[gradsusr] Still chasing the elusive leap-year...

Christopher Gilroy chris.gilroy at gmail.com
Sat Nov 21 01:14:24 EST 2015


Alright, so I've been trying to wrap my head around plotting with
leap-years in general... this is against 6-hour climo data from 1981-2010,
so each DDhhZ is a tdef over 29 years.

1.) Even if I had a climo dataset that didn't have leap-years come february
(during a leap-year) you'd have to start including leap-year data around
february 15th and keep using the leap-year climo until march 1st, no?
Considering on february 15th it's going to start having the leap-year day
at the end of the run getting closer and closer each subsequent run?

2.) I tried sitting down with CDO tonight and while I did: cdo
delete,month=2,day=29 ifile ofile all that seemed to do is literally strip
feb 29th out of the actual grib2 ofile... g2ctl ofile >ofile.ctl still has
tdefs for leap-year? Obviously they would spit out undefined...

3.) Regardless of HOW I can get about the actual grib and control file I'm
really confused on how something like this works, on Nov 21 00z run you
would need to do something like this, no? ave(hgt500mb,t=1297,t=43228,1460),
since 1297 is the day/run value for Nov 21 00z. That then brings up how do
you a.) figure out what the current dates run t= starts off as and b.) you
would need a loop or something to increment said t=1297, right? *This is
one of the things I really just don't get how that would need to work if
you were plotting Nov 21 00z's run. Nov 21 00z hour 00 = day value 1297,
Nov 21 00z's hour 06 = 1298, Nov 21 00z's hour 12 = 1299, etc.

4.) Is there anyway to simply (using CDO or other means) "insert" feb 29
with all 0 values into each year? That would at least allow you to make
1460 always be 1464 without fumbling with other oddities like having two
climo datasets, or symlinkig crazy things, etc.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://gradsusr.org/pipermail/gradsusr/attachments/20151121/bade25d8/attachment-0001.html 


More information about the gradsusr mailing list