Wouldn’t an underground test be easily faked, since the exact circumstance ,and hence seismic profile, of the explosion can’t be known, and only few of the results would be expected to be found in the atmosphere… (the ones so rare that the small leakage is still detectable… )
Even better, a bleve would be a great way of creating a very big boom. Construct a very large and very strong pressure vessel in an underground mine. Fill with water and weld it solid shut. Pipe in gas and air, and, provide a flue. Proceed to heat the pressure vessel. And keep heating, and heating and heating. The actual temperature of the vessel won’t get really high, but the water will adsorb a huge amount of energy. Eventually the vessel will let go. All the energy the water adsorbed will turn the water into superheated steam very very rapidly. City blocks get leveled with only a smallish bleve. The square/cube relationship would allow you to create something insane without taking a really huge amount of steel for the vessel. (Most belves are LPG tanks, which makes it look as if the explosive force is all about the burning gas, but it is the boiling liquid and expanding gas that provides the overpressure, even before it has found any oxygen to burn with.)
But, whether you could get the blast to approximate the profile of a nuke is another matter. A nuke delivers its energy into the earth faster than anything else one could conceive of, and it may be that no matter how clever you are at making a very big boom, you won’t get the frequency response right without very careful tailoring. It would probably take a few goes to get right, if at all, after which you have probably blown the secrecy of the idea.
The whole point is that the time vs. pressure profile of a nuke going off is very different from massed conventional explosives or an aerosol-plus-oxidizer explosion. These differences are inherent in the different physics of how the explosions work.
The US & the Russians have lots of experience listening to their own nuke tests and have developed elaborate networks of listening stations specifically designed to make this distinction.
The whole and entire question is whether there’s some way to arrange enough conventional explosives in the right kind of chamber to fool the sensors. Nobody who knows for sure is going to say here.
One thing’s for sure at the unclassified level: Even if it is theoretically possible, it darn sure isn’t easy. Which is another way of saying it might be easier to build a functioning nuke than it is to build a convincing fake.
So I tried putting some numbers to your scenario.
Say take one of the higher pressures used in the industrial pressure vessel - say around 175 bar (2161 psig) from urea synthesis. The boiling point of water at 175 bar is about 660F. Say we start with water at 100F and we superheat the whole pressure vessel so that the water+vapor contains around 50TJ (this is approximately the same yield as Little Boy). Then the volume of water needed is around 5 million US Gallon. This is the size of a large municipal water tank, except you will need to make the L/D higher to save on metal. Certainly doable - but will be very expensive - my gut feel is around a few billion dollars.
The Vela Incident, is still being debated, whether or not it was natural or nuclear, so it’s possible for something to go boom and be debated.