ISCCP data on the Mass Store.

  The Stage C2 data from ISCCP have been archived from July 1983
through June 1988, which gives five full years of monthly analyses.
The data include monthly means for 00, 03, 06, 09, 12, 15, 18 and 21 UTC, 
as well as a complete monthly mean obtained by averaging the
eight ``hour-monthly" means.  The total number of days of observations
used in constructing the average values is recorded as a variable.
Hour-monthly mean values which consist of less than three daily
observations were excluded from the complete monthly mean.
  Some adjustments were applied to selected parameters in the C2 data,
and it is  useful to briefly summarize them.  The adjustments have all
been made to the general version of the C2 data available to the
research community.  They are not adjustments made specifically for
this archive.  Parameters that have been adjusted are so marked in the
table which follows.  A more detailed discussion of the adjustments and the
Stage C2 data can be found in the  data set desiption of Rossow and
Walker (1991).
  As discussed in section 2.2, cloud amount for nighttime conditions can
only be obtained from IR radiances, while daytime cloud conditions are
obtained from combined IR and VIS radiance data.  This is especially
important in regions of low cloud where IR radiances are generally
insensitive and, thus, the combined VIS/IR estimates are superior.
Similarly for cloud top temperatures and pressures, it is possible to
adjust the IR radiance data to be consistent with the value of cloud
optical thickness retrieved from the VIS data.  This adjustment is
labeled ``A1" in the following table and is significant only for
optically thin clouds which falsely appear to be warmer and at higher
pressures since they transmit IR radiation from below.
  A second adjustment is derived from the mean differences between the
VIS/IR and IR only daytime values of total cloud, cloud-top pressure
and cloud-top temperature.  The mean differences are used to adjust
the nighttime results by linearly interpolating between dusk and dawn
values.  The interpolated values are then added to the nighttime IR
radiances.  This adjustment is labeled ``A2" in the following table.
Values of cloud-optical thickness are also interpolated over the
period between dusk and dawn values (A3).  Rossow and Walker report
that these ajustments are generally small.  For cloud amounts, they
are nearly uniformly distributed over the globe (see Fig. 12) and are
largest in marine stratus regimes and in low latitudes.  Cloud-top pressure corrections are positive where low clouds predominate and
are negative where there are high thin clouds.
  Rossow et al. (1987) desibe the procedures used to normalize the
radiances from different satellites over the data collection
period.  However, small residual differences left after the
normalization procedure are amplified when physical quantities are
retrieved.  An attempt to correct for these biases has been made and
is labeled as ``A4".  Essentially, differences in overlapping
measurements are compared with reference values from a polar orbiter.
The adjusted parameters and the range of the adjustments are cloud-top
temperature (+/- 2.5 K), surface temperature (+/- 3.0 K), cloud
optical thickness and water path (+/- 0.08%), and surface visible
reflectance (+/- 8%).  A second calibration adjustment (A5) is
made to the surface reflectance because the spectral response of the
METEOSAT ``visible" channel differs from the other radiometers used in
the analysis.  More details are given in Rossow and Walker (1991).
  Finally, before the hour-monthly means are combined to form the
complete monthly mean, adjustments are made to some parameters to
correct for diurnal sampling biases.  This adjustment is denoted
``A6" in the following table.  An incomplete sample is defined to be less
than four hour-monthly values in polar regions and less than eight
hour-monthly observations elsewhere.  The effects of sub-sampling are
determined using zonally averaged variations of the variables in local
time from locations with all eight hour-monthly mean values available.
Obviously, this adjustment affects only the complete monthly mean
estimates and not the hour-monthly means.
  The six aforementioned adjustments have been applied to the Stage C2
ISCCP data generally available to the research community (i.e., prior
to the archival desibed in this note).  In the Stage C2 data,
special count values were also included to flag-derived values that
were nonphysical.  Nonphysical values result from input data errors
and model errors.  In the ISCCP analyses archived here, the
nonphysical values have been removed and replaced with a missing
flag (1.0E+36).

  A climatology for each month has been established from the five
archived years of ISCCP data.  Only the complete monthly mean data
have been averaged.  The averaged variables have the same processor
names as in the archived monthly data, but not all variables have been
included in the climatologies. (See the listing of Time Average Tapes for
the specific variables)

Additional information on this dataset may be found in:

NCAR Tech Note TN-371, "Monthly Mean Satellite Data Sets Available in CCM
        History Tape Format", Hurrell and Campbell