The Shape of Days, and a Fall Mystery

In mid-November Seattle had a series of unexpectedly clear sunny days between storm fronts that provided a great opportunity to compare my Fall solar production with sunny Summer days. In the chart below I've averaged data for three July days and compared it to an average of three November days, all of which had clear cloud-free weather.

Click to enlarge.

The first this that jumps out is the marked decrease in production for both the span of time and power output. I think it's safe to assume that the shorter span is due to the shorter days in the Fall, as the times are closely correlated with sunrise and sunset. My guess is that the change in the power output is likely due to the significantly lower sun angle in the Fall, peaking at only 25 degrees in mid-November versus nearly 65 degrees in mid-July. The plot below (generated here) shows the sun angles throughout the day for these periods, and seems roughly proportional to the power output I'm getting.

Sun angle chart generated at

An oddity that I'm at a loss to explain, however, is the plateau that happens in the mornings in November between about 8:00am and 9:30am, after which production jumps quickly up to a peak before gradually declining throughout the afternoon. I had speculated that this was due to some kind of shading in the morning from the trees to the East, but a check out the window showed me that the sun was already clear of the trees when it rose.

My next thought was that the jump was due to the sun rising high enough to start hitting the panels facing West (I have eight facing East and six facing West). The panel data doesn't support this, though. It shows that the output from the West facing panels increases fairly steadily throughout the morning, while the East facing panels have a significant jump in production mid-morning before declining through the afternoon. Adding to the mystery is the lack of a large jump or drop on the West facing panel output in the afternoon or evening, and the noticeably lower peak output for those panels.

Click to enlarge.

I'm really not sure what to make of this chart. If you have an idea please let me know!