Re: Ignoring Land in an EOF analysis of SST

[Ansley Manke <ansley@xxxxxxxxxxxxx>]

Hi Andrew,
What version of Ferret are you running?  I see that in the release
notes for version 5.33  there is a note in the bug fixes that "EOF
functions had inconsistency in treatment of bad-data flags."   You
might try explicitly setting the MISSING data flag on your variable
with the command SET VARIABLE/MISSING= before calling the
EOF function just to be sure the flag is being properly understood

The EOF functions are written to deal with places where there is
no data. This example works with both version 5.22 and 5.33.
See if you can run it.

  USE coads_climatology
  SET REGION/X=67w:1w/Y=11S:11N
  LET eofxyfcn = eof_space(sst[X=67w:1w,Y=11S:11N], 0.8)
  set view ul; fill/l=1/title="eof 1" eofxyfcn; go land
  set view ur; fill/l=2/title="eof 2" eofxyfcn; go land

Note that the calling sequence for eof_space was slightly different
prior to version 5.22.

Ansley Manke

gendasza@carleton.edu wrote:

> Hi,
> I'm working to create an EOF analysis in Ferret, but I'm having trouble
> hiding the land values from the eof_space function. My land values are set
> to missing_value and are ordinarily masked in other Ferret plots, but as
> soon as I perform an EOF analysis on a portion of my data including land
> and create a nc file from the eof_space function, there is no data in that
> file. Any help would be greatly appreciated.
> Thanks,
> Andrew Gendaszek