[sancarlos]

From the theoretical standpoint we should consider the overall optical system performance, not just isolated element.
In this case system consist of laser+eye. So we need take into account the spectral sensitivity of an eye as well as absorbtion spectrum of the water. You right and green of 532nm (green lasers) is absorbed almost twice faster then blue laser (I think it is around 470nm), but human eye is almost twice more sensitive to green then to blue of those wavelengths.
So both effects are cancelling each other. BUT! We also need to concider the sun light since it is illuminating the target and produce the noise in our case, which makes the target recognition harder. And you got me on the paradoxal thought here. To make the laser dot more visible it might be benefitial to use red laser, not the green one since there is a lot of green illumination deep in the water, but almost no red! I need to do more research on that. So it just might be that in this particular application red might be better solution for optimal system performance, contrary to the initially obvious green laser.
Thanks for the great comment!