Friday, December 03, 2010

The sky for all

kw: book reviews, astronomy, virtual astronomy, astronomers, telescopes

In the title of yesterday's post about the Sloan Digital Sky Survey, I called the result the Galactipedia. The original vision for the SDSS was to take the spectra of a million galaxies, determine their redshift and thus their distance, and use that data to outline the large-scale structure of the Universe. The key figure in this effort is Jim Gunn, who had that original vision and eventually saw the SDSS through to completion.

Jim Gunn is the central figure of A Grand and Bold Thing, by Ann Finkbeiner, a blow-by-blow history of the SDSS and its hardware and software systems. Called Jim throughout the book (no doubt because he usually says, "Call me Jim"), Dr. Gunn, given time, might have done the whole thing himself. A premier instrument maker, his major material contribution to the survey telescope was the camera.

But this is not just about Jim Gunn. It is about the trend in astronomy, as in all the sciences, toward larger collaborations. While Jim abhorred the NASA "project over-management" method used to produce the Hubble Space Telescope and STScI (that's where the data goes), in the end, the effort now called SDSS duplicated the NASA model, because it had to. Long gone are the days of Lord Rosse building his giant six-foot-diameter telescope and surveying the heavens in splendid solitude. Big efforts become big projects, require big money, and need the complex interaction of big teams of people. A "lone ranger" is either brought to heel or becomes exiled, a "lone stranger".

Taking the spectra of a million galaxies in a finite time period, and storing them in a usable database, required imagining a telescope like none other before it; unique robotic controls; software to run the telescope, select targets (in groups of several hundred), store and manage and supply the data; and a human infrastructure to first produce and then tend all these systems. It is a good thing I already knew the outcome of the story, or I'd have been chewing my fingernails to the wrist at the unlikely series of troubles that plagued the project. Parts of the book read almost like cliff-hanger Perils of Pauline tales.

People don't behave. Astronomers in particular are all about the sky, tend to be loners, at least when at work, and are hard to manage. The software effort was huge, which means it required the services of a number of "superprogrammers". We used to call ourselves hackers when that was an honorable term. Superprogrammers are genuine lone ranger types, happier with machines than with people. The bane of any business manager, the most direction they will accept is "See that target? Hit it!". A manager'd better be quick, because they "Hit it" inordinately soon and immediately look for another target. There were targets aplenty as the SDSS project unfolded, and quite a bit of duplication of effort. Sometimes, there'd be two or three complete pieces of software, and they'd hold a benchmark (AKA face-off) test. The project burned through three Project Managers. But it did get completed.

Hardware doesn't behave. Both the mirrors for the telescope suffered near-catastrophic problems, which required much ingenuity (and extra money) to solve. The environment doesn't behave. Miller moths at the Apache Point site like to hide in tight, dark spaces, so they tend to grease the controls, then jam them.

One complication was money. Jim Gunn's initial guess was about $50,000 for a telescope and CCD camera, and the hope that he could afford lots of disks to hold the data. He was only off by a factor of about 1,000. Just the CCD's for the wide-field camera wound up costing $2 million. So he and the other astronomers learned to get along with the fund-raising establishment as the circle of donors and supporters was widened and widened again to raise $10 million, $20 million…I think the total cost to date approaches $100 million.

The telescope itself came to half a million. This is a story in itself. It isn't a huge telescope by modern standards, with a main mirror diameter of 2.5 m (98"). But a wide-field design requires optics a bit more costly than your typical Meade or Celestron star gazer. The field of view is about three degrees, and the CCD pack that covers this area takes a "picture" of about 200 Mpx. But getting light to focus well across the entire camera is daunting. Like all large-scale professional telescopes, including the Hubble and the two Keck 10m (34-foot) instruments, the design is Ritchey-Chrétien, or R-C. Older instruments with parabolic main mirrors and ordinary Cassegrain optics, such as the Mt. Wilson and Mt. Palomar telescopes, are considered passé.

Let's explore this just a bit. Here I step outside the book because R-C optics were not even mentioned. This diagram shows the classic Cassegrain optics as used at Palomar and in many amateur telescopes:


A parabolic mirror focuses on-axis light to a very sharp point. However, light coming at a slight angle, such as from any other part of an image, is not all focused to a point, but to a teardrop-shaped area. This kind of aberration is called coma. The secondary mirror increases the effective magnification of the system, but does not correct coma. A single-mirror system with an f/12 figure, and a Cassegrain system with an f/4 primary and a 3x secondary, will have exactly the same level of coma distortion in an image.

A Ritchey–Chrétien system looks exactly like a Cassegrain system to the eye, but there is a subtle difference. The primary is not parabolic, but hyperbolic, and the curvature of the secondary, while still hyperbolic, is matched to the primary in a way that eliminates coma. However, the smallest focal point is not quite as sharp as it is with the Cassegrain, but it is still very small, and is the same over a large area, three degrees in the case of the Sloan telescope. As we'll see, the images it produces are quite spectacular, and plenty "sharp enough". And the bottom line? Making that hyperbolic optical system cost much, much more than a parabolic one. But it is just so darn much better for modern astronomical needs that it is used for all professional telescopes since the 1950s.

The power of the SDSS is that all the data are public, and include an almost game-like interface for looking at images and retrieving data. As I noted yesterday, the SDSS maintains a SkyServer web site for viewing images and getting data.


This shows the galaxy known as M109, the 109th object in the Messier Catalog of objects that Charles Messier wanted to remember so he would not confuse them with comets. He was primarily a comet hunter. For this screen shot I opted to show objects that have had their spectra recorded. Note that M109 is not among them. It is much too close. The Sloan survey is intended to look at large-scale structure as outlined by distant galaxies. The Xes mark fourteen galaxies that you can hardly see at this scale. Their spectra tell their distance and other interesting facts. All this can be downloaded, either one-by-one or in whole reams of data.

If you want to "spelunk the sky" and see neat stuff, it isn't hard to do. If you want to do more and be a "Sloan Astronomer" (my term), make sure you have plenty of disk available. The entire SDSS archive currently comes to 40TBy (that's 40,000 GBy). It covers a quarter of the total sky, centered on the North Galactic Pole. Not just (more than!) a million galaxies, but spectral data on hundreds of thousands of stars is also included.

It is the second-largest data repository I know of; Google Earth at about 3 PBy (3,000 TBy) is the largest. But Google Earth provides an even easier way to view the sky. There is a Google Sky option in Google Earth, which uses the Sloan and other sources to map the sky and make it viewable and searchable. It was even quicker for me to find M109 in Google Sky than it was in SkyServer. If you are interested in some particular kind of star or galaxy, SkyServer has suggested projects and tutorials on how to do the work.

The book is a great example of a contemporary history. The author's Reference section begins with a five page list of interviews she conducted. The list of published works is correspondingly shorter. There hasn't been time for many of the principals to decide their experiences warrant a memoir. But it is all the more immediate for it. I imagine our view of the American Revolution would be significantly different if a historian could interview Washington and Adams and others. To me, the establishment of this virtual sky and the tools to easily use it is almost as revolutionary as the founding of a nation. One need not apply a year in advance for two or three observing nights on an overbooked large telescope and hope you don't get rained out.

Thousands of professional journal articles have been authored as a result of the SDSS and its current data release, called DR7. And what do I hope for next? A clone of the Sloan telescope and camera situated in southern Chile, which could double the total sky coverage.

No comments: