The distributions you show are not at all unsmooth. And they don't look at all uncommon for problems of that kind. So I'm not sure what you would expect to see that is better.
A serious problem however, is that any simple smoothing tool will very possibly introduce distortions that are non-physical. For example, a spline could easily allow the curve to go below zero, and that would be a bad thing in this context.
I'd suggest you accept what you have as reasonable distributions.
