so anyway i finally got it mostly sorted but wanted to post in case anyone else hits this. doing a large commercial roof inspection job today, its a big distribution warehouse, maybe 400,000 sq ft of TPO membrane, client wants condition mapping and wants to identify ponding areas and potential seam failures. flying a phantom 4 RTK with the D-RTK 2 base station set up about 80m from the structure

about halfway through the second grid pass i started getting really inconsistent RTK fix ? dropping from fixed to float constantly and it was screwing up my overlap consistency. checked everything, base station had clear sky view, rover link looked fine. eventually figured out theres a massive HVAC array on the roof that i think was causing some kind of multipath issue on the GPS signal when i was flying low passes maybe 30ft AGL. pushed up to 60ft and it mostly stabilized but now im second-guessing whether my ground resolution is still good enough to actually identify seam issues at that height

anyone dealt with multipath on large commercial roofs with a lot of mechanical equipment up there? feels like its going to be a recurring thing for this type of work

multipath from HVAC units is a real pain and youre not imagining it. the metal housing on big rooftop units can absolutely mess with your base station signal if the geometry is bad. one thing that helped me on a similar job was repositioning the base station to get better angular separation from the equipment clusters ? even moving it 40-50m in the right direction made a noticeable difference on fix quality. also worth checking if your NTRIP corrections are available in your area because using a VRS instead of a physical base can sometimes sidestep the local multipath issue entirely depending on your correction source

as for the 60ft vs 30ft AGL question on TPO seams ? honestly 60ft with the P4RTK camera is probably still workable for identifying obvious seam lifting or pooling areas but yeah youre going to lose some of the finer detail. if the client specifically wants seam-level condition assessment you might need to do a secondary lower pass just over the areas the orthomosaic flags, kind of a tiered approach

im curious what GSD you were targeting at 30ft. for TPO work ive found anything under about 3mm/px starts giving you more data than you can realistically process and the orthomosaic stitching gets weird around anything reflective. white TPO membrane is a nightmare for photogrammetry in direct sun btw if you havent hit that yet, the auto-exposure fights you the whole time and you end up with blown highlights right where you need the detail