Galaxy Zoo Starburst Talk

Quench Sample - Fluxes?

  • JeanTate by JeanTate

    When I load the Quench Sample, and view the data as a Table, I see the following "Flux" fields (columns):

    1. Halpha
    2. Hbeta
    3. Nad Abs
    4. Nii
    5. Oii
    6. Oiii

    Presumably the first three are, respectively, H-alpha (6562Å), H-beta (4861Å), and the Na D blend (~5891Å). But what are the other three? I guess "Oii" is the [OII] blend (3726Å and 3728Å).

    Also, each of the six has an associated "Err" field, presumably "error". However, in every case, the values are the same as the corresponding "Flux" values! Which doesn't make sense.

    Posted

  • mlpeck by mlpeck

    The flux errors = flux values issue was apparently a mistake in coding an SQL query. I had a brief exchange about that with Laura Trouille here: http://quenchtalk.galaxyzoo.org/#/boards/BGS0000007/discussions/DGS00000vv.

    The flux ratios that are usually used in BPT diagrams are [N II]6584/H alpha and [O III]5007/H beta, so 4 and 6 probably are fluxes for those two lines. There's a good example of a BPT diagram (figure 4) in the "massive blue ETG" paper released on the arxiv today.

    Posted

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

    Thanks!

    I'd read that thread, but forgotten about it. And its title - Recommendations & Advice Coming Soon - didn't make me think, "hmm, I guess I should look in that thread, to find stuff about fluxes" 😉

    Now that I know how to do logs - earlier today I checked that I could do division - I can have a go at creating a BPT diagram! 😃

    Posted

  • mlpeck by mlpeck

    You're right, that probably wasn't the best place to put that exchange. I'm still trying to figure out the best location for any particular topic. I'm also trying to figure out why there's an area for "The Objects" under all the main headers.

    Posted

  • wassock by wassock moderator

    Why are the scales on the various plots seemingly so out of whack with the values? Halpha Vs redshift for example gives every point on the H alpha baseline.
    Is it possible to change the scales on the axis in the scatterplots?

    Posted

  • Nax by Nax in response to wassock's comment.

    What I've been trying to do to solve that problems is filter out the very high values of the spectral lines, to try and get a meaningful graph. But when I try to filter some of them out, ALL of the data vanishes. For instance I'll type in "filter .oiii < [number]", and regardless of that number, all the data will dissapear. Not sure if I'm doing something wrong, or it's a problem with the software.

    Posted

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

    Nax, I think the last part of your post went AWOL because you used one of these symbols: < >

    If so, here's a thread that might help: How to write the symbols "less than" and "greater than"?

    I too found that, sometimes, a filter would make all the data disappear; but only sometimes. I have no idea why ...

    Posted

  • wassock by wassock moderator

    Having no joy with the filter to remove the halpha outlyers, tried it always and nothing seems to happen to the data. Did find the stats for the halpha_flux as follows:

    Halpha Flux

    Mean: 6,449,248.338
    Median: 81.164
    Mode: 0.000
    Min: -438.234
    Max: 19,333,900,000.000
    Variance: 124,475,455,514,100,000.000
    StandardDeviation: 352,810,792.797
    Skew: 54.763
    Kurtosis: 2,999.992

    Now that's some seriously skewed data - anyone understand how the mode can be zero if the values for mean and median are correct?

    Posted

  • wassock by wassock moderator

    OK sorted out the filter thingy, stupidly I was following the instructions! - don't need the "New Filter" bit. Filtering for Halpha less than 2000 gives a sensible scale, but doesn't reduce the number of pages in the data set - must only be a couple at the top end

    Posted

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

    You may find it easier to work with the data offline (there's instructions on how to download somewhere).

    I think there may be only three, maybe four, Halpha_flux outliers, of the "too huge to be real" kind, In addition to the one Nax found, there's AGS000002i (DR7 ObjId 587722983363117377), AGS00001h9 (587738409797419159), and AGS000022s (587741726574444657).

    For the first two, the spectra look OK (so it's the data reduction which produced nonsense results); however, the last one has a 'bad spectrum':
    enter image description here

    I think know at least one of the Halpha_flux < 0 galaxies also has a wonky Halpha_flux value, AGS00002ak (758877274402460510):

    enter image description here

    Why? Because it's not even a galaxy! 😄

    Posted

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

    You beat me to it. There are 3 gross errors in H-alpha flux estimates, and one that could be physically plausible but is still hugely wrong, and JeanTate identified them all.

    It appears the flux estimates came from the table "galSpecLine", which is the "MPA-JHU spectroscopic re-analysis" according to the documentation.

    There's another set of estimates from the SDSS spectroscopic pipeline that are much more reasonable for these 4 objects at least. Those are the ones you can pull up by going to the Navigate tool (http://skyserver.sdss3.org/dr9/en/tools/chart/navi.asp), finding your object, clicking Explore and then clicking "Interactive spectrum".

    That data and more is in one of the HDU's in the FITS file containg the spectra (at least in DR9 or DR10).

    I still haven't figured out how to do basic operations with Zooniverse tools. Can data tables be edited? Is there a symbol for "NA" or "NaN"? If not, filtering on halpha_flux would work. You could also filter on log_mass (a value of -1 indicates missing or not computed).

    Posted

  • wassock by wassock moderator

    Thanks Jean, seem to have that sorted now. Sadly the corresponding data for the control set seems to be absent completely 😦

    Posted