Propagation of electromagnetic waves on Earth is highly wavelength dependent, which is also true for the question whether the atmosphere has anything to do with it. What you are looking at is actually a formula for the geometric attenuation (aka free space or path loss), which has absolutely nothing to do with the atmosphere. As ACuriousMind points out, your units are wrong, and so is your mental model of wave propagation between antennas. Antennas have a near and a far field and both can be complicated. What you are looking for is the far-field pattern. By far-field we mean the radiation patter of the antenna that develops many wavelengths away from it. The effective far-field power density created by an antenna at a certain distance and angle is a function of the antenna's design. If you know this power density for a given antenna at a given distance, then you can estimate the further signal loss due to geometry for another distance.
The technological way to approach this properly is called "link budget" and I would suggest you start with that to understand the fundamental influences of different factors on the relationship between transmitter and receiver antenna power. http://en.wikipedia.org/wiki/Link_budget