Re: Descriptor files

[Steve Hankin <hankin@xxxxxxxxxxxxx>]

Hi Richard,

In version 5.22 of Ferret the files that are associated together into a time
series by an MC descriptor must each internally contain valid netCDF time
axes. Ferret performs sanity checks to ensure that the time steps in each
file match the values that you set in the STEPFILE records of your descriptor
file. It sounds like these may be mismatched in  the case that you describe.

In Ferret V5.3 BETA this restriction has been relaxed, enabling you to
associate together files which lack time axes internally or have incorrect
ones. (Also support for aggregating DODS URLs.)  However, neither V5.22 nor
V5.3 can handle 28,000 files in a single MC data set.  My recollection is
that the limit on number of files is 500  (I am on the road at the moment and
do not have ready access to source code.)

    - steve

===========

Richard Dixon wrote:

> Dear Ferret Users,
>
> A relatively new person to ferret here!
>
> I've created some descriptor files to concatenate 50 years of NCEP
> re-analysis dat - some 28,000 12-hourly data files. It loads into ferret
> fine and looking at files is also fine - however when I try to manipulate
> the data (for example just take the square of the field), I often get the
> error:
>
> **TMAP ERR: 1st timestep in datafile different from descriptor
>             Expected: 12344 Found 0
>
> It would seem that this is just a "warning" rather than an error, and I
> fear that it may well be slowing down calculations.
>
> Does anyone know how to solve this problem?
>
> Cheers
> Richard