The FCC is looking for quantitative noise floor data (ET Docket No. 16-191). Does anyone have any? There certainly appears to be a problem of the noise floor increasing, but the FCC wants to see data. Taking such data today might be fairly straightforward, but it appears they want historical data as well. Qualitatively, there seems to certainly be an increase. For example, here is a link to an aircheck of KOMA in Oklahoma City taken in 1964, at night from South Dakota. Partly because of its high frequency (1520 kHz), there are deep signal peaks and valleys. On the valleys, the spectrum is far quieter than it would almost certainly would be today, I'm pretty sure. And overcoming the noise floor to produce an intelligible signal required less strength than it would today, most likely. While South Dakota is not exactly downtown Philadelphia, on the nulls the spectrum is ghostly quiet.
https://www.youtube.com/watch?v=vQHPEFw0zHI
https://www.youtube.com/watch?v=vQHPEFw0zHI