When I was a C130 Captain, I would often have a bet with the navigator as to who could make sunrise/sunset prediction most accurately.
It is easy enough (Air Almanac or whatever it’s called) to calculate the time over a fixed point at a particular altitude. But you never know exactly where you will be at the moment of sunrise/sunset. So I would plot a simple linear graph of aircraft position against time and another using same scale of sunrise (set) for particular localities along track. The intersection of the lines would be the point of sunrise (set). The intersection of the two lines on the graph can alternatively be determined by the solving simultaneous equations but I found the graphical method much simpler.
I only made the challenge when the sun could be seen from the flight deck, ie sunrise when heading (vaguely) eastwards or sunset when westwards. Far greater accuracy can be achieved with the sunrise than with sunset due to a steeper angle of intersection in the sunrise case. I would invariably achieve +/-2 minutes for sunrise but the error with sunset could be well over 5 minutes. Of course, when flying at very high latitudes, sunset might in practice be virtually unpredictable if the aircraft’s groundspeed approximates to the rate of the earth’s rotation.
An interesting phenomenon most easily visible from the air (far less often from the ground) is the Green Flash. Wiki explains here:
Green flash - Wikipedia, the free encyclopedia
Jack