I analysed the problem with the solar information as found in the datafiles of Segur-Le-Chateau and I must conclude that several switches of the
DataLogInterval have been made over the years. This leads to an interesting effect on the data where I assumed a fixed step for the integration process but apparently it is a variable step.
I thought I was safe by taking the value of that parameter from the Cumulus.ini file, but that is clearly a wrong assumption. One can not assume anything about the sampling frequency as
it can be changed, so it will be changed (any CMX user will do that one time or more - except me

). The basic reason probably will be checking the effect on file size but that why is unimportant.
Anyway, I will adjust the algorithm to a dynamic determination of the sampling frequency to use a variable step, even if it will be fixed most of the time.
This adjustment will also account for data gaps (which occur while in maintenance or hardware failure). If any gap larger than 30 minutes occurs (the max sampling frequency one can choose) the data is skipped until sampling restarts (with whatever frequency has been chosen). If a shorter gap exists the data is extrapolated.
I will release a beta2 when ready.
The above may also have an effect on the wind sampling used for the WindRose btw, if so I'll get back on that one.
NB: Interesting effect of the way the user can influence the data. I wonder how this effects the results in the SQL database and the calculations done with those numbers. Anybody can shed a light on that one? Post here or send me a PM please! All data and graphs look great but that does not mean they are valid
