihood.
AWS
On-glacier AWS at 1078 m — hourly air temperature + rain.
2025 temp/RH sensor failed 2025-07-13 (intermittent); post-clean-hour data masked, rain stays valid.