I did another experiment to investigate how cooling the solar panels improves production.
I built this contraption to cool the panels down:
It is a popup lawn sprinkler mounted on the side of the roof and attached via a garden hose.
Here is todays production til about 4:45PM
The blue and red lines are the DC watts from each of the 2 different strings. The green line is the total output in AC watts ( after a small 4-5% inverter loss ).
5 times between 10:30am and about 1pm I turned on the sprinkler for several minutes, then once again at 3:45pm.
The red string is 6 panels on one part of the roof and 4 on another part – and the sprinkler only hits 6 of them. The blue string has all 10 panels on the main roof and the sprinkler can cool them all. The red string has a minor improvement, but the blue string is much better.
This chart includes a purple line that shows what the production would have been if the panels didn’t receive any water cooling.
The blue string peaked out at a 15.7% improvement from this line when it was being cooled ( significant variation can be attributed to fine tuning the sprinkler to not miss any panels )
This chart includes an orange line that is a prediction of what the improvement would have been if both strings had each received the full cooling.
The light blue line is a prediction of what output would have been all day if the cooling could have been applied consistently and continuously all day long.
High Speed Charging 
Great data, thanks for doing the experiment,
I wonder what effect water droplets on the panel surface and moisture evaporating off them had; if it blocked and/or scattered some small percentage of photons that otherwise would have fallen on the panels? And still total power output was boosted by a significant amount; 15%, quite encouraging. I also wonder if a water film temporarily defeats anti-reflective coatings? These musings only reinforce the positive results documented above. If PV panels could be cooled effectively without a front surface water spray then my guess is output might increase a hair more beyond these already good results.
Does anyone have a guess regarding the thermal conductivity coefficient across the PV panel sandwich, from the front wafer surface facing the sun to the backside of the panel, which I think is often a plastic material except for Silicon Energy Inc panels with a glass back surface? Also important to know how well coupled thermally the interior wafers are to the front and back sealed surfaces in a typical PV panel sandwich. This information would be useful in any attempt to engineer a rear cooling system for PV panels.