Files
aimsight-labeling/auto/analyses/LENZSPITZE_2026-10-23T2355_2026-11-01T1140/analysis.md
T

23 KiB
Raw Blame History

Labels: LENZSPITZE, 2026-10-23T23:55Z → 2026-11-01T11:40Z (UTC)

Request: verify the 32 existing labels of the range (work/existing.jsonl): keep, correct, remove, add. DETECTORS: no. No spanning label (work/spanning.jsonl empty).

The key fact of the range: the existing label LENZSPITZE-40 says "more than two days without any data" from 23/10 23:55 to 26/10 00:00. The exports of 24/10 and 25/10 were pumped on 05/11, after that label was written: inventory.py now finds 2498 instants on the range, a pose at each one, the metrics of the ten days and an image every hour from 24/10 00:01. The only absent steps are 26/10 14:30-15:10, 27/10 08:55-09:25 and 29/10 15:05-15:15 (labels 43, 46, 58). So 40 is removed, and the two days 24/10 and 25/10 were read from scratch.

Weather (data/LENZSPITZE/weather/): MeteoSwiss station Pully (PUY), 8.9 km from the station position (Crissier). Light rain the night of 23-24/10 and the evening and night of 24-25/10, rain 25/10 09:00-14:00 with humidity at 95 %, dry from 26/10; sunny 27/10 from 10:00, overcast 28-31/10, sunny 01/11 from 08:00.

Method. Inventory on the range; overview.py on the range and on 24-26/10 (common, residuals); days.py --bounds --night 22-4 --smoothing 5; metrics.py (only temperature_deg_c is filled, as stations.md says); quality.py --common --step 1d (one image regime over the range: calm reference = the whole folder) and --labels existing --outside; my scripts in scripts/: scan.py (per instant: rejections, references retained, pose, flux below half the sensor's median, sensors whose corr leave the median of their ±6 neighbours by more than 6 own noises; whole range in work/scan_all.txt), edges.py (the scan lines around each bound, work/edges_existing.txt), amp.py (jumps in px at given instants), forgotten.py (instants where ≥ N sensors jump by ≥ 0.02 px, or a rejection, outside the non-thermal labels), drift.py (15-min medians of corr against a base window), build_labels.py (writes labels.jsonl). Images: whole images and thumbnails in work/thumbs/.

Caution on scan.py: a deviation from the rolling median also appears, with the opposite sign, on the calm measurement just before a step; I read the sign before taking an instant as touched (e.g. LENZSPITZE-50 at 22:00).

Verdict on the 32 existing labels

Old id Verdict Evidence
LENZSPITZE-40 removed The data of 24-25/10 exist now (pumped 05/11): measurements, poses, metrics, images every hour. Replaced by the labels of these two days (below).
LENZSPITZE-41 corrected Start moved from 26/10 00:00 ("resumption after the gap": there was no gap) to 25/10 17:30: from 17:35 the corr_x of 1024 shifts by up to two tenths of a px and stays shifted until 22:30, the flux of 1024, 1002, 1004 fades from 18:25, the corr_x of 1002, 1004 and of the reference 8014 drift from 18:30 (+0.1 px for 8014 at 22:15), 1002 rejected from 22:20, 8012 from 22:35 (drift.py). End 09:20 kept (09:15 last measurement with 4 sensors out, 09:20 calm). Description rewritten for the evening, the rest kept; 22:01 image added.
LENZSPITZE-42 kept Pose turning point after the night without references, return of pan/tilt in the evening; 21:30-21:35 small reference y jumps not distinct (section 2).
LENZSPITZE-43 kept Gap 14:30-15:10 confirmed by the inventory (7 steps), calm curves on both sides.
LENZSPITZE-44 kept Edges calm at 22:15-22:45; at 08:40-08:55 the curves still jump (101B up to 0.06 px), the bound is the last measurement before the gap.
LENZSPITZE-45 kept Bounds on the pose; the 1020 step at 16:05-16:10 is the "1020 all at once in late afternoon".
LENZSPITZE-46 kept 5 steps absent 08:55-09:25.
LENZSPITZE-47 kept Agitation of 25-35 sensors per measurement, sharpness down / dominance up (quality 27/10 09:30-12:00).
LENZSPITZE-48 kept 10:55: 8004, 1008, 8003, 8010 rejected, flux of 8010, 1006, 8003 cut.
LENZSPITZE-49 kept Flux of 1016, 1022, 1024 cut 17:40-17:50, last touched 17:55 (1014).
LENZSPITZE-50 kept 22:05 flux of 1007 cut; 22:00 is the calm side (opposite sign); 1007 not quite back at 23:15, as written.
LENZSPITZE-51 kept 8012/8014/101B rejections 03:35-04:45, flux of 1002/1004 low until 05:15; corr of 1004 a little up to 05:30 (borderline, not enough to move a validated bound).
LENZSPITZE-52 kept Turning point 05:40, return 21:00.
LENZSPITZE-53 kept 10:05 8003, 8010 rejected, flux cut.
LENZSPITZE-54 kept 10:55 the even group rejected, three references.
LENZSPITZE-55 kept Agitation around noon, 28/10.
LENZSPITZE-56 kept Turning point 07:45, return 20:00.
LENZSPITZE-57 kept 07:55 the even group with 8003, 8010 rejected.
LENZSPITZE-58 kept One step absent at 15:10.
LENZSPITZE-59 kept 17:05-17:15 flux of 1007 cut, 1021/1015 jumps, back at 17:20.
LENZSPITZE-60 kept 20:05 8010 rejected.
LENZSPITZE-61 kept Weak wave of 30/10; calm night after.
LENZSPITZE-62 kept 07:55 the even group rejected with four references.
LENZSPITZE-63 kept 13:55-14:00 flux of 92H cut, 53 then 35 sensors jump; 14:10 8014 flux cut without effect, as written.
LENZSPITZE-64 corrected End 01/11 00:00 was the end of an older batch ("at the end of the batch"); the pan keeps moving all night (housing temperature 25.5 °C at 01:00, 24.2 °C at 06:00; pan 0.0142° → 0.0164°) and the next cycle starts at sunrise: end moved to 01/11 06:55, the start of the next cycle, without rest (labeling-rules.md §2). Description of the night corrected.
LENZSPITZE-65 kept 10:55 small common jump of 1001, 1009, 1013.
LENZSPITZE-66 kept 11:01 8004, 8003, 8010 rejected.
LENZSPITZE-67 kept 15:01 8003, 8010 rejected.
LENZSPITZE-68 corrected Start 10:30 was the start of a short batch ("the cycle started early in the morning"). Now in the range: at 07:00 the pan drops at once (0.0163° → 0.0139° → 0.0081° at 07:35), the tilt and the scale rise, the housing temperature rises (24.3 → 26.6 °C, hourly): start 06:55. Description completed with the sunrise; the rest kept.
LENZSPITZE-69 kept 11:00-11:01 jumps, train seen on the 11:00 image (as labeled).
LENZSPITZE-70 kept 8012 sudden light at 11:15; it now sits inside the sunny-morning label, as a separate event.
LENZSPITZE-71 kept 11:20 burst; separate from the long sunny-morning label, as it states.

1. My labels (added)

Bounds checked with scripts/edges.py and the scan; all bounds are measurement instants (checked in the session). Precedents: precedents.py returned the validated LENZSPITZE labels of 16-23/10 for the night and rain signatures (LENZSPITZE-13, 15, 16, 26, 27, 30, 35, 37, 38), and WYOMING-13/17 (validated), FLORIDA-72, JUNGFRAU-110/112 for a far reference that shifts alone (all Displacement, cause not established).

2026-10-23T23:55Z → 2026-10-24T12:35Z

  • Phenomenon of the catalogue: Visibility drops over the whole scene, without an obstacle (rain or water)
  • Precedents: LENZSPITZE-38 (the same episode until 23:55, validated), 35, 16: same uneven pattern, odd-numbered targets extinguished, nearby spots jumping; same wording of the cause
  • Family: Visibility: the nearby scene loses light, nothing in front
  • Measured: flux < 0.5 of median for 1009, 1017, 1013, 1015, 1012, 1010, 1011, 1007, 1005, 1001 at 23:55-01:00; lull 01:30-03:15 (only 1012, 1015 weak); 03:20 drop on almost all nearby targets (1009 at 0.01); 1008 rejected 03:50, 04:15, 05:10-09:35, 10:05-10:40; 10:05-10:20 odd ones at 0.01-0.02; 1015 back above 0.5 at 12:25. corr_x of 1011, 1013, 1015 shifted by up to 0.2-0.25 px, 1022 jumps up to 0.5 px, 1020, 1021, 1023 jumps; last touched 12:30 (1022y, 101B, 1015), calm 12:35
  • Images: 04:00 night, spots present, nothing in front; 09:00 and 10:01 overcast, nothing in front
  • Confidence: high for the label, medium for the cause (rain at Pully, no drops seen)

2026-10-24T05:45Z → 2026-10-24T05:55Z

  • Phenomenon of the catalogue: a reference dropped: the base changes (one measurement)
  • Precedents: LENZSPITZE-63 (92H hidden, base jump; mentions 8014 going out)
  • Family: Obstruction: flux of 8014 cut to 0.03 alone, one measurement; Visibility would need a wider loss
  • Measured: 8014 rejected at 05:50 only; 102H x +20 / y −17 noises, 201B, 101B ±12-15
  • Images: none at 05:50 (05:00, 06:00 only)
  • Confidence: medium (object not seen)

2026-10-24T08:55Z → 2026-10-24T09:05Z

  • Phenomenon of the catalogue: a pass-by hides the targets for the length of one measurement
  • Precedents: LENZSPITZE-19, 24, 39 (same group of even-numbered targets and references)
  • Family: Obstruction
  • Measured: 09:00 flux 1012 0.03, 8019 0.07, 1024 0.08, 92H 0.09, 1022 0.13, 1020 0.17, 1014 0.23, 1018 0.24, 7013H, 7013B, 1016 ~0.3, 8002 0.38; 1012 rejected (1008 already); 09:01 8010 0.01 rejected, 8003 0.26; jumps 1022x −113, 1023y +82, 1020x −74, 1019x +53 noises
  • Images: 09:00 whole image: overcast, track empty, nothing in front
  • Confidence: high

2026-10-24T09:20Z → 2026-10-24T22:15Z

  • Phenomenon of the catalogue: the camera moves through a thermal effect
  • Precedents: LENZSPITZE-29 (weak cycle under overcast sky)
  • Family: Thermal displacement: the pose follows the housing temperature and comes back with it
  • Measured: housing temperature 24.0 → 26.4 °C (16:00); pan departure 09:35 (days.py), minimum 14:35 (−12.5 noises), back to the night level ~22:00-22:15; tilt departure 09:25, small dip 11:05; scale maximum 16:00 (+19 noises), back ~22:00; corr_y of 1007, 1005 down by ~0.04 px in the afternoon and evening. The loss of visibility from 21:15 touches no reference (all 9 retained): it does not stop the cycle
  • Images: 10:01 overcast
  • Confidence: medium (weak cycle, start mixed with the visibility)

2026-10-24T13:00Z → 2026-10-24T13:15Z

  • Phenomenon of the catalogue: partial occultation of a few targets
  • Precedents: LENZSPITZE-59, 50
  • Family: Obstruction
  • Measured: 13:05 flux 1006 0.58, 1016 0.69; 1006y −0.096 px, 1014y, 1016x, 1002y 0.02-0.05 px; 13:10 1014x −0.041; back 13:15
  • Images: none at 13:05
  • Confidence: low (small, cause not seen)

2026-10-24T20:55Z → 2026-10-24T21:05Z

  • Phenomenon of the catalogue: a pass-by hides the targets for the length of one measurement
  • Precedents: LENZSPITZE-39, 24, 62
  • Family: Obstruction
  • Measured: 21:00: 14 targets (8003 included) and 4 references (8002, 8004, 8019, 92H) rejected, flux 0.00-0.03; 1020 (0.02) and 1024 (0.10) kept with a false point; 27 sensors jump (1009y +0.074, 1023y +0.062 px); back 21:05
  • Images: 21:01 whole image, night, nothing in front
  • Confidence: high

2026-10-24T21:15Z → 2026-10-25T03:35Z

  • Phenomenon of the catalogue: Visibility drops over the nearby scene (rain or water)
  • Precedents: LENZSPITZE-26, 15 (night episodes of the same signature)
  • Family: Visibility
  • Measured: corr_x of 1002 and 1004 rise from 21:15-21:30 (+0.03-0.06 px), 1013x +0.063 px at 21:25; flux of 1004, 1002, 1010, 1012, 1006, 1008 down from 22:50; 23:35 almost all nearby targets (1008 rejected 23:40-00:05); corr_x of 1010 and 1012 up to +0.19 px, 1013, 1011, 1005, 1019 up, 1020, 1022, 1015 down (−0.18 px for 1020); back by 03:00-03:30, 1005 last (flux < 0.5 until 03:30)
  • Images: 21:01 night, nothing in front
  • Confidence: high for the label, medium for the cause

2026-10-25T02:20Z → 2026-10-25T18:30Z

  • Phenomenon of the catalogue: the camera moves through a thermal effect
  • Precedents: LENZSPITZE-17 (a cycle that starts with the temperature rise)
  • Family: Thermal displacement
  • Measured: housing temperature 24.2 °C (01:00) → 29.4 °C (14:00) → 26.4 °C (18:00); the pan leaves 0.0153-0.0156° at 02:20, 0.0130° at 04:20, minimum ~0.004-0.005° at 10:35 and 15:00; scale 0.99964 → 0.99985 (15:10); tilt minimum 14:20; the rotation dips to −0.0078° at 11:00 while 8004 is rejected (base change, said in the description). End: the reference 8014 starts drifting at 18:30 (drift.py), a nuisance that distorts the pose stops the cycle (§2). Start at 02:20 rather than at the tilt departure (06:50 by days.py): the pan and the scale clearly leave their night level with the temperature, the tilt shows only noise before; a judgment, stated here
  • Images: 10:01, 13:00 overcast
  • Confidence: medium

2026-10-25T05:05Z → 2026-10-25T05:15Z

  • Phenomenon of the catalogue: none (a single target jumps alone)
  • Precedents: none useful
  • Family: Displacement: jump then return without established cause
  • Measured: 1022x −0.111 px at 05:10, flux 0.89, error_value normal; others < 0.04 px
  • Images: none
  • Confidence: medium

2026-10-25T05:55Z → 2026-10-25T06:30Z

  • Phenomenon of the catalogue: part of the scene goes dark at night (a few neighbours lose their light, nothing in front)
  • Precedents: LENZSPITZE-37 (half an hour of partial dimming, Visibility)
  • Family: Visibility: no object, several measurements, spots shrunken
  • Measured: flux 1006 0.13, 1002 0.34, 1018 0.38, 1008 0.40 at 06:05; 1023y +0.092, 1006y +0.075, 1008 x/y +0.05-0.07 px at 06:05-06:15; error_value of 1023 0.66; last touched 06:25
  • Images: thumbnails 05:01 / 06:01 / 07:00 (1006, 1018, 1002, 1009): at 06:01 1018 and the neighbours of 1006 smaller and fainter, nothing in front
  • Confidence: medium

2026-10-25T07:55Z → 2026-10-25T17:40Z

  • Phenomenon of the catalogue: Visibility drops over the whole scene (rain)
  • Precedents: LENZSPITZE-35, 13, 16
  • Family: Visibility
  • Measured: 08:00 13 sensors jump, 08:05 flux drop; 8004 rejected 09:40-11:05 (8 references retained at times), 12:35, 12:45-12:50; 1008 09:05-13:01; 8003 09:35-09:50, 10:25-10:50, 11:00; 1001 10:35, 11:01; 8010 11:00; 1012 09:25, 10:05, 12:35; corr_x of 1022 and 1024 down to −0.6/−0.7 px (09:30-13:30), 1010, 1011, 1012, 1013, 1019 up; flux of 1013 and 1009 back above 0.5 at 16:50-17:00; 1010 back at 17:10, 101B 17:25, 1013 and 1009 settled 17:40. The drops of a group in the middle of the rain (11:00: 8010, 8003) cannot be told from the general loss: no separate passage label. Overlaps LENZSPITZE-41 for 17:30-17:40 (two events)
  • Images: 10:01, 13:00 overcast and hazy, nothing in front
  • Confidence: high for the label, medium for the cause

2026-10-26T09:50Z → 2026-10-26T10:00Z

  • Phenomenon of the catalogue: a reference dropped: the base changes (pass-by)
  • Precedents: LENZSPITZE-63 (92H hidden by the arm of a machine, seen)
  • Family: Obstruction: flux of 92H 0.00, alone, one measurement
  • Measured: 92H rejected 09:55; 9 targets jump in corr_x by 6-8 noises
  • Images: 10:01 whole image, nothing in front
  • Confidence: medium

Partial occultations: 2026-10-26T13:00Z → 13:15Z, 2026-10-26T14:01Z → 14:15Z, 2026-10-29T14:30Z → 14:50Z, 2026-10-30T20:00Z → 20:10Z, 2026-10-31T04:35Z → 04:50Z, 05:05Z → 05:15Z, 06:01Z → 06:10Z, 2026-11-01T05:05Z → 05:15Z

  • Phenomenon of the catalogue: an obstacle covers part of a few spots for one measurement
  • Precedents: LENZSPITZE-59 and 50 (validated: flux of 1007 cut, 1021/1015 drop, nothing visible, Obstruction "cause not seen")
  • Family: Obstruction: a few nearby targets lose part of their light, the rest intact, abrupt
  • Measured (flux ratio, largest jumps): 26/10 13:05 1012 0.23, 1016 0.33; 1022x −0.25, 1020x −0.15, 1012x +0.11 px; 13:10 1020y, 1021y +0.07 px. 26/10 14:05 1017 0.44, 1015 0.60, 1023 0.61, 1019 0.74; 1019x +0.17, 1017y +0.10 px; 14:10 still +0.12/+0.11 px. 29/10 14:35-14:45 1022 0.77, 1009 0.69; 1022x −0.12, 1020y −0.08, 1009x −0.07 px. 30/10 20:05 1021 0.55, 1019 0.64; 1019x +0.12, 1021x −0.12 px. 31/10 04:40 1006 0.42, 1022/1023 0.66, 1018 0.77; 1022y +0.13, 1023y −0.09, 1018y +0.08 px; 04:45 1014y −0.07 px. 31/10 05:10 1018 0.62, 1020 0.78; 1020y +0.08, 1002x +0.07 px. 31/10 06:05 1006 0.24; 1004x +0.08, 1006y −0.06 px. 01/11 05:10 1006 0.11, 1018 0.52; 1006x +26 noises, 1018x −15
  • Images: thumbnails 26/10 13:01 and 14:00, 31/10 06:00, 01/11 05:00 (a few minutes before): nothing in front; none at the other instants
  • Confidence: medium (partial flux, object never seen; the small ones low)

Far reference 8012 shifting alone: 2026-10-28T15:00Z → 15:15Z, 2026-10-29T06:01Z → 06:10Z, 2026-10-29T11:20Z → 11:35Z

  • Phenomenon of the catalogue: the base changes without a reference lost
  • Precedents: WYOMING-13/17 (validated), FLORIDA-72, JUNGFRAU-110/112: a far reference jumps alone, Displacement, cause not established
  • Family: Displacement: nothing in the image quality of the scene; the far reference 8012 shifts alone
  • Measured: overview.py --rigid 29/10 base 11:00: residual dx of 8012 +0.08 px at 11:25, +0.10 px at 11:30, back 11:35, other references within ±0.03 px; corr_x of 8012 +0.065 px, 18 sensors ≥ 0.02 px; 8012 error_value 0.26-0.30 against 0.18-0.23, flux +7 %. 28/10 15:05-15:10: 8012x −0.084 px, 16 sensors, 8012 flux −15 %, dominance up. 29/10 06:05: 8012x +0.061 px, error_value 0.218 against 0.14
  • Images: none at these instants
  • Confidence: medium

2026-10-29T06:55Z → 2026-10-29T07:20Z

  • Phenomenon of the catalogue: part of the scene loses its light at the change from night to day
  • Precedents: LENZSPITZE-37
  • Family: Visibility (several measurements, nothing in front)
  • Measured: 07:05 flux 1011 0.21, 1005 0.41, 1009 0.49, 1001 0.59; 07:00 1011x −0.046, 1001y −0.040 px; 07:05 1009x −0.061, 1011y −0.061, 1005y +0.056 px; 07:10 1011y −0.079 px; 07:15 1005 x/y +0.05 px; 07:20 calm
  • Images: 06:00 and 07:01 thumbnails: nothing in front
  • Confidence: low-medium (Visibility rather than Obstruction is a judgment)

2026-11-01T06:55Z → 2026-11-01T07:15Z

  • Phenomenon of the catalogue: the sun in the field or behind a spot
  • Precedents: LENZSPITZE-70 (the sun on 8012, validated)
  • Family: Visibility (sun, seen)
  • Measured: 202H rejected 07:00 and 07:05 (flux 0.03, 0.04), kept at 07:01 with x +0.139, y −0.106 px, 07:10 flux 0.42, y −54 noises; 1006y −0.086, 8010y −0.053 px at 07:00
  • Images: 07:00 whole image: the sun in the field at the far end of the track, large glare
  • Confidence: high

2026-11-01T07:15Z → 2026-11-01T11:40Z

  • Phenomenon of the catalogue: the sun in the field, then atmospheric turbulence
  • Precedents: LENZSPITZE-47, 33, 18, 6 (image trembling in the sun, Visibility)
  • Family: Visibility: sharpness down, dominance up over the scene, flux up
  • Measured: quality 15-min medians targets/references: sharpness 0.93/0.89 at 07:00, ~0.95-1.00 until 10:45, 0.86-0.89 after 11:00; dominance 1.40 at 07:00 then 1.1-1.3; snr 0.67-0.94; flux 1.1-1.5. Up to 34 sensors jump per measurement; 8012x up to 0.15 px (10:10), 8019, 8002, 8013, 8017 ±0.05-0.07 px, targets 1011 up to 0.12 px, 1020 0.10 px
  • Images: 07:00, 08:00 sun in the field (glare), 09:00 sun out of the field, full sun
  • Confidence: high for the label; the passage from the sun in the field to the turbulence is not dated (one label, the description says both). Overlaps 70 and 71, labeled apart

2. What I left without a label

  • One-measurement jumps of the camera tilt (24/10 13:45, 25/10 02:35 and 07:01, 26/10 12:00-12:05, 27/10 19:25, 28/10 02:25, 28/10 16:05, 26/10 21:30-21:35, 01/11 00:30): the tilt jumps by a few thousandths of a degree for one measurement and 10-20 sensors move by 0.02-0.06 px in corr_y together; in own noise of corr_y it stays at a few noises (the scan finds 1-7 sensors above 6 noises, except 25/10 07:01). The validated labels do not label them (27/10 19:25, 28/10 02:25, 28/10 16:05 lie between validated labels); I treat them the same, 07:01 included (a first draft labeled it; removed for consistency).
  • Small y agitation of 8010, 1005, 1006, 7013B, 8003 by day (26/10, 29/10, 30/10, 5-8 sensors at 0.02-0.04 px): within the noise of corr_y.
  • Flux drops without effect: 27/10 18:00-18:15 (1004, 1002, 1016), 27/10 19:40-20:35 (1012, 1006), 28/10 05:10-05:20 (1002, 1004, 1011), 29/10 05:05 (1004), 29/10 08:05 (8003, 8010 at 0.34-0.37, one corr jump of 1023 only), 30/10 14:10 (8014, mentioned in 63), 30/10 14:55 (8003), 31/10 18:20 (8011): corr within a few noises.
  • 24/10 13:15-13:30: the corr_x of 1014 stays a little shifted (0.04 px) after the 13:05 dip.
  • The night pose drift after the thermal labels of 27/10, 28/10, 29/10: the pan rises again at night with the cooling (28/10 21:00 0.0109° → 29/10 04:00 0.0149°) after labels that end in the evening. labeling-rules.md §2 says that without rest the cycle ends at the next start; all the validated LENZSPITZE thermal labels end in the evening. I did not change validated thermal ends (section 3).

3. The documentation

  • Helped: families.md boundaries (Visibility vs Obstruction by extent; a hindered reference takes the family of its cause), labeling-rules.md §2 (bounds at the edges of a gap, a nuisance at night stops the cycle only if it distorts the pose, without rest the labels follow one another), tools/DATA.md quality metrics (flux cut vs flux whole), the validated LENZSPITZE labels of 16-23/10 as precedents, and the WYOMING / JUNGFRAU / FLORIDA reference-jump precedents.
  • Contradicted by the data: LENZSPITZE-40 (no data) is contradicted by data pumped later; a No data label should be re-read when a re-pumping fills its gap. LENZSPITZE-64 and 68 carried the edges of older batches ("at the end of the batch", "started early in the morning").
  • To clarify: "the camera cycle starts when the tilt and the pan leave their night level, the later of the two" does not say what to do when one axis shows only noise (25/10: pan and scale leave at 02:20 with the temperature, the tilt not clearly). The station practice (thermal labels ending in the evening) differs from "without rest, the cycle ends at the next start" on nights when the pan keeps following the cooling: one sentence in labeling-rules.md would settle it.

Difficulties

  • thumbnails.py --sensors with 1023 (no x, y, tools/DATA.md) stops with a Python traceback ("min() iterable argument is empty") instead of saying so.
  • quality.py --outside lists only flux/snr/dominance; the reference shifts (8012) and the jumps with a whole flux were found with my own scan (scripts/forgotten.py).

Sources read

agents/agents/analyst.md, instructions/analyst.md, work/prompt.txt, work/existing.jsonl, work/spanning.jsonl (empty), tools/GUIDE.md, tools/DATA.md, labeling/families.md, labeling/labeling-rules.md, labeling/reading-rules.md, labeling/stations.md, labeling/auto/LENZSPITZE.jsonl (labels at the edges of the range; LENZSPITZE-1 to 39 for comparison), precedents.py results; data data/LENZSPITZE/: displacement exports, camera poses, metrics, weather, images of 24/10 04:00, 09:00, 10:01, 21:01, 25/10 05:01-07:00 (thumbnails), 10:01, 13:00, 18:01, 22:01, 26/10 10:01, 13:01 and 14:00 (thumbnails), 29/10 06:00 and 07:01 (thumbnails), 31/10 06:00, 08:01, 01/11 05:00, 06:01, 07:00, 08:00, 09:00.