Synthetic Hourly Generation errors

[Helios210]

When using SolarGIS prospect data and the PVsyst synethetic generation algorithm I frequently get:

"December has a great discrepancy of time shift with respect to the average."

This might make sense if I was using hourly data from SolarGIS but this is hourly data that PVsyst itself has generated!

Looking at the best clear days graph for the site clearly shows something wrong for December but I don't know how to fix this.

Anyone else noticed similar problems/behaviour?

H.

Edit 1:

Added pictures.

Edited December 8, 2020 by Helios210

[Helios210]

Investigating further it seems generating from MN7 data imported within PVsyst does not present this problem, the graph for best clear days' for December is much better but there is still a significant timeshift and no warning is given this time (??) - see the attached pictures.

I have noticed that PVsyst seems to add MN7 data to imported datasets that don't include Linke Turbidity etc. however it leaves GHI and DIFF alone, so I would not expect this to have such a dramatic effect in a single month, though perhaps I am missing something...