Well, it seems fairly obvious that at a very shallow setting, where the gas bubble vents to the surface as it expands initially, you will just get a lot of water thrown up and scattered widely (and not so much shock wave transmitted through the water). Deeper down, the bubble collapses completely before it surfaces (into a set of small bubbles), and you only get a bit of turbulence and bubbles coming up.
I've had a quick look around for papers on underwater explosion dynamics - they are actually rather interesting and I think the wiki is surprisingly poor on this. Here http://www.abdn.ac.uk/~eng907/teaching/explosion-engineering/Underwater-I.pdf
is a little undergrad paper which explains things a bit better. It points out that my simplistic initial assumption that the gas bubble just expands is wrong - in fact the gas bubble expands and contracts rhythmically as it rises. Snay (1956) seems to be the authority here: here is a preview of some of his work:http://books.google.co.uk/books?id=4IEXG6GRxOUC&pg=PA3&lpg=PA3&dq=snay+1956+bubble&source=bl&ots=3LeIv0jNyy&sig=anM5zm-A44iVGkFKUVf5xFO_X9E&hl=en&sa=X&ei=WuuNT-a0IMul8gPLhojBCw&ved=0CCoQ6AEwAQ#v=onepage&q=snay%201956%20bubble&f=false
So my revised view is that an explosion creates a gas bubble and a shock wave. The shock wave takes some energy from the bubble, but by no means all. A typical initial size for a torpedo explosion seems to be a gas bubble 10-20 mt diameter, which then collapses as hydrodynamic pressure reasserts itself.
This collapse continues until the bubble is small and highly compressed (since gas is elastic). At that point it begins to re-expand, and a second shock wave is generated, losing some more energy. At each expansion pulse the bubble moves upwards; step-like rather than continuously, as I has assumed. This oscillation cycle may continue for 10 or more times, with the gas bubble adopting a toroidal rather than elongated shape, as I had pictured. For a small explosion the cycle duration may be 1/10th of a second - for a larger one it may be of the order of several seconds. These rhythmic pulses can cause considerable damage to an underwater structure, especially if they coincide with a natural vibration frequency.
The above describes an explosion in open water. Other documents I have glanced at suggest that this oscillating bubble will tend to move towards a fixed structure close to it, and, if it strikes it, it will collapse onto that structure with a jet of water coming from the opposite wall of the bubble striking the structure - rather like an 'explosively formed penetrator' for anti-tank use. That, at least, seems to match what I had assumed in my diagram! This will cause local damage if it hits a structural surface - and this jet is probably what is seen when the bubble surfaces and collapses into the air/water interface.
Comments from underwater explosives experts will be gratefully received...