Data Set Overview : Details of the observations and the data processing used are contained in DONES_ETAL_UNPUBLISHED-2004.PDF (DONESETAL2004) which can be found in the DOCUMENT subdirectory.  This data set contains images of the Saturn system from the Wisconsin-Indiana-Yale- NOAO(WIYN) using the WIYN Imager in late November 1995. Saturn system was observed on five half-nights, 19- 23 November 1995 (The second half of each night, after Saturn set, was used to search a selected sample of X-ray binary stars, Sterzik et al 1997). The Sun crossed the ring plane during 17-21 November 1995. The very low solar elevation angle, combined with the use of coronagraphic masks, helped to reduce scattered light from the main rings.  Every image showing Saturn, its rings or the region of the inner satellites has been included in this data set, regardless of the original intended purpose.  The WIYN observing program whose results are included in this dataset is was headed by:  PI: Richard H. Durisen (Indiana University)  The bulk of the following information has been extracted from DONES_ETAL_UNPUBLISHED-2004.PDF (DONESETAL2004) which can be found in the DOCUMENT subdirectory.  Objectives : The original goals of the observations were to determine the radial extent and color of Saturn's faint E and, possibly, G, Rings, and to investigate the main rings and determine the orbits of ring moons such as Prometheus and Pandora. However, the high signal-to-noise obtained for the small moons Helene, Telesto, and Calypso made it clear that the images were well- suited to detecting previously unknown small moons.  Observation summary and conditions :  The Instrument used to study the Saturn Ring Plane Crossing event was a special application of the WIYN Imager(WI). The WI consists of a Filter/Shutter Assembly (FSA) that mounts on the Instrument Adapter Subassembly (IAS) at the Nasmyth focus of the WIYN telescope, plus the CCD detector. The IAS provides for target acquisition, autoguiding, and for the wavefront sensing that is used to adjust the active optics of the WIYN telescope. The CCD used in this work was a backside-illuminated STIS 2048 x 2048 pixel detector with 21-micron square pixels.  As noted above, the team observed the Saturn system during five nights, 19-23 November, 1995. This was around the time when the Sun crossed Saturn's ring plane, so the rings were especially dark. The rings were edge-on to the Sun, and were open to Earth by 2.7 degrees. The phase angle was 5.5 degrees. Saturn was only up until around midnight local time.  A total of 165 science images were obtained. Most observations of the Mimas-Hyperion region were taken in the R filter to maximize signal-to-noise. However, observations of satellites interior to Mimas were taken through the 890-nm methane filter to minimize scattered light from Saturn. Exposure times ranged from 5 seconds to 5 minutes, with 1-2 minutes being typical.  The night of 19 November 1995 was devoted to instrumental checkout and the next four nights were devoted to Saturn observations. The night of 22 November 1995 was devoted exclusively to narrowband methane imaging of the main rings and inner satellites; this work will be described elsewhere. On the other three nights (20, 21, and 23 November), the team took both broadband and narrowband frames.  The team recognized early on the night of the 20th that the tiny satellites Helene, Telesto and Calypso were extremely bright in the images. These bodies have radii of 10-16 km. This raised their awareness that, in principle, they should be able to detect moons in the data that are substantially smaller than 10 km. For this reason, they devoted much of the 20th, 21st and 23rd to long exposures of the region outside the main rings. Typical exposure times were 30--140 seconds and the R filter was emphasized because it had the finest sensitivity.  NOTE: Inspite of careful image processing, ghosts from previous images remain in some images.  Parameters : The PDS label for each file contains a broad variety of additional parameters enabling the user to determine image geometry and to convert pixel values to physically meaningful quantities.  Data : The data provided here are images in FITS format. For each data file, a detached PDS label is provided containing additional parameters.  Ancillary Data : Additional calibration files are provided to assist in the analysis and interpretation of the data.  As mentioned above, the appropriate subdirectories under the CALIBRATION subdirectory contain bias and flat field files.  Coordinate System : All geometric quantities appearing in the labels are in J2000 coordinates. In this coordinate frame, the z-axis points northward along the Earth's J2000 rotation axis and the x-axis points toward the First Point of Aries.   Media/Format : This data set is archived on DVD media. Organization and formats are according to PDS and ISO 9660 level 2 standards.  Most binary data files are in least-significant-byte first, which is the native format for PCs and Digital workstations. Users of Suns and other workstations may need to swap bytes in some data files before use. Note that the software tools provided on this volume swap the bytes automatically if this is necessary.

Confidence Level Overview :  Inspite of careful image processing, ghosts from previous images remain in some images.  Calibration of the images involved several steps. For the known satellites and for three stars that repeated from image to image, the brightness was integrated and divided by each image's exposure time, yielding values of DN/sec for each target. DN is the integer data number found in each pixel of an image. These measurements were obtained from the original, unfiltered images to ensure that the filtering had not truncated the DN sums. In each case, a region surrounding the target body was used to determine the local background, which was then subtracted from the DN summation. Although this calibration carried out for the R, B and V filters, it became apparent quickly that the R filter was by far the most sensitive, and the other filters were excluded from further analysis.  From these individual determinations, the team then did a series of careful comparisons to verify that the visible moons and stars maintained constant brightness ratios. Because the moons are elongated, their brightness from one night to the next could change significantly, but this could be roughly matched to their known axial ratios (Thomas etal 1983) and this could be further complicated because one or more of the stars might be variable. The brightest star, identified as S1 and as 5249-00969 in the Hubble Space Telescope's Guide Star Catalog, was indeed found to vary significantly on time scales of hours. As a result, it was eliminated from further consideration as a calibration reference.  In the end, after allowing for the non-sphericity of Helene, Telesto and Calypso, all three moons were found to have very constant brightness ratios relative to the next brightest stars, S2 and S3. They chose S2 as their reference because it is the brighter of the two. Scaling to the dimensions of the three moons, the team inferred that S2 is equivalent in brightness to a moon of 21 km radius. The primary source of uncertainty arises from the fact that the three moons have different albedos, providing a size uncertainty of ~20%.  The other piece of the question is how bright an object needs to be in order to say with confidence that it should not have escaped notice in the search procedure. For this estimate, the team generated images in which false moons were added to the data prior to the filtering procedure. These simulated moons have the same point spread function as the actual moons and stars, which is found to be well matched by a two-dimensional Gaussian with a standard deviation of 1.34 pixels. They concluded that a moon with total DN : 3000 would be consistently visible, even if it falls rather close to the planet. Further out where the noise is lower, a total DN : 2000 should be detectable.  These can be translated into sizes assuming that they match the other moons in albedo. As noted above, S2 is the equivalent of a 21-km moon. The instrument records at least 1200 DN/sec for this star, corresponding to 144,000 DN in the minimum exposure time for satellite search images of 120 sec. This is can be interpolated to a 3000-DN moon in a 120 second exposure corresponds to a radius of 3.0 km and a 2000-DN moon corresponds to a radius of 2.5 km. These detection limits fall well below those of previous ground-based searches for moons in the Saturn system (Kuiper 1961, Baron and Elliot 1983).  Review : This data set passed peer review on 1/25/2010. The members of the peer review panel were J. Bauer, L. Dones and C. Olkin.  Data Coverage and Quality : The team have demonstrated that unknown moons larger than 2.5--3.0 km in radius should have been detectable in the data set, yet none were. The question remains of how thorough the orbital coverage was during the three nights of observing.  The team suggest that the minimum requirement for detection is that a moon appear in three different images. For each image, they tabulated the range of distances from the planet where a moon could fall without being obscured by the occulting mask or one of the known moons. They then generated a large set of hypothetical moons, orbiting Saturn on circular, equatorial orbits, positioned at radial intervals of 2000 km in semimajor axis and 2 degrees in initial longitude. Each simulated moon is advanced along its orbit and the number of images in which it should have been visible is determined. If that number exceeds three, then the team assume that a moon 3 km or larger should have been detected. By counting how many of the 180 moons at each orbital radius should have been detected, we determine the detection probability.  The requirement for detecting a 2.5-km moon is more stringent. For these tinier moons, it is necessary that a minimum of three detections all fall at least 300,000 km from the center of the planet. This is the rough boundary where the background noise and variations settle down to a low enough level that a 2.5-km moon could be reliably detected.  Fractional coverage fluctuates as a function of radius, with typical values in the 60-80% range between Enceladus and Titan. The overall coverage between the orbits of Mimas and Hyperion is 71% for 3-km moons and 57% for 2.5-km moons. While the possibility remains that the team missed a few moons above the detection threshold, they feel it is unlikely that they missed several. If the fractional coverage is f, then the probability of missing N moons is (1-f)^N. Thus, the probability of missing one 3-km moon is 29%; for two, the number drops to 8.4%; for three, it drops to 2.8%. For 2.5-km moons, the probabilities are 43% that one was missed, 19% that two were missed , and 8% that three were missed. The team concluded that the inner Saturn system beyond the orbit of Enceladus is relatively free of undiscovered moons 2.5-3 km or larger in radius.   Citing this dataset : The following is the recommended information to include in a journal citation of this dataset: Dones, L., M.R. Showalter, R.H. Durisen, R.K. Honeycutt, J.S. Jurcevic, R. Tripoli, and C. Strom, and D. Olson WIYN Observations of the November 1995 Saturn Ring Plane Crossing, WIYN-S-WI-2-RPX-V1.0, USA_NASA_PDS_RPX_0401, NASA Planetary Data System, 2010.