SCIENTIFIC ADVISORY COUNCIL OF THE
CANADA-FRANCE-HAWAII TELESCOPE

The SAC held an e-mail discussion on Feb. 3 - 7, 1997. The agenda of this discussion was:

1. Scientific drivers for UV wide-field imaging
2. Scientific need for an atmospheric dispersion corrector for Megacam
3. Agenda for the next SAC meeting (May 1997)

1. Scientific Drivers for UV Wide-field Imaging

There is a strong scientific case for UV wide-field imaging. By UV here, we mean down to at least 370-360nm, and possibly below. This scientific case is summarized in a report appended to this text.

All efforts should be made to provide such a capability with good throughput for Megacam. In particular, the specs for the CCDs should include a reasonable QE down to at least 370nm.

There seems to be technical solutions, for instance based on hybrid BK-7 / fused silica for the corrector. The quantum efficiency of the RGO 2kx4k EEV chip is also quite encouraging.

The access to the UV should not compromise significantly the quality of the images and the throughput at redder wavelengths. We feel that we do not have enough information, both on technical and on scientific grounds to provide a more quantitative guideline. We request such information when available (i.e. after the Megacam science meeting for the scientific part) to review this issue.

We will also need a better understanding of the cost of the UV access (including corrector and CCD), and will need to review the project when more information is available.

2. Scientific Need for an Atmospheric Dispersion Corrector for Megacam

We have identified some scientific merits for an ADC for Megacam (for instance pushing the limiting magnitude for point source work at U and B bands), but they are marginal. Most of the science programs which require an exquisite image quality are in the red where the effects of atmospheric dispersion are smaller than the expected image quality, while the programs identified so far requiring imaging in the B and U band, which would benefit from an ADC, are mostly of the photometric type on extended objects, and in general do not require the best image quality.

However, Megacam is supposed to be still operational in 10 years from now, and it is difficult to decide what observations will be needed by then. It is possible that excellent image quality in B or U will be required by future scientific programs, unidentified today.

On the technical side, we still need further studies to determine what level of degradation (if any) is produced by a properly designed ADC in the red wavelength range. We also need a proper cost estimate for the ADC, including its optics, mechanics and control system, as well as an estimate of the savings that an ADC will bring by facilitating the guiding.

We therefore conclude that it is desirable to keep the option for an ADC in the preliminary design of the corrector, but that this option is of lower scientific priority. The final decision to include it or not will depend on one hand upon the feasibility of an efficient ADC with unsignificant impact of imaging at red wavelengths, and on the other hand upon budgetary constraints.

3. Agenda for the Next SAC Meeting (May 1997)

The next SAC meeting will be held in Toulouse, on May 15,16,17, 1997. The preliminary agenda for this meeting is:

1. Director's report
2. Megacam:
   • report on the Toronto meeting and discussion on scientific programs of Megacam
   • technical report and discussions on major choices
3. KIR and OSIS detector developments: current status
4. OASIS acceptance
5. AOB/Argus: status of the project
6. spectropolarimeter: status of the project
7. block scheduling
8. long term future of CFHT
9. Any other bussiness

Appendix: Interim Science Considerations for Megacam Performance: Jan 1997

Note: this text was prepared by one member of the SAC. Although the entire SAC basically agrees on the arguments presented here, this text must not be considered as SAC's official and definitive position.

Since the Megacam requires a new wide field corrector to produce good images over ~1 degree field, and also be capable of fast guiding and automatic focus, it is important to examine the science drivers for coverage below the B-band. It is intended that the science workshop in May will address this in more detail, but the technical studies require a starting point, and also the science planning needs to know what performance is technically feasible and also within the budget.

The following are the scientific reasons for wanting to observe shortward of 4000A in wide field programs.

1. A major new breakthrough in cosmology has the been the discovery of a population of galaxies at redshift 3 - 3.5 and higher. The key signature of these objects is the Lyman discontinuity at 912A, and the most effective means of discovering these galaxies is by looking for "U-dropout" objects. The technique, pioneered by Steidel & Hamilton (AJ, 105, 2017) requires multi-band imaging (they used UGR) with one band as short as possible in wavelength. Steidel et al used a U_N band centered on 3570A with a passband of 700A, some 100A bluer than the Johnson U. The Lyman break of a z = 3 galaxy occurs at 3650A, at z = 3.5, it is at 4100A. It is already becoming clear that the number of galaxies is declining beyond z = 3.5 so to detect substantial numbers of high z galaxies to measure their clustering properties (a critical test of cosmological models) the long wavelength of ultraviolet filter should be less than about 4000A, yet yield substantial flux since the galaxies are faint. A standard U filter (3550A, 700A wide) is obviously best but a filter that covers the the range from 4000A down to whatever reasonably common optical glasses transmit (~3500-3600A) is acceptable. It should be emphasized that superb image quality at the lowest wavelengths is not required, but substantial flux below 4000A is essential.
2. Photometric redshifts have recently been recognized as extremely important tools for virtually all types of galaxy surveys. The critical requirements are twofold (Gwyn, 1996): long wavelength coverage and good signal to noise. Gwyn's simulations and CFRS data indicate that for z < 0.7 U, B, V or R and I are essential; U is critical (similarly, for 0.7 < z < 1.3, BVIK are required). With good signal to noise, redshifts with accuracies of less than 0.2 are easily obtained. In addition to pre-selecting samples of galaxies and/or clusters for spectroscopic follow-up, Brainerd et al. 1995 have recently shown that photometric redshifts can raise the galaxy-galaxy lensing signal by an order of magnitude, allowing velocity dispersions and masses of galaxies, including their dark haloes, to be determined to unprecedented accuracy.

   Although the above 2 kinds of imaging surveys are already being pursued, Megacam with good image quality and field size would be a strong player in this field, with 8m class telescopes for spectroscopic follow-up. This will be a major research area for the next decade.

3. Another break of major current interest is the 4000A break in stellar spectra. This is due to the line spectrum of different types of star, and the quantitative measure of the discontinuity is a powerful method for estimating the age of stellar populations. A large amount of work on galaxies of all kinds - from local dwarfs, to starburst galaxies, hosts of AGN, to distant cluster and field galaxies, all use this measure. The initial imaging surveys that Megacam would conduct would need to cover this break. In objects of redshift 0.1 or higher, the B-band will cover it, but there are major programs that will address objects at redshift lower than this.

   The connection between stellar populations and the evolution of galaxies has become a major topic of current research. Much of it awaits large area surveys and the 8m telescope spectroscopy that will make Megacam a key instrument for its years of operation. This thus requires a band of a few hundred angstroms reaching to 3700A or lower.

4. All surveys for hot objects benefit greatly from short wavelength coverage, since flux rises steeply to shorter wavelength, while that from old stars drops, particularly below the Balmer discontinuity at 3650A. This class of objects includes QSOs and active galaxies of all kinds, knots of star-formation, white dwarfs, cataclysmic variables and X-ray sources. Megacam will be a key instrument for such surveys and really needs some coverage below the B-band. At redshifts to 0.1 or so, this lever becomes more powerful, since we observe further into rest frame UV. At higher redshifts, many of the objects become too faint to be interesting, even with this natural gain, except for bright galaxies themselves, as covered in item 1 and 4 below.

We first consider the actual atmospheric dispersion from Mauna Kea. The extreme case that seems likely is the Galactic centre, at -29deg. The dispersion rises strongly towards shorter wavelengths. We are unlikely to have optics that transmit the full U-band, so I suggest an extreme bandpass to consider is 3500-4000. We then have the following results for total dispersion across the image, for a square bandpass filter. The angles are the declination, HA is the hour angle, and the numbers are in arcsec.

Hence, the wider bandpass of the B filter makes it the limiting case. With actual non-square bandpass cutoffs, these numbers will be reduced by about 20%.

We now consider the image quality we expect without atmospheric dispersion. The fast guiding of Megacam will essentially remove telescope tracking jitter, but do very little to correct atmospheric distortions over this large field. Experience with HRCam and CFHT statistics indicate that we should expect best images at 0.4 arcsec FWHM. These will occur in V and R bands. In B and U they will be 0.1 to 0.2 worse.

This means our worst case for atmospheric dispersion is in B-band on the Galactic centre, 2 hours off the meridian. Here we will be convolving 0.5 arcsec images with the same amount of atmospheric dispersion - images will be 0.7 x 0.5 arcsec. At V-band in this situation we will have 0.4 x 0.46, and in R you will not notice the atmospheric effects. In any case there may be variations of isoplanicity across the field of order 0.1 arcsec.

Finally, we need a realistic estimate of how well the ADC will work. There ar e optical surface figuring errors, mounting errors, and errors in the movement of the ADC optics in use. It seems likely that these will combine to produce their own distortions (that are hard to trace and model), of order 0.1 arcsec or more, at all wavelengths. Thus, the ADC may actually worsen images in bands V and longer, for all projects.

Thus, the need for an ADC should be considered in the light of specific science programs. In R-band, and probably V-band, there is probably zero (or negative) gain from an ADC. Do programs that require undistorted images also require U or B-band filters? Given that B and U-band images are worse anyway, will they be used for flux measures on images whose structure is primarily derived from longer wavelengths? For example, studies of the galactic centre are unlikely to be done at short wavelengths due to the heavy obscuration .

Virtually all extragalactic studies will likely be done either at the zenith or near the equator to provide access to both N and S 8m telescopes.

Here is the generic list of SAC May 1996 with comments on which programs require UV coverage or ADC. The list is based on the Megacam group science proposals, and the discussions at SAC and with the Megacam group.

Each program or set of projects are qualified by W, S, L, U, A, Z, G indicating whether they benefit significantly from:

Where these are uncertain or of lower priority they are given in parentheses

Notes on the science programs:

1. This covers many different programs, but all are aimed at a full understanding of the star-formation and evolution history of our own galaxy, and the nature of dark matter. This will be a major arena of work in the next few years.

2. It will be possible to search for low mass components of the haloes of M31 and M33 by microlensing. This will provide a direct comparison with the results for our own galaxy halo now being accumulated, and relate to galaxy formation and history.

3,4 These programs will seek to understand the stellar populations and astration processes in the general galaxy population at the current (z=0) epoch.

5,6 There are studies of the outer stellar and cluster populations in galaxies and clusters of galaxies. These trace the history of star-formation and merging of galaxies. Current results show great promise but lack statistics of sufficient numbers. Dust content of galaxies by measuring background galaxies.

7. The study of diffuse stellar populations is another aspect of understanding star-formation processes and their effects on galaxy histories.

8,9 These studes will seek to understand the history and workings of galaxy clusters, which are the largest bound systems known, and probably the oldest. Cosmic evolution of these structures will be studied to lookback time fractions of up to ~50%.

10 This is similar to 8,9 but the study of field galaxies which do not belong to major structures. Thus their formation and history is different and traces undisturbed galaxy evolution.

11 The statistics of supernovae which can be seen to large lookback times, offers a very powerful way to study the distance scale and cosmology. This is already under way, but requires better instrumentation to make significant progress.

12. Weak lensing is a powerful way to study the distribution of all matter (including invisible matter) to very large scales and distances. Excellent images over wide fields are essential.

13,14 There are surveys which will give us the statistics on the earliest times of galaxy formation, and the cosmic evolution of nuclear activity. Present results are very exciting but lack statistics of large umnbers. XMM survey follow-up for large numbers of distant clusters.

15. The search for objects in the outer solar system bears on our understanding of the formation of planetary systems in general. Observations of young stars are a link between our own and other systems.