MJD's answer for the square appearance is the classical one and very good, but I wanted to just add some points around your wider question.
First, physicists for hundreds of years have explained the square as the area of the sphere of influence of the force at that distance. Kant even, in his Prolegomena, says that natural laws MUST have this form by logical necessity even though it is a synthetic proposition, and is one of the big examples of synthetic a priori. When electrical forces were found to have the same form, many took it as obvious and well understood.
Now days, though, we know these laws are actually wrong. General relativity gives corrections to gravity that change the expression, as does (the relativistic) Maxwellian unification of electromagnetism (due finite propagation of the force). Kant is pretty much a laughing stock of hubris, and physicists no longer believe you can give those kinds of simplistic explanations based just on what you do to numbers. This is why you won't find many mathematicians willing to jump on answering your question.
Instead, science works the other way. It builds models that describe how to think about phenomena, with an ontology of things and their transformations. Then they derive equations for that model. By selecting different models and checking predictions, they get better equations over time. But they try not to ascribe some magical reason or meaning to why the equations take a given form outside the model.
There is however, some sense in looking at multiplying and dividing quantities. Unit analysis tells scientists what they are measuring. You have to be able to measure things consistently, or measurement is meaningless. If you are measuring mass and want to test it, you must test it against a theory that predicts a quantity convertible to mass. So when you see the $v^2$ in the kinetic energy, understand that you will expect units of kg $m^2 s^{-2}$ for energy, so it's at least understandable. But to get these equations, you need a model, and classical physics uses Hamiltonian or Langrangean models on Poisson manifolds.
Of course, like the gravity equation, the kinetic energy equation was also found wrong years ago, first with special relativity. So again, just be careful looking for meaning outside a model, and expect to be wrong in the long term.
So if you have a sound wave, then its total energy is the integral over time of the amplitude squared. But because of the Pythagoras thing, at also ends up being one half of the sum of the squares of the Fourier coefficients.
– Stephen Montgomery-Smith May 30 '14 at 04:00