Galaxy Zoo Starburst Talk

Major redshift measurement errors in the SDSS spectroscopic pipeline

  • mlpeck by mlpeck

    I'm not sure where to put this, so I picked an out of the way corner. This doesn't directly relate to the quench project, but it does relate to post-starburst galaxies so it's more or less on topic.

    The other day I decided to recreate the Goto (2007) catalog of 564 "E+A" galaxies using the object list in NED. I downloaded a selection of DR10 data for the sample from CasJobs and soon noticed some rather disturbing discrepancies in redshifts:

    enter image description here

    This picture shows the difference in measured redshift between the DR8+ measurement and Goto's adopted values, which I think came from DR5. There are a total of 29 discrepant redshifts, which is just over 5% of the sample. I checked all 7 with Δz > 1 and all are clearly erroneous in DR8+. Here is the first on the list, which happens to be a cataloged galaxy (I Zw 136), and a fairly interesting merging pair:

    enter image description here

    enter image description here

    The fiber was centered on the bright blue clump, which is probably the nucleus of the disrupted galaxy in the center and is definitely not a background QSO.

    Here is an example from the cluster of smaller, but still very significant discrepancies. I just picked the one with the smallest true redshift:

    enter image description here

    enter image description here

    Finally, the one discrepancy in the other direction is a little more ambiguous. The spectrum doesn't show up in Navigate, so here is a link to the interactive spectrum. In the Goto catalog the redshift is listed as 0.062. It was reclassified as an A star with redshift 0 in DR8 (or later). I don't see an obvious foreground star in the finder chart, but some of the apparent absorption lines aren't in the right position for a z=0.06 galaxy either:

    enter image description here

    Does anyone have thoughts about what to do about this? A 5% catastrophic error rate on an important and uncommon class of objects seems like something that ought to be addressed, but I - a lowly citizen scientist - wouldn't know who to contact or how to go about it.

    Posted

  • zutopian by zutopian

    This doesn't directly relate to the quench project, but it does relate to post-starburst galaxies so it's more or less on topic.

    Maybe it relates actually to GZQ. Many G05 sample galaxies are also in the QS sample. Does the QS sample contain erroneous G05 galaxies?

    Does anyone have thoughts about what to do about this? A 5% catastrophic error rate on an important and uncommon class of objects seems like something that ought to be addressed, but I - a lowly citizen scientist - wouldn't know who to contact or how to go about it.

    Here is an advice by me, an ordinary zooite.:

    I looked at "Contacting SDSS-III" and found the name of a well-known anstronomer.:
    Education and Public Outreach Director Jordan Raddick
    He is one of the authors of the Galaxy Zoo Post-Starburst paper by Wong et al.
    Maybe you don't need to contact him, if the GZQ scientists can answer your question. Or maybe they can inform him or suggest someone else, whom you should contact.

    Posted

  • zutopian by zutopian

    The fiber was centered on the bright blue clump, which is probably the nucleus of the disrupted galaxy in the center and is definitely not a background QSO.

    That GPair is one of the galaxy pairs, which are presented in below paper and one of the authors is Goto. So maybe you will find answer there.:

    E+A and Companion Galaxies - I : A Catalogue and Statistics

    (...)We selected 660 E+A galaxies with .... at a redshift of... 0.167 from the Data Release 5 of the Sloan Digital Sky Survey (SDSS). We selected their companion candidates from the SDSS imaging data, and classified them into true companions, fore/background galaxies and companion candidates using the SDSS and our KPNO spectra. We observed 26 companion candidates of E+A galaxies at the KPNO to measure their redshifts. Their spectra showed that 17 targets are true companion galaxies. The number of spectroscopically-confirmed E+A's companions are now 34. (...)

    Chisato Yamauchi (ISAS/Jaxa), Masafumi Yagi (NAOJ), Tomotsugu Goto (Ifa/Uh)

    (Submitted on 4 Sep 2008)

    http://arxiv.org/abs/0809.0890

    Posted

  • mlpeck by mlpeck

    Thanks for the link. The example I posted was number 27 on their list of interacting pairs. They also did a quite extensive study of I Zw 136: http://adsabs.harvard.edu/cgi-bin/bib_query?2008MNRAS.391..700G

    Posted

  • zutopian by zutopian

    I did following post in the "Sample Selection" topic today.:

    E+A and Companion Galaxies - I : A Catalogue and Statistics

    In this paper, we use publicly available true E+A galaxies (without Halpha nor [OII] emission) selected from the Sloan Digital Sky Survey (...) hereafter DR5) by Goto (2007b)

    Chisato Yamauchi (ISAS/Jaxa), Masafumi Yagi (NAOJ), Tomotsugu Goto (Ifa/Uh)
    (Submitted on 4 Sep 2008)
    http://arxiv.org/abs/0809.0890

    I think, that there is Halpha and OII in some cases in DR8. e.g.: pair ID 16:
    http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237658917995610179
    They took another spectrum at KPNO.: KPNO Observation #8

    Posted

  • JeanTate by JeanTate in response to mlpeck's comment.

    Does anyone have thoughts about what to do about this?

    Are any of the Goto et al. authors zooites? Quench SCIENTISTs? If so, then send them PM/Messages. If not, then send Goto an email (you can get the 'official' email addy from the arXiv preprint, if not elsewhere).

    zutopian: Many G05 sample galaxies are also in the QS sample. Does the QS sample
    contain erroneous G05 galaxies?

    That should be fairly straight-forward to check: compare the list of 'galaxies with erroneous redshifts' in Goto with the QS objects. There are several ways to do this, with fairly minimal effort; would you care to try yourself? Yes, mlpeck can certainly do this; however, as should be screamingly obvious by now, I have been using the Quench project in part as a means by which ordinary zooites can learn things, such as data analysis and 'getting answers efficiently and effectively'. 😉

    Posted

  • zutopian by zutopian in response to mlpeck's comment.

    Curiously, that GPair isn't in the Wong et al sample, though it is within the redshift range specified in the Wong et al paper.:
    Well, the DR8 redshift is wrong (z=1.35, zERR=0.00025, zWarning=Negative_Emission), but Wong et al had used DR7.
    I guess, that it was z=0.034.: NED says 0.034, but DR7 is however currently not available.
    However, that GPair is also missing in the QS sample.

    Here is the DR8 link.: http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237665357237518394

    Posted

  • zutopian by zutopian

    Here is a statement from "What's New in DR8?" :

    1D Analysis: This step produces redshifts and classifications for each spectrum.
    For all data, we rely on the idlspec2d pipeline for redshifts (also known as "specBS").
    This pipeline differs in detail wth the previous pipeline used in DR7, though for virtually all objects (> 99%) they agree.

    http://www.sdss3.org/dr8/whatsnew.php

    Posted

  • mlpeck by mlpeck

    Chisato Yamauchi (ISAS/Jaxa), Masafumi Yagi (NAOJ), Tomotsugu Goto (Ifa/Uh)
    (Submitted on 4 Sep 2008)
    http://arxiv.org/abs/0809.0890

    For some reason there are 74 NED objects associated with that paper even though their main data table lists 68 galaxies -- 34 E+A galaxies with one companion for each.

    By my count 8 of those are in the quench sample. 4 of the eight were classified as "merging" by us, the remaining 4 "neither".

    Are any of the Goto et al. authors zooites? Quench SCIENTISTs? If so, then send them PM/Messages. If not, then send Goto an email (you can get the 'official' email addy from the arXiv preprint, if not elsewhere).

    Goto seems to have hopped around a bit since 2007-2008, but I'm sure he can be tracked down. It's not really his problem though unless he's on the SDSS team at present. This is an SDSS spectro pipeline problem that happened to show up in a sample of Goto's E+A galaxies.

    Many G05 sample galaxies are also in the QS sample. Does the QS sample contain erroneous G05 galaxies?

    I'll answer anyway even though JeanTate advises to find out for yourself. There are no problematic redshifts in the quench sample. Or there might be one, but for a completely different reason in an object that should have been discarded from the sample long ago.

    Posted

  • zutopian by zutopian in response to mlpeck's comment.

    For some reason there are 74 NED objects associated with that paper even though their main data table lists 68 galaxies -- 34 E+A galaxies with one companion for each.
    By my count 8 of those are in the quench sample. 4 of the eight were classified as "merging" by us, the remaining 4 "neither".

    There are following other tables in the paper.:
    Table 5. List of backup targets with successful observation with the KPNO 2.1-m telescope.
    Table 6. List of fore/background galaxies with successful observation with the KPNO 2.1-m telescope.

    Maybe NED lists those galaxies also?

    Yesterday, I informed in below topic, that I noticed a merger, which was missed by GZQ, is presented in the "E+A with companions" paper.
    Could you please post the other IDs?
    http://quenchtalk.galaxyzoo.org/#/boards/BGS0000008/discussions/DGS0000219?page=2&comment_id=5260402672c1092ddf0000d6

    Posted

  • mlpeck by mlpeck

    Could you please post the other IDs?

    Sorry for the cut & paste, but here they are:

    [1] "AGS000006m" "AGS00000yv" "AGS00000s9" "AGS0000172" "AGS00001oq"

    [6] "AGS00000cq" "AGS000004b" "AGS00000u1"

    6 of these were from the main data set, 1 from the "backup" faint set, and 1 had a foreground neighbor.

    Posted

  • zutopian by zutopian in response to mlpeck's comment.

    There are no problematic redshifts in the quench sample.

    That's surprising.: In the "Sample Selection" topic you had informed as follows.:

    Out of 564 in the Goto (2007) catalog I count 290 that are in the quench sample, based on position matches. There's also one in the control sample.

    I had guessed, that probably among those 290 galaxies there would be some of the 29 galaxies with discrepant redshifts. So they are just among the 274 galaxies, which aren't in the QS sample.

    I copied following comment, which you did in the "Sample Selection" topic.:

    One thing I discovered is that there are some massive errors in the DR8+ spectroscopic redshift estimates. 2 of the 80 in the Wong sample were affected and 29 of the Goto sample. The spectroscopic pipeline seems vulnerable to huge errors in objects with strong Balmer and H+K absorption.

    Posted

  • zutopian by zutopian

    Below there is a further paper, in which Goto is listed as an author. The paper is from 2011/2012. They used DR7.:

    The current star formation rate of K+A galaxies

    We derive the stacked 1.4 GHz flux from FIRST (Faint Images of the Radio Sky at Twenty Centimeters) survey for 811 K+A galaxies selected from the SDSS DR7.

    Danielle Nielsen (Wisconsin), Roberto De Propris (CTIO), Susan E. Ridgway (CTIO), Tomotsugu Goto (IfA, Hawaii)
    (Submitted on 3 Jun 2011 (v1), last revised 22 Oct 2012 (this version, v2))
    http://arxiv.org/abs/1106.0757

    Citations to the Article:
    2013MNRAS.431.2034M by Melnick, J.; De Propris, R.
    The spectral energy distributions of K+A galaxies from the UV to the mid-IR: stellar populations, star formation and hot dust

    They used the sample from the other paper. De Propris, R. is listed as an author in both papers.

    Posted

  • zutopian by zutopian

    The Fundamental Planes of E+A galaxies and GALEX UV-excess early-type galaxies: Revealing their intimate connection

    In particular, we examine the FP scaling relations of the largest sample of ~1,000 E+As selected from the SDSS and ~20,000 morphologically-selected SDSS ETGs with GALEX UV data.

    Yumi Choi (1), Tomotsugu Goto (2,3), Suk-Jin Yoon (1) ((1) Yonsei University, (2) ISAS/JAXA, (3) University of Hawaii)

    (Submitted on 7 Apr 2009)

    http://arxiv.org/abs/0904.1209

    Posted

  • zutopian by zutopian in response to zutopian's comment.

    My comment:

    I had guessed, that probably among those 290 galaxies there would be some of the 29 galaxies with discrepant redshifts. So they are just among the 274 galaxies, which aren't in the QS sample.

    Why aren't any of those 29 in the QS sample? Just because, that they didn't meet the QS selection criteria (concerning spectral lines) like many of the G05 galaxies?
    Did GZQ use DR8 or DR7? Is that relevant? Are "just" the redshifts of those 29 galaxies different ("DR7 versus DR8") or are also the spectral lines different? (DR7 is currently not available) Has GZQ a redshift selection range?

    Posted

  • mlpeck by mlpeck in response to zutopian's comment.

    Why aren't any of those 29 in the QS sample?

    Did GZQ use DR8 or DR7? Is that relevant?

    Yes, it's relevant because the redshifts were correct in DR7, I think. On the other hand the Chen et al. population models didn't make their way into the SDSS database until DR10, so it's possible the models that were the basis for the sample selection were run on a later data release than DR7. If that were the case none of the 29 would likely have met the selection criteria since the assumed restframe wavelengths would be wrong.

    On the third hand it could just be an accident that none of them appeared in the quench sample. AFAIK the quench sample does not purport to be a complete census of "post-quenched" galaxies.

    Of course I am just speculating. The exact selection process remains an impenetrable mystery.

    Posted

  • zutopian by zutopian in response to mlpeck's comment.

    Thanks for your reply! So this topic might be related to the QS sample. Since the GZQ scientists however "withheld" some information from the volunteers, you can't know it. I think, that it was actually interesting to speculate/discuss about the unknown selection process here and also in the "Sample selection topic", but now it is time, that they inform the volunteers.

    PS: Besides I think, that it is actually now time for me to quit this project.: I started just one month ago and it was really interesting, but I have neither the knowledge nor the motivation to continue.

    Posted

  • zutopian by zutopian in response to zutopian's comment.

    Finally, there is more information about the selection process available in the "sample selection" topic. As far as I understood, they had used just DR7 for the PCA. If so, this topic isn't relevant for the QS sample. On the other hand, I wonder, what the actual reason is, that the GPair Zw 136 wasn't selected for the QS sample?

    Posted

  • zutopian by zutopian

    I found another case.: Curiously, it has no "PSG paper" as Ref and it isn't in the QS sample, though the spectrum looks like a PSG.
    So I wonder, why it hadn't been selected? It is also available in DR5, but Goto however had rejected it.
    Concerning the redshift discrepancy.: DR7 says z=0.07, DR8 says z=1.431 (QSO)
    I speculate, that the 29 from the Goto sample, the 2 from the Wong et al. sample and the below one, might be scientifically interesting, because they are special outliers.: They have similiar spectra and also redshift discrepancies "DR7 vs DR8".(redshifts in DR8 wrong)

    enter image description here
    http://cas.sdss.org/dr7/en/tools/explore/obj.asp?id=587732134844366853
    http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237657856069271557

    Posted

  • mlpeck by mlpeck in response to zutopian's comment.

    Nice catch. How did you find it? Zooming out a little this nearby ringed galaxy is at the same redshift (z=0.069). It sure looks like they're interacting.

    enter image description here

    Posted

  • zutopian by zutopian in response to mlpeck's comment.

    I had done a query in the SDSS Image List Tool.: Select "QSO Broadline".
    While checking some of the 1000 displayed objects, I found that galaxy.
    I wonder, how many other cases there are? Maybe you or a scientist could find out.

    Posted

  • zutopian by zutopian

    GZ astronomer Edd did following posts in the corresponding forum topic today.:

    I would recommend emailing the SDSS 3 help desk. We regularly get issues from there passed on to our internal mailing lists for discussion if it does turn out to be an issue. If not you should I think get a reply about the nature of the difference anyway (I can't look myself right now all that easily).
    Data releases are continuing for SDSS 3 and the spectroscopic pipeline can and is changed between releases.

    and

    On a general point - do make sure to pay attention to the ZWARNING flag. It is typical to throw out anything where this value is not zero and it may not be considered a priority to try to get a small number of objects with ZWARNING set to have it unset with a correct redshift down the line as it may be a disproportionate investment of effort.

    http://www.galaxyzooforum.org/index.php?topic=281438.msg644669#msg644669

    Posted

  • zutopian by zutopian in response to zutopian's comment.

    Here is the zWarning information on the SDSS III webpage.:

    http://www.sdss3.org/dr8/algorithms/bitmask_zwarning.php

    Posted

  • zutopian by zutopian

    I found 2 further cases.:

    enter image description here
    http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237661949707223150

    enter image description here
    http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237668348143796408

    Posted

  • zutopian by zutopian

    enter image description here
    http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237667912753020966

    Posted