[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Status

On Tue, 21 Dec 1999, Ekaterina Radeva wrote:

> Hello Linda and Robb,
> 
> In response to your e-mail, here is where we stand: the LDM software was
> installed on one of our external servers which is frequently used as host
> for outgoing data. Now we would like to run some tests by sending data to
> another IDD server, some of your machines for instance. 

Ekaterina,


It sounds like you have a handle on getting this operation in gear.  Yes,
I think it's a good idea to send data to one of our machines as a test.
The machine will be shemp.unidata.ucar.edu.  You will need to put an ALLOW
statment in your etc/ldmd.conf file for our machine, ie

allow ANY shemp.unidata.ucar.edu

Also, I will need the name of your machine so I can request the GEM model.
> 
> Technicalities
> 
> In general, we would like to send over the IDD system output from our
> regional GEM model in grib format. As a starting point, we propose the
> following subset of the GEM output to be sent to the UNIDATA users. Feel
> free to feedback your suggestions/comments.
> 
> Fields (49 in total):
> 
> - geopotential height at 1000, 850, 700, 500, 400, 250, 150, and 100 hPa
> - air temperature at the same pressure levels + at the model surface 
> - wind components projected on the x- and y- axies of the grid at the
>   same pressure levels + at the model surface 
> - dew-point temperature at the same pressure levels + at the model
>   surface 
> - sea-level pressure 
> - pressure vertical velocity at 700 hPa 
> - surface pressure 
> - accumulated precipitation 
> - instantaneous rainfall rate 
> - total column cloud cover
> 
> Grid:
> 
> A polar-stereographic grid over Canada, the U.S. and the adjacent waters.
> 135x94 points, horizontal resolution of 60km at 60N.
> 
> The regional operational model is run twice per day based on 00Z and 12Z
> data, to produce 48-hour forecast. We would like to send the the 0-48 hour
> forecasts of the above fields from both runs, at a temporal resolution of
> 6 hours. This will make for 882 fields/products with a volume of ~ 17Mb
> per day. This seems reasonable, having in mind that the model output from
> the NMC and ECMWF models you disseminate accounts for 20,000 products and
> 75 MB per day. However, you may have restrictions for experimental data.
> 
> As far as I understood from the documentation, each field is sent as a
> separate file/product so that the users can select the fields they need.
> Is this so, or can we group several fields as one product ? Say, all
> fields for a certain forecast hour ? It seems to me that this will make
> for less trafic.
> 
You are correct 17Mb is not much data for a 24 hour period. I would
suggest that you break the GEM model into products that are the most
useful to the end user that are about 16k apiece. If its by parameter or
time it's up to you.  The only restriction would be not to send the whole
GEM model as one product because if there is sending problem then the
whole product would have to be resent again.  Also, the product name
should reflect the contents and time but don't add the complete directory
structure to the product name. 


> As for GRIB encoding, we use an in-house encoder which is consistent with
> the WMO GRIB (edition 1) standards. I will get your GRIB decoder from the
> UNIDATA site and check whether it works with our GRIB data. Is your GRIB
> decoder decoding into ASCII or some other internal format ? If so, the
> decoding test should probably be run at your site, as we may not be
> familiar with the internal format.
> 
Unidata GRIB decoder "gribtonc" will read WMO GRIB (edition 1) standards
and the output is a netCDF file. If you have a sample GEM GRID file, I
could test it for you or you can install the Unidata decoders package,
that requires installing netCDF, udunits, and the decoders packages. All
the packages are available by ftp.unidata.ucar.edu or their respective web
pages.  Another idea, I could make available a gribdump routine in binary
format for you platform. I don't know the platform type.



> As a simple test, I guess I can use the pqinsert/pqsend utilities to send
> you a few files using the LDM software ? I can use the -f option to
> specify an EXP feedtype. I will also need the address of the remote
> server I am to send the data to. 
> 
For the test, you could use the pqinsert utility to put the product in the
LDM queue and the request from our machine will take care of the sending
part. Yes, use the -f flag with -f EXP.


> I'll appreciate any advice about how to use the LDM, what test we can make
> and how to finish the set-up. Thanks in advance,
> 
If you need any more info or if I confused you on some of the brief
explanations, let me know.

Robb...



> Ekaterina
> 
>  
> 
> ______________________________________________________
> 
> Division de l'implementation et services operationnels
>    Centre meteorologique canadien
> Implementation and Operational Services Division
>    Canadian Meteorological Centre
> 
> e-mail: address@hidden
> tel: (514) 421-4646
> ______________________________________________________
> 
> On Fri, 3 Dec 1999, Linda Miller wrote:
> 
> > Hi Ekaterina,
> > 
> > I wondered if  you've made any progress with the installation and testing of
> > the LDM for distribution of the GEM data to U.S. universities.  I talked 
> > with
> > Robb Kambic of Unidata and he said that he answered an initial email from 
> > you,
> > but we were unsure if you are having problems with it.  Be sure to contact 
> > us
> > if there's anything we can do to assist.
> > 
> > It would be wonderful to get the data moving to university users.
> > 
> > Thanks, in advance, for any update you can provide.
> > 
> > Linda
> > 
> > Linda Miller - address@hidden
> > External Liaison, Unidata
> > University Corporation for Atmospheric Research
> > P.O. Box 3000
> > Boulder, CO 80307-3000
> > 303 497-8646 fax: 303-497-8690
> > URL:  http://www.unidata.ucar.edu/staff/lmiller/un.act.html
> > 
> > 
> 

===============================================================================
Robb Kambic                                Unidata Program Center
Software Engineer III                      Univ. Corp for Atmospheric Research
address@hidden             WWW: http://www.unidata.ucar.edu/
===============================================================================