If dimpling/buckling is unavoidable, then making sure that its effects do not contribute to larger RCS (or even positively contribute to smaller RCS) through structuring the dimples would seem a good thing. But, I'm no engineer ;-)
I am thinking someone analysed the effects of using longer wavelengths with enough post processing (esp available as computational power has increased significantly) and figured out (with final in-situ validation) the utility of the tradeoff between mie resonances at regular microwave range that would be in play with the panel size/dimple geometry...and rayleigh for longer wavelengths (and thus lower effective RCS of the body at those ranges given higher effective diffusivity and lower amounts of other modes) that would be introduced because of the "dimpling".
It would ultimately depend on the summation of all the modes that operate at each range (Esp with low incidence angle reflection the water provides which will act as another retro mode). Thus having a perfectly planar, continuous surface next to a refractive medium interface (sea/air) I think actually creates backscatter modes (among others) that need to be accounted for passively.
Basically you inherit a situation and make the best of it for each regime (i.e might as well make it more resistant to detection at longer wavelengths rather than pure optimisation at lower wavelengths which would call for large perfectly planar, continuous surfaces...which are difficult to fabricate and maintain anyway in a non-ideal world with other forces at play).
When you don't have this issue of a refractive medium interface, the issue drops greatly (i.e the mode channels)....and you can optimise much more effectively depending on which modes dominate. Hence we wouldn't see as much utility of having a slightly dimpled (panel-based) surface for aircraft or spacecraft which are enveloped in entire homogeneous mediums. Aircraft of course have major aerodynamic considerations anyway that preclude this significantly to begin with.
This was the line of thinking (there could easily be others...I am no high level theory expert, I worked on materials application mostly) I ruminated on when it came up in the earlier thread....when claimed it was something totally unrelated to the RCS....when I think its much more complicated and involved than that.
I will have to look up later if there is any paper in the open domain about what I am speculating here though....it would need a large number of real world testing + validation which I doubt are non-classified.
@Vergennes @Taygibay
Do we know any real radar ppl in this forum anywhere btw?