Open mFulton07 opened 2 years ago
thespacedoctor
commented
2 years ago The bright star reported is in the bottom left of this image (the transient location is the red circle).
@genghisken is there a way to view all ATLAS objects associated with classified TNS sources? It would be good to quantify these Sherlock-triggered misses to see how common they are. I could also run Sherlock directly against the coordinates of classified TNS sources and see what results I get.
smarttgit
commented
2 years ago Seems too far from the transient to be classified as a BS ? Maybe we should reduce that radius. Agree that a TNS match may tell us what we have missed, that's a good test Dave.
Also important is sherlock's text statement which doesn't seem to be coherent. A 16.3 mag galaxy but only in PS1 ? And this possible host seems too far from the transient to be associated.
Classified as BS, at 88.46 arcsec. Best crossmatch is galaxy The transient is possibly associated with 164242902141422599; an r=16.32 mag galaxy found in the PS1 catalogue. Its located 48.63" N, 73.89" W from the galaxy centre.
smarttgit
commented
2 years ago All seems strangely wrong, the marshall seems to point at a star, not a PS1 galaxy, and the blue square on Lasair does not have an object. https://www.pessto.org/marshall/transients/search?q=2022rhl Think this is an edge case that has mixed up information (perhaps bogus info in the catalogues) and the usual algorithm gets caught out ?
thespacedoctor
commented
2 years ago Agreed I may need to revisit the BS radius, but I think this is an edge case.
The BS to the bottom left is identified as a galaxy in PS1, but Sherlock treats all 'galaxies' above a given brightness as a BS ... and I think it was correct to do so in this case. Saturated stars mimic the morphology of galaxies and trip up the source classification machine learning algorithms.
Best crossmatch is galaxy
This is generated by the Lasair code, not Sherlock, and is misleading. It doesn't appear in the new pages ;)
smarttgit
commented
2 years ago Yes, so edge case, indeed bright objects which are only in the PS1 catalogue, and classified as galaxies would likely be spurious galaxy classifications.
The radius for BS is probably too large for this case, agree the the solution here is probably to revise the mag dependent BS radius
mFulton07
commented
2 years ago Although this galaxy is not known to just Pan-STARRS, it is also in 2MASS/2MASX according to NED and has a photometric redshift (photoZ~ 0.053211). Should the multiple catalogue entries not encourage the "galaxy" association?
genghisken
commented
2 years ago I can pull out any TNS crossmatches in the stars list (currently 48 million) - restricted to (currently 21 million) BS objects if necessary, but I might as well do them all. Might take a while!
thespacedoctor
commented
2 years ago Don't worry Ken, the direct download of coordinates from TNS will be quicker and I will be able to run the latest version of sherlock against them. The ATLAS Sherlock classifications could be from a few different versions.
This ATLAS object: https://star.pst.qub.ac.uk/sne/atlas4/candidate/1192044031465256100/ which is also in Lasair: https://lasair-ztf.lsst.ac.uk/object/ZTF22aasoali/ has been classified as a bright star by Sherlock.
This is not right. Should have been flagged as an Orphan. There are no bright sources nearby. There is a galaxy crossmatch also, but it is ~80 arcsecs away. The object has been classified as a Type IIn: https://www.wis-tns.org/object/2022rhl.