> Set up the sampling as many years in advance as possible to establish good reference data. Then create hydraulically moveable tunnel rings that push against a

> Set up the sampling as many years in advance as possible to establish good reference data. Then create hydraulically moveable tunnel rings that push against a static ring, to re-create the desired movement. Use machine learning to refine the model during the pre-construction sampling period, and to refine results during operational period. Yeah, it's definitely going to be cheaper to just pay for subsidence. Not to mention that it's literally impossible to perform the kind of sampling you're talking about. It's basically the ultimate wet dream of the oil, gas and mining industries and they've poured trillions of dollars into developing subsurface sampling for over a century, with nowhere near the resolution you're thinking about.