1) Earth_atm_01.png shows the effect of standard atmosphere (well, only troposphere + isothermal stratosphere, up to 20 km altitude) on a ray of light at 589.3nm (Na D). psh_ray_tracing2, [-3,1.0000002], prof='earth', coeff=589.3, taumax=1d2,/PLOT, prec=1d-6 yields a deviation of 35.3' (41.8' with precision 1d-7, 50.05' with 1d-8, 53' with 1d-9) with our crude model atmosphere, a well-known astronomical result (~35') (i.e., when Sun or Moon appear just above horizon, they are in fact just below it!) 400nm: 36.12' 750nm: 35.52' Deviation is 0.55 degrees at most (grazing ground). No differences (at least at 2x10^-8 level) between 400 and 800 nm. 2) Using Euler-Lagrange, putting dx=ds, one obtains the following approx: d^2y/dx^2 = 1/n dn/dy Typical atmosphere: 1/n dn/dy = 2.71x10^-8 rad/m (Or 5.6"/km) [~independent of lambda (!)]. psh_ray_tracing2.pro yields a deflexion of 2.71x10^-5 for 1 km, (slab2) (hurray!), 750nm light is deflected slightly more than 400nm light, by about 10^-5 . See also: http://www.bibliotecapleyades.net/sociopolitica/condonreport/full_report/s6chap04.htm for even more details. 3) It seems that in a dry atmosphere, from HF to UHF, the index of refraction is pretty similar to the one for optical light. ( see http://mst.nerc.ac.uk/refract_index.html ). High water vapor content further increases n... (at saturated level: , refractivity (n-1) is up to ~30% of dry air density's value)