Empirically, when I wash the half dozen seagull splatters and random dirt off of my panels I can't see a difference in my power production graphs.
I'm sure the output goes up, but it is too small for me to bother with the somewhat uncomfortable chore of washing them with the mop and squeegee on a 16 foot pole.
I snapshotted a picture of the data for those interested. http://jim.studt.net/solar.gif (there are some missing data sections, sorry, the data collection port has some interesting quirks).
The green line is solar panel current delivered to the 24v battery bank (actual panel voltage is several times higher). The red and yellowish lines are various inverter and telemetry loads. In the bottom half the red line is battery bank voltage and the blue is "state of charge" as indicated by the somewhat mysterious Outback DC monitor.
For the sake of cleaning differences, note that I cleaned the panels around 10am on the 11th. Compare the 9am solar output, the first faint vertical line to the left of the noon line on the 10th, 11th, and the 13th(post cleaning). All very similar, a few percent change at best.
I can't let you into the live system to browse, it has an expensive query and the intertubes would crush my server.
I'm sure the output goes up, but it is too small for me to bother with the somewhat uncomfortable chore of washing them with the mop and squeegee on a 16 foot pole.