Hi Craig,
The volume calculations are done using the generated map data, which is always produced with all offsets (keel + tide) applied, so the differences you are seeing do not come from this.
There are, however, some differences that you might see between the min/max depths for a generated map versus the depths shown in the volume module.
The process for map generation looks like this:
grid depth data points -> triangulate -> interpolate areas between points -> smooth -> (volume calcs) -> downsample -> create contours -> downsample again -> create 3D model.
Note that volume calculations take place before down-sampling (creating a lower resolution grid for contouring/display), and that the minimum/maximum values for depth are calculated at this point. This gives (in theory) more accurate results for the volume calcs than the result you see for the finished map, although the differences should be small.
I am not clear on whether the 6.6m maximum you mention is an adjusted depth or not, but a maximum calculated depth of 6.85 m could make sense in this situation because the keel offset is added to the recorded depth, to account for the depth of water between surface and transducer. This maximum depth could be a local maxima which was lost in the downsampling, which is why you don't see it in your map depth. The zero upper depth also makes sense when you consider that you have a zero depth shoreline that ultimately forms part of the map.
Hope that makes sense,
Matt