i think i understand your argument.
first, to address a minor - and i think incorrect - point, which even if wrong doesn't necessarily affect the legitimacy of your argument:
Now, imagine a train on those tracks. Given your actual position, and for the purpose of my idea, you're in such a position that the train is below the horizon line on the tracks. If it starts to move, it will eventually (depending ultimately on your POV) rise above that horizon line before it eventually converges with the rest of the picture you're viewing and moves beyond the vanishing point.
in this scenario, as i've understood your description, the train wouldn't rise above the horizon line, ever - as long as it is below the horizon line at any time during your viewing it from a fixed perspective.
and like i said, i think i understand the main thrust of your argument now, that was helpful clarification. i'll quickly summarize what i think you said, you can tell me if i got it wrong: we see the vanishing point literally higher in the sky, than we do any part of the train (as long as we can see the train). the closer the train is to us, the better we will be able to see it because the reflected light has less atmo to travel through before reaching us. however, the vanishing point is also visible, because it is effectively "higher" relative to our elevated viewing position, and thus is less affected by the atmospheric gradient you have proposed. the reflected light has less scattering density or whatever to deal with. the net result of this is that, somewhere in between us and the vanishing point, is (for two competing reasons) the most opaque. closer [lower] is clearer, and higher is clearer. it is in this middle area, that things can disappear.
am i understanding? if so, i would say the photos i showed above disproves it, because we can clearly see the horizon, and we can clearly see that it is not the vanishing point. unless you want to argue that it's not the horizon, it's nothing more than a common surge of water between us and the ship. (so common that it's in every shot.)
but the fact that we can see a ship beyond the horizon, demonstrates 1) the "laws" of perspective are great help in art, and are an obvious phenomenon, but some of it is irrelevant to reality. [e.g. why are there never any fe discussions about
two-point perspective?]
for example, there is no "vanishing point" in reality. thing just appear smaller, eventually to the point where the angular resolution of our eyes fail to detect them. but don't suddenly disappear like a singularity. [i don't think that is what you are arguing.] if a pea disappears thanks to perspective, replace it with an elephant. or a pryamid. or a planet. or a galaxy. etc. secondly, even on a flat earth, the gradeschool concepts of perspective lines and vanishing points are of limited usefulness, because there are hills, mountains, huge convex and concave areas, etc., getting in the way of the vanishing point at the end of a perfectly flat plane of perspective lines mapped to the earth. similarly, on a re, a hill of ocean gets in the way. if you imagine yourself standing on a plane, then the earth falls away in every direction and you have no plane, no perspective lines, and no vanishing point as they relate to the earth. or if instead, you pick a distant object and make that an integral part of your viewing plane, then a hill of water or/or land will be between you, with a fake horizon. just like the pictures above.