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

[THREDDS #OOA-314138]: ncep model run posting schedule

Hi Evan,

> Hello Unidata support,
> I am scripting some volcano ashfall maps to query the following address
> for wind data:
> 
> http://motherlode.ucar.edu:8080/thredds/catalog/fmrc/NCEP/NAM/Alaska_11km/catalog.html?dataset=fmrc/NCEP/NAM/Alaska_11km/NCEP-NAM-Alaska_11km_best.ncd
> 
> The documentation summary says ' Model runs are made at 0, 6, 12, 18Z
> with analysis and forecasts every 3 hours out to 84 hours'. Are the
> models actually posted slightly ahead of that schedule so that queries
> from my scripts on the same schedule actually get the most recent data,
> or should I delay my runs to give your server time to update? If so, by
> how many minutes is sufficient?

For the NAM Alaska 11km data, each model run has generally fully arrived about 
3 hours after the run time. The data starts arriving about an hour before that. 
The "Best Time Series" dataset you are looking at shows all the time steps from 
an arriving dataset returning "missing data" values for those steps that 
haven't yet arrived.

Some of the other NCEP model data are a bit slower to arrive. GFS, for 
instance, is is fully arrived more like 4 or 4.5 hours after the run time. It 
mostly depends on how long the model takes to run and how long it takes to 
post-process the results into the GRIB products we receive.

One more note, if you go to the "File Access" dataset page, it lists all the 
model run datasets and includes the size of the file that's on disk. You can 
check (approximately) if the latest dataset has fully arrived if the size is 
the same as the older datasets.

Hope that helps,

Ethan
 
> Thanks
> Evan
> 
> --
> 
> Evan Thoms
> U.S. Geological Survey, Alaska Science Center
> 4210 University Drive, Anchorage,  AK 99508


Ticket Details
===================
Ticket ID: OOA-314138
Department: Support THREDDS
Priority: High
Status: Closed