File this under "perspective is weird". Last summer on a camping trip, we were stargazing just after sunset. I remarked how—bizarrely—the lit region of the moon appeared to point above the horizon, despite the fact the sun was below the horizon.
What the heck is happening?
A recent blog post by Randall Munroe (which I can't find right now) asserted this was impossible, which made me question my memory. I have, however, just seen precisely the same phenomenon tonight, which makes me confident it's a real thing and perspective is just plain weird.
Here's a couple of stills out of a video I shot to prove it. Some things to notice:
1. The moon is way higher in the sky than the sun is. (The moon is perhaps 30 degrees above the horizon, the sun is setting.)
2. Nevertheless, the lit portion of the moon is the upper right quadrant.
3. Without any other markers to suggest what's going on, this is pretty weird.
Fortunately, there are some other markers I didn't realize until I was annotating the photographs: the horizontal lines on the long hardware store that appears in both images.
In the first picture, the roof lines appear to go "up" to the west, in the direction of the sun, but in the second picture, they go back down again. In fact, the horizontal roof line in the first image makes it clear that it's not really the "upper" portion of the moon that's lit at all: the line to the sun is straight south, following the roof line. It's just that in that linear perspective, horizontal lines which are above you appear to go "up" as you get to the edge of your field of view.
Since we're "below" the moon (at least from my Earthen reference point) we can see the partially unlit 'bottom' of the moon and none of its partially unlit "top".
The root error in thinking this scenario is possible is assuming that these lines are straight. You'd actually need a fish-eye lens to see both at the same time, which would force "straight" lines to be curved: