It was a PG&E document, but I had to get if off google cache. I think it is some sort of deposition, not an article per se, but the info is there at the start of the document.
I don't think they plucked it from thin air exactly, but made a reasoned study - there is mention that some consulting engineers were involved - to turn what was agreed to be "typical" low density - 1 - low to medium -2 - and high density from the aerial photos into numbers people could use. There also seems to have been some investigation into how this would affect design in terms of percent of a route that would be affected.
Remember - More details = better answers
Also: If you get a response it's polite to respond to it.