Two-box, maybe. PI with a monster searchcoil, maybe. VLF metal detector with monster searchcoil, maybe. Geophysical VLF apparatus, maybe. Mag, might detect iron metal associated with the cache if not the gold itself. Underground radar, it "all depends". All these might detect the soil profile disturbance associated with the burial of the cache, even if not detecting the cache itself.
That's what made Marc's proposal a tempting "LRL" proposal. If the problem were to find an iron manhole cover a foot deep, it wouldn't have been an interesting problem.
* * * * *
~EE~
If it were my challenge, I'd get some decent (less than $100) dowsing rods and dowse the site; and would also do a map dowse. All as blinded as I can figure out how to do it. Early in the game, before forming too many opinions about where the caches could or "should" be located since that kind conscious selection can adversely influence dowsing. If I knew someone experienced in successful cache locating who works for cheap or for free, I'd drag them into it, and if they dowse I'd ask them dowse it non-blinded to take advantage of the knowledge that's buried in their subconscious.
I'd do the dowsing stuff early on because dowsing is a simple shortcut method requiring little investment. Dowsing "evidence" is pretty flimsy stuff, proving nothing; therefore it is the kind of "evidence" that's not likely of much use late in the game when you are trying to analyze the meaning of factual observations.
Dowsing and other early results should be followed up with more plain old site research, and research and critique of the stories that led to all this interest. Then instrumented field work not just to "find the cache" (which may require shovel work to actually locate), but to establish insofar as possible the pattern of human activity that prevailed at the time of the supposed cache. And save the hard work of digging until last, unless earlier on in the project there is a very promising spot worth going ahead and digging.