As hunt_mat said, you could just do the direct thing: parametrize the surface and integrate. He suggests polar coordinates; I'm not sure if he really meant cylindrical coordinates or spherical coordinates, though. It might be worth considering plain ordinary rectangular coordinates; the three summands are pretty much already set up as ordinary double integrals.
Normally you'd consider the generalized Stokes' theorem to integrate this. (Green's theorem?) But, alas, the origin is a problem.
So what if you put a tiny sphere around the origin, and used Stokes' theorem on the region between them? Or alternatively a huge sphere. If you can say something useful about the behavior of the integral on very small or very large spheres, this approach could work.