IMHO the worst atmosphere effect in FM is a form of “tropo ducting” where moisture between the antenna (usually 900 ft or more) and the ground “blocks or reflects” the local signal (less than 10 miles from the tower) and “bounces” it a couple of hundred miles away. This can happen to stations on the American Gulf Coast. There was a post several years ago about a station actually switching to their back up facilities at lower high because they couldn’t pick up the station at the studio. I personally thought circular polarization was supposed to fix this (and have better building penetration too), but I guess the atmosphere sometimes does things that can’t be engineered for. I do know from personal experience at Eglin AFB sometimes the severe fog caused all kinds of issues with Trac 97 line of site shots over water (swamps and bays) during JCS exercises during the early 1970’s. But these were 1 watt shots that used really sensitive receivers* and a signal comparer to pick up signals as low as -102 dBm. which was pretty good at the time for equipment that bounced around the countryside in the back of a ton and a quarter jeep/pickup.
Here are my to favorite sources which show where FM can behave “badly”
http://aprs.mennolink.org/
http://www.dxinfocentre.com/tropo.html
*I could trouble shoot and repair the receivers using the test points and the manual, but the Trac 97A used a weird “tunnel diode” circuit in the receiver which I really never understand of how it worked. I do know that the tunnel diodes were very sensitive to static charges.