You need a few bombs and some places of varying geology to set them off. You take those data points, cross reference with all your historical knowledge and should be able to say whether a bunker of given construction a given depth under a given geology can be breached.
I hate how allergic to just testing and prototyping things modern engineering culture is.
Yeah, the bomb is expensive, but you gotta test it too so if you do it all right you get two birds with one stone.
The US Navy's torpedo station in Newport, RI produced torpedos that were really prone to failure during the first few years of WW2.
IIRC, the problem persisted so long because an admiral in charge refused to provide enough torpedos for adequate testing.
(Sorry if there are any errors here, I can't easily fact check at the moment.)
Never mind the fact that bomb damage assessment is one of the most difficult problems in photograph interpretation -- it's hard enough when the target is above ground, worse when it isn't.