Tom Droege is completing the Mark IV camera system, designed to address several of the constraints of the Mark III. The Mark IV will contain 2048x2048 CCDs, can be pointed anywhere in declination, and can track for about 2 hours near the meridian. Improved optics will give finer resolution and deeper imaging capability.
In order to use these cameras effectively, we need a survey plan. Given below are details of the hardware constraints, plus a description of one such plan.
| CCD | Loral CCD-442A |
| Active Pixels | 2032 x 2032 |
| Full well | 300,000 e- |
| MPP Full well | 80,000 e- |
| Gain | 2.89e-/ADU |
| Read Noise | 14e- |
| Dark Current | 35e-/pix/sec |
| MPP Dark Current | 0.35e-/pix/sec |
| Quantum Efficiency | 0.35 |
| Pixel Size | 15 microns |
| Diameter | 100mm |
| Focal Length | 400mm |
| f ratio | f/4.0 |
| bandpass | 400-1000nm |
| resolution | 10micron 80percent ECC |
| Pixel size | 7.5 arcsec |
| Field of View | 4.23 x 4.23 degrees |
| Dec range | -30 to +90 degrees |
| HA range | +-1hr |
| Tracking accuracy | 0.3 arcsec |
| Dec slew rate | 12deg/sec |
| Dec step size | 3.6arcmin/step |
| HA slew rate | 1deg/sec |
| HA step size | 0.9arcsec/step |
While non-MPP mode is tempting due to the larger full well, I think the decreased dark current of the MPP mode is a much larger consideration, and I recommend only using MPP mode for the survey.
Each Mark IV system will have either two or four cameras (VI or BVRI). The initial concept (July 8, 1998) is to have 5 BVRI and 10 VI sites, with two BVRI sites each in the north and south located at good photometric sites. Most of the VI sites will be in the North as is the case for the Mark III survey cameras, but it would be good to have several VI cameras in the south.
Hipparcos/Tycho. This satellite surved the sky in the early 1990's, with the Hipparcos mission providing milliarcsec astrometric precision for 100K stars and the Tycho mission giving BV photometry for 1M stars, complete to V=10.5. There is currently a reprocessing effort to extend Tycho to a V=11 completeness, with about 2M stars in the final catalog. Note, however, that photometry for stars fainter than V=10 is very crude, with errors greater than 0.1mag.
Guide Star Catalog (version 1.2). The GSC was a photographic survey that gave positions and magnitudes for all stars down to about V=14. The astrometry is reasonably good, but the photometry is typically +-0.3mag and is a mix of B, V, and R.
USNO-A2.0. This is a photographic catalog based on the POSS-I blue and red plates. The astrometry is reasonably good, but the magnitudes can have large errors and are not on any standard system.
There are several upcoming surveys. Most of the GRB rapid response systems using their nightly data to compile catalogs of known objects, so that they can determine in real time if a new object appears in a given field. An example is the LOTIS system, where a bank of cameras survey the sky four times nightly down to roughly V=15. Note, however, that these surveys know that they need to go faint and so are unfiltered (wide R-like response). The Sloan Digital Sky Survey (SDSS) is currently underway, and will provide 5-color photometry from V=14+ to V=20 or so over 1/4 of the sky. QUEST, using the Venezuela Schmidt, will handle +-6 degrees of the equator in 3 passbands. LONEOS and other asteroid search systems are again wide-band R, but deeper than the really wide-field systems like LOTIS. There are some small area surveys, like the Macho groups, covering specific areas of the sky (in this case, the galactic bulge) on a nightly basis.
Since this will be a photometric survey, we should use Landolt standards as our primary references. We should use photometric nights for the master catalog and use nonphotometric nights for either finding variables or other programs.
The Mark IV is a pointed/tracked system, unlike the Mark III. This means that drift scanning will not be used. The shortest exposures that can be used are limited by the shutter to about 10 seconds (100ms open/close + errors); the longest exposures are limited by the sky background and tracking accuracy. Tom's initial frames indicated a sky background of about 45 e-/s/pixel. This means that Tom should be able to make 1000-second exposures and only fill pixels to half-well. Tracking errors are unknown at this time.
Michael Richmond did a signal-to-noise calculation for typical Mark IV parameters. I've modified the results somewhat for the exposure times considered above, by placing a bright limit assuming the star's flux is peaked into 4 pixels and assuming a signal-to-noise of 10 for the faint limit (and using Tom's initial Mark IV frames as a reality check). A table of the predicted V-band magnitudes is given below.
| Exposure | Suburban | Dark | ||
|---|---|---|---|---|
| Seconds | Bright | Faint | Bright | Faint |
| 10 | 7.2 | 11.5 | 7.2 | 12.5 |
| 100 | 9.7 | 14.0 | 9.7 | 15.0 |
| 1000 | 12.2 | 15.5 | 12.2 | 16.5 |
The current read time for a camera is 50 seconds. Since this is also about the time necessary to move the camera to a new field, it makes sense to overlap the readout/move operation after every exposure to maximize the efficiency. While exposures of roughly 900 seconds plus a rewind would follow the motion of the sky at the equator, once you progress north or south of the equator you would have to wait many minutes for the next consecutive image in a given declination zone to be positioned correctly. This means a strip-scan survey is not the most efficient method to use the Mark IV.
I went through and calculated a sky map, where each pointing is on a 4-degree grid. This means that frames overlap about 7 arcmin on each edge, which should be sufficient. This sky map is very similar to the POSS-I (6 degree centers) and POSS-II (5 degree centers) maps, but since the spacing is less, therre are more fields to cover (2600 in fact). The entire sky map is available on my ftp site (ftp://ftp.nofs.navy.mil/pub/outgoing/aah/tass/skymap.txt).
I suggest that no observation be made very far from the meridian, and at no larger airmass than X=2. This provides very good overlap between northern and southern sites. For Flagstaff, this means that 1786 fields would be observable. You might want to place limits such that there is adequate overlap between the north and south, but not so much overlap that the survey efficiency goes down. An example would be to only cover down to the -4 degree zone from the north and up to the +4 degree zone from the south, for a total of 1435 zones per site, with a three-zone overlap. However, my guess is that there will be many fewer southern sites than northern ones, and therefore you might want the northern sites to work a little closer to their southern horizon.
The 10-second survey is the quick-and-dirty survey to get our feet wet (and yet is scientifically interesting). Since each exposure plus move is about one minute, every field that transits should be obtainable on a dark night. Therefore, this part of the total survey need only be done a night or two per month, and after six months, the sky will have rotated enough to overlap the start of observations.
The 100-second survey will take about three times as many nights. It should be doable under any moonlight conditions, and should be completable in under a year.
The 1000-second survey is just a wish. It depends greatly on the tracking ability of the mount. My guess is that it would be easier to use the 100-second survey to set frame zero points than to take time to observe Landolt standards. This is another step removed from a proper photometric survey, but means you can use some non-photometric weather to take frames, and therefore complete sky coverage that much earlier. From a site like Flagstaff, you would be able to cover the entire sky about twice per year if all conditions were exactly right.
This covers the photometric nights well, but what should we do with the non-photometric nights? I suggest a survey for discovering variable stars. The 100-second exposure time is about the most convenient setting to maximize sky coverage, yet get reasonable time resolution. You could either hit one region of the sky as often as possible, or try to cover the entire sky. My preference is the latter, since with 15 sites active, each sky field is likely to be covered several times per month already.
I would suggest only the 'dark' sites attempt the 1000-second exposures, and then only when the moon is not visible. Therefore, one possible strategy would be:
The 1000-second exposures look for faint variables, and together with the high-precision 100-second photometric survey, can extend that survey about 2.5 magnitudes fainter. The 100-second exposures during non-photometric weather will look for variables down to about V=12, and give higher time resolution than the longer exposures.
After the first year, the 10-second survey should be complete and those nights that would normally be used for that survey can be reassigned to other projects.
The Tycho-catalog equivalent will be the output of the 10-second survey, and should be available about 6 months after the start of the survey. It will test the photometric and astrometric pipelines, and will be a rapid method of advertising the survey. It avoids the problem of blended images and crowded fields that will have to be addressed for the later segments of the survey. One suggested output format would be something similar to USNO-A, in that you divide the catalog into 7.5degree zones, and for every star in a given zone, list its RA/DEC (from Tycho), B,V,R,I magnitudes and errors, and sources for each of the magnitudes (since you might want to add the Bright Star Catalog to the product to make it complete, for example). Since Tycho contains 1M stars, and the TychoII catalog contains perhaps 2M stars, this output product would be about 64MB in size and could be made available over the Web site.
The GSC-equivalent requires the 100-second survey, which might take a year to complete. This is a much larger catalog, and would require about a full CDROM. Therefore, while you could place it on the Web site, it might be logical to create a CDROM for distribution.
The variable star survey would be a dynamic one, and I'd suggest placing the output on the Web site and continually update it as new variables are identified.
Arne Henden / USNO / aah@nofs.navy.mil