Hi again,
I was just looking into this and I think we will be able to offer it
in the next release. It has been requested before on occasion and
we've been working to allow users to write data from datasets,
preserving the characteristics of the original dataset.
To continue with what we do for file variables, by default Ferret
will write coordinate data using the data type of the coordinate
axis from the file. This will include subsets of data writing
subsets of original axes. User-defined axes will be written in
double precision by default. For any axis, the data type used to
write it may be changed with a SET AXIS/OUTTYPE command.
The SET LIST/OUTTYPE and SAVE/OUTTYPE qualifiers that will be new to
Ferret v6.83 will still apply only to non-coordinate variables.
-Ansley
On 10/23/2012 9:51 AM, Ansley Manke
wrote:
Hi
We don't offer a setting for the coordinate variables, even with
the (upcoming) release of Ferret v6.83. As it says in the
documentation you point to here, "Coordinate data is always
written in double precision". The SET LIST/OUTTYPE and
SAVE/OUTTYPE will apply only to file variables. (My apologies
that the documentation got ahead of the Ferret releases.)
In Ferret v6.82 and previous versions, you can control the output
type, still only for file variables, with SET VAR/OUTTYPE.
We would really need separate syntax to offer control over the
type of coordinate variable data. Even if you wanted floating
point data for a file variable, it might be important to list the
coordinate data using double precision; if for instance it
represents time in minutes for a number of years.
Ansley
On 10/23/2012 3:29 AM, Yu, Hao-Cheng
wrote:
Dear ferret user
I also notice "set list/outtype=float" seems start to support
from v6.83
(at http://ferret.pmel.noaa.gov/Ferret/documentation/users-guide/commands-reference/SAVE#SAVE_OUTPUT_TYPE)
However, the latest version seems is 6.82?
Is this means I can't change type right now?
Regards, dan
2012/10/23 Yu, Hao-Cheng <danish.url@xxxxxxxxx>
Dear all
ferret users,
I have to sub-sample from a netcdf file and keeps its type
as original file.
The file contains 2 variables with time axis which has
"float" type.
I use script like following:
----------------------------
use input.nc
CANCEL MODE UPCASE_OUTPUT
set list/outtype=float
list/KEEP_AXISNAMES/i=320:482/j=294:428/form=cdf/file=test.nc/clob lon,lat,dlwrf,dswrf
-----------------------------------------------------
All variable saves with "float" type but not for "time",
it's "double".
Any ideas to easily convert this will be appreciated!
Regards, dan
|