Your aspect ratio is not too bad-1.8m at 3m. If the tunnel is closed and not breathing there may not be much, if any, IR signataure--I guess it would be worth a cheap try, however.

Regarding Whites TM808 (and/or Discovery TF900). It sort of depends on the nature of the ground-bedrock, soil over bedrock, wet/dry, type of material, etc.. It is very doubtful that you just set to "cave" setting and walk over it and it sounds. What you need to do is to train and adjust the unit to optimal settings for the area over a known tunnel (maybe there is a mine or something nearby. Then, you might have a chance of finding it.

Better for this would be EM31 which measures variation in conductivity of soil in upper 6m (still depends on nature of site).

Resisitivity does have a good chance of working and we have found that best for deeper tunnels. Just sticking two-pin electrodes around may or may not do it (depends on conditions). You need to use, well it would be better to use, four pin, two for current input, two for voltage drop measurement in standard arrays. Modified Bristow array is best for tunnels and I have used it around the world, although it is time consuming. THers are always looking for something cheap and quick-the earth does not often work that way.

Seismic can work but again depends on ground conditions. Same with radar but that would stand a fair change although at 4m getting a little deep.

Steve