My solar generation has fallen off significantly due to tree shading at different parts of the day. I have very tall pine trees which don’t shed much needles in the fall/winter. PVWatts was very accurate in August, but September was a little lower by 10%, October 25% lower, and November was about 40% lower. Some of this reduction was due I think in more cloudy days than the 30 year average that PVWatts uses but I don’t have any multiple years of experience to try and segment this out from shading.
I used the phone app SunQuest to gain a better perspective on the impact of shading. It shows the path of the sun during the various times of the day and allows you to use your phones camera to pan the horizon and follow the sun position at various parts of the day and see which trees are going to impact shading. While it does not quantify the impact of shading on power impact, it helps you determining what parts of the day are going to have lower power output.
On a sunny day in November, my 400 watt panels peak at about 325 watts/panel with no tree shading. Once the sun goes behind the pine trees, that drops to 100-150 watts/panel on the panels on a higher pitch section of my roof, but drop to about 25 watts/panel on a lower pitch section of the roof. The peaks and valleys line up very well with what you see on SunQuest.
If I would have used SunQuest before I put my system in, I probably would have put more panels on the higher pitch section of the roof and less on the lower pitch. Now that I know this, I might move some panels around to generate better performance.