For the US and Canada, this is a very useful site for determining degree of light pollution and moonlight, and likely observing conditions (cloud cover, "seeing", etc) - uses Bortle scale for light pollution. Light pollution data are contributed by amateur astronomers at the various featured sites, who determine what magnitude stars are visible at the site. The likely observing conditions are predicted using weather reports.
The hardest part for landscape astrophotography is getting nice dark skies. I am doing a lot of experimental shooting at the Bortle orange-red site (Brommelsiek Park astronomy area) that is 45 minutes from where I live and that is public and safe. There is a big fat light dome, and there are a lot of aircraft from a small civilian light-aircraft airport approximately 4 miles away. Still, it is good enough for determining the characteristics of lenses, determining ISO with particular cameras, learning to use Astrotrac sidereal tracking. Light painting can be practiced anywhere except Brommelsiek Park, where it would bother the astronomers. At any rate, I have done a fair amount of practice, but haven't done the desired dark sky landscape astrophotography yet, because I need to scout for genuinely dark (Bortle blue) or dark-ish (Bortle green) sites that have good foreground photo potential and have enough of a clearing to make it worthwhile, and which allow rough camping. Weekend, clear Sat. night/ Sun. pre-dawn, no moon, minimum of 90 minute drive, most sites about 2.5 to 4 hours away.
My copy of the 14mm f/2.8 Samyang is used at f/4 full frame, because it lacks critical sharpness in the corners.
Thanks for all the tips regarding post-processing. I have not scratched the surface of post-processing landscape astrophotography. One problem is the annoying orange gradient from the light dome - I am not entirely thrilled with my simple anti gradient color correction. I have tried pushing some star field sets, and also the "control" slides (darks, biases) needed to deal with noise, through Nebulosity. There is definitely a learning curve to this.