Forecast Model Run Datasets; was: More comlicated data structure

Hi Lukas:

You might want to look at this:

The first part talks about these kinds of Forecast Model Run Dataset; the 
remaining is about aggregation that may not be useful.

Also, the CF group is debating now whether a convention is needed for this 
case. If you are following that group, you might want to put your opinion in.

Lukas Lehner wrote:
Hi Don

Thank you for quick and satisfactory answer :)
More stuff below.

Don Murray napsal(a):

Hi  Lukas-

Lukas Lehner wrote:

The new version 2.2 (beta) solved some of my issues, but not all of them. All of the datasets i can find are basically grids put on timeline. Also CF convention and other sources are pretty much based on this scheme.

Which is why we took that route in the IDV. ;-)

Yet is there any way to make dataset - and view it with IDV of course - for prediction environment as this?
time axe - actual time of prediction
horizont axe - time step in prediction
2D or 3D grid.

Are you talking about having a 2D time dimension - model run time
and forecast hour?  Some work has been done on this at Unidata to
view collections of model data where you could say:

- show me all forecasts that are valid at a particular time
- show me all the  12 hour forecasts for the collection.

However, this is a fundamental change in the way we handle
time in the IDV and is going to take some work to get that
functionality in the  IDV.

I see....
So I have redesign the archive.

One can make this kind of structure quite easily , but to view it with common tools seems impossible.

What would this structure look like to you in a netCDF file?

like this

Dataset {
    String time[time = 881];
    Float32 t2[time = 881][rdim = 79][dimi = 67][dimj = 79];

time is run time
rdim is forecast hour
them just the grid


This feels to me like usual situation in storing prediction data.

The default view of the data is to show the valid time, not
the model run+forecast hour.

Don Murray                               UCAR Unidata Program
dmurray@xxxxxxxxxxxxxxxx                        P.O. Box 3000
(303) 497-8628                              Boulder, CO 80307

To unsubscribe idvusers, visit:

  • 2007 messages navigation, sorted by:
    1. Thread
    2. Subject
    3. Author
    4. Date
    5. ↑ Table Of Contents
  • Search the netcdf-java archives: