N.B.: This data set is saved for historical purposes only; it is
      not considered to be of archival quality.
 
      The 'raw data' mentioned in this description are saved as PDS data
      set PHB2-M-TS-2-THERM/VIS-IMGEDR-V1.0
 
      Data Set Overview
      =================
      The first part of this description repeats general information
      from the raw data description and the latter part refers
      specifically to this data set.  See the raw data set template for
      more description of the raw data.
 
      In February and March, 1989, the Termoskan instrument on board the
      Phobos '88 spacecraft of the USSR acquired a limited set of very
      high resolution simultaneous observations of the reflected solar
      flux (hereafter referred to as the visible channel) and emitted
      thermal flux (thermal infrared (IR)) from Mars's equatorial
      region.  These are, so far, the highest spatial resolution thermal
      data ever obtained for Mars.  Four slightly overlapping thermal
      panoramas (also called scans or swaths) cover a large portion of
      the equatorial region from 30S to 6N latitude.  Simultaneous
      visible panoramas were taken during each of the four observing
      sessions; due to spacecraft memory limitations, visible channel
      processing was stopped early relative to the thermal channel for 2
      of the sessions (2 and 4). Thus, the visible channel panoramas are
      shorter than the thermal panoramas for these sessions.
 
      The instrument was fixed to the spacecraft with the optical axis
      pointing in the anti-solar direction.  As a consequence, all
      observations are at approximately zero degrees phase angle and
      only daytime observations were acquired.  Scan lines were acquired
      approximately going from North to South on the planet at a rate of
      1 line per second.
 
      In the first session (taken Feb 11, 1989), the periapse altitude
      of the spacecraft's elliptical orbit was 1150 km and the
      resolution at nadir was approximately 300 m per pixel.  The
      thermal and visible channel panoramas from this session exhibit
      longitudinal gaps of varying size between scan lines. Within each
      scan line (acquired in the North-South direction), however, full
      resolution and coverage were maintained.  In the remaining three
      sessions (one taken on March 1, 1989 and two on March 26, 1989),
      the panoramas were acquired from a circular orbit of altitude 6300
      km with a resolution at nadir of approximately 1.8 km per pixel.
      In these panoramas, line and frame scanning correspond; therefore,
      there are not significant gaps between scan lines and geometrical
      distortions primarily occur only because of the sphericity of the
      planet.  The dark West-East bands in observing sessions 3 and 4
      are from the shadow of Phobos.\n Each image consists of 384
      samples.  The number of lines varies depending upon how long the
      instrument was on in any given panorama.  The data is 8 bit data
      with dn values ranging from 0 to 255 for both the thermal and the
      visible channels.  West is towards the top of each image file and
      North is to the right.
 
      There are three major differences between this data set and the
      raw data set  First, the raw data has been stripped of all non-
      image samples, leaving 384 samples.  Second, all panorama
      fragments of the raw data files have been recombined so that each
      of the full length edited files (designated by file names ending
      with ed) contain one entire panorama (either thermal or visible).
      Thus, there are 8 full length files corresponding to the 4
      Termoskan observing sessions.  For ease of display on some
      systems, each of these full length files has also been chopped
      into 512 line fragments (with file names ending in f# where # is
      the fragment number for that panorama).  Note that the last
      fragment file of each panorama may have less than 512 lines
      depending upon the length of the full length file.  Third, data
      from the visible and thermal channels have been aligned so that a
      given line and sample in a thermal image should correspond to the
      same location on Mars as the same line and sample in the
      corresponding visible channel image.
 
      Various actions were required to align the thermal and visible
      data files.  Three factors affected the alignment of lines between
      the thermal and visible files.  First, there were offsets at the
      beginning of files, probably caused by data lines added on Earth.
      This initial offset was determined by comparison of surface
      features near the beginnings of the panoramas.  Then, non-data
      lines were removed from the beginning of the appropriate file.
      Second, there are dropped (missing) data lines in some of the
      thermal data and in some of the visible data.  In most of the raw
      data, no extra lines were added to fill these gaps. Comparison of
      the same surface features in both visible and thermal raw images
      on either side of the dropped lines was used to determine the
      number of lines missing.  This number of black (dn = 0) lines were
      inserted to represent the missing lines in the edited files.
      There were rare occurrences of seemingly superfluous dn = 0 lines
      in the raw data.  These lines were removed.  Within the raw data
      set there are also occasional lines with dn = (a constant value
      not equal to zero).  Most of these lines appeared to preserve
      geometry, so were left in the edited files.
 
      There was also a constant offset between thermal and visible
      samples (North-South direction).  The offset ranged from 2 to 3
      samples but appeared to be constant for any given observing
      session.  This offset was corrected for in the edited data by
      adding either 2 or 3 dn = 0 samples to the beginning of each line
      in the thermal channel.
 
      All thermal files within the edited data set have 388 samples per
      line, with either 2 or 3 leading dn = 0 samples, 384 data samples,
      and either 1 or 2 trailing dn = 0 samples.  All visible channel
      files have 384 samples per line.
 
      The UT start and stop times given in the full length edited file
      labels are times at the spacecraft, not earth receive times.
 
      PDS1 Data Set Parameters
      ========================
      SAMPLING_PARAMETER_NAME = PIXEL
      SAMPLING_PARAMETER_RESOLUTION = 1
      MINIMUM_SAMPLING_PARAMETER = N/A
      MAXIMUM_SAMPLING_PARAMETER = 384
      SAMPLING_PARAMETER_INTERVAL = 1
      MINIMUM_AVAILABLE_SAMPLING_INT = N/A
      SAMPLING_PARAMETER_UNIT = N/A
      DATA_SET_PARAMETER_NAME = DN
      NOISE_LEVEL             = UNK
      DATA_SET_PARAMETER_UNIT = DIMENSIONLESS

Confidence Level Overview
      =========================
      The alignment of thermal and visible edited files should be good
      to within approximately +/- 1 pixel.
 
      Lengths  of the scans derived from the start and stop times and
      the 1 line per second scan rate do not always agree with the
      lengths of the scans, probably due primarily to lines added or
      subtracted at the beginnings and ends of the 'original' (raw)
      files.  However, the magnitude of the difference in all cases is
      less than 5 minutes (300) lines and thus an insignificant error
      for most applications.
 
      In addition, the following confidence notes remain from the raw
      data: noise in every eighth sample of the thermal channel
      amounting to a positive 1 or 2 dn increase (probably tied somehow
      into the calibration that was going on at a similar rate);
      sporadic single pixel spikes that occur occasionally within the
      data; occasional partial lines which appear corrupted; occasional
      whole lines with single dn values that appear to preserve
      geometry, but have no value as data; depending upon the panorama
      either the first or last of the 384 sample pixels is corrupted in
      the thermal channel, leaving at most 383 good samples per line;
      and two bright West-East lines (affecting the same samples in each
      line) in visible panorama 1 (possibly caused by reflections off
      the spacecraft?).  The thermal channel appears to be very well
      calibrated to at least 3 K (Murray, et al. 1991).