My English is very bad.
I'm just trying to show that a comparison of shelters with a single measurement every 5 minutes is no good. It's too random.
This method is sufficient for stable weather conditions.
Look at the fluctuations in temperature (minimum, average, maximum) and radiation in the 10-minute interval on Helix.
With a measurement every 5 minutes (i.e. 1 measurement every 8 minutes) it's impossible to know whether this measurement is before/during/after the fluctuation.
If we add to this the ‘personality’ of each shelter (slower, faster, etc.), we have too many uncertainties to draw any conclusions using this method.
A simple example, 3 different days, data from the same sensor (Pro3 SHT45) but Ecowitt export from 2 different gateways.
It's easy to see the problem of consistency due to the single value every 5 minutes.
Have a nice weekend.
PS: please, no polemics.

[ You are not allowed to view attachments ]