jwst-gc pipeline monitor

Stage ladder

Products per filter

* = the product name carries no _o<obs> token and this field has more than one observation, so the count cannot be attributed to 3958/o007.

Astrometry — m2 checkpoint

“tiles ok” is not a tolerance. measure_offset_grid runs with no max_off_mas, and astrometry_offsets sets off_ok=True whenever that is None — so N/N counts tiles whose offset histogram had a coherent peak, however large the offset. The column that carries the gate is worst tile, against 15 mas.

Provenance

Queue

No jobs in the queue for this field.

Findings

• fail
  m8: products come from 2 different pipeline tags

  2026-09-04_PR677_f404dcc-dirty (1), 2026-08-21_PR460_3fd4e04-dirty (1). Image and catalog must be deployed from the SAME pipeline run; a mixed-tag set is not releasable as one product.

  catalogs/*m8*.prov.json

  what is affected, and why

  This happens when filters are re-run at different times and only some are regenerated — a partial rerun, or a merge that picked up an older per-filter product. The fix is to re-run the lagging filters at the current tag, not to re-stamp: the tag records which code wrote the numbers.

  Diagnostic writeupfull diagnostic writeup (PDF)

  pipeline tag	products
  2026-09-04_PR677_f404dcc-dirty	1
  2026-08-21_PR460_3fd4e04-dirty	1

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• fail
  F187N_o007: 17/96 exposures misaligned vs their visit consensus

  An exposure is misaligned when its offset exceeds 2.0 mas AND is significant against the peak error bars. At m2 this CORRECTS the offsets table and stops the run; the crf frames must be regenerated before cataloging continues.

  checkpoint_m2_F187N_o007_latest.json

  what is affected, and why

  Affected: 4/4 detectors — spread across detectors, so the frame as a whole moved: look at the offsets table and whether these exposures were regenerated from _cal after the table changed. fix_alignment SKIPS a frame that already has a RAOFFSET header, so correcting the table alone leaves stale frames in place; regenerate the working copy from _cal instead of re-applying on top.

  Misaligned by detector

  nrcb25/24

  nrcb14/24

  nrcb34/24

  nrcb44/24

  Diagnostic writeupD2 — internal astrometric repeatability · full diagnostic writeup (PDF)

  visit	detector	dRA (mas)	dDec (mas)	off (mas)	contrast	pairs	window ("")
  1	nrcb1	0.9	1.9	2.1	71.0	1076	3.0
  1	nrcb1	1.1	2.4	2.7	72.0	1091	3.0
  1	nrcb1	0.7	2.0	2.1	65.0	11670	10.0
  1	nrcb1	0.4	2.1	2.2	61.0	11840	10.0
  1	nrcb2	2.5	-0.7	2.6	68.0	8716	10.0
  1	nrcb2	2.7	-0.7	2.8	72.0	8874	10.0
  1	nrcb2	-1.1	-2.9	3.1	104.0	14247	10.0
  1	nrcb2	-1.0	-2.4	2.6	102.0	14823	10.0
  1	nrcb2	-0.5	-2.4	2.4	97.0	15603	10.0
  1	nrcb3	1.3	2.8	3.1	76.0	12557	10.0
  1	nrcb3	1.1	2.7	2.9	76.0	12431	10.0
  1	nrcb3	1.2	2.8	3.0	97.0	13635	10.0
  1	nrcb3	1.0	3.0	3.2	99.0	13865	10.0
  1	nrcb4	2.6	-0.9	2.7	83.0	1212	3.0
  1	nrcb4	2.6	-0.8	2.7	82.0	1239	3.0
  1	nrcb4	-1.5	-2.4	2.8	95.0	1521	3.0
  1	nrcb4	-1.5	-1.7	2.3	79.0	1482	3.0

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• fail
  F210M_o007: 44/96 exposures misaligned vs their visit consensus

  An exposure is misaligned when its offset exceeds 2.0 mas AND is significant against the peak error bars. At m2 this CORRECTS the offsets table and stops the run; the crf frames must be regenerated before cataloging continues.

  checkpoint_m2_F210M_o007_latest.json

  what is affected, and why

  Affected: 4/4 detectors — spread across detectors, so the frame as a whole moved: look at the offsets table and whether these exposures were regenerated from _cal after the table changed. fix_alignment SKIPS a frame that already has a RAOFFSET header, so correcting the table alone leaves stale frames in place; regenerate the working copy from _cal instead of re-applying on top.

  Misaligned by detector

  nrcb316/24

  nrcb110/24

  nrcb210/24

  nrcb48/24

  Diagnostic writeupD2 — internal astrometric repeatability · full diagnostic writeup (PDF)

  visit	detector	dRA (mas)	dDec (mas)	off (mas)	contrast	pairs	window ("")
  1	nrcb1	-1.8	1.3	2.2	398.0	27423	3.0
  1	nrcb1	-1.9	1.3	2.3	410.0	28539	3.0
  1	nrcb1	-1.2	1.8	2.1	416.0	27578	3.0
  1	nrcb1	-1.1	1.8	2.1	433.0	28036	3.0
  1	nrcb1	-2.2	0.4	2.2	297.0	29030	3.0
  1	nrcb1	-2.2	0.4	2.2	275.0	29325	3.0
  1	nrcb1	1.3	1.8	2.2	314.0	20704	3.0
  1	nrcb1	1.2	2.0	2.3	318.0	21288	3.0
  1	nrcb1	1.3	1.8	2.2	328.0	20650	3.0
  1	nrcb1	1.3	1.9	2.3	303.0	20704	3.0
  1	nrcb2	2.8	-1.0	3.0	406.0	27327	3.0
  1	nrcb2	2.8	-0.9	2.9	388.0	28004	3.0
  1	nrcb2	1.7	-2.1	2.7	424.0	28702	3.0
  1	nrcb2	1.6	-2.1	2.6	394.0	28884	3.0
  1	nrcb2	-1.3	-2.1	2.5	324.0	25272	3.0
  1	nrcb2	-1.2	-2.2	2.5	337.0	24721	3.0
  1	nrcb2	-1.5	-3.1	3.4	402.0	24774	3.0
  1	nrcb2	-1.5	-3.1	3.4	387.0	23820	3.0
  1	nrcb2	-0.7	-2.0	2.1	368.0	26999	3.0
  1	nrcb2	-0.7	-2.1	2.3	365.0	25936	3.0
  1	nrcb3	-2.5	1.6	2.9	464.0	29419	3.0
  1	nrcb3	-2.5	1.7	3.1	451.0	28801	3.0
  1	nrcb3	-1.6	1.5	2.2	395.0	29942	3.0
  1	nrcb3	-1.8	1.4	2.3	392.0	29071	3.0
  1	nrcb3	-2.1	1.3	2.5	398.0	27090	3.0
  1	nrcb3	-2.2	1.2	2.5	439.0	29482	3.0
  1	nrcb3	-2.8	1.2	3.1	480.0	30433	3.0
  1	nrcb3	-2.8	1.3	3.0	469.0	30385	3.0
  1	nrcb3	2.0	2.8	3.4	436.0	26591	3.0
  1	nrcb3	2.0	2.7	3.4	441.0	26450	3.0
  1	nrcb3	2.3	1.7	2.8	357.0	25573	3.0
  1	nrcb3	2.3	1.9	3.0	337.0	25995	3.0
  1	nrcb3	1.5	2.4	2.8	409.0	24045	3.0
  1	nrcb3	1.6	2.5	3.0	402.0	23886	3.0
  1	nrcb3	1.8	2.9	3.4	452.0	27067	3.0
  1	nrcb3	1.9	2.9	3.5	443.0	26424	3.0
  1	nrcb4	-1.4	-2.7	3.0	427.0	27043	3.0
  1	nrcb4	-1.5	-2.5	2.9	430.0	26236	3.0
  1	nrcb4	-2.4	-2.6	3.5	468.0	27803	3.0
  1	nrcb4	-2.2	-2.4	3.3	451.0	27191	3.0

  showing 40 of 44

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• fail
  F480M_o007 visit 1: weakest tile peak contrast 4.50 < 5

  At least one tile has no real tie, so its offset is noise.

  checkpoint_m2_F480M_o007_latest.json

  what is affected, and why

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  m7: 6 product(s) stamped with a -dirty tag

  The tree had uncommitted changes, so the tag does not identify the code that ran. Fine for a development run; not for a release.

  catalogs/*m7*.prov.json

  what is affected, and why

  Diagnostic writeupfull diagnostic writeup (PDF)

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  m8: 2 product(s) stamped with a -dirty tag

  The tree had uncommitted changes, so the tag does not identify the code that ran. Fine for a development run; not for a release.

  catalogs/*m8*.prov.json

  what is affected, and why

  Diagnostic writeupfull diagnostic writeup (PDF)

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F187N: 20/96 exposures misaligned vs their visit consensus (checkpoint predates the frames)

  An exposure is misaligned when its offset exceeds 2.0 mas AND is significant against the peak error bars. At m2 this CORRECTS the offsets table and stops the run; the crf frames must be regenerated before cataloging continues. NOTE: the reduced frames are NEWER than this checkpoint, so the correction it triggered has probably already been applied — re-run the m2 checkpoint to confirm rather than treating this as current.

  checkpoint_m2_F187N_latest.json

  what is affected, and why

  Affected: 4/4 detectors — spread across detectors, so the frame as a whole moved: look at the offsets table and whether these exposures were regenerated from _cal after the table changed. fix_alignment SKIPS a frame that already has a RAOFFSET header, so correcting the table alone leaves stale frames in place; regenerate the working copy from _cal instead of re-applying on top.

  Misaligned by detector

  nrcb17/24

  nrcb25/24

  nrcb34/24

  nrcb44/24

  Diagnostic writeupD2 — internal astrometric repeatability · full diagnostic writeup (PDF)

  visit	detector	dRA (mas)	dDec (mas)	off (mas)	contrast	pairs	window ("")
  1	nrcb1	-1.8	1.1	2.1	66.0	9188	10.0
  1	nrcb1	-2.0	-0.7	2.1	71.0	11436	10.0
  1	nrcb1	-2.1	-0.6	2.1	66.0	11051	10.0
  1	nrcb1	0.9	2.2	2.4	74.0	1076	3.0
  1	nrcb1	1.1	2.6	2.8	74.0	1091	3.0
  1	nrcb1	0.8	2.3	2.4	67.0	11670	10.0
  1	nrcb1	0.5	2.3	2.4	65.0	11839	10.0
  1	nrcb2	2.5	-0.7	2.6	68.0	8716	10.0
  1	nrcb2	2.6	-0.7	2.7	75.0	8873	10.0
  1	nrcb2	-1.1	-2.8	3.0	104.0	14248	10.0
  1	nrcb2	-1.0	-2.3	2.5	104.0	14821	10.0
  1	nrcb2	-0.5	-2.3	2.3	94.0	15603	10.0
  1	nrcb3	1.3	2.8	3.1	73.0	12558	10.0
  1	nrcb3	1.1	2.6	2.8	74.0	12431	10.0
  1	nrcb3	1.2	2.7	3.0	97.0	13635	10.0
  1	nrcb3	1.1	3.0	3.2	99.0	13865	10.0
  1	nrcb4	2.6	-0.9	2.8	83.0	1212	3.0
  1	nrcb4	2.7	-0.8	2.8	83.0	1239	3.0
  1	nrcb4	-1.5	-2.4	2.9	95.0	1521	3.0
  1	nrcb4	-1.4	-1.8	2.3	80.0	1482	3.0

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F210M: 44/96 exposures misaligned vs their visit consensus (checkpoint predates the frames)

  An exposure is misaligned when its offset exceeds 2.0 mas AND is significant against the peak error bars. At m2 this CORRECTS the offsets table and stops the run; the crf frames must be regenerated before cataloging continues. NOTE: the reduced frames are NEWER than this checkpoint, so the correction it triggered has probably already been applied — re-run the m2 checkpoint to confirm rather than treating this as current.

  checkpoint_m2_F210M_latest.json

  what is affected, and why

  Affected: 4/4 detectors — spread across detectors, so the frame as a whole moved: look at the offsets table and whether these exposures were regenerated from _cal after the table changed. fix_alignment SKIPS a frame that already has a RAOFFSET header, so correcting the table alone leaves stale frames in place; regenerate the working copy from _cal instead of re-applying on top.

  Misaligned by detector

  nrcb316/24

  nrcb110/24

  nrcb210/24

  nrcb48/24

  Diagnostic writeupD2 — internal astrometric repeatability · full diagnostic writeup (PDF)

  visit	detector	dRA (mas)	dDec (mas)	off (mas)	contrast	pairs	window ("")
  1	nrcb1	-1.8	1.3	2.2	398.0	27423	3.0
  1	nrcb1	-1.9	1.3	2.3	411.0	28539	3.0
  1	nrcb1	-1.2	1.8	2.1	416.0	27578	3.0
  1	nrcb1	-1.1	1.8	2.1	434.0	28036	3.0
  1	nrcb1	-2.2	0.4	2.2	296.0	29030	3.0
  1	nrcb1	-2.2	0.4	2.2	275.0	29325	3.0
  1	nrcb1	1.3	1.8	2.2	314.0	20704	3.0
  1	nrcb1	1.2	2.0	2.3	318.0	21287	3.0
  1	nrcb1	1.3	1.8	2.2	328.0	20650	3.0
  1	nrcb1	1.3	1.9	2.3	303.0	20704	3.0
  1	nrcb2	2.8	-1.0	3.0	406.0	27327	3.0
  1	nrcb2	2.8	-0.9	2.9	388.0	28004	3.0
  1	nrcb2	1.7	-2.1	2.7	423.0	28702	3.0
  1	nrcb2	1.6	-2.1	2.6	393.0	28884	3.0
  1	nrcb2	-1.3	-2.1	2.5	324.0	25272	3.0
  1	nrcb2	-1.2	-2.2	2.5	337.0	24721	3.0
  1	nrcb2	-1.5	-3.1	3.4	402.0	24774	3.0
  1	nrcb2	-1.5	-3.1	3.4	387.0	23820	3.0
  1	nrcb2	-0.7	-2.0	2.1	369.0	26999	3.0
  1	nrcb2	-0.7	-2.1	2.3	365.0	25936	3.0
  1	nrcb3	-2.5	1.6	2.9	464.0	29419	3.0
  1	nrcb3	-2.5	1.7	3.1	451.0	28801	3.0
  1	nrcb3	-1.6	1.5	2.2	395.0	29942	3.0
  1	nrcb3	-1.8	1.4	2.3	392.0	29071	3.0
  1	nrcb3	-2.2	1.3	2.5	398.0	27090	3.0
  1	nrcb3	-2.2	1.2	2.5	439.0	29481	3.0
  1	nrcb3	-2.8	1.2	3.1	480.0	30433	3.0
  1	nrcb3	-2.8	1.3	3.0	469.0	30385	3.0
  1	nrcb3	2.0	2.8	3.4	436.0	26591	3.0
  1	nrcb3	2.0	2.7	3.4	441.0	26450	3.0
  1	nrcb3	2.3	1.7	2.8	357.0	25573	3.0
  1	nrcb3	2.3	1.9	3.0	337.0	25995	3.0
  1	nrcb3	1.5	2.4	2.8	408.0	24045	3.0
  1	nrcb3	1.6	2.5	3.0	402.0	23886	3.0
  1	nrcb3	1.8	2.9	3.4	452.0	27067	3.0
  1	nrcb3	1.9	2.9	3.5	443.0	26424	3.0
  1	nrcb4	-1.4	-2.7	3.1	427.0	27043	3.0
  1	nrcb4	-1.5	-2.5	2.9	430.0	26236	3.0
  1	nrcb4	-2.4	-2.6	3.5	468.0	27803	3.0
  1	nrcb4	-2.2	-2.4	3.3	452.0	27191	3.0

  showing 40 of 44

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F480M_o007 visit 1: reference tie recorded as could-not-verify (apply_ok=false)

  A correction is applied only when the tie is coherent AND the gross cross-check (100 mas) passes AND the per-tile map is clean. This one was not, so no correction was made — the frame keeps whatever it had.

  checkpoint_m2_F480M_o007_latest.json

  what is affected, and why

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F187N visit 1: worst tile 31.3 mas at cell (5,5) (> 15 mas)

  36/36 tiles are reported "ok", but that counts tiles whose offset histogram had a coherent PEAK — not tiles within tolerance. The m7 cross-band gate is no significant 2" cell above 15 mas, and this cell exceeds it, so the bulk tie being ~0 does not mean the field is flat.

  checkpoint_m2_F187N_latest.json

  what is affected, and why

  20/36 cells exceed 15 mas. They are not confined to the edge, so this is an interior residual: a distortion or per-detector alignment problem, not a coverage artefact. The bulk tie for this visit is 2.08 mas — a small bulk value cannot cancel a local one.

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  cell	off (mas)	dRA	dDec	contrast	pairs
  (5,5)	31.3	-30.5	-7.2	13.0	36589
  (2,1)	28.7	10.6	26.7	14.0	59563
  (4,5)	28.2	-11.2	-25.9	12.0	36402
  (0,4)	28.0	-22.1	17.1	14.0	41556
  (3,5)	27.1	-9.6	25.3	15.0	42943
  (1,5)	26.0	25.9	-2.1	10.0	28355
  (4,3)	22.9	-19.9	11.4	14.0	44133
  (4,0)	22.8	14.4	-17.6	11.0	47355
  (0,2)	22.5	-0.2	22.5	14.0	58241
  (2,3)	22.4	-12.1	-18.9	18.0	49870
  (5,3)	21.7	-13.1	-17.2	14.0	36181
  (2,4)	20.9	-1.9	-20.8	20.0	67992
  (0,1)	20.1	-7.8	18.5	17.0	60420
  (3,4)	19.8	-7.6	-18.2	15.0	54764
  (5,0)	19.2	-12.6	-14.4	10.0	40995
  (4,2)	17.1	-16.2	-5.7	13.0	37687
  (0,5)	17.1	14.7	8.7	14.0	33091
  (4,4)	17.0	-10.9	-13.0	19.0	52625
  (3,0)	16.7	-16.1	4.1	17.0	53600
  (5,4)	15.6	0.6	-15.6	11.0	38220

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F187N_o007 visit 1: worst tile 29.5 mas at cell (5,5) (> 15 mas)

  36/36 tiles are reported "ok", but that counts tiles whose offset histogram had a coherent PEAK — not tiles within tolerance. The m7 cross-band gate is no significant 2" cell above 15 mas, and this cell exceeds it, so the bulk tie being ~0 does not mean the field is flat.

  checkpoint_m2_F187N_o007_latest.json

  what is affected, and why

  19/36 cells exceed 15 mas. They are not confined to the edge, so this is an interior residual: a distortion or per-detector alignment problem, not a coverage artefact. The bulk tie for this visit is 0.16 mas — a small bulk value cannot cancel a local one.

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  cell	off (mas)	dRA	dDec	contrast	pairs
  (5,5)	29.5	-28.2	-8.7	13.0	36589
  (2,1)	29.2	12.6	26.3	15.0	59565
  (1,5)	28.2	28.1	-2.2	9.0	28358
  (4,5)	27.7	-8.8	-26.3	13.0	36401
  (3,5)	26.2	-10.9	23.8	14.0	42941
  (0,4)	26.1	-20.0	16.7	14.0	41554
  (4,0)	24.4	16.4	-18.1	11.0	47354
  (0,2)	22.1	1.9	22.0	14.0	58243
  (2,3)	21.7	-10.0	-19.3	15.0	49868
  (5,3)	20.8	-11.1	-17.6	14.0	36181
  (4,3)	20.6	-17.4	11.0	14.0	44131
  (2,4)	20.2	2.0	-20.1	19.0	67989
  (3,4)	19.5	-5.9	-18.6	13.0	54759
  (0,1)	18.9	5.4	18.1	18.0	60422
  (0,5)	18.7	16.8	8.3	13.0	33091
  (5,0)	18.2	-10.6	-14.8	11.0	40996
  (5,4)	16.2	2.8	-16.0	11.0	38218
  (4,4)	16.0	-8.7	-13.4	18.0	52622
  (4,2)	15.3	-14.0	-6.1	13.0	37687

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F210M visit 1: worst tile 23576.4 mas at cell (1,5) (> 15 mas)

  36/36 tiles are reported "ok", but that counts tiles whose offset histogram had a coherent PEAK — not tiles within tolerance. The m7 cross-band gate is no significant 2" cell above 15 mas, and this cell exceeds it, so the bulk tie being ~0 does not mean the field is flat.

  checkpoint_m2_F210M_latest.json

  what is affected, and why

  23/36 cells exceed 15 mas. They are not confined to the edge, so this is an interior residual: a distortion or per-detector alignment problem, not a coverage artefact. The bulk tie for this visit is 2.56 mas — a small bulk value cannot cancel a local one.

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  cell	off (mas)	dRA	dDec	contrast	pairs
  (1,5)	23,576.4	22,913.8	-5,550.4	6.5	175315
  (2,1)	5,573.4	-5,502.5	886.0	6.0	34371
  (4,0)	2,248.0	-2,115.3	760.9	6.0	25859
  (5,5)	37.0	-36.1	-8.4	7.0	23842
  (0,4)	34.5	-30.5	16.2	14.0	27047
  (4,5)	31.3	2.1	-31.3	9.0	21965
  (0,1)	27.1	-19.4	18.9	11.0	36736
  (3,2)	27.0	-14.4	-22.9	11.0	27931
  (5,3)	26.9	-10.2	-24.9	11.0	20655
  (0,2)	26.4	-14.7	21.8	10.0	30757
  (1,4)	26.3	-25.5	-6.6	7.0	21885
  (1,2)	25.9	22.7	12.5	10.0	24743
  (1,3)	24.2	13.3	20.2	10.0	26768
  (3,0)	23.7	-17.6	-15.9	9.0	27839
  (4,2)	22.5	13.1	-18.3	9.0	23045
  (5,0)	21.8	9.8	-19.5	8.0	21197
  (5,1)	21.7	-10.2	-19.2	9.0	28758
  (3,4)	21.0	-6.0	-20.2	9.0	31723
  (0,5)	20.1	16.3	11.6	10.0	15130
  (2,3)	19.2	-17.6	-7.6	10.0	29031
  (4,1)	16.7	-12.7	10.8	9.0	30079
  (2,5)	16.2	-15.9	2.8	9.0	22293
  (0,3)	15.0	-12.6	8.2	7.0	28042

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F210M_o007 visit 1: worst tile 5570.9 mas at cell (2,1) (> 15 mas)

  36/36 tiles are reported "ok", but that counts tiles whose offset histogram had a coherent PEAK — not tiles within tolerance. The m7 cross-band gate is no significant 2" cell above 15 mas, and this cell exceeds it, so the bulk tie being ~0 does not mean the field is flat.

  checkpoint_m2_F210M_o007_latest.json

  what is affected, and why

  24/36 cells exceed 15 mas. They are not confined to the edge, so this is an interior residual: a distortion or per-detector alignment problem, not a coverage artefact. The bulk tie for this visit is 0.10 mas — a small bulk value cannot cancel a local one.

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  cell	off (mas)	dRA	dDec	contrast	pairs
  (2,1)	5,570.9	-5,500.0	885.9	6.0	34372
  (4,0)	2,245.5	-2,112.7	760.8	6.0	25858
  (4,4)	41.0	-34.5	-22.1	14.0	27553
  (5,5)	35.1	-34.1	-8.2	7.0	23842
  (1,5)	34.3	30.6	-15.6	6.0	18871
  (0,4)	32.2	-27.9	16.0	13.0	27052
  (4,5)	31.7	4.6	-31.4	9.0	21964
  (0,3)	30.3	8.8	-28.9	9.0	28039
  (3,2)	28.3	-15.8	-23.5	9.0	27930
  (5,3)	28.1	-7.7	-27.1	8.0	20655
  (1,2)	26.8	25.1	9.5	10.0	24744
  (1,3)	25.4	15.7	20.0	11.0	26770
  (0,1)	25.2	-16.8	18.8	10.0	36733
  (0,2)	24.9	-12.2	21.7	10.0	30751
  (4,2)	23.7	14.9	-18.5	9.0	23042
  (5,0)	23.2	12.3	-19.6	9.0	21196
  (0,5)	22.1	18.9	11.5	10.0	15130
  (5,1)	20.8	-7.7	-19.3	7.0	28755
  (1,4)	19.9	-19.0	6.0	7.0	21885
  (2,2)	18.6	11.2	-14.8	9.0	26097
  (3,0)	18.2	-13.5	-12.3	9.0	27840
  (3,4)	17.9	-4.6	-17.3	9.0	31721
  (1,1)	17.1	13.0	11.1	10.0	32436
  (2,3)	17.0	-15.1	-7.8	9.0	29032

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F335M visit 1: worst tile 27011.0 mas at cell (5,3) (> 15 mas)

  36/36 tiles are reported "ok", but that counts tiles whose offset histogram had a coherent PEAK — not tiles within tolerance. The m7 cross-band gate is no significant 2" cell above 15 mas, and this cell exceeds it, so the bulk tie being ~0 does not mean the field is flat.

  checkpoint_m2_F335M_latest.json

  what is affected, and why

  27/36 cells exceed 15 mas. They are not confined to the edge, so this is an interior residual: a distortion or per-detector alignment problem, not a coverage artefact. The bulk tie for this visit is 3.08 mas — a small bulk value cannot cancel a local one.

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  cell	off (mas)	dRA	dDec	contrast	pairs
  (5,3)	27,011.0	-9,883.8	25,137.7	9.0	71550
  (3,3)	26,094.6	-23,104.4	12,129.1	8.0	75605
  (2,1)	53.3	27.3	45.8	11.0	53553
  (4,5)	52.3	-10.9	-51.2	15.0	81405
  (3,0)	45.0	-28.1	35.1	11.0	73477
  (0,3)	41.8	-20.1	-36.7	9.0	65481
  (4,4)	41.8	-34.0	-24.4	10.0	10795
  (4,2)	33.5	-24.2	-23.1	12.0	66000
  (2,2)	30.3	13.1	-27.3	12.0	63438
  (0,1)	28.9	13.0	25.8	11.0	77000
  (1,3)	27.9	7.6	26.8	12.0	83278
  (3,4)	25.3	-23.9	8.3	8.5	94563
  (0,4)	25.0	13.4	21.1	8.0	67407
  (5,1)	24.7	-14.5	-19.9	11.0	80000
  (5,0)	23.0	-13.6	18.6	9.0	50416
  (0,0)	22.0	14.8	16.2	13.0	82127
  (1,0)	21.8	21.8	-0.8	12.0	83457
  (0,2)	21.5	-5.8	20.7	12.0	66851
  (1,4)	21.4	-18.7	10.4	8.0	69409
  (2,4)	20.7	-12.9	-16.2	6.0	11487
  (0,5)	19.8	13.6	14.4	7.0	39145
  (3,2)	18.2	-13.8	-11.8	8.0	9571
  (3,5)	16.6	-11.8	11.8	14.0	73694
  (4,0)	16.0	14.9	5.8	8.0	74647
  (4,3)	15.9	-13.1	9.0	10.0	52911
  (1,1)	15.5	8.9	-12.7	12.0	68845
  (5,4)	15.0	8.9	-12.1	11.0	60312

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F335M_o007 visit 1: worst tile 27010.6 mas at cell (5,3) (> 15 mas)

  36/36 tiles are reported "ok", but that counts tiles whose offset histogram had a coherent PEAK — not tiles within tolerance. The m7 cross-band gate is no significant 2" cell above 15 mas, and this cell exceeds it, so the bulk tie being ~0 does not mean the field is flat.

  checkpoint_m2_F335M_o007_latest.json

  what is affected, and why

  27/36 cells exceed 15 mas. They are not confined to the edge, so this is an interior residual: a distortion or per-detector alignment problem, not a coverage artefact. The bulk tie for this visit is 0.02 mas — a small bulk value cannot cancel a local one.

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  cell	off (mas)	dRA	dDec	contrast	pairs
  (5,3)	27,010.6	-9,880.8	25,138.5	9.0	71550
  (3,3)	26,128.3	-23,134.6	12,144.0	9.0	75608
  (1,4)	15,227.9	-14,493.5	4,672.0	7.0	69411
  (4,5)	61.1	-8.9	-60.5	15.0	81404
  (2,1)	55.5	30.2	46.5	11.0	53553
  (3,0)	43.8	-25.3	35.8	11.0	73473
  (0,3)	39.7	-17.2	-35.8	9.0	65479
  (2,2)	31.0	16.0	-26.5	14.0	63439
  (4,2)	30.8	-21.3	-22.3	11.0	66003
  (1,3)	29.6	10.5	27.7	13.0	83278
  (0,4)	29.1	-23.3	17.4	8.0	67409
  (0,1)	26.6	13.1	23.2	11.0	76998
  (5,0)	25.8	-16.9	19.5	9.0	50414
  (0,0)	24.7	17.8	17.1	14.0	82128
  (4,4)	23.1	12.8	-19.2	10.0	10794
  (3,4)	22.9	-21.0	9.1	8.0	94555
  (0,5)	22.5	16.6	15.3	7.0	39148
  (5,1)	22.3	-11.6	-19.1	10.0	79996
  (0,2)	21.8	-2.8	21.6	11.0	66850
  (1,0)	21.7	21.7	-0.6	12.0	83456
  (2,4)	20.4	-13.7	-15.0	6.0	11488
  (4,0)	19.1	17.9	6.7	8.0	74645
  (4,3)	17.5	-17.1	-3.5	9.0	52912
  (1,1)	16.8	11.8	-11.9	13.0	68847
  (5,4)	16.4	11.9	-11.3	11.0	60315
  (3,2)	15.5	-10.9	-11.1	6.0	9574
  (3,5)	15.1	-9.5	11.7	13.0	73688

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F470N visit 1: worst tile 37.8 mas at cell (2,1) (> 15 mas)

  36/36 tiles are reported "ok", but that counts tiles whose offset histogram had a coherent PEAK — not tiles within tolerance. The m7 cross-band gate is no significant 2" cell above 15 mas, and this cell exceeds it, so the bulk tie being ~0 does not mean the field is flat.

  checkpoint_m2_F470N_latest.json

  what is affected, and why

  19/36 cells exceed 15 mas. They are not confined to the edge, so this is an interior residual: a distortion or per-detector alignment problem, not a coverage artefact. The bulk tie for this visit is 1.46 mas — a small bulk value cannot cancel a local one.

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  cell	off (mas)	dRA	dDec	contrast	pairs
  (2,1)	37.8	20.4	31.9	12.0	37350
  (0,3)	30.3	9.9	-28.6	13.0	44734
  (3,3)	29.4	-6.7	28.6	12.0	42764
  (5,3)	28.7	-17.3	-22.9	8.0	31931
  (1,4)	28.1	-28.0	-1.7	10.0	37768
  (3,4)	27.3	-14.6	-23.0	12.0	45990
  (0,4)	26.6	-20.6	16.8	8.0	36599
  (4,2)	25.0	-13.8	-20.8	12.0	39428
  (4,5)	24.2	-8.0	-22.9	13.0	37259
  (1,5)	24.1	18.4	-15.5	10.0	28559
  (0,1)	23.7	11.6	20.7	16.0	44053
  (4,3)	21.9	-21.8	2.4	10.0	41562
  (3,0)	21.4	-18.4	10.8	13.0	38923
  (4,0)	18.3	13.0	-12.9	7.0	37775
  (1,0)	17.8	15.1	9.3	18.0	52140
  (3,5)	17.8	-13.7	11.3	15.0	39561
  (2,3)	17.5	4.2	-17.0	13.0	39264
  (4,4)	16.8	-9.8	-13.7	17.0	47429
  (1,3)	15.7	1.6	15.6	13.0	43888

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F470N_o007 visit 1: worst tile 37.8 mas at cell (2,1) (> 15 mas)

  36/36 tiles are reported "ok", but that counts tiles whose offset histogram had a coherent PEAK — not tiles within tolerance. The m7 cross-band gate is no significant 2" cell above 15 mas, and this cell exceeds it, so the bulk tie being ~0 does not mean the field is flat.

  checkpoint_m2_F470N_o007_latest.json

  what is affected, and why

  19/36 cells exceed 15 mas. They are not confined to the edge, so this is an interior residual: a distortion or per-detector alignment problem, not a coverage artefact. The bulk tie for this visit is 1.46 mas — a small bulk value cannot cancel a local one.

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  cell	off (mas)	dRA	dDec	contrast	pairs
  (2,1)	37.8	20.4	31.9	12.0	37350
  (0,3)	30.3	9.9	-28.6	13.0	44734
  (3,3)	29.4	-6.7	28.6	12.0	42764
  (5,3)	28.7	-17.3	-22.9	8.0	31931
  (1,4)	28.1	-28.0	-1.7	10.0	37768
  (3,4)	27.3	-14.6	-23.0	12.0	45990
  (0,4)	26.6	-20.6	16.8	8.0	36599
  (4,2)	25.0	-13.8	-20.8	12.0	39428
  (4,5)	24.2	-8.0	-22.9	13.0	37259
  (1,5)	24.1	18.4	-15.5	10.0	28559
  (0,1)	23.7	11.6	20.7	16.0	44053
  (4,3)	21.9	-21.8	2.4	10.0	41562
  (3,0)	21.4	-18.4	10.8	13.0	38923
  (4,0)	18.3	13.0	-12.9	7.0	37775
  (1,0)	17.8	15.1	9.3	18.0	52140
  (3,5)	17.8	-13.7	11.3	15.0	39561
  (2,3)	17.5	4.2	-17.0	13.0	39264
  (4,4)	16.8	-9.8	-13.7	17.0	47429
  (1,3)	15.7	1.6	15.6	13.0	43888

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F480M visit 1: worst tile 19082.4 mas at cell (0,5) (> 15 mas)

  36/36 tiles are reported "ok", but that counts tiles whose offset histogram had a coherent PEAK — not tiles within tolerance. The m7 cross-band gate is no significant 2" cell above 15 mas, and this cell exceeds it, so the bulk tie being ~0 does not mean the field is flat.

  checkpoint_m2_F480M_latest.json

  what is affected, and why

  23/36 cells exceed 15 mas. They are not confined to the edge, so this is an interior residual: a distortion or per-detector alignment problem, not a coverage artefact. The bulk tie for this visit is 2.14 mas — a small bulk value cannot cancel a local one.

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  cell	off (mas)	dRA	dDec	contrast	pairs
  (0,5)	19,082.4	463.6	-19,076.7	7.0	48454
  (2,1)	44.8	19.3	40.4	13.0	61519
  (5,1)	37.4	-20.8	-31.1	13.0	82582
  (4,5)	37.1	-10.8	-35.5	8.0	10560
  (4,4)	33.8	-29.4	-16.7	12.0	11331
  (1,5)	33.0	30.7	-12.1	10.0	62144
  (0,3)	30.5	-7.2	-29.6	12.0	79253
  (5,3)	28.4	-18.8	-21.3	8.0	66336
  (0,4)	27.3	-20.2	18.3	13.0	83464
  (4,2)	25.8	-17.5	-19.0	15.0	78017
  (0,0)	24.7	-11.0	22.1	10.0	11637
  (3,0)	24.1	-19.9	13.6	15.0	78648
  (2,3)	22.8	4.8	-22.3	13.0	81903
  (1,4)	22.2	-21.5	-5.7	5.0	8988
  (4,0)	21.0	14.7	-15.0	8.0	77269
  (2,4)	19.5	10.4	-16.5	9.0	13205
  (0,1)	19.5	2.1	19.4	9.0	11415
  (3,2)	18.7	-13.4	-13.0	10.0	10116
  (4,3)	16.3	-15.0	6.5	11.0	68454
  (3,5)	15.8	-12.4	9.8	16.0	83317
  (0,2)	15.7	-0.0	15.7	11.0	76352
  (1,3)	15.7	7.8	13.7	9.0	9807
  (3,4)	15.2	7.8	-13.1	7.0	10590

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• warn
  F480M_o007 visit 1: worst tile 19082.1 mas at cell (0,5) (> 15 mas)

  35/36 tiles are reported "ok", but that counts tiles whose offset histogram had a coherent PEAK — not tiles within tolerance. The m7 cross-band gate is no significant 2" cell above 15 mas, and this cell exceeds it, so the bulk tie being ~0 does not mean the field is flat.

  checkpoint_m2_F480M_o007_latest.json

  what is affected, and why

  21/36 cells exceed 15 mas. They are not confined to the edge, so this is an interior residual: a distortion or per-detector alignment problem, not a coverage artefact. The bulk tie for this visit is 0.14 mas — a small bulk value cannot cancel a local one.

  Diagnostic writeupD3 — absolute tie to the reference frame · full diagnostic writeup (PDF)

  cell	off (mas)	dRA	dDec	contrast	pairs
  (0,5)	19,082.1	465.8	-19,076.4	7.0	48454
  (2,1)	46.0	21.4	40.8	13.0	61517
  (4,5)	36.2	-8.7	-35.1	8.0	10562
  (5,1)	36.0	-18.7	-30.7	13.0	82583
  (1,5)	34.9	32.8	-11.7	10.0	62144
  (0,3)	32.9	9.3	-31.5	13.0	79254
  (4,4)	31.8	-27.3	-16.3	12.0	11329
  (0,4)	27.8	-18.9	20.4	14.0	83465
  (5,3)	26.8	-16.7	-20.9	8.0	66336
  (1,4)	25.8	-25.3	-5.4	4.5	90363
  (0,0)	24.2	-8.9	22.5	11.0	11637
  (4,2)	24.2	-15.3	-18.6	15.0	78016
  (2,3)	23.0	6.9	-22.0	13.0	81906
  (3,0)	22.6	-17.8	14.0	15.0	78652
  (4,0)	22.3	16.8	-14.6	8.0	77268
  (2,4)	20.5	12.5	-16.2	9.0	13205
  (0,1)	20.2	4.2	19.8	9.0	11416
  (3,1)	17.7	14.7	9.9	12.0	86861
  (3,2)	17.0	-11.3	-12.7	8.0	10116
  (1,0)	16.8	14.1	9.2	11.0	12220
  (5,4)	15.5	8.5	-13.0	14.0	69604

  Existing figures

  • astrometry_offsets_2026-07-10.png audit_plots
• ok
  F335M: all 24 exposures within 2.0 mas of consensus

  checkpoint_m2_F335M_latest.json
• ok
  F335M_o007: all 24 exposures within 2.0 mas of consensus

  checkpoint_m2_F335M_o007_latest.json
• ok
  F470N: all 24 exposures within 2.0 mas of consensus

  checkpoint_m2_F470N_latest.json
• ok
  F470N_o007: all 24 exposures within 2.0 mas of consensus

  checkpoint_m2_F470N_o007_latest.json
• ok
  F480M: all 24 exposures within 2.0 mas of consensus

  checkpoint_m2_F480M_latest.json
• ok
  F480M_o007: all 24 exposures within 2.0 mas of consensus

  checkpoint_m2_F480M_o007_latest.json
• ok
  m7 present (1 file(s))

  catalogs/
• ok
  m8 present (2 file(s))

  catalogs/
• ok
  8 recent log(s), no error signatures

  logs/
• ok
  m7: single tag 2026-08-21_PR460_3fd4e04-dirty (6 products)

  catalogs/*m7*.prov.json