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

20030715: UDUNITS "hours since 1-1-1"



Bob,

> To: address@hidden
> From: Bob Haxo <address@hidden>
> Subject: "hours since 1-1-1".
> Organization: Carnegie Institution of Washington
> Keywords: UDUNITS

The above message contained the following:

> I read in the UDUNITS known problems that versions prior to 1.10 incorrectly 
> handle "hours since 1-1-1".  I have installed 1.11.7 and am extracting "hours 
> since 1-1-1" from a NetCDF file, and comparing the results of the udunits 
> utility with the results of the IDL CDF_EPOCH routine.  With "17522904 hours 
> since 1-1-1" I am finding discrepancies between the IDL routine, the udunits 
> utility and with the FERRET program.  (I like the output of the Ferret 
> program ... Jan 1, 2000 ... so I'm believing it :-).
> 
> Has the unidata code been fixed, or perhaps is the IDL routine known to be in 
> error?  

Unfortunately, the date "1-1-1" is ambiguous in that it doesn't
represent a fixed, unique point in time; rather, the time in question
depends on what calendar system or systems you choose to use and how you
choose to apply them.  As a consequence, different time-handling systems
can legitimately arrive at the different conclusions you encountered.

This is definitely true for dates prior to October 15, 1582 (when the
Gregorian calendar was adopted in Rome) and can be true for later dates
as well because the Gregorian calendar was adopted at different times in
different countries.

Your best bet would be to convince the provider of the data to use a
more recent date as the temporal origin.  Personally, I like
"2001-01-01".

> Thanks .... bob
> 
> -- 
> Bob Haxo
> Carnegie Institution of Washington,
> Department of Global Ecology
> Stanford, CA
> 
> address@hidden
> 1.650.325.1521 x367

Regards,
Steve Emmerson