Cavechat.org

Would it be possible to discover an unknown cave without a large opening by means of the type of exploration we use to discover oil deposits, such as sonic, seismic or electrical scanning?

Yes. I have no specific examples to provide you with at the moment, but I've read of this being done many times.

Here's something I found googling cave images:
http://www.geo-sense.com/GPR.htm

It's possible, but you'll notice that using ground penetrating radar (GPR) is difficult for deeper penetration. May commercially available ones have a max depth of 15 meters in Limestone. In this research http://www.uvm.edu/~lewebb/CCLI/various%20papers/Smith%20and%20Jol%201995%20GPR.pdf their max depth was 52 meters (with any clarity).

One example I have is a mountain room where the floor is 72 meters below the surface and the top of the passage is 52 meters below the surface. This is an enormous passage that could conceivably have collapse events. And it would be very hard to detect with current GPR.

So if you are looking for caves near the surface to dig into...yes. If you are looking for deep caves...maybe, but probably not with off the shelf hardware.

This technique also is used to pinpoint "other" entrances to caves...ie, to find (or confirm) that a passage in a known cave runs at a certain spot so a new entrance can be dug, often saving hours of transit time to a remote location in a cave. I am familiar with two such cases, one where it worked to confirm that the passage was close to the surface and the other where the intervening layers were too thick or the equipment too weak to get a confirmation (but we still think the spot is over known passage).

Look for an article in JCKS (maybe as soon as the December 2014 issue but perhaps sooner on-line) called "Ground Penerating Radar Investigation of Limestone Karst at the Odstrzelona Cave in Kowala, Świętokrzyskie Mountains, Poland". It is a solid academic-style piece on GPR.

http://userpage.fu-berlin.de/geodyn/tut ... aeoSci.pdf
says
There are a number of geophysical techniques that can be used to detect the presence of caves and other voids below the surface. Microgravity is popular and effective since there is clearly a large difference in density between the surrounding substrate and the void (Butler, 1984; Smith & Smith, 1987; Linford, 1998). However, gravity methods have the limitation that they do not detect the actual shape of the void but rather require the surveyor to postulate a possible shape, run a simulation, and see whether the simulated outcome matches the observed data. With a priori knowledge of likely shape (when verifying documen- tary evidence of a mine or tunnel, for example) this can be very effective, but when prospecting for an unknown cavern it is a major disadvantage. Resistivity tomography can also be used since the resistance of the void will be higher than the surrounding substrate (Noel & Xu, 1992; Manzanilla et al., 1994). Unfortunately, limestone it- self has a very high resistance which means that this technique is less likely to be successful. A third tech- nique is geophysical diffraction tomography using low- frequency sound waves, an approach which has been used with some success (Levy et al., 1996). The main disadvantage with this technique is that the placing of the geo- and hydrophones required can be difficult and time consuming which makes it less appropri- ate for prospecting.

Also, https://gsa.confex.com/gsa/2011AM/final ... 192658.htm