Galaxy Zoo Talk

Request for help: Galaxy Zoo literature search.

  • klmasters by klmasters scientist, admin

    Dear volunteers,
    Here at Galaxy Zoo we know that a fraction of you are looking for ways to be more involved in the entire process of making science from your clicks.

    So we had an idea.....

    The team are currently in the process of writing a paper which in its introduction discusses some of the current assumptions/errors/lazy approximations common among our fellow astronomers when thinking about galaxy morphology and classification. As such we’d like to collect as many papers as possible which do the following things:

    1. Claim that colour and morphology are equivalent
    2. Defines “early-type” galaxy as any galaxy without visible spiral arms (e.g. our “smooth” category, which can include elliptical galaxies, and smooth disks), rather than as a galaxy that isn't a disk.
    3. Defines “early-type” galaxy as including Sa spirals as well as lenticular and ellipticals.
    4. Define “late-type” galaxy as only late-type spirals (e.g. excluding Sa spirals)
    5. Use colour or spectral type to split galaxies into “early-“ or “late-“ types (or “elliptical” and “spiral”)
    6. Use the bulge-to-total ratio (or some proxy for it like concentration, or the SDSS “fracDeV” parameter) to place spiral galaxies in a sequence.

    The current draft text in the paper which talks about these assumptions is:

    “The morphology of a galaxy encodes information about its formation history and evolution through what it reveals about the orbits of the stars in the galaxy, and is known to correlate remarkably well with other physical properties (e.g. Roberts & Haynes 1994). These correlations, along with the ease of automated measurement of colour or spectral type, have resulted in a recent trend for classification on the basis of these properties rather than morphology per se (e.g. Weinmann et al. 2006, van den Bosch et al. 2008, Zehavi et al. 2011). Indeed the strength of the correlation has led some to authors to claim that the correspondance between colour and morphology is so good that that classification by colour alone can be used to replace morphology (e.g. Park & Choi 2005, Faber et al. 2007). Meanwhile the size of modern data sets (e.g. the Main Galaxy Sample of the Sloan Digital Sky Survey, SDSS, Strauss et al. 2002) made the traditional techniques of morphological classification by small numbers of experts implausible. This problem was solved making use of the technique of crowdsourcing by the Galaxy Zoo project (Lintott et al. 2008, 2011). One of the first results from the Galaxy Zoo morphological classifications was to demonstrate on a firm statistical basis that colour and morphology are not equivalent for all galaxies (Bamford et al. 2009, Schawinski et al. 2009, Masters et al. 2010) and that morphology provides complementary information on galaxy populations useful to understand the processes of galaxy evolution. “

    and later when talking about the spiral sequence:

    “Modern automatic galaxy classification has tended to conflate bulge size alone with spiral type (e.g. Laurikainen et al. 2007, Masters et al. 2010a), and automatic classification of galaxies into “early-” and “late-” types, referring to their location on the Hubble Sequence and based on bulge-total luminosity ratio (B/T ) or some proxy for this through a measure of central concentration, or light profile shape (e.g. Sersic index, as reviewed by Graham & Driver 2005) has become common (e.g. van der Wel et al. 2011). Indeed, Sandage (2005) says this is not new, claiming ”the Hubble system for disk galaxies had its roots in an arrange- ment of spirals in a continuous sequence of decreasing bulge size and increasing presence of condensations over the face of the image that had been devised by Reynolds in 1920.””

    We’d like to ask for your help in searching for more examples of these behaviour. We have made a simple Google form, and we ask that you submit any examples you find in the next few weeks.

    Some of the papers you find might end up cited in the Galaxy Zoo team paper (please be aware there are rules/guidelines about the appropriate number - we don’t want to have too few; it doesn’t make the point about how widespread this is, and we don’t want to single out specific astronomers, but the journal won’t accept too many either). If there are more papers found than we can use, they will be kept in a list on the Galaxy Zoo website (and we can continue to add to them if needed).

    I want to reassure you that helping with this does not mean you have to read the entire extragalactic astronomy literature, or even the entirety of a paper! The best place to look for this information in a paper will be the “Sample Selection”, or “Data” sections. Modern online PDF papers also have excellent search facilities - so searching the text for key words (e.g. “spiral”, “early-type”, “colour/color-selected”) may work extremely well.

    We’re happy for you to do this however you like (e.g. Google Scholar is fine), but we’d like you to return the NASA ADS (Astrophysics Data System) URL for the paper you find. You can search ADS here: http://adsabs.harvard.edu/abstract_service.html, and I give examples below of the URL I mean. This makes it easy for us to get the full bibliographic data to add the reference to the paper.

    One tip - there are some papers in extragalactic astronomy which are cited my most/many results. A good place to start to look through recent papers would be the citation and reference lists of such papers, which can be found in ADS.

    For example:

    Strateva et al. 2001 “Color Separation of Galaxy Types in the Sloan Digital Sky Survey Imaging Data”
    http://adsabs.harvard.edu/abs/2001AJ....122.1861S is cited by 926 papers, and references 25 - this would be an excellent starting place, and the more papers you read the more mentions you may find other other papers doing similar things.

    Other good starting places:
    Strauss et al. 2002: “Spectroscopic Target Selection in the Sloan Digital Sky Survey: The Main Galaxy Sample” http://adsabs.harvard.edu/abs/2002AJ....124.1810S

    Ironically the papers which cite some of our Galaxy Zoo papers where we demonstrate there are galaxies which are not in the normal correlation between colour and morphology may also be good starting points (some citations to these are along the lines of saying things like: “most galaxies fall into blue=spiral; red=elliptical, a few don’t (cite Galaxy Zoo here), but we’re going to use this definition anyway”.

    The initial papers on colour not being the same as morphology are:

    Bamford et al. 2009 (281 citations): http://adsabs.harvard.edu/abs/2009MNRAS.393.1324B
    Schawinski et al. 2009 (81 citations): http://adsabs.harvard.edu/abs/2009MNRAS.396..818S
    Masters et al. 2010 (125 citations): http://adsabs.harvard.edu/abs/2010MNRAS.405..783M

    We hope that if several of you take up the challenge, you’ll find different paths through the literature and find lots of different examples for us. Again here's the link to our Galaxy Zoo Literature Survey.

    Karen Masters (Galaxy Zoo Project Scientist)

    (PS. I also posted this to the blog).

    Posted

  • klmasters by klmasters scientist, admin

    Link to blog post: https://blog.galaxyzoo.org/2017/09/28/galaxy-zoo-literature-search/

    Link to form to submit paper links: https://goo.gl/forms/2YyNtFLtU0x7NeRH3

    Posted

  • klmasters by klmasters scientist, admin

    Just added a question so people can indicate their name and email (optional). For those who've filled it in, if you want to resubmit with those that's fine, but it's not necessary. 😃

    Posted

  • JeanTate by JeanTate

    I've just submitted one (excellent) example, "Revisiting the Hubble sequence in the SDSS DR7 spectroscopic sample: a publicly available Bayesian automated classification", Huertas-Company+ 2011. Of course, this is well-known to the GZ team, and astronomers generally; it's discussed in the main GZ2 paper (Willett+ 2013).

    This paper to some extent conflates color and morphology, and uses proxies for B/T to stage spirals; it also, by omission, does not use the presence (or absence) of spiral arms to distinguish ellipticals from S0s from spirals (an elliptical with spiral arms could have a high P(E), and a spiral with no spiral arms could have a high P(Scd)).

    Here's how I filled in the form:

    Claims colour and morphology are equivalent (Please provide quote)

    Three types of parameters used to estimate morphology probabilities; "we decided to include three types of parameters: (1) color (g − r, r − i) k-corrected with Blanton et al. (2005) code; (2) shape (isoB/isoA in the i-band and deVAB_i); and (3) light concentration (R90/R50 in the i-band). For color measurements
    we use modelmagnitudes corrected for galactic extinction. isoB and isoA are the isophotal minor and major axes respectively, and deVAB_i is the DeVaucouleurs fit b/a. R90 and R50 are the radii containing 90% and 50% of the petrosian flux, respectively    "

    Defines “early-type” galaxy as any galaxy without visible spiral arms (e.g. our “smooth” category, which can include elliptical galaxies, and smooth disks), rather than as a galaxy that isn't a disk. (Please provide quote)

    By omission: the criteria for any type - elliptical, S0, spiral - do not include the presence (or absence) of spiral arms

    Defines “early-type” galaxy as including Sa spirals as well as lenticular and ellipticals. (Please provide quote)

    N/A

    Define “late-type” galaxy as only late-type spirals (e.g. excluding Sa spirals)

    N/A

    Use colour or spectral type to split galaxies into “early-“ or “late-“ types (or “elliptical” and “spiral”)

    See above

    Use the bulge-to-total ratio (or some proxy for it like concentration, or the SDSS “fracDeV” parameter) to place spiral galaxies in a sequence.

    See above

    I'd like some feedback please; is this the kind of response you're looking for?

    Posted

  • JeanTate by JeanTate

    There's one question in the survey I'm particularly curious about:

    Use the bulge-to-total ratio (or some proxy for it like concentration, or the SDSS “fracDeV” parameter) to place spiral galaxies in a sequence.

    Why?

    Because, as I understand it, this is precisely one of the core features used to stage spirals (from S0/a to Sm)! 😃

    For example, Buta+ 2015 (he's using the Comprehensive de Vaucouleurs revised Hubble-Sandage” (CVRHS) system, which is "a modified version of the de Vaucouleurs (1959) revised Hubble-Sandage (VRHS) system that is described in the de Vaucouleurs Atlas of Galaxies"):

    The positioning of galaxies in stage depends on specific morphological characteristics: [...] spirals form a sequence (S0/a-Sa-Sab-Sb-Sbc-Sc-Scd-Sd-Sdm-Sm) of decreasing bulge-to-total luminosity ratio, increasingly open spiral arms, an increasing degree of star formation, and increasing asymmetry

    He also writes, a bit earlier (my bold):

    Because stage correlates with several basic physical properties of galaxies (e.g., average surface brightness, color, HI mass-to-blue light ratio), it is considered the fundamental dimension of the system.

    To what extent is this literature survey looking (perhaps not deliberately) to provide a critique of the criteria used to classify galaxies according to one version or another of the Hubble tuning fork? In other words, should I write up Buta+ 2015 in the Google form?

    Posted

  • Ghost_Sheep_SWR by Ghost_Sheep_SWR in response to klmasters's comment.

    First off I think there should be a cut-off at 2009 when the first GZ galaxy colour-morphology papers started appearing?

    Secondly, for my reference could you please provide some advice if these two example papers are what is looked for (should I submit?)?;

    Relationship between Hubble Type and Spectroscopic Class in Local Galaxies Almeida+ 2011

    From abstract:

    We compare the Hubble type and the spectroscopic class of the galaxies
    with spectra in the Sloan Digital Sky Survey Data Release 7. As has
    long been known, elliptical galaxies tend to be red whereas spiral
    galaxies tend to be blue; however, this relationship presents a large
    scatter, which we measure and quantify in detail for the first time.

    but also

    A considerable fraction of the red galaxies are spirals (40--60 %),
    but they never present very late Hubble types (Sd or later). Even
    though red spectra are not associated with ellipticals, most
    ellipticals do have red spectra: 97 % of the ellipticals in the
    morphological catalog by Nair & Abraham, used here for reference,
    belong to ASK 0, 2 or 3. It contains only a 3 % of blue ellipticals.
    The galaxies in the green valley class (ASK~5) are mostly spirals,

    https://arxiv.org/abs/1104.4864

    A catalogue of 2D photometric decompositions in the SDSS-DR7 spectroscopic main galaxy sample: preferred models and systematics Meert+ 2015

    From paper (my bold):

    2.1 SDSS CasJobs data

    The data used in this analysis were drawn from the spectroscopic
    sample of the Legacy area of the Sloan Digital Sky Survey Data Release
    7 (DR7; Abazajian et al. 2009). The spectroscopic sample provides a
    well-established sample with well-defined and tested selection
    criteria. The criteria are presented in Strauss et al. (2002)

    https://arxiv.org/abs/1406.4179

    PS. it seems that NASA ADS isn't working properly for me atm, so that makes searches really difficult. Am I the only one?

    Posted

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

    it seems that NASA ADS isn't working properly for me atm, so that makes searches really difficult. Am I the only one?

    No, you're not. I am having the same difficulty. 😦

    The two papers you mention - Almeida+ 2011, and Meert+ 2015 - are among those I've been considering for inclusion too! 😮 So your question is very timely (and pertinent), I feel.

    Posted

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

    Ok thanks for the feedback!, I'll postphone searching then because it's really unworkable right now 😕

    Secondly, good of you to mention considering those two papers, makes me think it's really necessary to have a publicly available list of submissions. Would be a waste of time and effort if multiple people are researching papers they are considering for submission when those papers might have been submitted multiple times already.

    Posted

  • JeanTate by JeanTate

    Tal&van Dokkum (2011) "THE FAINT STELLAR HALOS OF MASSIVE RED GALAXIES FROM STACKS OF MORE THAN 42000 SDSS LRG IMAGES" may be an example of what is being sought, or it may be irrelevant; some guidance please!

    Here's how I filled out the form for this paper (ADS seems to be working OK again):

    Claims colour and morphology are equivalent (Please provide quote)

    Throughout this paper, the authors conflate "massive red" with "elliptical". This formally begins with the selection criteria ("We selected galaxy images for this study from the Sloan Digital Sky Survey (SDSS, Abazajian et al. 2009) including all objects classified as Luminous Red Galaxies (LRG) that have a spectroscopic redshift measurement. LRGs are intrinsically red and luminous objects that were identified as such from their central surface brightness and location on a rotated color-color diagram (for full details see Eisenstein et al. 2001). This selection is aimed at finding the most luminous red galaxies in the nearby Universe (L  3L⋆) out to a redshift of z = 0.5.")") which does not distinguish between red ellipticals and red spirals (nor red S0s). However, the first sentence in the Discussion section strongly implies that massive red S0s, massive blue ellipticals, and massive red spirals are irrelevant ("The first and foremost result that arises from this study is that faint, gravitationally bound stellar light can be traced in massive elliptical galaxies out to a radius of 100 kpc."), to take just one example.

    Defines “early-type” galaxy as any galaxy without visible spiral arms (e.g. our “smooth” category, which can include elliptical galaxies, and smooth disks), rather than as a galaxy that isn't a disk. (Please provide quote)

    N/A

    Defines “early-type” galaxy as including Sa spirals as well as lenticular and ellipticals. (Please provide quote)

    N/A

    Define “late-type” galaxy as only late-type spirals (e.g. excluding Sa spirals)

    N/A

    Use colour or spectral type to split galaxies into “early-“ or “late-“ types (or “elliptical” and “spiral”)

    Not really applicable, but see above

    Use the bulge-to-total ratio (or some proxy for it like concentration, or the SDSS “fracDeV” parameter) to place spiral galaxies in a sequence.

    N/A

    On the one hand, it seems as if the authors carefully avoided any of the claims or definitions this literature survey is looking for; however, the lack of any explicit mention of S0s, red spirals, or blue ellipticals, and the frequent conflation of LRGs with "ellipticals", surely makes this paper just the kind this survey is looking for, right?

    One concern I have is that this kind of omission and conflation is surely to be found in almost all observational papers on galaxies (other than our own), and maybe a large percentage of theoretical papers too. Perhaps it might be simpler to find examples of papers in which none of the claims/definitions are made/used, even implicitly?

    Posted

  • klmasters by klmasters scientist, admin in response to JeanTate's comment.

    Hi Jean. Thanks for all of this. You are very astute in picking up the theme of the paper from the questions! It's not so much a critque of the spiral sequence on the Hubble tuning fork, but a look at how Galaxy Zoo morphological classifications fit into it.

    I do have Buta+2015 cited elsewhere already - an excellent review of course. 😃

    Posted

  • klmasters by klmasters scientist, admin in response to Ghost_Sheep_SWR's comment.

    Hi Ghost_Sheep - I actually don't think either of those are what we're looking for. The first actually is looking into how good the correlation is - and commenting it's not perfect. The latter uses a selection, but from the bit you quote, not to make assumptions about morphology.

    Posted

  • klmasters by klmasters scientist, admin in response to Ghost_Sheep_SWR's comment.

    Oh and I'm fine with post 2009 - in fact that's almost worse for many of these assumptions, as the Galaxy Zoo team have been working hard to point out they don't always work! 😉

    Posted

  • klmasters by klmasters scientist, admin in response to Ghost_Sheep_SWR's comment.

    I'll look into getting good examples up on a public list as soon as is practical. I'm on travel this week for a conference (Gas in Galaxies in Malta - lucky me!) so may not be until next week.

    There are a couple of ADS mirrors I think - so if one isn't working try the others? It's usually pretty reliable, but all websites have their moments!

    I think LRGs are the perfect example of the assumption that red=elliptical and not thinking about it further. I see it done so much in that type of extragalactic paper! Great to have several concrete examples of that. 😃

    Posted

  • Ghost_Sheep_SWR by Ghost_Sheep_SWR in response to klmasters's comment.

    Ok thanks for the feedback, I'll await the the public list / example papers since it is clear I am not getting it right. Not familiar with the mirrors but I should be able to figure it out.

    Have fun in Malta 😃

    Alexander

    Posted

  • klmasters by klmasters scientist, admin

    I came across two potential candidates today in the conference I'm at. If you want to look into these a bit more that would be awesome:

    Romanowsky & Fall (2012) - morphology = B/T

    Wang et al. (2008) - red/blue split

    Working on the public list of entries to date for next week.

    Posted

  • JeanTate by JeanTate

    One thing which would be of great help (to me, at least) would be the ADS links for the papers you mention, klmasters, either in the GZ blog post or in this thread. While some do have links - e.g. Strateva et al. 2001 - most do not. Here's a list that I compiled:

    • Roberts & Haynes 1994
    • Weinmann et al. 2006
    • van den Bosch et al. 2008
    • Zehavi et al. 2011
    • Park & Choi 2005
    • Faber et al. 2007
    • Lintott et al. 2008, 2011 (both easy enough to find, obviously!)
    • Bamford et al. 2009 - is this http://adsabs.harvard.edu/abs/2009MNRAS.393.1324B?
    • Schawinski et al. 2009 - is this http://adsabs.harvard.edu/abs/2009MNRAS.396..818S?
    • Masters et al. 2010 } these two illustrate why links are needed
    • Masters et al. 2010a } obviously two different papers, but which is which?
    • Laurikainen et al. 2007
    • Graham & Driver 2005
    • Wel et al. 2011
    • Sandage (2005)
    • Romanowsky & Fall (2012)
    • Wang et al. (2008) - ADS returns 1,825 hits for author "Wang" in 2008! 😮

    One thing I was confused by (and I doubt that I'm alone) is that you seem to be after papers which make a one-to-one match between a single feature - color, say, or B/T ratio - and morphology, rather than ones which use two or more characteristics - e.g. color AND B/T ratio - to map to morphology.

    So, in Buta+ (2015), one reads:

    The positioning of galaxies in stage depends on specific morphological characteristics: [...] spirals form a sequence (S0/a-Sa-Sab-Sb-Sbc-Sc-Scd-Sd-Sdm-Sm) of decreasing bulge-to-total luminosity ratio, increasingly open spiral arms, an increasing degree of star formation, and increasing asymmetry

    Thus it's clear there are four, independent, aspects which go into deciding a morphological type for spirals (stage, in this case). And as four is greater than one, this paper is not what you're after.

    Huertas-Company+ (2011), a paper I've filled in a form on, is a bit different: while it does use three independent things in its determination of "morphology" (color, shape, and light concentration), and it explains why no more than these three are needed - so it wouldn't be the kind of paper you are looking for anyway - it produces what it calls probabilities, that a galaxy will belong to one of four morphological classes (E, S0, Sab, Scd). So, even if it had used just a single characteristic, it would not be a paper worthy of consideration, because it explicitly does not do one-to-one matching (how HC11's "probabilities" relate to what you'd find in a modern, Bayesian, statistics textbook is an entirely different discussion).

    I'd appreciate some feedback please; how close to the mark am I?

    Posted

  • klmasters by klmasters scientist, admin

    Hi all,
    I made a summary of results so far here: https://blog.galaxyzoo.org/2017/10/28/galaxy-zoo-literature-search-initial-results/

    Posted

  • klmasters by klmasters scientist, admin

    Jean - to get those links, all I would do is go here: http://adsabs.harvard.edu/abstract_service.html put ^author in the author box (e.g. ^Lintott - this just searches on papers with Linott as the first author), and the year in the final publication date box (this puts ones from that year at the top of the search). Would be super helpful if you could do that. 😃

    I am indeed looking for examples where astronomers are being overly simplistic in their morphological definitions. You'll hear no arguments from me that it's more complex than that! You are completely right that Huertas-Company et al. 2011 recognises that it's more complicated, so it's not quite what I'm looking for.

    Posted