Hi Erik,
If the mean is a straightforward average of the 600 per-second samples
(at 23:10:00, 23:10:01, .... 23:19:59), then this most accurately
represents (according to the "Midpoint Rule") a mean for the period
extending from 23:09:59.5 to 23:19:59.5, so those should be your bounds.
cheers,
Karl
On 3/27/18 12:41 PM, Erik Quaeghebeur wrote:
I think we've been thinking about this wrong. You apparently want to
report the mean for the period 23:10:00 - 23:20:00 from 1-second
samples (sampled on the second). […]
So, your bounds should define the interval represented by the mean
you are reporting (23:10:00 - 23:20:00), and you should estimate that
mean using 601 samples, but weighting the first and last sample half
as much as the others.
I do not compute the average and neither can I influence that. Each
statistic is based on 600 samples. Either 23:10:00 to 23:19:59,
labeled 23:10, or 23:10:01 to 23:20:00, labeled 23:20. So either
closed-open or open-closed if you would use 23:10 and 23:20 as bounds.
Best,
Erik
_______________________________________________
CF-metadata mailing list
[email protected]
http://mailman.cgd.ucar.edu/mailman/listinfo/cf-metadata
