The Galileo Shoemaker-Levy 9 Data Set 1. Data Description and Acquisition Method During Shoemaker-Levy 9 (SL9), multiple exposures of Jupiter were acquired on the CCD array by simultaneously slewing the scan platform and activating the shutter while suppressing CCD read-out. The resulting multiple exposure image is called an on-chip mosaic (OCM), and is referred to as the readout frame or playback image. Each of the single exposures within the OCM is called a shutter event. The intent of the SL9 Data Return Strategy was to return a key subset of the acquired data based upon successive refinement of the comet fragment impact timing uncertainties. This goal was achieved through the use of two types of OCM observation patterns, diagonal scan and discrete shutter events, and the use of Jail Bar data playback techniques. These data acquisition approaches are described below. Diagonal Scan: Impacts "K" and "N" were observed with a pattern of 6 25-second diagonal scans, one scan per OCM shuttering, followed by CCD readout in the HCM mode. This pattern used the CCD as a tracing photometer, as the images of Jupiter were deliberately smeared across many pixels. Impact signatures can therefore be observed as brighter pixels in the smear, and at finer time resolution than the fastest 2-1/3 second shuttering mode. The pattern occupies approximately 1/3 of the SSI field of view. The SCLKs of the six playback images for the impact "K" OCMs are: 248806645, 248807000, 248807300, 248807700, 248808045 and 248808400. The SCLKs of the three playback images for the impact "N" OCMs are: 248949900, 248950200 and 248950600. Discrete: Impact "W" was observed with a discrete 8x8 pattern of 2-1/3 second OCM shutterings, followed by CCD readout in the IM4 mode. This pattern leaves (nominally) 40 pixels between each 60 pixel image. Each discrete image of Jupiter represents one point in a 2-1/3 second time series of data points, with one readout (or playback image) providing up to 64 discrete data points. In this manner, the time domain is translated into the spatial domain in the image. Fifty-six of the shutter events are within the field of view (FOV) in each playback image. The SCLKs of the playback images for the four discrete OCMs are: 249220555 (40 shutter events within the FOV played back), 249221000, 249221400 and 249221800. Jail Bar Techniques: The Jail Bar Searches (JBS) were used to locate the actual image shuttering and scanning pattern within each SSI FOV (the locations of which varied from the nominal due to scan platform pointing uncertainties) and also to provide a correspondence between the image data and locations on the spacecraft tape. The JBS guaranteed return of two image lines out of every 80 for the diagonal scans and two lines of 40 for the discrete images. The images returned during JBS were selected based upon ground observations and timing information available from other spacecraft instruments. Based upon the results of the JBSs, the discrete shuttering events were subjected to a Jail Bar Return (JBR) of up to eight 40 to 50 line swaths per frame for the four frames returned. The JBRs for the diagonal scan observations returned four lines (guaranteeing two) out of every eight. See Section 3 ("Special Data Processing") for information about additional data recovery of lines for the diagonal scans. For additional information about the Jail Bar Searches and Returns, consult "EJ-7 Shoemaker-Levy 9 DMS Data Return Strategy", Bruce McLaughlin, May 13,1994, GLL-MDT-94-096.IOM) 2. Using the SL9 Data Set Graphics file: A graphics file corresponding to each playback image has been generated to simplify the identification of shutter event PICNOs within playbacks. The graphics files were generated by contrast enhancing a copy of the playback image to make the target more visible, then overlaying the text information onto the image. Due to the quantity of shutter events within the playbacks, in order to make the graphics file readable, only the last four digits of the PICNOs are used in the graphics file to identify shutter events. Although the files have been called "graphics files" they are still VICAR files with the same data format as the REDR described in the AAREADME.TXT. Each graphics file and its detached PDS label is located in the same directory as the corresponding playback image, and will have essentially the same filename as the playback, except the last character will be a "G" instead of an "R". For example, for the REDR playback image [SL9.C024880]6645R.IMG, the associated graphics file is [SL9.C024880]6645G.IMG and its PDS label is [SL9.C024880]6645G.LBL. Index files: The IMGINDEX.TAB (in the INDEX directory) provides PDS label information for each shutter event and playback image. Only those shutter events within the FOV are included. In the case where shutter event field values varied within a single playback image, values were averaged. Those averaged fields are: TWIST_ANGLE RIGHT_ASCENSION DECLINATION All PDS label formats and documentation are based on the Planetary Data System Data Preparation Workbook (D-7669, May 1991) and the Planetary Science Data Dictionary (D-7116, Rev. B, May 1991). A brief description of the PDS label is described in the PDSLABEL.TXT file. The VICAR label is described in the VICAR2.TXT file. Both text files are located in the DOCUMENT directory. An additional text file, SSIINDEX.TXT (also located in the INDEX directory) has been generated which provides the spacecraft clock rim and mod91, Earth Observation Time (EOT), playback image filename, graphics filename, and a (line,sample) location of each shutter event. The EOT is the time (UTC) that an Earth observer would have witnessed an event on Jupiter that was observed by the spacecraft during the specified shutter event. The EOT is calculated by starting with the time at shutter midpoint, subtracting one-way light time to the subspacecraft point on Jupiter, and adding one-way light time from Jupiter to the Earth. For the diagonal scans, the (line,sample) is a point roughly in the center of the slew, and for the discrete shutter events, the (line,sample) is a point on Jupiter. Signal to Noise Ratio: In order to improve the quality of data returned, the Deep Space Network sometimes used Block V Receivers to capture the data. This improved the quality of data returned, but impacted the reliability of the signal to noise ratio value recorded in the binary line prefix of the REDR images. 3. Special Data Processing To maximize the data return from the diagonal scan playbacks, the telemetry data was massaged so the MIPS Real-Time Subsystem would process embedded lines not normally recovered from HCM playbacks. As a result, the number of returned image lines doubled. A side-effect of this process introduced artifacts which appeared in the form of an entire (formerly blank) line of pixels set to DN value 255. These lines were nullified by setting the affected pixel DN values to 0. Nullified lines are identified in the VICAR label of the affected images. Dominating Reed-Solomon errors also existed in the diagonal scan playback images. At the request of the SSI team, partial lines were nullified to minimize the distracting visual effects of these errors. These nullified partial lines are also recorded in the VICAR label. 4. Special Handling of Ancillary Information Due to the OCM images in the SL9 data set ancillary information required special handling. In the past, the telemetry format for the SSI instrument data matched the telemetry format for the spacecraft, so the spacecraft telemetry format was automatically recorded in the data labels. During the OCM process, the camera operated independently and under a different telemetry format than the spacecraft. In the PDS and VICAR labels for the SL9 data set, the SSI telemetry format is provided. The imaging mode provided by the NAIF mini-e kernel for the first shutter event included in each of the "discrete" playback images had the value of the imaging mode of the playback image (8-2/3), not the shutter event. This value was manually modified to reflect the shutter events' imaging mode (2-1/3). In theory, the exposure duration of the playback images should be 0, but this would hinder subsequent image processing, such as radiometric correction. Therefore the EXPOS_TIME for the playback images has been set to match the exposure time of the shutter events.