That gels pretty well with the general observation that the sky becomes "bluer" at angles further away from the sun. IR scatters even less than visible red, so

That gels pretty well with the general observation that the sky becomes "bluer" at angles further away from the sun. IR scatters even less than visible red, so you'd expect its angular distribution to be more strongly skewed as well. I'm having a hard time trying to qualitatively estimate the angular distribution of perceived spectrum at a particular location, as opposed to the distribution of the angles of scattering events. I may have to try simulating this when I have some time.