I've never actually seen full design calcs for that type of roof. I suspect the detailed design would be considered proprietary.
For symmetrical loading, you can isolate one rafter, neglect circumferential plate stresses, draw free-body diagrams, sum forces and moments in various directions and find the forces in the different elements.
I've not seen an analysis for unbalanced loading and would expect that to be a numerically generated design of one kind or another.
You can make approximations for the combined section properties of rafters welded to the roof plate, but it is not known how closely those approximations reflect reality.
The steel codes include design provisions for straight members, but not so much for curved members.
You can find design data for buckling of circular beams loaded in the plane of curvature. But for these rafters, they are basically prevented from deflecting outward due to the roof plate, but not restrained from deflecting inwardly, and so there is no design information available for their buckling strength.
The roof plate itself will be carrying a fair bit of the load. There is design information available for spheres in compression, but not for stiffened spheres in compression.
It's hard to tell, but it looks like some of those weld details are showing stitch-welding of the rafters to the plate, and that further complicates design.
This looks like an old drawing, and the original design may not have included any provision for unbalanced loading. I think that is a fairly recent addition in API-650.
For wind uplift, you could likely neglect the structure and just check the plate.
All in all, I think a rather involved FEM problem due to thin plates, buckling, large deflections, nonsymmetrical loading, stitch welding, etc.