In fact the normal ray tracing and winding rule approaches work just fine on the surface of a sphere, with a little adjustment.
On the surface of a sphere a 'straight line' is a great circle and distances are measured in angular units rather than metres or inches. To draw a ray from an arbitrary point on the surface of the sphere simply form a great circle through that arbitrary point and any other point on the surface of the sphere. To keep the maths clean choose a second point about pi/2 away from the point whose location you are testing. Apply the usual even-odd rule to the great circle and your test polygon.
The winding rule also translates directly from straight lines in the plane to (segments of) great circles on a sphere.
All you need now are Java implementations of basic spherical geometry operations. I don't have any recommendations on that front, but I guess the Internet will help. For the maths start with Mathworld.
Another approach would be to project your points and polygons from the surface of the sphere to the plane -- which is what map projections do -- the topological relationship of insideness will not be affected by such a transformation.
Oh, and you'll have to decide what to do if your polygon describes a great circle