Hi Ryo,
The seconds you specify are being used; this is just an issue
with the default display of calendar time. Try changing the
seconds you ask for, and the result should also change. Here's a
quick example.
yes? define axis/t=1-jan-2020:4-jan-2020:10/units=seconds taxis
yes? let func = randu(L[gt=taxis])
yes? list
func[t="2-jan-2020:10:15:10":"2-jan-2020:10:30:05"@ave]
VARIABLE : RANDU(L[GT=TAXIS])
TIME : 02-JAN-2020 10:15 to 02-JAN-2020 10:30
(averaged)
0.4963
yes? list
func[t="2-jan-2020:10:15:10":"2-jan-2020:10:30:55"@ave]
VARIABLE : RANDU(L[GT=TAXIS])
TIME : 02-JAN-2020 10:15 to 02-JAN-2020 10:30
(averaged)
0.4896
Actually, there's a setting that allows for the whole time spec
to be shown in the listing. MODE CALENDAR turns off or on a
setting for whether the times shown in listings and on plot labels
as a calendar date/time or as time coordinates, and when turned
on, it lets you set the resolution used to list the date and time
yes? show mode calendar
MODE STATE ARGUMENT
CALENDAR SET minutes
yes? set mode calendar:seconds
yes? list
func[t="2-jan-2020:10:15:10":"2-jan-2020:10:30:05"@ave]
VARIABLE : RANDU(L[GT=TAXIS])
TIME : 02-JAN-2020
10:15:10 to 02-JAN-2020 10:30:05
(averaged)
0.4963
Ansley
Dear Ferreters,
I was wondering how the seconds are handled in the date-time specification?
In the example below, I get from LIST
TIME : 04-AUG-2015 03:06 to 18-AUG-2015 00:26
as the averaging period whereas I specify "2015-08-04T03:06:40":"2015-08-18T00:26:40"@AVE . Is this just a display issue? Or are the seconds actually ignored?
It's not that the seconds matter in my particular calculation at hand. I'd just like to know.
I'm working with colleagues who just use seconds for their time axis and when I mechanically converted their time range to a date-time range, I got these date-times.
Regards,
Ryo-----set data tmp.nc ! !<-- I can send this small file on request.
let t1 = "2015-08-04T03:06:40"
let t2 = "2015-08-18T00:26:40"
let slpmean = slp[t="`t1`":"`t2`"@AVE]
list slpmean