steps1-5-update-notes

Notes on GISTEMP changes between 2008-09-11 and 2009-12-03.
Nick Barnes, Ravenbrook Limited, 2009-12-17

The only things we need to do for this are:

- step1.py needs to get the appropriate ushcn.tbl file, now we are using USHCNv2.

- correct typo in our v2.inv file.

For STEP0 see step0-update-notes

STEP1:

Ts.strange.RSU.list.IN and ushcn.tbl both come from STEP0 now.

There is a typo correction in v2.inv:  40371964000 is called WHITEHORSE, Y, not WITHEHORSE, Y.

The driver script has changed to use more consistent directories for
input, output, and temporary files.

STEP2:

A typo correction in v2.inv, matching the one in STEP1:  40371964000 is called WHITEHORSE, Y, not WITHEHORSE, Y.  Note: the v2.inv files are identical.

Directory renames for intermediate files.

text_to_binary.f has a check for special station IDs in Antartica,
which Reto explains thusly:

    The modification of text_to_binary.f has nothing to do with the
    current input files and has no effect whatsoever (no IDs start with
    90); it should never have been in the public version and it is not
    there anymore.  What happened is that somebody wanted us to use some
    gridded Antarctica data that he somehow obtained. To minimize the
    changes in our code, I treated each grid box as a station whose ID
    started with 90 and which contained lat and lon of that grid
    box. Since I did not want to add many new lines to the inventory file
    v2.inv, I had to slightly modify the various programs to add the
    missing information.

So we don't need to do anything with that change.

STEP3:

Directory renames for intermediate files.

to.SBBXgrid.f:
  236c236
  <          DO 110 N=1,NCM
  ---
  >          DO 110 NC=1,NCM

This corrects a bug in the handling of boxes which contain no
stations: in that case sub-box records are written out for all the
sub-boxes containing only BAD values.  The previous code incorrectly
put identical lat/long coordinates in each written sub-box record.  In
the Python we don't currently bother with an "early bath" in this
case, so the original bug didn't affect us.

STEP4_5:

Fixed the ocean-weighting bug we found in SBBXotoBX.f