templates and fixed forecast times for different runs

[Bill Bua]

Austin --

See below.

Bill

Austin Conaty wrote:
> |
> |Bill,
> I almost understand what you're trying
> to do.  Just a few more questions and
> then hopefully I'll get the full picture.
>
> In your original message, you indicate that you have a bunch of
> files
>
> pgbf06.%y4%m2%d2%h2 (for six-hour forecasts)
> pgbf12.%y4%m2%d2%h2 (for 12-hour forecasts)
> pgbf18.%y4%m2%d2%h2 (for 18-hour forecasts)
> all the way up to
> pgfb180.%y4%m2%d2%h2
Yes, and the values after the "f" in pgbf are forecast lead times.
>
> So it seems to me you must have a separate .ctl and .idx file
> for each case (06 up to 180).
Yes, I've done that.  I ran into trouble right away with f06 files in
template form so I stopped there.  By trouble, I mean, plan view maps
that look like scrambled eggs rather than anything resembling a real
state variable field.
> It seems also you would need to
> run gribmap on each case and create a uniquely named .idx file
> for each case.
Did that as well.
>
> Do you have separate .ctl files AND separate .idx files
> for each case?  Have you run gribmap for each case?
>
For f00 (analysis) worked fine.  For f06, not so much ... see "scrambled
eggs" comment above.
> Have you tried changing the %h2 to %f2 in the template string?
That merely increments the forecast run cycle time up to 800*6 = 4800 at
the end of the forecast initial time.  Doesn't work either,
unfortunately.  I am wondering if there is some option after gribmap I
need to use.
>
>
>
>
> Bill Bua wrote:
>> Hey Austin --
>>
>> (long time no see!)
>>
>> Anyway, I want to be able to examine the forecast error at a fixed
>> forecast
>> time from different runs, and look at that error in a time series
>> based on
>> the forecast INITIAL time, rather than the forecast VALID time),
>> compared to
>> the analysis for the GFS (I'll use f00 for the forecast times).  The
>> first
>> forecast was from 00z24Jan2008, the last will be the latest forecast
>> (hence
>> the large number of times in tdef, and the 6 hour time interval
>> representing
>> the time interval between forecast initial times).
>>
>> Wouldn't I want ONE .ctl and .idx file with the template set up to
>> look at
>> the initial time for each forecast run?  I figured this would allow
>> me to
>> look at all the variables at a particular lead time (f06, f12, f18,
>> f24, f30
>> and so on) sequentially by forecast run times, so that I could get my
>> time
>> series of forecasts to compare to the verification (and then get my RMS
>> error time series).
>>
>> Does that help explain what I'm attempting to do?
>>
>> Thanks for helping me clarify ...
>>
>> Bill
>>
>>