Radio Galaxy Zoo Talk

What does "radio loud" mean?

  • JeanTate by JeanTate

    From the Bagchi+ (2014) preprint (this Journal Club thread discusses it), I learned that extragalactic radio sources come in basically two flavors, "radio-loud" and "radio-quiet"; i.e. the distribution of intrinsic luminosity, in "the radio", is bimodal.

    This is completely unlike the distribution of intrinsic luminosity of AGNs in the optical, for example; that's a cline with the distinction between "QSO" and "Seyfert" (for example) a purely arbitrary one, a legacy of a half-century or so of optical observations*.

    What is the 'radio green valley', the range of intrinsic luminosity of extragalactic sources which is a near desert? And what strange creatures live there, relics and ghosts perhaps? Lobes and jets that are fading, but because they are so huge take a very long time go do so. I guess there are no AGNs in this near-wasteland; per Bagchi+, AGNs in spirals are never loud, and AGNs in ellipticals are either on or off (radio-wise). Being so small, the time taken to go from 'on' to 'off' (or vice versa) must be incredibly short, so we'd never see any in between, even among 10 million AGNs.

    Of course, within "loud" (and "quiet"), there must be a considerable range of intrinsic luminosity; are there any good, empirical, analytic descriptions of these distributions? For example, do they closely resemble Gaussians?

    And is this dichotomy true across the whole radio spectrum, from the long wavelengths Grote Reber was fond of in his later years, to the mm wavelengths of the Event Horizon Telescope?

    *I wonder if anyone has looked at the language of astronomy from a paleological or historical anthropological perspective? Key terms as fossils, sort of thing ...

    Posted

  • JeanTate by JeanTate

    As you might expect, if you've been reading my posts here, I have not been sitting around waiting for someone to come along and answer my question. 😃

    From Morganti+ 2011:

    Unlike all objects mentioned above, the radio power of PKS 1814-637 (P5GHz = 4.1 × 1025 W Hz−1, Tadhunter et al. 1993; Morganti, Killeen, & Tadhunter 1993; Morganti et al. 2001) falls well above the radio power boundary between FRI and FRII radio sources (P5GHz ~1025 W Hz−1; Fanaroff & Riley 1974). For comparison, its radio power is two orders of magnitude higher than the most powerful radio Seyfert (NGC 1068, P5GHz ~1023 W Hz−1), and more than a factor of four higher than the next most powerful radio source hosted by a disk galaxy (see Table 1). Interestingly, radio-loud narrow line Seyfert 1 (NLS1) galaxies can reach radio power comparable to PKS 1814-637 (see e.g. Foschini 2011 and refs therein) but because their radio emission is likely dominated by a beamed jet emission, their intrinsic jet (and extended radio lobe powers) might be orders of magnitude lower, making them more similar to Seyferts or low power FRIs.

    Of course, this only hints at the answer to my question, but they're good hints, right?

    My guess is that the "radio power" mentioned is somewhat comparable to the absolute magnitude optical astronomers use*; however, how the robustly estimate 'apparent integrated magnitude' (especially for doublelobe sources) I do not know, given that source flux/beam from an interferometric observation will have an unknown zero point. And what's the radio astronomy counterpart to k-correction? Surely far from trivial; in the paper I cite, the 'wavelength' is 5GHz, but FIRST is 1.4GHz, so it'd be kinda like taking u-band estimates and extrapolating them to J or H-band 😮

    Anyone?

    *crudely, observe a galaxy, measure its apparent magnitude; from its redshift, calculate its distance modulus, turn the handle. Complications include deredening (not relevant for radio astronomy, right?), k-correction (surely there's a radio astronomy counterpart), and choice of values for parameters in the cosmological model of your choice (~'737' are fairly popular values, LCDM the almost universal choice of model).

    Posted

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

    Making some slow progress ... if only there were a nice, general introduction to radio astronomy, with a section on calculating luminosity (etc) for distant galaxies ...

    I did find RADIO SOURCES AND COSMOLOGY, by James J. Condon (who I gather is one of the 'greats'), from NED (source). Here are some extracts from 2. BASIC RELATIONS (I've done my best to transcribe it; needless to say, Talk doesn't support anything resembling LaTeX 😦):

    The flux density S as a function of frequency and the angular size of a radio source are the observables directly relevant to most cosmological problems. They are related to the intrinsic source luminosity L and projected linear size d as described below.

    Consider an isotropic source at redshift z with spectral luminosity L at frequency (measured in the source frame). Its spectral flux density S measured at the same frequency ν (in the observer's frame) will be:

    S = L/(A(1+z)1+α)

    where A is the area of the sphere centered on the source and containing the observer and α ≡ - ln(S / S0) / ln(ν /ν0) is the two-point spectral index between the frequencies ν and ν0 = ν/ (1 + z) in the observer's frame. (Note that the negative sign convention for α is used throughout this chapter.) The (1 + z)1+α term expresses the special relativistic Doppler correction; the geometry and expansion dynamics of the universe appear only in A. An "effective distance" D (Longair 1978) can be defined by A ≡ 4π D2. Since the area of the sphere centered on the observer and containing a source at redshift z is always A / (1 + z)2, the relation between (projected) linear size d and measured angular size θ is

    θ = d(1+z)/D

    The "angular size" distance is defined by Dθ ≡ d /θ = D / (1 + z). The "bolometric luminosity distance" Dbol defined by Sbol = Lbol / (4π Dbol2) is given by Dbol = D(1 + z).

    In Friedmann models (cosmological constant Λ = 0) with zero pressure, density parameter Ω = 2 q0, and current Hubble parameter H0 the effective distance is [...]

    Yes, this is a rather old source, as you can tell by the use of q; otherwise seems pretty straight-forward, though the terminology isn't quite the same as used in contemporary optical astronomy, I think (and "expresses the special relativistic Doppler correction" seems a bit odd).

    Anyway, once I find a way to map the last bit onto what seems to be standard use today - or decide if I can use one of the three simplifying formulae (not copied; for Ω = 0, 1, 2) - I should be ready to start turning handles. Oh, and making sure that α is the same "spectral index" α that appears in papers like Morganti+ (2011); it wouldn't be the first time, in astronomy, that a key symbol is used with two (slightly) different meanings, despite the fact that the English language gloss is (nearly) identical! 😮

    Posted

  • akapinska by akapinska scientist

    Hi @JeanTate,

    let me give you first part of the answer. Historically radio loud were objects that had more radio power for a given optical luminosity, i.e. all objects for which ratio between radio luminosity at 1.4 GHz and optical luminosity at M_B (careful on the units!):

    R> L(1.4GHz) / L(MB) => radio loud

    It was defined by Kellerman+ 1989: http://adsabs.harvard.edu/abs/1989AJ.....98.1195K

    However, as things in astronomy are never that clear, this is a historical definition that some people still use, but there has been a lot of work done between 1989 and 2014, and often people would use just radio luminosity to decide if a source is radio quiet. To make things more complicated: you may still have faint radio loud objects! I would say border is somewhere around ~10^20 W/Hz at 1.4 GHz, but this is very fuzzy border!

    As for the bimodality of the radio loudness distribution - yeah, that's not so clear now either, as we recognise now that initially we had there quite a bit of selection bias. I'll try to find a few good papers that discuss this issue and post it here shortly.

    Posted

  • akapinska by akapinska scientist

    Oh, one more thing!

    If you are to combine equations from various papers make sure the spectral index α is defined in the same way. Just to (again) make it more confusing, radio astronomers define spectral index either as:

    S_nu \propto nu ^(-α)

    or

    S_nu \propto nu ^α

    In the first case you'll have positive α values, but i the second you'll end up with negative!

    Each paper should have this defined though

    Posted

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

    Thanks very much! 😃

    I would say border is somewhere around ~10^20 W/Hz at 1.4 GHz, but this is very fuzzy border!

    Indeed. About the only positive thing I can say, based on my very limited reading of the literature, is that each author/paper does spell out the threshold they use.

    And there's also the distinction between FRI and FRII, based on estimated radio power/spectral flux density (e.g. in the Morganti+ 2011 paper I quoted from, above).

    make sure the spectral index α is defined in the same way.

    I noticed that too; and radio astronomers seem to like putting 'blue' (higher frequencies/shorter wavelengths) on the right in plots/charts/graphs/figures, the opposite of optical astronomers (blue is always on the left). How all this mish-mash came to be is, of course, a fascinating historical narrative; however, I must say that it's a seemingly needless learning burden for newbies. Have there been any meters/feet oops! events like the Mars Climate Orbiter?

    As for the bimodality of the radio loudness distribution - yeah, that's not so clear now either, as we recognise now that initially we had there quite a bit of selection bias.

    It seems to me, a complete ignoramus, that this is something quite important to get right ... for example, models of galaxy evolution can't really be well tested if their predictions on the distribution(s) of intrinsic radio luminosity don't match the observational data, because there are unknown/unmodeled selection effects!

    On top of such selection effects, there's the "S_nu \propto nu ^(-α)"* assumption, behind which is the "either synchrotron OR thermal" assumption? Reminds me of the only-half-in-jest 4-choice answer (perhaps c) and d) should be swapped?):

    • a) Yes
    • b) No
    • c) Both the above
    • d) None of the above

    What if, within the beam, a source has both synchrotron and thermal components?

    I'll try to find a few good papers that discuss this issue and post it here shortly.

    That would be very much appreciated, thank you.

    *nice try with the LaTeX; I'll bet no one ever asked the Zooniverse developers to include 'compatible with LaTeX' (or something similar) as a requirement for Talk! 😛

    Posted

  • akapinska by akapinska scientist in response to JeanTate's comment.

    And there's also the distinction between FRI and FRII, based on estimated radio power/spectral flux density (e.g. in the Morganti+ 2011 paper I quoted from, above).

    Yes, there is this fundamental distinction between FRI and FRII type radio galaxies, but the distinction is morphological. That is FRIs have bright turbulent jets that end in a diffuse emission extending far beyond the jets (can be turbulent diffuse emission or lobed, but there are no hotspots are jets are clearly decelerating!), while FRIIs have well defined, often one sided jets that terminate in hot spots and create radio lobes. There may be asymmetry in FRIIs but no turbulent jets as seen in FRIs. (By the way, #wat, #nat, #plume are all FRIs).

    OK, originally (FR74, http://adsabs.harvard.edu/abs/1974MNRAS.167P..31F), indeed, luminosity division was introduced - in fact that's how the two types were found. But now we know that there are powerful FRIs (quasars) and there is also veeery many low luminosity FRIIs. So the luminosity division between the two classes is not valid any more in my opinion. You may be interested in Ledlow-Owen diagram who showed that this luminosity division may be dependent on optical magnitude ( http://adsabs.harvard.edu/abs/1996AJ....112....9L) and Laing paper who pointed out existence of low luminosity FRIIs hosts of which are low-excitation galaxies ( http://adsabs.harvard.edu/abs/1994ASPC...54..201L)

    I know that people use sometimes the old FR74 luminosity division to separate out "FRIs", especially if the sources are unresolved, but personally I disagree with that, for the above reasons.

    On top of such selection effects, there's the "S_nu \propto nu ^(-α)"* assumption, behind which is the "either synchrotron OR thermal" assumption?

    Synchrotron. Especially at low radio frequencies thermal is very much negligible.

    Posted

  • akapinska by akapinska scientist in response to JeanTate's comment.

    Ah, @JeanTate, by the way thanks for asking all those question, just went off to search for some radio-loudness articles for you and stumbled upon this one: http://adsabs.harvard.edu/abs/2010MNRAS.406..493M which suggests to define radio loudness as the ratio of radio to mid-infrared luminosity.

    Don't know how good, solid the paper is as I haven't read it yet (just printing..) but hey, I even didn't know up until this very moment that such definition was recently suggested! 😃

    Having a look at it now...

    Posted

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

    Thanks! 😃

    I've now read that there are hybrid sources, FRI on one side, FRII on the other*; I do hope they're using consistent definitions! With your clear definition of morphology in hand, I'll now keep a sharper eye out mixed morphology objects. I guess the activity of the core (or lack of it) is independent of lobe/jet morphology?

    Synchrotron. Especially at low radio frequencies thermal is very much negligible.

    Yes, but ... I've now read papers in which the estimated α is quite different from (-)0.7 (Bagchi+ 2014 is one such, Tsai+ 2013 another), and not just because a source may have several (unresolved) components. Maybe I just don't know what "low radio frequencies" are 😦

    which suggests to define radio loudness as the ratio of radio to mid-infrared luminosity

    Interesting ... but not very practical, at least not for this ordinary RGZ zooite (estimates of [O IV] λ25.89 μm emission line strength aren't as easy to come by as those of S1.4GHz! 😛 ).

    *e.g. Tsai+ (2013)

    Posted

  • akapinska by akapinska scientist in response to JeanTate's comment.

    I've now read that there are hybrid sources, FRI on one side, FRII on the other*; I do hope they're using consistent definitions! With your clear definition of morphology in hand, I'll now keep a sharper eye out mixed morphology objects.
    *e.g. Tsai+ (2013)

    YES @JeanTate! In fact I am almost finished with paper on those hybrid morphology radio galaxies from RGZ! I hope to have it formally out by the end of the year which is really exciting! I'll follow it up with a blog entry 😄

    As for Tsai+2013 I disagree with their finding (it's not a hybrid), but have a look at these papers (which are classics on the subject now): Gopal-Krisha+Wiita 2000 ( http://adsabs.harvard.edu/abs/2000A%26A...363..507G) and maybe also Gawronski+2006 ( http://adsabs.harvard.edu/abs/2006A%26A...447...63G)

    Posted

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

    Thanks! 😃

    As for Tsai+2013 I disagree with their finding (it's not a hybrid), but have a look at these papers

    Hmm ... when I get a chance, I'll have a look at what WISE J233237.05-505643.5 looks like in ... well, I guess SUMMS (it's too far south for FIRST, etc). And read the papers too.

    Back to the topic of this thread.

    Here's a kinda histrogram of the radio luminosity, L1.4GHz, with the log-scale x-axis binned (0.25 dex; zero is set at 1022 WHz-1):

    enter image description here

    This is ~400 galaxies with FIRST and/or NVSS sources within ~10" of the galaxies' SDSS positions (so there'll be few lobes, but all the cores), in a random ~percent region of the sky*.

    To my ignorant eye, that's nothing like a bimodal distribution, and may even be close to a Gaussian (I'll have a go at fitting a Gaussian later, and getting some sort of goodness-of-fit statistic).

    *71/27/73 ΛCDM cosmological model, and SDSS spectroscopic redshifts; however, the calculations are a bit, um, rough and ready

    Posted

  • akapinska by akapinska scientist

    The bimodality was/is seen in radio loudness parameter R distribution, not in luminosity density distribution.

    Besides, the drop off at the fainter luminosity densities is predominantly due to the flux density limit of NVSS survey, not the lack of radio sources with such luminosity density. To go around this selection bias you may want to plot radio luminosity function instead. Though again, this will be for luminosity density distribution only, not for the radio loudness parameter.

    Posted

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

    Thanks

    The bimodality was/is seen in radio loudness parameter R distribution, not in luminosity density distribution.

    Trying to keep all these terms clear, in my mind, as I read various papers, is not easy! 😦

    For example, is this R the same as "the ratio of core to extended luminosity at a fixed emitted frequency of 5 GHz" in the Laing+ (1994) paper you suggested earlier*?

    *"and Laing paper who pointed out existence of low luminosity FRIIs hosts of which are low-excitation galaxies ( http://adsabs.harvard.edu/abs/1994ASPC...54..201L)"

    Posted

  • akapinska by akapinska scientist in response to JeanTate's comment.

    The bimodality was/is seen in radio loudness parameter R distribution, not in luminosity density distribution.

    Trying to keep all these terms clear, in my mind, as I read various papers, is not easy! 😦
    For example, is this R the same as "the ratio of core to extended luminosity at a fixed emitted frequency of 5 GHz" in the Laing+ (1994) paper you suggested earlier*?

    No. R is ratio of radio 1.4 GHz to optical in B (Kellerman+1989)

    Posted