Re: MC netCDF datasets

[Jonathan Callahan <callahan@xxxxxxxxxxxxx>]

Jean-Louis,

Jonathan Callahan here, developer of the new LAS.

If you're using the latest release of LAS (called LAS2 and written in
perl), everything you want to do is available through configuration of
the server.  The interface can display a dataset with many variables,
each of which lives in a separate file.  When the interface sends a
request to the server, the variable sent in can have a particular
dataset name override the one that was sent in.

An example:

On our demonstration server

http://ferret.wrc.noaa.gov/las-bin/LiveAccess

there is a dataset titled 

    ETOPO topography/bathymetry

with variables titled

    60 minute resolution
    20 minute resolution
     5 minute resolution


The file which configures the web interface, TMAP_DATA.js, specifies

etopo = new Dataset("etopo",
  "ETOPO topography/bathymetry",
  null,
  "/las-doc/data_documents/etopo.html",
  null, null);

with (etopo) {
  aV("rose60", "60 minute resolution", "XY", "m");
  aV("rose20", "20 minute resolution", "XY", "m");
  aV("rose05", " 5 minute resolution", "XY", "m");
}


The file which configures the server, TMAP_DATA.pl, specifies

#########################################
# etopo
#########################################

$dset_list{"etopo"}{"default"}{"land_type"}     = "contour";
$dset_list{"etopo"}{"default"}{"palette"}     = "dark_land_sea";
$dset_list{"etopo"}{"default"}{"fill_levels"}  = 
"(-10500,-5500,5000)(-5500,0,500)(0,2000,250)(2000,6000,1000)";
$dset_list{"etopo"}{"rose60"}{"dataset_name"}     = "etopo60";
$dset_list{"etopo"}{"rose60"}{"variable_name"}     = "rose";
$dset_list{"etopo"}{"rose20"}{"dataset_name"}     = "etopo20";
$dset_list{"etopo"}{"rose20"}{"variable_name"}     = "rose";
$dset_list{"etopo"}{"rose20"}{"land_type"}     = "none";
$dset_list{"etopo"}{"rose05"}{"dataset_name"}     = "etopo5";
$dset_list{"etopo"}{"rose05"}{"variable_name"}     = "rose";
$dset_list{"etopo"}{"rose05"}{"land_type"}     = "none";
$dset_list{"etopo"}{"rose05"}{"fill_levels"}  = "20c";


As you can see, I've used the "dataset_name" and "variable_name"
properties to convert the incoming request for
(dataset=etopo,variable=rose60) into a Ferret request for
(dataset=etopo60,variable=rose).

You can use this mechanism to create an interface which displays a
single dataset with 21 variables, yet associate separate dataset names
wich each of your variables.

I hope that helps.

And I'd LOVE to take a look at your server when you get it up and
running.  Our goal is to put LAS2 sites in touch with each other so that
they can ultimately access each other's data.

Salut!


-- Jonathan Callahan



> Hi Ferret users,
> 
> I use NCEP reanalysis monthly means files from noaa/cdc.
> The filenames are as follows :
>         air.mon.mean.nc 
>         hgt.mon.mean.nc 
>         omega.mon.mean.nc 
>         pottmp.mon.mean.nc 
>         ...
> (21 variables).
> 
> Each file contains 1 scientific variable.
> All files share a common time axis which spans from 1958 to 1999, with a timestep
> of 1 month.
> 
> I would like to build a Multi netCDF dataset with all these files, to allow
> my Live Access Server to point to it as a single dataset with 21 variables.
> 
> I have browsed the manual, the FAQ, the mail archive, the .des sample files included in the
> distribution and also the gt_des_annot.txt . 
> 
> I am still unable to understand how to build the .des file.
> Are MC datasets useful only for concatenation along the time axis ?
> 
> Can anyone help me ?