Postby bert_the_turtle » Fri Dec 10, 2010 11:00 pm
Basically, physics. You calculate what the explosion energy can do. The fireball size would be the radius of a sphere just small enough so that the energy can heat up what's in the sphere to, say, the melting point of lead. It would grow with the third root of the energy, roughly, with a prefactor you most conveniently get from past overground tests. For the overpressure, you assume the energy goes into the shockwave. A shockwave's energy dissipates with one over the distance to the center, squared, the energy density is proportional to the pressure squared, so the pressure should be proportional to the ratio of the square root of the blast energy and the distance to the center, again with a prefactor you take from tests. Temperature is essentially thermal energy density, that would be proportional to the blast energy divided by the square of the distance. Of course, with a prefactor. The prefactors are hard to predict, they depend on how the blast energy gets distributed over the different channels. And they'll still have some dependence on the blast energy; for really large blasts, a lot of the energy just gets radiated off into space.
And we only had to release enough radioactivity to give a hundred thousand people cancer to get the calibration data that now serves your amusement!