I have noticed that our best days for generating solar electricity are those when thin, high cloud develops in the late afternoon. At that time our panels are shaded from direct solar rays, and thin cloud can increase the indirect illumination.
The perfect generation conditions would not be a clear sky, but a layer of thin, bright cloud, with one small hole allowing direct sunlight to fall on the panels. Obviously these conditions will occur only rarely, and then, only very locally. But a quick estimate suggests that in these admittedly unusual conditions, a PV installation could generate up to 50% more power than under a perfectly clear sky.
You need to assume that the ideal thin cloud layer scatters sunlight equally in all downward or upward directions. (Thick cloud will absorb some radiation and scatter most of the rest upwards.) You also need to assume that the panels convert oblique rays efficiently.
Even patches of thin cloud - the kind of weather we are getting right now (in East Sussex) -could result in higher peak generation, though the average output would be reduced because, in reality, the clouds are sometimes going to get between the sun and the panels.
Has anyone investigated this phenomenon? Does it have any practical consequences for PV system design or operation? Am I right in thinking solar panels are rated strictly according to their 'clear sky' peak generation, or do manufacturers allow themselves a little extra for the possible concentrating effects of off-axis clouds?

Have I understood correctly...

In the late afternoon your panels are shaded from direct sun by a tree or similar? In which case not entirely surprising that output increases if high clouds scatter light around the tree onto the panels.