LAST DB columns by pipeline version

This page lists, for the LAST v1 database tables, the pipeline versions that added columns or changed the meaning of existing columns. The base is the first release, lv1.0.0; a version appears here only if it brings such a change. For what each version changes in the processing, see LAST production releases.

How the tables are created and filled: LAST DB tables: creation and ingestion.

Tables covered

• image_visit: one row per coadd sub-image, from the coadd image header (discussion #1409).
• image_proc: one row per epoch sub-image, from the epoch image header (discussion #1413).
• src_visit, src_proc: one row per source in the coadd and epoch catalogues (discussion #1411).

The image-table columns are header keywords (column name = keyword in lower case); the source-table columns are catalogue columns. Columns that the ingestion computes itself (id_coadd in image_proc, bjd, ingestion_time_jd, …) are listed only when a pipeline change alters them.

What counts as a change of meaning: a change of what a column measures, of its units, or of how its value is coded. Changes of the algorithms that only make the values better or different (for example the coadd changes of lv1.1.0) are not listed; they are described in the release notes.

Pipeline version of a row: the image tables have the column pipever, the header keyword PIPEVER (for example ':lv1.1.2'; the text before the colon is the git branch, empty for a release tag). Rows of the source tables get it through id_coadd / id_proc.

Status (2026 Oct 21): the tables exist only in the test database last_test. All products made before lv1.1.0 were deleted from the archive, so the archived data, reprocessed with lv1.1.0 and later, have no lv1.0.x rows.

Changes since lv1.0.0

Version	Date	Table	Column	Change
lv1.1.3	2026 Oct 22	src_proc	rows with forced = 1	rows, no column change: the code is that of lv1.1.2, but the forced-photometry catalogue ForcedPhotList has 85 more positions (fast radio bursts), so visits that contain one of them have forced-photometry rows there. About 220 visits started on 2026 Oct 22, 06:05–06:38 UT, have pipever = ':lv1.1.2' although they may have used the new catalogue.
lv1.1.2	2026 Oct 20	image_visit, image_proc	pt_normu, pt_fnorm	new: pt_normu = 'm2' gives the units of the normalisation pt_1_v1; pt_fnorm is the coadd's pt_1_v1 divided by the collecting area of a LAST telescope (π·0.1397² m²), about 0.31 (#1431). In image_proc both are copied from the coadd of the same visit and sub-image, so pt_fnorm is not the epoch's own pt_1_v1 divided by the area.
		image_visit, image_proc	pt_1_v1	meaning: an effective collecting area in m² (about 0.019, in the coadd and in the epoch headers); before, relative to the area of a LAST telescope (about 0.35), so old values compare with the new pt_fnorm, not with the new pt_1_v1 (#1431).
lv1.1.0	2026 Oct 19	image_visit	maxbckd	new: the largest difference of the background levels (MEDBCK) of the coadded epochs (#1355).
lv1.0.10	2026 Oct 18	image_proc, src_proc	id_proc	coding: the level part of the image ID is "proc" (3) instead of "raw" (2), so the IDs are larger by 16 (#1412). The header keyword LEVEL (not a column) is now proc.
		image_visit	id_procf, id_procl	coding: the same: the IDs of the first and last epoch carry the "proc" level (#1412). id_coadd is unchanged.
lv1.0.6	2026 Oct 17	src_proc	magerr_psf, sn (forced-photometry rows, forced = 1)	meaning: the error and S/N now include the photon noise of the source itself; before only the background noise, so the errors of bright stars were too small (1.5× at S/N 30–50, 13× at S/N 200–400) (#1408). Detected rows are unchanged.
lv1.0.4	2026 Oct 15	image_visit	coaddop	coding: the value is wrobust; before it was the typo wrbosut (same method).

The source tables have no new columns: the coadd catalogues have the same 54 columns and the epoch catalogues the same 51 columns in the samples of #1411 (coadd catalogues lv1.0.4, epoch catalogues lv1.0.5), in lv1.1.0 and in lv1.1.2.

File form (not a column change): since lv1.1.0 the epoch catalogues and epoch headers of a visit are saved in two MAT files, proc_Cat_1.mat and proc_Head_1.mat, instead of 480 proc_Cat FITS files (#1417). The content is the same, and the ingestion reads both forms.

Columns that come from the camera software

These keywords are written into the raw images by the camera control software (LAST_OCS) when the image is taken. So whether a row has them depends on the night of the raw image, not on the pipeline version: older nights reprocessed with lv1.1.2 (for example 2026 April) do not have them.

Since	Table	Column	Change
LAST_OCS 0c5189f, raw images from the night of 2026 Oct 19	image_visit, image_proc	schedulr	new: the scheduler that chose the target, e.g. last.survey (LAST_OCS #132).
LAST_OCS 67c3f4d, raw images from the night of 2026 Aug 20	image_visit, image_proc	quicfwhm	new: the camera software's quick estimate of the FWHM in the centre of the raw image, in pixels; in image_visit the value of one of the epochs.

Rows made before a column existed hold the column default ('' for text, NaN for numbers).