Atmospheric effects are mostly confined to altitudes where there is appreciable -er- atmosphere. Say below 100,000 feet (nonlinear, so probably 50,000 feet or less is where most turbulence occurs). Recon satellites orbit at ~150 miles or so (KH-11 figure, but commercial earth observation satellites like to orbit above ~400 miles to avoid residual drag). Imagine a line drawn between a ground telescope and the satellite (represents your sight line).
First case, ground to satellite, anchored on the ground... You're turbulence, so grab the line at the 10 mile up mark and wiggle. The satellite end of the (sight)line will be displaced by some distance.
Second case, satellite to ground, anchored at satellite... Again, you're turbulence, so grab the line at the 10 mile mark and wiggle. The ground end of the (sight)line will also move, but over a much smaller distance than the first case.
So my (possibly incorrect) conclusion is that observing the earth from a satellite is less affected by atmospheric conditions than observing a satellite from the earth.