Digitized Sky Survey POSS-1

[HoaxEye]

I mean according to Solano et al. 2022, they did identify interesting phenomena. The last sentence of the abstract is the bit about the lack of vanishing stars I think.

In this paper, we report a search for vanishing sources in POSS I red images using virtual observatory (VO) archives, tools, and services. The search, conducted in the framework of the VASCO project, aims at finding POSS I (red) sources not present in recent catalogues like Pan-STARRS DR2 (limiting magnitude r = 21.4) or Gaia EDR3 (limiting magnitude G = 21). We found 298 165 sources visible only in POSS I plates, out of which 288 770 had a cross-match within 5 arcsec in other archives (mainly in the infrared), 189 were classified as asteroids, 35 as variable objects, 3592 as artefacts from the comparison to a second digitization (Supercosmos), and 180 as high proper motion objects without information on proper motion in Gaia EDR3. The remaining unidentified transients (5399) as well as the 172 163 sources not detected in the optical but identified in the infrared regime are available from a VO compliant archive and can be of interest in searches for strong M-dwarf flares, high-redshift supernovae, asteroids, or other categories of unidentified red transients. No point sources were detected by both POSS-I and POSS-II before vanishing, setting the rate of failed supernovae in the Milky Way during 70 yr to less than one in one billion.

Agreed. MNRAS 2022 is interesting project, and that's why I started my project in the first place. Too bad I had to do this: https://github.com/jannefi/vasco/releases/tag/v.0.9.5-legacy-mnras-repro - I archived the repro work as a "historical" release. But I can continue the work with changed goals: https://github.com/jannefi/vasco/blob/main/README.md (published about 30 minutes ago).

There are some well known projects in this field like the Digital Access to a Sky Century @ Harvard (DASCH) project.
Link: https://dasch.cfa.harvard.edu/

DASCH was the project to digitize the Harvard College Observatory's Astronomical Photographic Glass Plate Collection for scientific applications. This enormous — multi-decade — undertaking was completed in 2024. Its legacy is DASCH Data Release 7, an extraordinary dataset that enables scientific study of the entire night sky on 100-year timescales.

Research involved investigating transient-like phenomena on early plates that could be caused by either rare astrophysical events or instrumental noise. E.g. "Extreme Transients Discovered with DASCH: New BH-LMXBs or new Classical Novae?" https://scixplorer.org/abs/2019AAS...23311206G/abstract (Grindlay, 2019).

This is a very interesting research area. There are challenges related to old glass plate images and producing good digital datasets from them, but that doesn't mean they cannot be studied. One "ghost image" is discussed here: https://dasch.cfa.harvard.edu/pipeline-overview/

The image at the left (from plate I31090) shows the center of M44 taken in 1903 and illustrates a special analysis problem presented by old photometric techniques. Note that the triangle on the right has a ghost image on the left generated by a "Pickering Wedge". This wedge was placed on the telescope objective in an attempt to extend the dynamic range of film. When the primary star image is saturated, a researcher can measure the ghost images and add a known wedge magnitude offset.

 

[Mendel]

Agreed. MNRAS 2022 is interesting project, and that's why I started my project in the first place. Too bad I had to do this: https://github.com/jannefi/vasco/releases/tag/v.0.9.5-legacy-mnras-repro - I archived the repro work as a "historical" release. But I can continue the work with changed goals: https://github.com/jannefi/vasco/blob/main/README.md (published about 30 minutes ago).

Good choice! Plate awareness facilitates more meaningful statistical analyses.

 

[HoaxEye]

About The Minnesota Automated Plate Scanner (MAPS) Catalog discussed by Watters et al. It's a catalogue derived from the first epoch of Palomar Sky Survey (POSS-I). It contains over 89 million matched stellar objects (stars and galaxies). There's a problem: it has been retired from Vizier. I couldn't find it from any online catalog search provider either.

But the data can be found from this website: https://aps.umn.edu/ Although it covers almost 90 million objects, it's not huge, around 13 gigabytes. I decided to make a local mirror and perhaps run some tests with it. Here's one simple download script: https://github.com/jannefi/vasco/blob/main/scripts/maps_mirror.sh - it uses aria2c for downloads.

The data is in proprietary format. Luckily there is a good enough description of it on their website. They talk about C and Perl source code examples, but those are not on their website - and never has been (according to archive.org). I use Python and Parquet a lot, so here's my Python converter: https://github.com/jannefi/vasco/blob/main/scripts/maps_decode_to_parquet.py Additional script for adding ICRS coordinates based on original FK4 B1950: https://github.com/jannefi/vasco/blob/main/scripts/maps_b1950_to_icrs.py

All this MAPS stuff, and the fact I'm facing big design and code changes, made me think about the STScI coordinates: does the FITS downloader service return good enough coordinates for 30x30 arcmin tiles? It actually does, for the purposes of source extractor. Based on FITS headers, they declare ICRS with EQUINOX 2000. Not B1950 or similar. However, tile WCS is a linear TAN approximation anchored at CRPIX/CRVAL It's a standard, but it's an approximation over a finite field of view. AMDX/AMDY are present, indicating there is a richer, non-linear astrometric model for the tile geometry that may provide better accuracy than the linear WCS alone.

But I can't test how accurate STScI tile coordinates are. This is why I've been using "wcs fixed" coordinates in the software, but only in downstream scripts, not in the pipeline.

The scripts doesn't actually "fix" anything, I just call it "wcs fixing". In reality it's a Gaia-tied polynomial correction of catalog RA/Dec, which is a very good coordinate correction method provided that you get enough Gaia matches. It's not time-dependent, which makes it even more suitable for this project.

Because I'm now free to explore, I made some tests with "wcs fixed" vs. original coordinates against latest Gaia and PS1 datasets. Difference was huge. I couldn't believe the numbers so I made a test set consisting of 60K rows of data from 200 randomly selected tiles. The confusion matrix results agreed, and today I made additional false-match tests. Details here: https://github.com/jannefi/vasco/wiki/Confusion‐table-summary

In the future, I will use canonical coordinates only. Together with newer optical catalogs like PS1/Gaia, the match rate will increase dramatically even at 2" arcsec (current MNRAS 2022 defined radius is 5" arcsec).

I also decided to store the test files in case anyone wants to re-test or check. Decision is to use 2" arcsec search radius with canonical / Gaia-tied coordinates. How to implement this in the pipeline... I don't know yet. It seems there's always something that needs be implemented and tested.