Hi Joerg,
Special dataset or variable commands are going to be applied using a
special script just for this dataset. In the new Armstrong versions of
LAS, the scripts are set up so that the initialization steps are in
LAS_initialize_data.jnl, which does these things:
- initialize the dataset (set data dataset_name)
- call any special initialization scripts needed for this
particular data.
- initialize the variable (apply region information and possibly
transformations)
The CANCEL AXIS/CALENDAR command would appear in the special
initialization script.
In the dataset configuration XML, the name of the initialization script
is a Ferret
property called init_script, so a dataset or an individual variable
might have the property given this way, where you would define your own
name, of course.
<properties>
<ferret>
<init_script>longdata_initialize_data</init_script>
</ferret>
</properties>
and the script sst_initialize_data.jnl would contain, say, the
following. The symbol data_var will have been already defined by the
call to the dataset initialization script.
! longdata_initialize_data.jnl
! This script is to initalize my long-time-series data from the dataset
xxx
CANCEL AXIS/CALENDAR `($data_var),return=taxis`
EXIT
! End of longdata_initialize_data.jnl ----------------------
This kind of init script is where the SET MEMORY command can be used to
ask for more Ferret memory if the dataset is large; If an axis needs
fixing with Ferret commands, or say a depth attribute is missing, that
can be added, and so on.
Ansley
Joerg Franke wrote:
Hi,
I think it might be a good idea to put CANCEL
AXIS/CALENDAR `var,return=taxis` in the ferret init scripts.
As I have no experience in editing the init scripts, can you please
tell me, where I should include this command. I tried it after SET DATA
($data_url) in
/opt/las.armstrong.1.0/JavaSource/resources/ferret/scripts/LAS_initialize_dataset.jnl
but that was not changing anything.
Regards,
Jörg
Manke wrote:
Hi Joerg -
We're thinking that perhaps a CANCEL AXIS/CALENDAR `var,return=taxis`
might just do the job; this would need to go in the Ferret init script.
Ansley Manke wrote:
Ferret has a hard limit for calendar time
axes of 4-digit years. It can deal with an axis that's not a calendar
axis, but just T=0 to 50000, but not formatted as dates. I'm not sure
how/whether this can be done with descriptor files or whether there's a
way to fit this in to a LAS configuration.
Ansley
Joerg Franke wrote:
Hi LAS experts,
I want to manage data for the past 50000 years in the new amstrong
version of the live access server. They are 10yr averages which make
5000 timesteps in a 4D netcdf file. Now I am searching for a nice way
to adjust the "Select time" menu. In the old LAS 6.5 I manipulated
/las_servlet/jakarta/webapps/las/constrains.vm to show the timestep in
the dropdown menue multiplied by 10. That worked alright, although in a
time series plot I had the wrong dates at the time axis of course.
If I write a descriptor file for my netcdf data in amstrong now:
...
&STEPFILE_RECORD
\hspace*{2ex} S_FILENAME = 'PATH and name of first netCDF
file'
S_AUX_SET_NUM = 0
S_START = 00000
S_END = 09990
S_DELTA = 10
S_NUM_OF_FILES = 1
S_REGVARFLAG = ' '
/
&STEPFILE_RECORD
S_FILENAME = 'PATH and name of second netCDF file'\newline
S_AUX_SET_NUM = 0
S_START = 10000
S_END = 19990
S_DELTA = 10
S_NUM_OF_FILES = 1
S_REGVARFLAG = ' '
...
I manage to see the year I want to see in the dropdown menue of LAS and
also to plot lon/lat maps for every time step, but time series plots
which go beyond 9999 crash.
Does anybody has an idea how to solve that problem?
Thanks for any help.
Cheers - Jörg
|