There are areas in the US that do not have cell/data coverage where important road information updates (think mountain pass conditions) are broadcast over AM radio.
Vast areas of the Americas are barren of any FM radio reception but receive AM well. The manufacturers that want to drop AM radio seem to find that fact inconvenient.
I would guess the first two have nothing to do with the type of modulation, but with the frequency bands being used and/or the transmitter power. That is: I would guess it'd work equally well for frequency modulated radio signals broadcasted on those bands with the same transmitters. So those arguments are in favour of using certain specs for broadcasting, but don't seem to require any specific modulation.
(Would love to hear if that's incorrect though)
I'm not sure about DIY'ing a radio - no clue how the modulation complicated things. That could well be a reason favoring AM.
1. Yes, with analogue signals, given equal ERP (Effective Radiated Power) levels, FM modulation breaks down with a ''cliff effect'' (suddenly) while AM blasts on. Typically, on the receiver end as the range increases, an FM Radio stereo signal will drop to mono for a while, then just disappear even if a carrier is still detected in a sensitive signal meter. With HD Radio digital FM, weak signal will cause an abrupt loss of any subchannel(s) and reversion to only analogue before the cliff effect takes place. Again, analogue AM signals barrel on until they simply diffuse into the background noise. HD Radio digital AM signals drop to analogue as the range increases, which is why none of them are run digital-only.
2. Yes, AM band signals bounce around a bit in the evening sky, so AM Radio stations often have nighttime power reductions specified in their licenses. The FM Radio band can experience its own effects like tropospheric ducting (see any number of radio and TV DXing sites for much deeper info) but with nowhere near the predicatability and stability of strong AM signals being bounced.
3. Yep. See the crystal radio and ferrite coil comments elsewhere in this thread.
Other:
Regarding the modulation being transmitted on a particular band, it has a lot to do with tolerance of interference. A perfect example comparing FM and AM on the same band is when the North American analogue TV standard NTSC was being devised, which originally used all AM modulation, but the audio portion of the signal was found to be so disrupted by the video portion that the relatively new FM modulation was used for the audio instead. In AM radio there is only just the audio, so in effect the problems of interference at the source are gone and pushed out to the receiver end as being an accepted norm. It used to be common for charging and ignition systems of cars and trucks that were of the old breaker point type to generate clicking, tapping, and whining on the AM band (the title of Frank Zappa's ''Dinah-Moe Humm'' is a reference to the problem of dynamo or magneto noise from DC generators) but shielding, AC alternators, and the use of condensers and better materials by the automakers solved those problems. Home AM radios would pick up refrigerator solenoid clicks and other such noises, which could be very problematic in apartment blocks, but again better shielding and other techniques in consumer electric devices were soon practiced. Nevertheless, environmental interference on the AM radio band has always been a problem to this day. FM radio has its own problems with interference, but typically in congested urban areas in which ''picket fencing'' from large structures can occur. Digital radio, like iBiquity's IBOC HD Radio system, uses error correction and buffering to combat such interference problems. So, you could transmit either AM or FM modulation on your signal on a given band, but you'd better be ready for the interference consquences of each.
> (think mountain pass conditions) are broadcast over AM radio
This isn't the scientific theory, but I drive the mountains a lot and AM consistently dies much sooner than FM.
Can't explain it, but 100% true.
EDIT: I live in Salt Like City area and drive through the Wasatch Range on a near weekly basis. # of receivable AM stations goes to zero nearly immediately.
AM can be untrustworthy in mountains too. Many factors play in such as how the landscape is actually laid out, the make-up of the rock in the mountain. How signals bounce off the rock, etc. It for sure isn't fool proof. I wouldn't think that much better than AM unless your close to the tower. AM's big advantage is being able to travel farther on less power, but the low frequencies don't have good penetration power. This is why the Navy uses radios that runs on 900+ MHz for shipboard radio, 900 MHz (UHF range) has less distance, but better penetrating power through steel decks. Compared to frequencies that AM operates (HF range). Then there are atmospheric conditions and the cycle of the sun. At a solar maximum sure, but at a solar minimum, probably not.
Generally AM is used in mountainous areas because it's MUCH cheaper than FM, and those areas don't have the population density to make FM stations viable.
It's also used for natural parks, and anywhere you might need updated, local, information, again, because it's so cheap and because receiving is so easy.
Day time vs night time? In the evening while camping in Canada's north in flat or mountainous terrain I've sat beside a fire with a portable radio and tuned to certain high power U.S. stations on the AM radio band. The locals will tell you it is not uncommon. As for reception in a moving vehicle in the daytime, any number of fields or other conditions may be causing the dropouts but in general AM radio propagates vastly farther than FM for a given Effective Radiated Power range specified in the station's license.