Files
aimsight-labeling/auto/analyses/long-labels/NEBRASKA/analysis.md
T
maximeandClaude Opus 5.5 ddb17c2189 auto: labels requalified under the 5-noise rules (only what stands out from the noise): 32 analyses refined, the 6 paused periods verified, the long labels requalified with the fixes the period analysts found; 1153 -> 961 labels
- 32 analyses: each label kept, rewritten or removed by its analyst (section Requalification of the report; mm magnitudes where the targets have them)
- SNCF01 x3, WILDSTRUBEL 2026-09-01 -> 2026-10-06 x3: verified from their existing labels under the new rules
- long labels (no analysis): auto/analyses/long-labels/<STATION>/; NORDEND-169 and -195 run to 07/11, so NORDEND-283 and -284 (their continuation) are removed
- reading-rules.md, stations.md: the citations of removed or rewritten labels point to their successor, or to [expert]

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-17 23:58:00 +01:00

2.0 KiB
Raw Blame History

Requalification of the long labels: NEBRASKA

Rules: labeling/labeling-rules.md section 1 (threshold of 5 own noises on corr_x, corr_y, or their mm), labeling/families.md. Cadence 4 h. Evidence in work/ (inventory.txt, ranges_*.txt).

  • NEBRASKA-2 (2026-10-09T08:00Z → 2026-11-01T00:00Z, Target issue): kept as is. The reference REFCOLH22 is rejected at 133 of 133 measurements from 2026-10-09T12:00Z to the end (100 %; the only gap is the station's 3 missing steps on 16 October). The compensation then runs on REFH22 and REFP''21 alone (their corr at 0). A reference rejected almost always is a Target issue, and its loss also moves the other curves: in the step 2026-10-09T08:00Z → 12:00Z, corr_x jumps by +13.2 noises (COLJ20), +11.8 (POUJ19), +10.3 (POUH19), and corr_y by +8.4 (COLH19), +6.7 (COLJ23). The bounds and description are correct.
  • NEBRASKA-6 (2026-10-16T20:00Z → 2026-11-01T00:00Z, Displacement): corrected. The drift is well beyond the threshold. Against the level of the range, corr_x is at −16.8 noises for COLJ20 at the start (28 measurements in a row beyond 5 noises), −12.5 for POUJ20, −12.3 for COLJ19 and −11.2 for POUJ19, then +5.7 to +8.8 for COLJ20, COLH20, COLH19, POUH21 and POUH20 on 29–31 October. corr_y: COLJ24 at −9.3, POUJ19 at +6.6. In mm (daily medians of corr_x_mm), it rises steadily from 16 October to 1 November: COLJ20 by about 1.2 mm, POUJ19 1.0, COLJ19 0.9, POUH19 0.85, POUJ20 0.8, COLJ24 1.5 (in px COLJ24 moves little, but it is far from the camera). Corrections: (1) the tool ensemble.py --rigide, which does not exist, is replaced by what it did, "the raw positions against the overall movement of the targets" (correction already found by the analysts of the periods); (2) COLJ24 is called weak only "in the image", and the description now says that in mm it drifts as much as the others. Start, end, family and the rest of the description are unchanged. The review field is removed.