PDA

View Full Version : Spectra from Extreme Red-Shifted Objects



Chip
2004-Sep-02, 08:03 AM
Here are a few questions about the tiniest dots in the Hubble Ultra Deep Filed, so small that they are not really visible in reproduced printed book or magazine pages and too small for the pixels on Internet images. These odd, early galaxies are extremely red-shifted. They are basically red-shifted just about beyond the limit the visible spectrum. So it seems very challenging if not impossible to somehow obtain the spectra of these early objects.

In reading and Googling around, I've found nothing yet about how spectra could be obtained for extremely remote objects. I know this is sloppy thinking, but would it be possible to reverse engineer the spectra? In other words, beginning with the fact that something is detected even if at a remote threshold, would it be possible to devise a framework based on known, more obtainable galaxies, and apply computer programs that theoretically embellish the known spectra into a artificial extreme red-shift representation? The goal would be to approach what little data is obtained from the barely seen object, and treat it as a code in place of the data from a well-known object.

Anyone interested might want to chime in with better information or ideas. For all I know, there is a much better way to approach the problem of spectra for extremely red-shifted objects.

Is there anything akin to taking a known technique, (Linear Regression perhaps?) between sets of closer galaxies, expressed as equations and then hypothetically moving them very far away within a mathematical computer application of forced extreme red-shifts? Then see what matches actual extreme red-shifts as if it were encoded to represent the distorted known data. If this were possible, one could say: if A =B, then B becomes C in the remote object, then C = B and A in the actual remote object.

Maybe the answer is also to build bigger space telescopes? :wink:

ngc3314
2004-Sep-02, 01:09 PM
Here are a few questions about the tiniest dots in the Hubble Ultra Deep Filed, so small that they are not really visible in reproduced printed book or magazine pages and too small for the pixels on Internet images. These odd, early galaxies are extremely red-shifted. They are basically red-shifted just about beyond the limit the visible spectrum. So it seems very challenging if not impossible to somehow obtain the spectra of these early objects.

In reading and Googling around, I've found nothing yet about how spectra could be obtained for extremely remote objects. I know this is sloppy thinking, but would it be possible to reverse engineer the spectra? In other words, beginning with the fact that something is detected even if at a remote threshold, would it be possible to devise a framework based on known, more obtainable galaxies, and apply computer programs that theoretically embellish the known spectra into a artificial extreme red-shift representation? The goal would be to approach what little data is obtained from the barely seen object, and treat it as a code in place of the data from a well-known object.

Anyone interested might want to chime in with better information or ideas. For all I know, there is a much better way to approach the problem of spectra for extremely red-shifted objects.

Is there anything akin to taking a known technique, (Linear Regression perhaps?) between sets of closer galaxies, expressed as equations and then hypothetically moving them very far away within a mathematical computer application of forced extreme red-shifts? Then see what matches actual extreme red-shifts as if it were encoded to represent the distorted known data. If this were possible, one could say: if A =B, then B becomes C in the remote object, then C = B and A in the actual remote object.

Maybe the answer is also to build bigger space telescopes? :wink:

One technique is in common use which is similar to what you describe is known as "photometric redshifts". If you have measurements of galaxy brightness in multiple, accurately calibrated bands (just what the Hubble deep data provide), you can run a consistency test against all known kinds of galaxy and quasar spectra (plus various amounts of foreground absorption) and work out not only the most likely redshift, but its error distribution. As a test, one of the groups getting Keck spectra in the original Hubble Deep Field obtained redshifts for a couple of hundred faint galaxies, made public which ones they were, and invited various groups doing photometric redshift modelling to submit their estimated redshifts, after which a comparison would be made when they revealed the spectroscopic values. It doesn't work badly - systematic errors are usually at the level of 0.01-0.03 in redshift z. Having more filters over a wider wavelength range improves things (there is a redshift range from about z=1.5-2 where the optical window includes only weak absorption lines and iso therwise nearly flat, not helpful).

Of course, to confirm the latest highest-z candidate, one always feels better having individual detected spectral features, but for the faintest objects in the Hubble fields we're a long way from being able to do that. Anything we see at all in the optical has to ave a redshift less than about z=10, because intergalactic hydrogen absorbs an increasing fraction of the light emitted shortward of 1216 Angstroms as one goes to higher redshifts. Hence the infrared emphasis of many new facilities, such as the James Webb Space Telescope.

Even for brighter objects, the most accurate redshift estimates are often generated by cross-correlation (for the nitpickers, in log-wavelength space) against high-quality spectra of stars and reference galaxies, so that all the weak and blended spectral lines contribute to the measurement as well as the few strong and unblended ones. This means that as soon as you can see spectral features in the spectrum, there is enough signal-to-noise for a good measurement of the redshift.

Chip
2004-Sep-02, 07:55 PM
One technique is in common use which is similar to what you describe is known as "photometric redshifts"...

Thanks for a very informed, interesting response. I read recently, perhaps implied through the photometric redshifts you've described, that some galaxies as dim as redshift 7 contain normal stellar populations. This might push the age of first star birth further back into the time of the dark age. I'm hoping you'll be able to "see" and deduce back as far as redshift 8 when the universe was just 600 million years old.

There is also the question of the ionization of neutral hydrogen in the early universe and what it implies. As an amateur who squints at nearby stars through a backyard telescope, I appreciate that there are a lot of new discoveries to be made by astronomers who are "squinting" at far more distant and fewer photons.