Author Topic: Spectrum class differences:DR7 versus DR8  (Read 8258 times)

zutopian

  • Hero Member
  • *****
  • Posts: 2852
  • REOPEN this forum!!!
    • View Profile
Spectrum class differences:DR7 versus DR8
« on: December 22, 2011, 04:20:42 pm »
Here are some objects, which have different spectral classes "DR7 versus DR8" from various threads: They have same or different redshifts,
For those objects, which have same specclass, but different redshifts, I started another thread:
http://www.galaxyzooforum.org/index.php?topic=279763.msg574940#msg574940

Mitch's Mystery "Star": http://www.galaxyzooforum.org/index.php?topic=277052.msg558699#msg558699
STAR: http://cas.sdss.org/dr7/en/tools/explore/obj.asp?id=587739406764540066
QSO:  http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237665127989444752

STAR: http://cas.sdss.org/dr7/en/tools/explore/obj.asp?id=588018091615125718
QSO:  http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237662337863319765

QSO    : http://cas.sdss.org/dr7/en/tools/explore/obj.asp?ra=137.309495&dec=57.402249
Galaxy: http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237651533875118264
Same redshift

QSO     : http://cas.sdss.org/dr7/en/tools/explore/obj.asp?id=587742863132983414
Galaxy  : http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237668584357888113
Same redshift

This means, that I report:
2 cases: specclass Star in DR7 have specclass QSO in DR8.
2 cases: specclass QSOs in DR7 have specclass Galaxy in DR8.


Here is an info from "What's New in DR8?" :

Quote
2.  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


PS: I haven't knowledge to examine spectrum charts, but I am interested in them. ;D
« Last Edit: December 28, 2011, 01:05:32 pm by zutopian »

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Spectrum class and/or Redshift differences:DR7 versus DR8
« Reply #1 on: December 23, 2011, 04:51:45 pm »
Here are some objects, which have different spectral classes and/or different redshifts "DR7 versus DR8" from various threads:

Mitch's Mystery "Star": http://www.galaxyzooforum.org/index.php?topic=277052.msg558699#msg558699
STAR: http://cas.sdss.org/dr7/en/tools/explore/obj.asp?id=587739406764540066
QSO:  http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237665127989444752

It it important to recognize that, in these charts, the specClass/Class and z are outputs of automated pipelines.

Perhaps more important is that even these brief summaries contain indicators of how 'good' the z outputs are; for example, in this case:

DR7: zConf 0.822783, zStatus   XCORR_HIC, zWarning   AB_INC NOT_QSO
DR8: Warnings: MANY_OUTLIERS

In the downloadable, fully accessible files that contain the full pipeline outputs there is a lot of data 'explaining' or 'describing' the pipelines' 'decisions'*. By going through that data you can - usually - work out how the outputs are consistent (or not) with the inputs; in many cases you can also see - sometimes quickly, sometimes not - where the pipeline went 'wrong'.

In this particular case, this kind of white dwarf has a very distinctive spectrum, but it's not one of the template spectra the pipelines use. No surprise, then, that the output is, um, unreliable.

Quote
PS: I haven't knowledge to examine spectrum charts, but I am interested in them. ;D

Have you read NGC3314's Tutorial bits on galaxy spectra?

It's a pretty good place to start!  :) And you can ask any questions you have!!  ;D 8) (Although it may take a while before someone with the appropriate knowledge answers  :( ).

There are some other fora where you could get answers to questions you might have about astronomical spectra; BAUT's Space/Astronomy Questions and Answers, for example.

* I'm putting all these in quotes because, of course, the pipelines are just software, which mindlessly crunches numbers; explanations, descriptions, and decisions are things done by people.

ETA: Also, the Newbies, post your spectrum questions here thread.
« Last Edit: December 23, 2011, 05:14:11 pm by JeanTate »

zutopian

  • Hero Member
  • *****
  • Posts: 2852
  • REOPEN this forum!!!
    • View Profile
Re: Spectrum class and/or Redshift differences:DR7 versus DR8
« Reply #2 on: December 23, 2011, 06:42:58 pm »
Here is an info from "What's New in DR8?" :

Quote
2.  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
« Last Edit: December 23, 2011, 06:48:29 pm by zutopian »

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Spectrum class and/or Redshift differences:DR7 versus DR8
« Reply #3 on: December 23, 2011, 11:27:36 pm »

STAR: http://cas.sdss.org/dr7/en/tools/explore/obj.asp?id=588018091615125718
QSO:  http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237662337863319765

It's a point source, whether a QSO or a star, that's for sure!

But not much else is (sure, that is); the spectra - esp. the DR8 one - are ratty (and I guess the specClass/Class comes from the estimated z). Perhaps this is a typical - if noisy - spectrum for an unusual type of star (I really don't know)?

Quote
QSO    : http://cas.sdss.org/dr7/en/tools/explore/obj.asp?ra=137.309495&dec=57.402249
Galaxy: http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237651533875118264
Same redshift

The DR8 Class might be a particularly good example of a 'bad chart'! The spectrum has pretty obvious signs of an AGN: the H-alpha/[NII] complex is broad, and the continuum has a 'UV rise' (and if it's STARFORMING, where is the strong [OIII]?). In any case, it's not a point source, which - as far as I know - the (DR7) spectroscopy pipeline pays no attention to, in deciding whether the specClass is QSO or not ...

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Spectrum class and/or Redshift differences:DR7 versus DR8
« Reply #4 on: December 26, 2011, 08:06:00 pm »
Here is an info from "What's New in DR8?" :

Quote
2.  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

From The H-beta aberration OOTD, we read "Of 1.053.144 objects with spectrum in SDSS DR7 ..."

1% of 1.053144 million is 10,531. At the 99.9% level - i.e. if the statement had been "virtually all objects (> 99.9%) they agree" - there would be >~1,000 objects which would "differ in detail".

Given the level of scrutiny of SDSS objects by zooites, and their propensity to report the unusual, the anomalous (etc), it may be more surprising that you've found only five (so far)!  ;D

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Spectrum class and/or Redshift differences:DR7 versus DR8
« Reply #5 on: December 26, 2011, 08:49:03 pm »
QSO     : http://cas.sdss.org/dr7/en/tools/explore/obj.asp?id=587742863132983414
Galaxy  : http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237668584357888113
Same redshift

The DR8 Class is GALAXY BROADLINE. In this case, GALAXY is the class, and BROADLINE is the subclass. The DR8 Schema Browser says this about (this particular kind of) subclass: "Schlegel subclass from PCA analysis -- not alwasy [sic] correct!! AGN/BROADLINE/STARBURST/STARFORMING" (The DR8 Class is "Spectroscopic class (GALAXY, QSO, or STAR)").

At the word level - i.e. ignoring the specifics of the actual definitions - a broadline galaxy, if sufficiently luminous, is a QSO!  8)

Quote
Here is a case, where the specclasses are same, but the redshifts are different:
http://cas.sdss.org/dr7/en/tools/explore/obj.asp?id=587731173306073230
http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237663543138910468
Discussed in these threads:
http://www.galaxyzooforum.org/index.php?topic=279013.30
http://www.galaxyzooforum.org/index.php?topic=6732.msg568650#msg568650

This is an example of both spectroscopic pipelines being correct!  ;)

There is a very clear emission line system at z~0.0524, with all the usual suspects (H-alpha, [NII]6585, [OIII]5007 and 4959, [OII]3726, and [SII]6718), and even H-beta.

There is also a clear absorption line system at z~0.209, with many of the usual suspects (H and K, G, Mg, and Na).

As the write-up of both pipelines makes clear, the high-level output from each, for an individual object, is a single redshift; for example: "There exist among the spectra a small number of composite objects. Most common are bright stars on top of galaxies, but there are also galaxy-galaxy pairs at distinct redshift ..."

This is one of those small number of composite objects:P  :o

zutopian

  • Hero Member
  • *****
  • Posts: 2852
  • REOPEN this forum!!!
    • View Profile
Re: Spectrum class differences:DR7 versus DR8
« Reply #6 on: December 28, 2011, 09:00:25 am »

STAR: http://cas.sdss.org/dr7/en/tools/explore/obj.asp?id=588018091615125718
QSO:  http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237662337863319765

It's a point source, whether a QSO or a star, that's for sure!

But not much else is (sure, that is); the spectra - esp. the DR8 one - are ratty (and I guess the specClass/Class comes from the estimated z). Perhaps this is a typical - if noisy - spectrum for an unusual type of star (I really don't know)?

SIMBAD says : SDSS J154104.67+360252.9 -- Cataclysmic Variable Star
« Last Edit: December 28, 2011, 09:02:40 am by zutopian »

zutopian

  • Hero Member
  • *****
  • Posts: 2852
  • REOPEN this forum!!!
    • View Profile
Re: Spectrum class differences:DR7 versus DR8
« Reply #7 on: December 28, 2011, 09:32:51 am »
I copied my below post from another thread:

DR7: specClass Galaxy:  z=1.389    (click ALL Spectra)
DR8: specClass QSO    :  z=1.394
NED: QSO                     :  z=1.395

I quote my own post:Here is the QSO in DR8:


http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237663543139566456
In DR8 the image of the quasar is revised: In DR7 it was faint, but in DR8 it isn't. (It is a 2SLAQ Quasar:The 2SLAQ survey has used imaging from the Sloan Digital Sky Survey (SDSS) and spectra from the 2dF-spectrograph on the Anglo Australian telescope to obtain redshifts for 13,000 luminous red galaxies with 0.4 < z < 0.7 and 6000 faint QSOs.)
In DR8 the class is QSO!

2SLAQ J210117.60-000429.0 -- Quasar
In SDSS: - Classified as STAR, but the specClass is GALAXY! -TARGET_QSO_FAINT TARGET_SERENDIP_BLUE
In NED and Simbad there is given QSO!
The 2SLAQ survey has used imaging from the Sloan Digital Sky Survey (SDSS) and spectra from the 2dF-spectrograph on the Anglo Australian telescope to obtain redshifts for 13,000 luminous red galaxies with 0.4 < z < 0.7 and 6000 faint QSOs.



http://cas.sdss.org/dr7/en/tools/explore/obj.asp?id=587731173306729090



« Last Edit: December 28, 2011, 10:06:07 am by zutopian »

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Spectrum class differences:DR7 versus DR8
« Reply #8 on: December 28, 2011, 07:53:34 pm »
A few things:

TARGET_QSO_FAINT TARGET_SERENDIP_BLUE is the value in the field PrimTarget; as I understand it, the photometric pipeline assigns at least one (and sometimes more than one) value to PrimTarget, based on various automated algorithms. In words, you can think of these as giving guidance to the astronomers (and operators) as to what objects should have spectra taken, and why.

Stars and (most) QSOs are 'point sources'; at the angular resolution and dynamic range of SDSS' photometry, they both 'look like' points of light. QSO candidates are selected, by the photometry pipeline, based on their colors.

With a spectrum in hand, it is often very easy to tell that a point source is a QSO and not a star (and vice versa); however, if the spectrum is really ratty (etc), it may not be so obvious.

No matter how you look at it, this object is faint! It's certainly no brighter than 19, in all five SDSS bands, and fainter than 20 in most.

The DR7 and DR8 pipelines are independent; that's one of their great strengths. That independence includes definitions; the definition for one class of thing in DR7 may not be the same as that for the same (named term) in DR8; indeed, DR8 may not even include a definition/term that is used in DR7! Apparent inconsistencies may be the result of slightly different definitions; they may, of course, be different conclusions (based on the same data).

DR7's z is 1.3892±0.0055 (0.57)
DR8's is 1.39360±0.00065 (no estimate of confidence)

The two estimates are consistent (compatible), aren't they?

zutopian

  • Hero Member
  • *****
  • Posts: 2852
  • REOPEN this forum!!!
    • View Profile

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Spectrum class differences:DR7 versus DR8
« Reply #10 on: January 01, 2012, 09:07:17 am »
Talk about an "epic fail!" for the DR8 spectroscopic pipeline!  ::)

DR7 has it right:



The continuum red-ward of the H-alpha/[NII] complex - at least to the [SII] doublet - is definitely strange, but I can't tell if it's an instrumental problem or something to do with the overlap (or both). H-alpha is quite broad, so it seems this is a broadline AGN; however, it's quite asymmetric, so maybe something else is going on?

zutopian

  • Hero Member
  • *****
  • Posts: 2852
  • REOPEN this forum!!!
    • View Profile
« Last Edit: January 10, 2012, 09:24:12 pm by zutopian »

JeanTate

  • OotD posters
  • Hero Member
  • *****
  • Posts: 2976
    • View Profile
Re: Spectrum class differences:DR7 versus DR8
« Reply #12 on: January 11, 2012, 02:15:59 am »
Star    : http://skyserver.sdss.org/dr7/en/tools/explore/obj.asp?id=588015508739063842     <click> "All Spectra"
Galaxy: http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237663783667040348
Good find zutopian!  ;D

This is an overlap. DR8 picked out one redshift system (z=0.19558), probably of the yellow elliptical/lenticular. Whether the blue is local star (per DR7) or something else I can't say (just by looking at the spectrum).

It's not in the overlapping galaxy pairs catalog, so it may be that NGC3314 decided it's a star. Of course, it's also possible he hasn't seen it before ..

zutopian

  • Hero Member
  • *****
  • Posts: 2852
  • REOPEN this forum!!!
    • View Profile
Re: Spectrum class differences:DR7 versus DR8
« Reply #13 on: April 19, 2012, 06:20:32 am »
Same redshift (z=0.003), but different specclasses in DR7 and DR8:


DR7: http://skyserver.sdss.org/dr7/en/tools/explore/obj.asp?id=587732769983889425      specclass: GALAXY (photometric objtype:Galaxy)
DR8: http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237658491208794124  specclass: STAR     (     "                  "           "     )



zutopian

  • Hero Member
  • *****
  • Posts: 2852
  • REOPEN this forum!!!
    • View Profile
Re: Spectrum class differences:DR7 versus DR8
« Reply #14 on: April 19, 2012, 07:18:31 am »
Same redshift, but different specclasses in DR7 and DR8:


DR7: http://skyserver.sdss.org/dr7/en/tools/explore/obj.asp?id=587739721906520183       specclass:UNKNOWN
DR8: http://skyserver.sdss3.org/dr8/en/tools/explore/obj.asp?id=1237665443131424882   specclass:QSO