Open jarcangeli opened 5 years ago
Yep, probably lots of ways to go about this. I was thinking we could do something with transparency to de-emphasise regions that are poorly constrained, not sure how that would look.
The degeneracies/covariances between pixels/spherical harmonic coefficients make it really hard to come up with an economical visualization of the uncertainties on an actual image of the map. My suggestion would be to show the maximum likelihood map and then a bunch (10, 20, 30?) of independent posterior draws to convey the spread in the possible solutions...
I imagine we could brainstorm in some way on this, but I am definitely interested in thinking about this!