I am doing the same, Mainly I wanted to record where pipes and such are located but might as well record buildings, fences and trees while there. I was not worried about geologic movement but all the global coordinate systems felt very clunky when locally mapping at the centimeter level, I finally settled on having a well defined surveyed reference point, and using meters north and east off that.
Perhaps that would work for you. because while the plate is shifting with respect to the globe, everything on that plate will maintain the same relative position.
I set up my own NTRIP base station at a fixed point in the middle of my roof with a hefty bracket, did a PPP survey to determine its location once, and am considering everything relative to that. This spring when I'm back at the surveying, I'll likely do another PPP run and make sure it hasn't moved too much. If it has, I guess I'll have to figure out how to reconcile that. From what I've gathered, I don't think this is too far off from how real surveying works.
I didn't see much reason to use a different unit than degrees. Although while I haven't gotten super in depth learning QGIS, it feels like there's an impedance mismatch in that it seems to be a 2D program that treats degrees as linear units (and then applies a fudge factor to degrees longitude), rather than a native 3D program. So I'm doing all of my collection, point storage, and calculation with scripts outside of QGIS, and then only pushing the cooked results to QGIS for visualization.
I don’t have a good sense for how far one needs to be from one’s NTRIP base station, but I’m pretty happy with the UNAVCO data I’m getting. The nearest station is operated by Serious Professionals, and it won’t move if I mess around with my roof. :)
I've got a professional base station that's decently close, but still far enough to degrade the accuracy a bit (I forget the exact figure but you can find it if you search). When I used the professional base station to figure out the point and compared it to what PPP had found, it seemed to be within that amount of error.
I chose the part of the roof least likely to be disrupted. But if I do need to move it, then I'll have to figure out exactly how I want to map to points going forward.
Instead of a reference + north/east meters, how about two reference points, and then everything is referenced as polar coordinates from there, point A to point B being 0 degrees. My concern would be if the direction of north/east changed over time from the plate movement.
Perhaps that would work for you. because while the plate is shifting with respect to the globe, everything on that plate will maintain the same relative position.