Ferret vs. NCEP dates

Alec,

>Date: Fri, 12 Dec 1997 12:16:24 +0200 (SAT) 
>From: Alec Joubert <address@hidden>
>Organization: University of the Witwatersrand    
>To: Steve EMMERSON <address@hidden>
>Subject: success! 
>Keywords: 199712121031.DAA14038 

In the above message, you wrote:

> Thanks for all the trouble over the last few weeks. The code I was
> given by the folk at NCEP was missing an integer function called "lnblnk".
> Adding it to the code made every bit of difference  - code compiles and 
> runs happily.

Great!

> 
> May I put a query about how netcdf and ferret handles dates past you:

The handling of dates lies outside the purview of the netCDF package.
One package that does handle dates is the UDUNITS package (also from us).

> 
> I have run the NCEP read code to print dates from the first dataset.
> Having printed all the time-steps in the file, the output indicates the
> following:
>                 year month day hour
> time-step    1: 1973     1   1    0
> time-step 1460: 1973    12  31   18 (ncdump says there are 1460 times)
> 
> But throughout ferret,
> 
> cont/t="01-jan-1973:00:00"    gives the first date as: 03-jan-1973 00:00
> and
> cont/t="31-dec-1973:18:00"    gives the last date as: correct
> 
> Why does the first date differ for ferret?

I'm afraid I can't answer that question as I know next to nothing about
Ferret.  I suggest that you contact the Ferret people.

> 
> Thanks
> 
> Alec
>   
> -------------------------------------------------------------------------------
> Dr Alec Joubert                   
> Climatology Research Group         Tel:   +27 11 716 2998
> University of the Witwatersrand    Fax    +27 11 716 3161
> P.O. Wits 2050                     Email: address@hidden    
> SOUTH AFRICA                       WWW:   http://crg.bpb.wits.ac.za
> -------------------------------------------------------------------------------
> 

--------
Steve Emmerson   <http://www.unidata.ucar.edu>