The SI beam may hit the surface, but it may also hit details on the stern even if they don't seem to obscure the beam. This is possible because the SI beam is not a single thin slice with sound as is shown in most sketches, it actually consists of a large number of thin slices which fill the angular space between the main beam and the plane of the piezo element. These sidelobes have much lower intensity than the main lobe has so they don't use to affect the image. This bundle with lobes act like one wide lobe that sometimes produces wide arcs around very reflective objects. If the stern has something very reflective on it, this can cause a line in the image.

The image below show a simulation of the SI beams in a transducer with 113 mm long piezo element at 455 kHz. This is a way to show what the beams would look like if they were visible from above. The beams are calculated down to -50 dB which is rather low so the image actually exaggerates the influence from sidelobes.

IMG]http://i477.photobucket.com/albums/rr134/rnvinc/HB/20FA515D-842E-4639-B477-AB3F2FB2C39F_zpsui9c3cin.png[/IMG]