Oh okay, well my mistake on the directionality, then.
But it is still the case that using any one set of relative sizes is incorrect no matter what -- whether they refer to area or to volume -- when applied to both side and frontal calculations.
Nor does adding some scalar multiplier bonus solve it, because the differences between front and side are quite non-linear, and a linear adjustment would be insufficient.
If you wanted to be as accurate as possible while still keeping memory and processing footprint as close as possible to the current footprint, then you should do something like this:
* Come up with some rough blobby 3d models of each creature. Some may use templates if differences aren't important.
* calculate projections from 5 angles: front, back, side(s), front diagonal(s), back diagonal(s), back, and make a table of cross sectional areas for each.
* Save those tables in RAWs. The 3d models can then be ignored and need not be packaged with the game.
* The game draws from whichever table is relevant for the current combat event as needed.
In the meantime, I would still maintain that interpreting them as cross sectional areas AT LEAST makes sense for one direction. Whereas interpreting them as volumes makes sense for no directions. Unless the game during combat always takes a cube root and then squares it to determine target size, which I very much doubt / hope it doesn't.