Gridded Humidity Fields
      in the
      DSS Archive
      D. Joseph
      8 Feb 89

      NMC humidity fields for atmospheric layers were not handled
      properly in the DSS archiving procedures for the period 1983
      through 1987. DSS duplicate elimination programs erroneously
      eliminated some of the layer humidity fields which were not
      duplicates. The results of this problem are summarized
      below. There is still a fairly complete set of humidity ana-
      lyses available from the archive.

      MISSING HUMIDITY FIELDS

      The only humidity fields available on the NMC 65x65 and the
      DSS derived 47x51 grids are layer humidities. These are
      represented as relative humidity for the boundary layer,
      bottom 1/3 of the troposphere, and middle 1/3 of the tropo-
      sphere. In the ON84 format these grids have a Q of 88. The S
      value is 144 for boundary layer and 145 for the other two
      layers. The surface values when S is 145 are .333 and .667
      for middle 1/3 and .667 and 1. for the bottom 1/3. Unfor-
      tunately, the sort and dupe elimination programs only exam-
      ined the first surface field and truncated the value to an
      integer which resulted in both layers with S of 145 having
      identical IDs. Since the sort randomly orders identical IDs,
      either of the two layers might be eliminated by the dupe
      elim. The sort problem was fixed before the dupe elim was
      fixed. This results in a predictable layer being present
      after that date. The boundary layer is unaffected.

      A similar situation exists in DSS holdings for the NMC glo-
      bal 2.5 degree grids. One addition layer for the bottom 2/3
      of the troposphere is present in these grids (surface values
      of .333 and 1.). Only one of the three levels will be
      present during the problem periods. Note that the global
      grids include humidities at pressure levels and neither
      these nor the boundary layer humidity are affected by this
      problem.

      The following table summarizes the dates for which fields
      are missing.

       NMC 65x65 and 47x51 grids.

         Jan 83 through Sep 86, only one random layer (of two) is present.
         Oct 86 through Dec 87, only the lower 1/3 layer is present.

       NMC Global grids.

         Jan 83 through Feb 87, only one random layer (of three) is present.










                                 - 2 -


         Mar 87 through Oct 87, only lower 1/3 layer is present.

      SOURCES FOR HUMIDITY FIELDS

      The humidity layer fields that are missing from DSS archives
      should be present in the NCDC holdings, but a large number
      of volumes would need to be read to recover these fields.
      Since boundary layer values are available for the 65x65 and
      global grids and values at pressure levels (1000mb through
      300mb) are available on the global grids, a humidity subset
      could be developed from these. This subset might include
      precipitable water and possibly some temperature fields.

      Relative humidity is also available on pressure levels from
      the the DSS holdings of ECMWF WMO analyses.