by George Rhee
We have to do systematic work to make progress in astronomy. Tycho showed us this with his work on the island of Hveen hundreds of years ago. By making careful and very accurate measurements of the planets, Tycho enabled Kepler to determine the true shape of planetary orbits. Galaxy surveys havet also revolutionized our view of the universe.
The Redshift Machines
To map the galaxy distribution in three dimensions we need to measure redshifts. Redshifts are measured by obtaining a spectrum of a galaxy using a spectrograph. The spectrograph uses a device such as a prism or a diffraction grating to split the light into its constituent colors. Isaac Newton used a prism to show that sunlight contained all the colors of the rainbow. With a prism and our naked eye we can see what colors are present in the light we examine. The spectrograph enables us to plot the intensity of light as a function of wavelength. It tells us how much light is present at red wavelengths compared to say green wavelengths. The output produced by putting the light through a spectrograph is called a spectrum. By analyzing the spectra of galaxies we can measure their redshifts and hence estimate their distances.
The spectra of galaxies at visible wavelengths consist of the spectra of lots of stars added together. Stellar spectra are known to contain absorption lines, narrow parts of the spectrum where the light is fainter. These can be easily seen in our sun’s spectrum. The absorption lines are due to the presence of atoms such as calcium and iron in the outer parts of stars. These absorption lines have known wavelengths that we can measure in the laboratory. Their location in a galaxy’s spectrum enables us to measure the redshift of that galaxy. Essentially we see a recognizable pattern of lines all shifted towards the red (Fig. 5.1). Computer programs have been developed to automatically measure redshifts.
The most common spectrograph design was for a long time the long slit spectrograph. Any object that fell on the slit would have its spectrum taken. Typically one could measure the spectra of 1–3 galaxies at one time. As we have noted above cosmologists are greedy for redshifts. They want to measure as many as possible. Wouldn’t it be great to measure hundreds of redshifts at one time? This has become possible through the design of fiber spectrographs. These ingenious devices still employ a long slit but use up to several hundred optical fibers to take the light from many galaxies and feed it into the slit. It is a trick to align all the galaxy images in a nice linet along the spectrograph slit. Two major nearby galaxy surveys made use of these fiber spectrographs. The two degree field redshift survey was carried out using the Anglo Australian
Fig. 5.1A spectrum of an elliptical galaxy, showing how much light the galaxy gives off at different colors. The horizontal axis labels shows the wavelength of the light with blue on the left and red light on the right. The vertical axis gives the intensity of the light. The spectrum is shifted towards the red relative to what it would be on earth. By identifying emission features labeled in blue, and absorption features labeled in red with the known wavelengths measured in the laboratory we can calculate the redshift of this galaxy. In this case the redshift is about 0.16 (Credit: Sloan Digital Sky Survey (SDSS) Collaboration, www.sdss.org)
Fig. 5.2The plugboard for the Sloan Digital Sky Survey fiber spectrograph. A thin plate of metal is mounted on the back of the telescope. This plate has a set of about 500 holes drilled into it at exactly the locations of 500 galaxies in the direction that the telescope is pointing. Each hole is plugged with an optical fiber, which carries the light from one galaxy down to a big spectrograph. In this way 500 galaxies can be observed at once (Credit: Fermilab Visual Media Services)
Telescope located in New South Wales, Australia. The Sloan Digital Sky Survey was carried out at Apache Point Observatory in southern New Mexico. The Sloan Digital Sky Survey project uses a thin metal plate mounted on the back of the telescope (Fig. 5.2) to hold the fibers in position. The plate has 500 holes drilled into it in exactly the locations of 500 galaxies in that area of sky. Each hole is plugged with an optical fiber which carries the light from one galaxy down to a big spectrograph. The spectrograph splits the light from each fiber into a spectrum and measures all the spectra at once.
The first step is to drill holes in metal plates. One then inserts a fiber in each hole and positions the metal plate where the image of the sky is formed in the telescope. This is hard work. You have to plug in all the fibers, note their numbers and note which galaxy’s light is going down which fiber. Once you’ve set the whole thing up some clouds may drift by. What to do? The plates are set up for observing at a given time on a given night. Once that window is gone you have to move to the next plate.
The two degree field project used a spectrograph that carried the fibers on arms. Writing a computer program to locate 100 or more fibers on galaxy positions without the arms colliding is a tricky undertaking.
Once the data have been obtained you have to use the spectra to obtain redshifts. A CCD image may contain a million numbers corresponding to the spectra of 100 galaxies. One thus reduces a million numbers to 100 numbers, the redshifts. Most of this work can now be automated and carried out by sophisticated computer programs that produce a list of redshifts together with measurement errors and confidence levels. Measurement errors aret always important in scientific experiments as we have outlined previously. Kepler argued in favor of an elliptical orbit for Mars despite the fact that a circular orbit gave a close fit because he understood the measurement errors in Tycho’s data. The confidence level is a similar number. It says how certain you are that you have the correct redshift.
The galaxy surveys are thus used to produce catalogs of redshifts that can be analyzed by cosmologists to test their theories of the formation of structure in the universe. The theories specify how many clumps of a given mass should form at a specified time, thus predicting what the universe should look like today and in the past. Comparing theory to observation is an art in itself. The theorists are good at using computers to calculate in detail how the clumping of dark matter particles evolves with time. The tricky part is identifying which clumps or halos host which kinds of galaxies.
Strategies for Surveying the Universe
Before one goes to the telescope one needs a strategy or plan. You might argue that to have a plan one should anticipate what one will find. If onet already knows what is to be discovered why bother, its not research. This is the dilemma of research. Research proposals are usually crafted in such a way as to argue that great progress has been made in an area of research and we can wrap it up by making this new crucial observation. However as the astrophysicist John Bahcall pointed out in making the case for the Hubble Space Telescope;
We often frame our understanding of what the space telescope will do in terms of what we expect to find and actually it would be terribly anticlimactic if in fact we find what we expect to find... The most important discoveries will provide answers to questions that we do not yet know how to ask and will concern objects that we have not yet imagined.
The musician Mickey Hart has put it more poetically
Magic doesn’t happen unless you set a place at the table for it.
In astronomy the key is to do something new. The pattern has been that if you don’t find what you are looking for you will find something more interesting. The studies of the structure of the universe originally focused on showing that the galaxy distribution in space is not random and how the distribution deviates from randomness. The large surveys revealed that galaxies populate a large filamentary structure which we call the cosmic web, since it resembles a spider’s web in some respects. The discovery of the cosmic web was completely unexpected just like the discovery of dark energy.
The Sloan Digital Sky Survey uses an unusual camera to make a photographic survey of the sky (Fig. 5.3). The camera contains six columns of CCD chips. Each column has a series of five chips, each
Fig. 5.3The Sloan Digital Sky Survey CCD camera. The image on the left shows the six columns of CCD chips each with the five filters arrang
ed in rows. The right hand image shows how a star image moves down a column during an exposure and is successively observed through different filters (Credit: Michael Carr and the Sloan Digital Sky Survey (SDSS) Collaboration, www.sdss.org)
with a different filter. The telescope is scanned across the sky so that stars move across the camera in straight lines, exactly along the columns of the CCDs. The traditional way of using CCDs on telescopes is to point the telescope at a target object. One then keeps the telescope tracking the sky so that the object remains at a fixed position on the detector. One exposes the CCD then closes the shutter then reads out the image to be stored in digital form on disk. The Sloan Survey camera operates in drift scan mode. That is to say the objects in the sky drift across the camera and the CCDs are read out in sync with the drifting. The technique produces a sharp image. An object takes about 1 min to drift across one of the CCDs. It then drifts to the next CCD in the column and so on. The result is a set of images of each object through five different color filters ranging from blue to near infrared. We can combine these five images to visualize the color of the object we are imaging.
Ideally one would want to measure the redshifts of all the objects in a galaxy catalog. In practice, it takes much longer to measure the redshifts of faint galaxies than it does to obtain an image. One thus chooses a brightness limit and an area of the sky. One then measures the redshifts of all the galaxies brighter than some limit in this specified area of the sky. Almost all redshift surveys are carried out in this manner. This method has the great advantage that it is simple. Simplicity of strategy is a virtue in cosmology. If you select galaxies for redshift measurements simply on a whim, because they look cool or whatever, it will be impossible to draw significant conclusions. The sober and systematic approach to observing is the one that pays off in the long term. It also ensures reproducibility. The authors explain what they did and how they did it and you can reproduce their results.
Redshift and imaging surveys can be divided into two main categories. The first aret all-sky or wide angle surveys. Astronomers often think of the night sky as a sphere. We can describe a survey as covering some fraction of the sphere. Redshift surveys that cover large areas of sky cannot go very deep. By this we mean that we cannot observe very faint galaxies. This is because (a) there are many more faint galaxies than bright galaxies and (b) it takes more time per galaxy to obtain redshifts for faint galaxies. We can also look to much larger distances by observing faint galaxies in a small part of the sky, say half a moon diameter in size. The Keck Observatory DEEP (Deep Extragalactic Evolutionary Probe) is a case in point. This survey used a 10 m telescope to survey distant faint galaxies that had been imaged by the Hubble Space Telescope.
The large area redshift surveys tell us about the galaxy distribution on large scales. The deep redshift surveys tell us about galaxy evolution. Let us recall that as we look back over large distances we are looking back in time. The deep redshift surveys thus enable us to compare the colors and shapes of galaxies as they were a long time ago with galaxies that we see around us today. The wide angle or shallow redshift surveys tell us about the galaxy distribution nearby. As we shall see both kinds of survey provide fascinating information. Table 5.1 lists a few examples of surveys. Some of these surveys make their data available online. Over one million distinct users have accessed the Sloan Digital Sky Survey data! Table 5.1Examples of galaxy surveys
Acronym
Name
Website
SDSS
Sloan Digital Sky Survey
www.sdss.org/
2dFRS
2 degree Field Galaxy Redshift Survey
www.mso.anu.edu.au/2dFGRS/
2MASS
2 Micron All Sky Survey
pegasus.phast.umass.edu/
FIRST
Faint Images of the Radio Sky at Twenty cm
sundog.stsci.edu/
DEEP2
Deep Extragalactic Evolutionary Probe
deep.berkeley.edu/
VVDS
The VIRMOS-VLT Deep Survey
cesam.oamp.fr/vvdsproject/
BOSS
The Baryon Oscillation Spectroscopic Survey
cosmology.lbl.gov/BOSS/
DES
The Dark Energy Survey
darkenergysurvey.org/
LSST
The Large-Aperture Synoptic Survey Telescope
www.lsst.org/lsst/
Results from Surveys of Nearby Galaxies
One of the first redshift surveys designed to map the universe was carried out by astronomers at the Smithsonian Astrophysical Observatory using a 1.5 m telescope in Arizona. The survey consisted of 2,500 galaxies brighter than a certain blue color (magnitude B = 14.5). Mathematical analyses showed that the galaxies were clustered.
A much more striking result was obtained by a now famous survey (known as CfA2) that was published in 1986 (Fig. 5.4). The strategy for this survey was to select galaxies in a strip on the sky. When one adds redshifts to the coordinates on the sky the volume of space surveyed looks like a slice of pizza. When the redshifts for this survey are plotted the results are striking. In such a plot one marks each galaxy as a point. One sees a clear network of filaments and voids. That is to say there are regions of space that have a spherical shape that contain no galaxies. In between these voids there are filaments and sheets. There are two other features that dominate this survey. One is the Coma cluster of galaxies, the large clump located in the center of Fig. 5.4. This is the most massive nearby cluster known. The second feature is known as the great wall. It is a filament of galaxies (a sheet or wall in three dimensions) that spans the whole survey and goes through the Coma cluster.
The fact that the CfA2 survey found this huge feature, the great wall raised an interesting question. We base our models of the universe on the assumption that the universe is homogeneous and isotropic. The CfA2 survey found a feature that was comparable in scale to the size of the survey. The fact that the slice survey was dominated by the great wall and the Coma cluster suggested the need to probe to even larger scales to get a fair sample of the
Fig. 5.4The “slice of the Universe” that represents the first set of observations done for the CfA Redshift Survey in 1985. These are spectroscopic observations of about 1,100 galaxies in a strip on the sky 6 ∘ wide and about 130 ∘ long. We are at the apex of the wedge. The radial coordinate is redshift, measured in kilometers per second. The outer arc of the plot is at a distance of about 700 million light years (Credit: Smithsonian Astrophysical Observatory)
universe. This has been a motivation for carrying out even larger and deeper surveys such as the Sloan and two degree field surveys.
The question of a fair sample can again be illustrated using a polling analogy. If I want to predict the outcome of a presidential election I could ask my neighbor who he will vote for. It would be irresponsible on the basis of one opinion to publish a statement saying that based on my research I expect Candidate A to win. I would do better if I asked a few people on my street. It would be even better to sample various neighborhoods in town and then go to neighboring towns and even other states. If I decided to poll people as a function of distance from my house, starting close by and working my way out, I could decide to stop once the results are no longer changing. That is to say I could stop when I have obtained a fair sample of the population. This is the issue facing the surveyors. The results of the Sloan Digital Sky Survey and two degree field survey show that a fair sample of the universe has finally been reached.
Both the Sloan Digital Sky Survey and the Two Degree Field Galaxy Redshift Survey produced spectacular maps of the galaxy distribution in redshift space (Fig. 5.5). The cosmic web of galaxies traced out by these surveys has been accounted for by the galaxy simulations. Notice the phrase ‘accounted for’ rather than predicted which is an important distinction in sc
ience.
Fig. 5.5The galaxy distribution obtained from redshift surveys. The smaller slice at the bottom shows the CfA2 survey, with the Coma cluster at the center. The upper ‘slice’ is section of the Sloan Digital Sky Survey in which a “Great Wall” has been identified. This is one of the largest observed structures in the Universe, containing over 10,000 galaxies and stretching over more than 1.37 billion light years (Credit: J. Richard Gott and Mario Juric, 2005, Astrophysical Journal, 624, 436, Reproduced by permission of the American Astronomical Society)
Many scientific findings resulted from these surveys. As well as mapping out the cosmic web as traced by galaxies, the nearby galaxy surveys helped put constraints on the dark energy density, the matter density and the baryon density in the universe. One can accurately measure the star formation rate in the local universe and use this to show that the star formation rate was higher a few billion years ago than it is now. One can also study the effect of environment on galaxy properties. The colors of galaxies and the rate at which they form stars depend on the environment in which a galaxy finds itself. The surveys discovered quasars at high redshifts and made it possible to study the intergalactic gas properties at early times.