PV panel temp. Wind and ambient temp control this. You can get quite a drop in power with no wind, as the panels heat up.

I just checked the history from the weather station 2 miles from here, and the difference seems small to me. Can the following be significant emough? There wasn't much wind either day, and the temps were close both days - the panels are "exposed" in the back since they're on trackers, which, I would think, help dissipate heat.
Feb 14th (33.58 KW):
Time.............. Temp.........Wind Speed
08 am............46.............2
10 am............53.............3
12 pm............60.............2
02 pm............60.............2
04 pm............60.............2
05 pm............60.............2
06 pm............55.............0

Feb 16th (30.1 KW):
Time.............. Temp.........Wind Speed
08 am.............35.............0
10 am.............50.............2
12 pm.............60.............1
02 pm.............63.............1
04 pm.............63.............2
05 pm.............63.............0
06 pm.............60.............0

Check out the spec on the panel

http://www.evergreensolar.com/upload/19 ... 010707.pdf

Notice the spec Temperature Coefficients, α Vmp (%/ ºC) -0.47

Meaning, your array will be reduced in Vmp ( voltage maximum power ) 0.47% for every 1 degree C above 25C ( 77F ). This is NOT the ambient air tempature, this is the panel tempature and typically panels can be 25-50C higher depending on mounting and airflow. Your racking is much better than roof mounting, my panels are in the 65-75C range here in Florida thanks to my roof mounting and warm climate.

So if your panels are 10C hotter than 25C, thats a reduction of 5% from the rated power in standard test condidtion

Also, the STC is 1000wmsq ... thats a rating of the sunlights intenstity, in typical clear condition, solar noon, you might see 930-950 max unless your on a mountain top somewhere.

Trust us, your array is performing excellent, there in nothing to concern yourself with

I've become comfortable with the production - at this point, it's more a curiosity. I can see that the panel temps can cause a difference, but a 10+% change is quite a bit. The curiosity is with an output difference between the two days where the temps were essentially the same and essentially windstill both days. Curiousity is hard to turn off - it just happens

Connect a PC and log data using SG-View using 10 to 30 second logging interval.

No two days are the same, wether is the tempature, clouds, crap in the air, hickups in the tracker that only data logging would catch.

Its a fun project ( data logging ) ... give it a try and you can nail down what caused the difference and no more guessing!

Good idea.