Digitized Sky Survey POSS-1

[Mick West]

You can download the plates they used as FITS

Well, yes. But how? They mention 635 plates. How do I get one of these?

 

[HoaxEye]

I added an Earth shadow visualization to Sitrec
https://www.metabunk.org/threads/transients-in-the-palomar-observatory-sky-survey.14362/post-355369
View attachment 85312

I think a useful next step would be to add a FITS viewer, which would take a FITS file and paste it in the correct position on the celestial sphere, with adjustable transparency so you can see the (larger) stars that match the plate. You could then also (based on the plate time) animate the shadow over it.

I think this would be useful for visualizing what is going on and maybe also for validating some things.

Can someone suggest a good FITS file to start with? Do we know what the 635 transient files they used were?

Alternatively, do we just have the metadata for them, so I can display their outlines all at once?

One identified plate is dss1red_XE186.fits. You can download it here: https://irsa.ipac.caltech.edu/data/DSS/images/dss1red/
That's from 19th of July 1952. It's mentioned in this paper, also visible in the figures section: https://www.nature.com/articles/s41598-025-21620-3
Coordinates from my own notes: RA: 21 02 52.28 DEC: +48 34 18.90

 

[Mick West]

One identified plate is dss1red_XE186.fits. You can download it here: https://irsa.ipac.caltech.edu/data/DSS/images/dss1red/
That's from 19th of July 1952. It's mentioned in this paper, also visible in the figures section: https://www.nature.com/articles/s41598-025-21620-3
Coordinates from my own notes: RA: 21 02 52.28 DEC: +48 34 18.90

Thanks. I've converted @boguesuser's Rust code to JS, and I'm downloading them in batches. The red images seem to be 14,000x14,000 pixels! 392MB each (all the same size as uncompressed).

Now, how to map this one to the celestial sphere. I'd need the ra/dec of the corners, then grid it into something like 20x20 squares

 

[HoaxEye]

Thanks. I've converted @boguesuser's Rust code to JS, and I'm downloading them in batches. The red images seem to be 14,000x14,000 pixels! 392MB each (all the same size as uncompressed).
View attachment 85315

Now, how to map this one to the celestial sphere. I'd need the ra/dec of the corners, then grid it into something like 20x20 squares

I suppose ChatGPT might help:

To find the coordinates of the top-left corner of a FITS image when the header gives the center coordinates, you'll need to use the World Coordinate System (WCS) information in the FITS header. Here's how it works conceptually:

Step-by-Step Guide

1. Extract WCS info from the header:Look for keywords like:
   
   • CRPIX1, CRPIX2: pixel coordinates of the reference point (usually the center).
   • CRVAL1, CRVAL2: world coordinates (e.g., RA/Dec) at the reference point.
   • CD1_1, CD1_2, CD2_1, CD2_2 or CDELT1, CDELT2 and CROTA2: transformation matrix or scale/rotation.
   • NAXIS1, NAXIS2: image dimensions in pixels.
2. Calculate pixel coordinates of the top-left corner:
   
   • Top-left pixel is usually at (1, 1) in FITS convention (or (0, 0) in Python/NumPy).
3. Use WCS transformation:
   
   • Apply the WCS transformation to convert pixel coordinates (1, 1) to world coordinates (RA/Dec or other system).

---

Example Using Python (Astropy)

If you have the FITS file, you can use astropy like this:



from astropy.io import fits
from astropy.wcs import WCS
# Load FITS file
hdul = fits.open('yourfile.fits')
header = hdul[0].header

# Create WCS object
wcs = WCS(header)
# Convert pixel (1, 1) to world coordinates
topleftworld = wcs.pixeltoworld(1, 1)
print(topleft_world)
``

---

 

[HoaxEye]

One identified plate is dss1red_XE186.fits. You can download it here: https://irsa.ipac.caltech.edu/data/DSS/images/dss1red/
That's from 19th of July 1952. It's mentioned in this paper, also visible in the figures section: https://www.nature.com/articles/s41598-025-21620-3
Coordinates from my own notes: RA: 21 02 52.28 DEC: +48 34 18.90

Here's another identified plate: dss1red_XE297.fits, it's the "candidate 1" from 1954 in the PASP paper.
Coordinates from the paper: RA 02:29:33.71, DEC +28:31:56.98

 

[Mendel]

http://svocats.cab.inta-csic.es/vanish-possi/

The SVO archive of vanishing objects in POSS I red images provides access to the catalogue of vanishing objects in POSS I red images described in Solano et al 2022. It contains 5399 sources detected in POSS I red images but not in any other archive.

In this sub-catalogue we provide information about 5399 sources detected in POSS I red images.

Article:

Coverage Map

Source: http://svocats.cab.inta-csic.es/vanish-possi/index.php?action=coverage

Article:

Figure 9.
Spatial distribution in galactic coordinates of the final list of 5399 candidates (blue dots). A coloured POSS-II images is displayed in the background.

Source: https://academic.oup.com/mnras/article/515/1/1380/6607509

 

[flarkey]

This is interesting. It lets us see the original POSS-I images with the 'transient' and the associated image where it has 'vanished'.

Edit - I don't think this list is showing only filtered 'transients', I think it includes both possible transients and plate artefacts.

http://svocats.cab.inta-csic.es/vanish-possi/index.php?action=search

 

[HoaxEye]

This is interesting. It lets us see the original POSS-I images with the 'transient' and the associated image where it has 'vanished'.

Edit - I don't think this list is showing only filtered 'transients', I think it includes both possible transients and plate artefacts.

http://svocats.cab.inta-csic.es/vanish-possi/index.php?action=search

Yes, this is quite useful, although the pictures appear to be screenshots from Aladin.
Too bad they didn't list the actual plate ID for each row: the FITS files are required for any automated analysis, probably also for Sitrec. FITS header contains important metadata that can be used to make various calculations.

I haven't found plate ID <-> transient mapping in any of the published or preprint papers. I also haven't found any software they used in the automated pipeline for identifying large number of transients. They talk about open-source and being transparent, but so far they haven't published enough info to even reproduce their findings.

 

[Mendel]

I haven't found plate ID <-> transient mapping in any of the published or preprint papers.

I was hoping that the RA/DEC sky coordinates would narrow down the plates it could be on from the plate metadata, ideally down to a single plate.

 

[HoaxEye]

I was hoping that the RA/DEC sky coordinates would narrow down the plates it could be on from the plate metadata, ideally down to a single plate.

There are no dates either. I don't think we can find the corresponding plate without the observation date.

Unless these transients are from a single plate and the date or plate id is published somewhere

Update: found the paper for this dataset https://academic.oup.com/mnras/article/515/1/1380/6607509?login=false They used scripting capabilities of Aladin to "tessellate the sky covered by POSS I in circular regions of 30-arcmin radius". I couldn't find the script(s)

 

[Mendel]

Edit - I don't think this list is showing only filtered 'transients', I think it includes both possible transients and plate artefacts.

Yes.
Plate artefacts have not been ruled out.

Article:

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)

Source: https://academic.oup.com/mnras/article/515/1/1380/6607509

Going into this, they had ruled out some transients that may be plate artefacts:

Other artefacts' removal: Morphometric parameters like the full width at half-maximum (FWHM) or the elongation were used to clean the SE xtractor catalogue from spurious sources.

 

[HoaxEye]

Update: found the paper for this dataset https://academic.oup.com/mnras/article/515/1/1380/6607509?login=false They used scripting capabilities of Aladin to "tessellate the sky covered by POSS I in circular regions of 30-arcmin radius". I couldn't find the script(s)

I had not seen this paper for some time https://academic.oup.com/mnras/article/515/1/1380/6607509?login=false Looking at it now, it could be that they used the described workflow in all their analysis, including the new papers. At least that would explain why they have not released e.g. plate IDs or software, often just coordinates and some Aladin screenshots. Most of the required sw can be found from this publication:
- Aladin
- SExtractor
- stilts-cdsskymatch
Only thing perhaps missing is the Aladin scripts they used.

Also: they used POSS-I red plates from the Digitized Sky Survey from ESO. https://archive.eso.org/dss/dss/ Not from Caltech. Perhaps they used the command line batch tool: https://archive.eso.org/cms/tools-documentation/the-eso-st-ecf-digitized-sky-survey-application.html

All workflow steps are described. Only thing that seems hard to reproduce is the tessellation of sky covered by POSS-I in circular regions of 30-arcmin radius, because they apparently used Aladin scripting capabilities. Paper refers to: "Boch & Fernique 2014", which can be found online: https://www.aspbooks.org/a/volumes/article_details/?paper_id=36231 - that article talks about web-based Aladin Lite, which is weird.

If this is the only software environment VASCO team used, it might be fairly easy to replicate their analysis after they release the Aladin scripts, or at least describe how the sky tessellation step was done.

 

[Mick West]

Thanks. I've converted @boguesuser's Rust code to JS, and I'm downloading them in batches. The red images seem to be 14,000x14,000 pixels! 392MB each (all the same size as uncompressed).

I've got most of them downloaded now, looks like it will be about 1.2TB total. There's 626 blue images, most of which are 1GB each, but also 14,000x13,999. I generated a client-side FITS parser which can decode them in about 20 seconds. I'll see if I can incorporate that into Sitrec, eventually. Bit busy now.

 

[Mendel]

I've got most of them downloaded now, looks like it will be about 1.2TB total. There's 626 blue images, most of which are 1GB each, but also 14,000x13,999. I generated a client-side FITS parser which can decode them in about 20 seconds. I'll see if I can incorporate that into Sitrec, eventually. Bit busy now.

Is that the Supercosmos scan, or the older one?

 

[Mick West]

Is that the Supercosmos scan, or the older one?

The older one, DSS.
https://www.ipac.caltech.edu/doi/irsa/10.26131/IRSA441

The Digitized Sky Survey (DSS) is a ground-based imaging survey of the entire sky in several colors. The survey, performed with Palomar and UK Schmidt telescopes, produced photographic plates that were later digitized at the Space Telescope Science Institute (STScI) to produce the Hubble Guide Stars Catalog (GSC).

Each plate produced by the Survey covers 6.5 x 6.5 degrees of the sky, and the plates have been digitized using a modified PDS microdensitometer. The digital images have a pixel size of either 25 microns (1.7 arcsec per pixel) or 15 microns (1.0 arcsec per pixel), and are 14000 x 14000 or 23040 x 23040 pixels per side.

Survey / Archive	Scanner	Pixel Scale	Data Access	Notes
DSS1 (STScI)	PDS (Photometric Data Systems) microdensitometer	1.7″/pix	MAST/STScI DSS	First public digitization; basis for most "DSS1" FITS files.
SuperCOSMOS (SSS)	SuperCOSMOS (Edinburgh)	0.67″/pix	WFAU/ROE SuperCOSMOS	Higher resolution and better astrometry; rescanned POSS-I and POSS-II plates.
APS (Minnesota Automated Plate Scanner)	APS microdensitometer	0.5–0.7″/pix	APS Project	Scanned for object catalogues; limited public FITS availability.
DSS2 (STScI)	POSS-II plates scanned at STScI	1.0″/pix	MAST DSS2	Deeper plates, later epoch (~1980s–1990s).
GSC-II	Combined POSS-II + UKST scans	1.0″/pix	Internal (STScI/ESO)	Used for Hubble Guide Star Catalog; not public as raw FITS.

 

[NorCal Dave]

The older one, DSS.

From your post:

The survey, performed with Palomar and UK Schmidt telescopes, produced photographic plates that were later digitized at the Space Telescope Science Institute

Does this mean the Brits were doing the same thing? In the same time period? If so, wouldn't these so-called techno signature transients appear on the UK plates as well? If they were supposedly in geostationary orbit, wouldn't the UK plates have recorded them as well? Maybe they were using a different red emulsion?

 

[Ravi]

From your post:

The survey, performed with Palomar and UK Schmidt telescopes, produced photographic plates that were later digitized at the Space Telescope Science Institute

Does this mean the Brits were doing the same thing? In the same time period? If so, wouldn't these so-called techno signature transients appear on the UK plates as well? If they were supposedly in geostationary orbit, wouldn't the UK plates have recorded them as well? Maybe they were using a different red emulsion?

UK and Palomar are about 120deg longitudinally separated..

 

[Mendel]

UK and Palomar are about 120deg longitudinally separated..

31°16′20″S 149°04′16″E
33°21′29″N 116°51′43″W
I'm seeing 94⁰ difference in longitude.
The UK Schmidt is in New South Wales, and covered the Southern sky for the catalogue.
It should obviously have noticed a similar amount of transients, being based on the same technology.

 

[HoaxEye]

I've got most of them downloaded now, looks like it will be about 1.2TB total. There's 626 blue images, most of which are 1GB each, but also 14,000x13,999. I generated a client-side FITS parser which can decode them in about 20 seconds. I'll see if I can incorporate that into Sitrec, eventually. Bit busy now.

Mick: could you share the FITS parses source code? I would like to try it.

 

[HoaxEye]

The older one, DSS.
https://www.ipac.caltech.edu/doi/irsa/10.26131/IRSA441

The Digitized Sky Survey (DSS) is a ground-based imaging survey of the entire sky in several colors. The survey, performed with Palomar and UK Schmidt telescopes, produced photographic plates that were later digitized at the Space Telescope Science Institute (STScI) to produce the Hubble Guide Stars Catalog (GSC).

Each plate produced by the Survey covers 6.5 x 6.5 degrees of the sky, and the plates have been digitized using a modified PDS microdensitometer. The digital images have a pixel size of either 25 microns (1.7 arcsec per pixel) or 15 microns (1.0 arcsec per pixel), and are 14000 x 14000 or 23040 x 23040 pixels per side.

Survey / Archive	Scanner	Pixel Scale	Data Access	Notes
DSS1 (STScI)	PDS (Photometric Data Systems) microdensitometer	1.7″/pix	MAST/STScI DSS	First public digitization; basis for most "DSS1" FITS files.
SuperCOSMOS (SSS)	SuperCOSMOS (Edinburgh)	0.67″/pix	WFAU/ROE SuperCOSMOS	Higher resolution and better astrometry; rescanned POSS-I and POSS-II plates.
APS (Minnesota Automated Plate Scanner)	APS microdensitometer	0.5–0.7″/pix	APS Project	Scanned for object catalogues; limited public FITS availability.
DSS2 (STScI)	POSS-II plates scanned at STScI	1.0″/pix	MAST DSS2	Deeper plates, later epoch (~1980s–1990s).
GSC-II	Combined POSS-II + UKST scans	1.0″/pix	Internal (STScI/ESO)	Used for Hubble Guide Star Catalog; not public as raw FITS.

Note that it's possible that VASCO team used POSS-I scans from the Digitized Sky Survey from ESO. https://archive.eso.org/dss/dss/ Not from Caltech. Not sure if they used the web interface or command line tools: https://archive.eso.org/cms/tools-documentation/the-eso-st-ecf-digitized-sky-survey-application.html

I don't know if ESO scans are same as the Caltech scans. I also don't remember how we ended up downloading FITS files from Caltech. This needs to be checked. If the goal is to reproduce their findings, we better use the same dataset.

 

[Eburacum]

Coverage Map
View attachment 85316

------------------------------

View attachment 85317
Spatial distribution in galactic coordinates of the final list of 5399 candidates (blue dots). A coloured POSS-II images is displayed in the background.[/article]

A superficial visual comparison between these two maps seems to show a number of places where the distribution of anomalies is consistent with the orientation of the plates, especially towards the centre and to the right of the centre of the map.
So it seems likely that the production process of the maps has affected the distribution of the anomalies, at least to some extent.

 

[Mendel]

So it seems likely that the production process of the maps has affected the distribution of the anomalies, at least to some extent.

Now I really want to see a plot of these 5399 transients in terms of "distance from the plate edge".

 

[jarlrmai]

I've been mentioning this a few times early on, lets see these 5399 (or a random sample of 100 of them) and lets see how strong the line correlation is, what the interval is, how regular its is, where they are in relation to the full plate, how far they go across the full plate and in what directions they appear to be moving.

 

[Mendel]

I've been mentioning this a few times early on, lets see these 5399 (or a random sample of 100 of them) and lets see how strong the line correlation is, what the interval is, how regular its is, where they are in relation to the full plate, how far they go across the full plate and in what directions they appear to be moving.

We have the data for that, I think.
They're not "moving", though.

 

[jarlrmai]

They're not "moving", though.

Appear to be, based on the assumption (of the authors) that a line of dots = a moving glinting satellite.

 

[HoaxEye]

Note that it's possible that VASCO team used POSS-I scans from the Digitized Sky Survey from ESO. https://archive.eso.org/dss/dss/ Not from Caltech. Not sure if they used the web interface or command line tools: https://archive.eso.org/cms/tools-documentation/the-eso-st-ecf-digitized-sky-survey-application.html

I don't know if ESO scans are same as the Caltech scans. I also don't remember how we ended up downloading FITS files from Caltech. This needs to be checked. If the goal is to reproduce their findings, we better use the same dataset.

Headers of same POSS-I image downloaded from ESO https://archive.eso.org/dss/dss (on the left side of the screenshot) and from stsci.edu (on the right side of the screenshot) may suggest the datasets are slightly different. FITS files from Caltech have similar headers compared to stsci.

Villarroel et al. said they used ESO dataset in the earlier paper (https://academic.oup.com/mnras/article/515/1/1380/6607509?login=false) but mentions stsci in the latest papers. However, they do not specifically mention from where they took the plate images.

Perhaps I shouldn't worry about this: all scanned images are part of some known Digital Sky Survey. Maybe all sky surveys used the same POSS-I plate scans. But based on FITS headers, I just can't be sure.

Note that the observation date is different by one day: 20 July 1952 (ESO) vs 19th of July 1952 (stsci). Perhaps due to timezones. But I don't know for sure. These are the same plates: original plate ID assigned by the observatory is E533. Also the Galactic Supernova and Stellar Survey (GSSS) Region Name XE186 matches (and that's also part of the filename in CalTech servers).

Everything related to this topic is just complex - or perhaps it's just me

 

[Mick West]

Mick: could you share the FITS parses source code? I would like to try it.

https://www.metabunk.org/sitrec/tools/fits.html
https://github.com/MickWest/Sitrec2/blob/main/tools/fits.html

 

[overlord]

Now, how to map this one to the celestial sphere.

Article:

Keywords other than those for standard FITS headers describe the details of the plate scans, including a polynomial solution for mapping X, Y pixel coordinates on the plate to right ascension and declination.

Source: https://archive.stsci.edu/dss/booklet_n.pdf

It's from the booklet for the original CD-ROM data release, but should be applicable to the online available versions as well.
See pages 13 and 14 for details.

And if things are working correctly, I'd like to suggest the following exercise:
The entire sky-sphere color coded based on how many plates cover the given area* (red plates only, I'd expect no area to be over single digit coverage) and calculating the percentage for each, then this projected to 2D using Hammer projection (https://en.wikipedia.org/wiki/Hammer_projection)
* - Ignoring that the image edge and the actual plate edge don't align.

[edit] Also please state the content of the 'Origin' tag of the headers, to make sure people understand which files you are using.

 

[HoaxEye]

I had not seen this paper for some time https://academic.oup.com/mnras/article/515/1/1380/6607509?login=false Looking at it now, it could be that they used the described workflow in all their analysis, including the new papers. At least that would explain why they have not released e.g. plate IDs or software, often just coordinates and some Aladin screenshots. Most of the required sw can be found from this publication:
- Aladin
- SExtractor
- stilts-cdsskymatch
Only thing perhaps missing is the Aladin scripts they used.

Also: they used POSS-I red plates from the Digitized Sky Survey from ESO. https://archive.eso.org/dss/dss/ Not from Caltech. Perhaps they used the command line batch tool: https://archive.eso.org/cms/tools-documentation/the-eso-st-ecf-digitized-sky-survey-application.html

All workflow steps are described. Only thing that seems hard to reproduce is the tessellation of sky covered by POSS-I in circular regions of 30-arcmin radius, because they apparently used Aladin scripting capabilities. Paper refers to: "Boch & Fernique 2014", which can be found online: https://www.aspbooks.org/a/volumes/article_details/?paper_id=36231 - that article talks about web-based Aladin Lite, which is weird.

If this is the only software environment VASCO team used, it might be fairly easy to replicate their analysis after they release the Aladin scripts, or at least describe how the sky tessellation step was done.

I managed to build a software environment with all required software. Aladin, SExtractor, stilts-cdsskymatch (via TOPCAT) and DSS batch tools from ESO. With a lot of help from GTP-5, I also managed to get a smoke test for the actual pipeline to work. I did this on OSX, which is probably bit harder compared to some Linux OSs like Ubuntu and Debian where all the required software is available via package repositories.
Lots of parameter and config fine tuning with SExtractor and other components is needed. GPT-5 didn't get them right in one go and actually needed some guidance.
But the tessellation of sky in circular regions of 30-arcmin radius remains a mystery. GPT-5 like any LLM wants to try, because they want to please users, but I don't think the end-result would be fully reliable for reproducing all transient related findings.
Again: Villarroel et al. really should release all software and exact data sources. They want others to reproduce their work and work on new datasets, but I don't see how that's possible right now.

 

[Mick West]

I've downloaded all the POSS-1 red and blue plates from DSS. (1.1TB), but some of them seem be missing. I wonder if there's a list of these - I just want to make sure I have the full set.

Blue is missing 3,6,7,8,9,11,12,15,68,197,205,211,295,392,503
Red is missing 1,722,853,876

 

[HoaxEye]

I've downloaded all the POSS-1 red and blue plates from DSS. (1.1TB), but some of them seem be missing. I wonder if there's a list of these - I just want to make sure I have the full set.

Blue is missing 3,6,7,8,9,11,12,15,68,197,205,211,295,392,503
Red is missing 1,722,853,876

Those are missing also from the DSS file list: https://skyview.gsfc.nasa.gov/surveys/dss/dss/DSSListFiles/all.v30.list
Search for XOnnn for blue plates and XEnn for red plates. For missing red plate 1, there are almost matching "XE001A" and "XE001B" - not sure what it means.
The list may turn out to be useful: it contains RA/Dec coordinates for each plate and the observation date.

 

[Mick West]

I've improved my FITS viewer
https://www.metabunk.org/sitrec/tools/fits.html

Originally just a testbed for my FITS loading code, it's actually proving useful, as I've not found a reliable viewer for the Mac (this is cross platform, just being a web page).

When I have the time, I'll create .PNG files at various resolutions (containing all the metadata), and munge them into Sitrec in some interactive way that can visualize the shadow being cast on them over the course of their exposure.

 

[Mick West]

a fully open approach

Speaking of that, I recognize not everyone can do this:

The huge size of the datasets has always been a problem for replication here. So establishing a common repository of reasonably sized proxies (down to just the metadata) would be useful. I know we've been doing our own ad-hoc approaches.

Anyway, a task for another weekend.

 

[sigmops]

I managed to build a software environment with all required software.

Is that setup reproducible via a docker file? (Not sure why the authors of the paper don't provide that in the first place if they want to disclose the truth)

 

[HoaxEye]

@HoaxEye you mentioned here:



Is that setup reproducible via a docker file? (Not sure why the authors of the paper don't provide that in the first place if they want to disclose the truth)

Unfortunately the setup isn't reproducible via docker file at least yet. The first setup is a bash script that downloaded one 30x30 arcmin fits file and then attempted to perform analysis with sextractor, Stilts/TOPCAT against Pan-Starrs, USNO‑B1.0 and what not. I had a lot of problems with getting things to run correctly on OSX. Dockerizing that mess won't help anyone.
I'm now trying to create a better version using Python which would be more OSX agnostic. But I'm now facing a lot of new technical challenges e.g. I had to install Conda to get few required sw to work with Python. My time is limited, but I'll keep trying.
Yes, it would have been awesome if authors would have provided a Docker file or just chunk of code in github project. Doing everything from scratch using complex astro-software is not easy although I've been coding since late 1980s

 

[NorCal Dave]

I'm not arguing you're p-hacking. I'm pointing out that without pre-registration, any findings will be vulnerable to that critique, and legitimately so

Isn't that what this thread is? It's public. It's transparent. It's a real time public record of what various members are doing in an attempt to understand, and perhaps ultimately, replicate Villarroel's findings. They are trying to identify the exact dataset used, download it and then attempting to recreate the software to replicate the workflow that identifies the transients.

Open code shows what analysis was run, but not how many were tried first. If 10 approaches are tested and the one showing significance is reported, the code for that approach looks fine...but it's still p-hacking.

Is any of Villarroel's code open? I'm assuming not, thus the people here are trying to recreate it. Do we know how many times she and others ran various bits of code on the datasets until they arrived at the final conclusions in their papers? If they ran multiple test on the data until it showed what they were looking for, is that "still p-hacking"? I honestly don't know.

That's not a distraction - it's the difference between exploratory analysis (which is fine if framed honestly) and rigorous replication (which requires committing to methodology upfront).

Yes, but how does one "commit to methodology up front", the software workflow, if the study they are attempting to replicate does not share the software workflow used? It becomes "exploratory" by definition it would seem, until one can create, and make multiple runs of software workflow configurations that replicate Villarreol's claims. Then one can comment on why the workflow used is meaningful or not.

 

[Mick West]

I'd appreciate it if people keep this thread to the more technical aspects of the discussion.

 

[ThickTarget]

Hello, I'm new here. I'm an astronomer, and I was trying to investigate some of the analysis myself when I came across this forum. I'm still catching up on the discussion. But I thought I would share some info:

Those are missing also from the DSS file list: https://skyview.gsfc.nasa.gov/surveys/dss/dss/DSSListFiles/all.v30.list
Search for XOnnn for blue plates and XEnn for red plates. For missing red plate 1, there are almost matching "XE001A" and "XE001B" - not sure what it means.
Tlist may turn out to be useful: it contains RA/Dec coordinates for each plate and the observation date.

I found this other catalog of the POSS-I plates with dates, times, coordinates and other info. I was using it to convert between observation times by matching to the catalog of transients (with Topcat). The downside is that in the overlapping regions, it's not clear which plate the event comes from.

https://vizier.cds.unistra.fr/viz-bin/VizieR?-source=VI/25

But the tessellation of sky in circular regions of 30-arcmin radius remains a mystery.

It's not exactly clear, but it sounds like they are using Aladin to generate cutout images rather than using the originals. Aladin can show you the surveys as a fancy all sky picture, which it stores as in a format called HiPS. Which is a hierarchical tessellation of the sky. This form can give you examples of generating a cutout. But it gives square cutouts. So not exactly sure what they mean.

https://alasky.cds.unistra.fr/hips-image-services/hips2fits

Edit: I think was mistaken. They talk about using Aladin to tessellate the sky covered by POSS I. Which sounds to me like MOC, a something related to HiPS which makes tessellated maps of the coverage of some dataset. One could then use the coordinates of the pixels as the centers of the tessellation scheme. One of the resolutions has 28 arcmin resolution. It is a bit arcane to convert these back to RA Dec, but I think I figured it out. Maybe then they just queried the ESO system with these centers.

This step may be unnecessary, if you are downloading the whole dataset anyway. It just seems like they did this to chop up the data into manageable chunks for retrieval. If you plan to query it from one of the services like they did, then this would be the way to get those coordinates. I would go ahead and create such a file, but I haven't found a MOC of POSS I (DSS R) data. I could make one, but I would need all the images, or to do it with the catalog.

 

[ThickTarget]

Spatial distribution in galactic coordinates of the final list of 5399 candidates (blue dots). A coloured POSS-II images is displayed in the background.

If you plot the larger list of transients from their webpage (171,753), you can see just how patchy the catalog is. As you said, you can clearly see the plate edges. You can also see the huge gouge taken out by the Galactic Plane, because they remove anything near a Gaia star. This patchiness is not accounted for in their estimate of the statistical significance of the shadow. They assume Possionian statistics, which assumes it's random. I also suspect they failed to account of the effect of cross-matching in their injection tests.

Also as an aside you can see they have incorrectly plotted the survey in normal coordinates, and not Galactic.

Unfortunately this is not the catalog they used for that analysis, these objects were removed as they match to WISE objects (approximately).

 

[John J.]

As you said, you can clearly see the plate edges.

Might this be (at least in part) due to overlapping plates?