Files
aimsight-labeling/auto/analyses/NORDEND_2026-08-10T0115_2026-08-23T0515/report.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

5.1 KiB
Raw Blame History

NORDEND: requalified labels, 10 to 23 August

You asked me to requalify the labels of 10/08 01:15 → 23/08 05:15 (UTC) under the new rules: keep only what stands out from the noise of the corrected curves.

In short

28 labels stay as they were; 21 are removed. Most of the removed ones are lost points of CTR01, which move no other curve. Do not trust the curves on sunny mornings (PR14 to PR18 dip by up to 7 mm), on the cobweb nights (18→19, 19→20, 21→22/08) and on the dull mornings of 17 and 18/08. No real movement of the quarry face is seen.

Requalification

  • Kept: 28. Rewritten: 0. Removed: 21.
  • NORDEND-2, 6, 8, 11, 14 (sunrise), NORDEND-4, 7, 10, 13, 16, 19, 22, 26, 43 (sunny afternoons), NORDEND-21, 34, 47 (dull mornings): CTR01 lost points only. CTR01 is not used to correct the other targets, so losing it moves no other curve. Deviation: none of its own (what moves at those hours is the camera cycle, already labelled).
  • NORDEND-30 (cobweb night 18→19/08): apart from the spider's passage and the hidden reference, both still labelled, it held only lost points. Every target stays within 4.6 noises; PR04 has its own label.
  • NORDEND-25 (17/08), NORDEND-28 (18/08), NORDEND-35 (19/08), thermal days: on these overcast days the camera turns little and the correction absorbs it. Deviation: 4.5, 3.7 and 4.9 noises at most.

What happened

The camera follows the heat of its housing, every day. At sunrise the sun faces the camera, and the housing heats up all at once. The camera turns, and the correction does not erase it all. The horizontal curves of PR14 to PR18 (85 to 145 m from the camera) dip in the morning and come back in the afternoon or during the night. It is strongest on the sunny days of 12 to 14/08: PR18 (116 m) loses up to about 7 mm, PR16 about 2.5 mm. It is weaker on 15, 16 and 20 to 22/08 (PR18 2.5 to 4 mm). It stays within the noise on the overcast days of 17 to 19/08. On the coldest night (22→23/08), PR06 and PR15 stay up by about 1.5 mm in the vertical until the morning. Confirmed (camera position, housing temperature, images of the sun). This is not a movement of the quarry face.

Lit cobweb threads at 23:00, each night Look at the last four images: from 18/08 the infrared lights cobweb threads in front of the lens, every night.

Cobweb in front of the lens, from 18/08. On 18/08 at 18:00 the spider itself hides the reference CTR02, and all targets jump together by up to 4.6 mm in the vertical. From 22:30 to 01:00 a thread hides the reference CTR03, and all curves shift by up to 5 mm. On 19→20/08 PR07 jumps by 2 mm and PR08 drifts by 1.6 mm. On 21→22/08 PR08 is off by 4 mm all night and PR14 jumps by 5 mm at dusk. Confirmed on the images.

Poor light. On the dull mornings of 17/08 (PR06 2.8 mm) and 18/08 (PR16 3.6 mm), and on the afternoon of 18/08 (PR16 2.6 mm), targets jump from one measurement to the next. The cause is not established, probably weak light on the spots. On 20/08 at 16:30 the sun reaching the wall makes PR14 and PR15 jump by 3.4 mm for a few measurements (confirmed on the images).

Small events. An animal very close to the lens moves PR12 by 2.4 mm on the night of 10/08 (confirmed on the images).

Gaps. Nine times the station delivers no measurement at all. There are a few one- or two-step gaps at night (15, 19, 20, 21/08), and longer ones on 19/08 in the afternoon (6 h) and on 20/08 at the end of the night and in the morning. The cause is not established. The curves resume without any offset.

There is no weather station for NORDEND in the data, so the images were the only source for sun, clouds and cobweb.

What to do

Clean the cobweb from the window in front of the lens: it comes back every night from 18/08. Check PR04 on site (its own existing label).

Not established

  • Why the measurements are missing during the gaps. Whether other stations missed them at the same time would settle it, but no August data from other stations is in the folder.
  • The cause of the jumps on the dull mornings of 17 and 18/08.

Labels

Start (UTC) End (UTC) Family Targets In a few words
10/08 01:15 10/08 02:30 Obstruction PR12 animal at the lens
10/08 05:30 → 17/08 07:30 each next morning Thermal displacement PR14–PR18 camera cycle, 7 labels (10 to 16/08)
20/08 06:15 23/08 05:15 Thermal displacement PR18, PR14–16; cold night PR06, PR15 camera cycle, 3 labels (20, 21, 22/08)
17/08 02:30 17/08 10:30 Visibility PR06, PR08, PR16 dark dull morning
18/08 06:15 18/08 12:15 Visibility PR16, PR07, PR08 uneven morning light
18/08 15:15 18/08 16:45 Visibility PR16, PR13 passing disturbance
18/08 17:45 18/08 18:15 Obstruction all (CTR02 hidden) spider in front
18/08 22:15 19/08 01:45 Obstruction all (CTR03 hidden) thread on CTR03
19/08 20:15 20/08 03:45 Visibility PR07, PR08 cobweb night
20/08 16:15 20/08 17:15 Visibility PR14, PR15, PR16 late sun on the wall
21/08 15:45 22/08 07:30 Visibility PR08, PR14, PR16, PR07 cobweb night
15/08 01:00 … 22/08 16:30 No data all 9 gaps