Info

Figure 10.3. Subtracting images to find supernovae. The image from a month ago is subtracted From last night's image to reveal a new supernova. The image area shown is about 1/1000 of the full area provided by the CCD camera Rapid processing of dozens of image pairs demands nimble software and a dedicated team of searchers. Courtesy of Brian Schmidt. Australian National University.

Figure 10.3. Subtracting images to find supernovae. The image from a month ago is subtracted From last night's image to reveal a new supernova. The image area shown is about 1/1000 of the full area provided by the CCD camera Rapid processing of dozens of image pairs demands nimble software and a dedicated team of searchers. Courtesy of Brian Schmidt. Australian National University.

curve comes in the first month after maximum, so that's where we concentrate our effort.

But cosmologicai expansion not only shifts spectrum lines to the red, it also slows the ticking of distant clocks, so the radioactive decay that powers distant supernovae appears to run more slowly at high redshift, making our follow-up job a little less urgent.4 For a redshift of one, 30 days elapsed at the supernova corresponds to 60 days for the observer at the telescope. So to find out how fast the supernova declined in the first month after maximum (with time measured at the supernova), you need to observe it several times in the next two months here on Karth.

Kven then you are not done. Supernovae are bright objects, but they are not ordinarily as bright as the galaxies of 100 billion stars in which they erupt. So, even if you are very careful, the host galaxy can add a significant amount of galaxy light to the supernova light you want to measure. You need to subtract the galaxy light. We wait for a year, then come back and take a really good "after" image that will show the galaxy, which, like the Cheshire Cat's smile, is still present after the supernova has faded. This makes the whole process quite sluggish. A supernova that you discover in 1995 needs to be revisited in 1996, and it wouldn't be surprising if it took well into 1997 to pass the strict photometric quality control of Nick Sunt-zeff. So even though we were working diligently starting in 1995, wc didn't have much to say until the end of 1997.

All our discoveries, all the spectra, and most of the light-curve data came from the ground. The Hubble Space Telescope is the most wonderful telescope of our time, but it makes its super-sharp images of only a small patch of the sky. It is an effective tool to search for supernovae only when you are interested in extremely distant galaxies that are cheek-by-jowl in a deep I lubblc field. To search wide areas to a moderate depth, we have used 4-meter telescopes with big CCD cameras at Cerro Tololo and the CFHT on Mauna Kea.

11ST does make beautiful images. It is above the blurring effects of the Earth's atmosphere and it is (now) a nearly perfect optical system. This can help solve the problem of measuring light from a supernova that is on top of a galaxy. The angle between the supernova and the galaxy is often less than 1 arcsecond—about the amount that the Earth's atmosphere blurs the image for both the supernova and the galaxy. We subtract the galaxy light measured later, but the results arc never perfect. We do better by using the Hubble Space Telescope to take a series of pictures of the fading supernova. In each of them, the image of the supernova is a small hard dot, only 1/100 the area of a ground-based star image, and the separation of galaxy light from supernova light is much more precise. Precision matters because we are using the apparent brightness of the supernova to measure the history of cosmic expansion, and the expected effects are small.

But there is a price for using the Hubble Space Telescope. Bureaucracy. The paperwork associated with 11ST observing is somewhere on the scale of personal inconvenience between doing your tax return and enduring a root canal. The Space Telescope is in a low orbit, circling Earth every 90 minutes or so. It operates as a robot—ground control loads a long list of instructions into an onboard computer every week, and then HST plods down its to-do list, moving to the objects of interest, locking on to guide stars, acquiring data with the cameras or spectrographs, and then sending

SN 1997cJ

Hubble Space Telescope

Figure 10-4. HSTand Ground-based images of SN I997cj. Sharp images from die Hubble Space Telescope make accurate measurements of Supernovae much easier Courtesy of Peter Garnavich; University of Notre DamefNASA,

Ground-Based D.7"

Hubble Space Telescope

Figure 10-4. HSTand Ground-based images of SN I997cj. Sharp images from die Hubble Space Telescope make accurate measurements of Supernovae much easier Courtesy of Peter Garnavich; University of Notre DamefNASA, the bits to the ground by radio. Since this intricate dance is taking place without human intervention, the crew at the Space Telescope Science Institute in Baltimore likes to get everything set and checked well in advance. They don't like surprises, and they don't like last-minute changes to that list of instructions. For some reason, they believe the safety of the telescope is more important than leaping immediately to implement our desires.

So their rule is: tell us where you want to observe a month in advance. Now, for ordinary observing with 11ST, this is a reasonable rule. Tt gives the scheduling wizards at the Institute time to build an efficient schedule and to check and double-check the telescope's instructions before they are sent up. Next week's instructions are reviewed during a work week in Baltimore, and the tlSMS Load" goes up to 11ST through NASA's communications network on the weekend.

But what if you want to observe a supernova? How are you going to schedule that a month in advance? We have some experience with that. I am the principal investigator for a program we call SINS (Supernova INtensive Study), whose aim is to use 11ST to learn more about nearby, bright supernovae by using I IST's unique ability to observe in the ultraviolet. We've had gpod success in working with the Space Telescope Science Institute to get new objects into the schedule on short notice, including the nearby SN la SN 1992A.

Hut these "Target of Opportunity" observations have the disruptive quality of a 2 a.m. fire alarm in a college dorm and the Institute, quite rightly, doesn't want too many of them because they demand so much staff attention.

As chief SINner, I have learned over the years that the natural rhythm of the Institute responds best to supernovae reported to them on a Tuesday. Then they can revise the instruction set during the balance of the week, send it up on the weekend and, if you are very lucky, get your target observed as soon as the next Monday. Elapsed time: as little as six days. They don't respond so well to a supernova found on a Thursday night. If you call on Friday morning, they'll say "too late" to change the instructions for next week. At best, they might put you on the docket for a week from next Monday and it might be a week from Sunday. Elapsed time: 11 to 18 days. Does this procedure challenge the limits of human intelligence? Not really. But it is rocket science.

The Space Telescope can't point just anywhere on the sky—it has to avoid the sun, the moon, and especially the Earth, all of which are changing position as seen from a low Earth orbit. So, the problem is to find supernovae at places on the sky and times specified a month ahead, and to report those to the Institute early in the week. This is not quite as crazy as it sounds. Luckily for us, Pete Challis at the CfA used his understanding of the inner workings of the Space Telescope system to puzzle out how to do this. We enlisted Ron Gilliland at the Space Telescope Science Institute to help us solve this problem. Of all the people who use HST, Ron is the most successful in thinking through exactly how the instruments and operations of this complicated machine can be used to do unusual and valuable science.

Here's what we do. The telescope time we get assigned at CTIO or in Hawaii for supernova searches comes through the usual mechanism of Time Allocation Committees and telescope scheduling for big ground-based telescopes. That's done six months in advance. So we know when we are going to find distant supernovae. If we find any at all, we are going to find them right after the nights when we observe. If we have a choice, we try to make the discoveries on a Saturday so we have time to sort things out before we choose our

11ST targets. And we know where we are going to find them. We are going to find them in the fields we observed in last month's dark run and for which we are going to make observations this month. We know where our target fields are. So we know we are going to find supernovae when we have observing time and where we point the telescope.

The search fields are about half a degree across, which is precise enough for the Space Telescope schedule to be constructed. This needs to take into account a zillion technical details. Ones I know of are the timing of the telescope's orbit, the direction to the sun, and the time it takes for the telescope to turn from the previous object to ours, moving at the stately pace of the minute hand of a clock. And there arc lots more T don't have the neurons to know (but Ron Gilliland does). We tell the Institute boffins the precise position of the newly discovered supernovae we want to observe, they insert those details into the schedule, check, double-check, transmit, and execute. Elapsed time: about a week.

Is it worth all this bother? Absolutely. In 1997, the high-2r team pulled this off, discovering supernovae on schedule at CFIIT or Cerro Tololo, getting their spectra at Keck and at the MMT in Arizona, early light curves from the University (if Hawaii's 88-inch telescope, and after delivering the precise target list on a weekday, we obtained a beautiful sequence of observations with 11ST starting one week after the Keck spectra and extending over the next 80 days.1* While our original motivation for using HST was the wonderful imaging that makes photometry more precise, we also benefited from the absence of weather and the fact that moonlight doesn't light up the sky when you are above the atmosphere. The observations took place exactly as planned, which hardly ever happens on the ground, and we could time them in the optimum way to learn about the light-curve shape.

One difficult part of these measurements was making certain that the measurements from 11ST and from the ground agreed. To do this, we carcfully matched ground-based and HST measurements of 15 background stars in the HST images that did not vary, which were bright enough to see from the ground, but not so bright they overwhelmed the HST's CCD detector.

getting it right

0 0

Post a comment