If the 20 ton figure is correct then deaths could largely be limited to first responders if the dock is reasonably far away from the nearest residential area. But that's a big if. I haven't looked at how the areas around the dock are built up.
You can plot it in Nukemap to get a rough perspective using the 20 t Davy Crocket as a stand-in. It's not perfect, with this explosion not being a nuke and all, but the explosive forces can be compared. That's why nukes are counted in equivalent tons of TNT, after all.
http://nuclearsecrecy.com/nukemap/
20 t gives you a kill zone of 20 psi overpressure out to 60 m and 5 psi overpressure with deaths and severe injuries widespread out to 120 m. Injuries beyond that are likely to be lighter but deaths and severe injuries are still possible.
Here's a visualisation for 50 tons if you think 20 sounds too little. It's from a test in 1956. I'm not comfortable making any suggestion if 20 t is enough, the videos don't give you a good enough sense of scale, but as you can see 50 t is pretty dramatic:
http://m.youtube.com/watch?v=2_efRhXGgSQ
50 t gives you a kill zone of 80 m and widespread severe deaths and injuries 170 m out. As with 20 t people further out are still at risk. As you might have noticed, twice the yield does not equal twice the range. Explosive force falls exponentially with distance. The most common injury will probably be from shattered glass. I imagine more than a few people will probably have suffered hearing damage too.
Not taking any bets on total deaths, too little is known at this time, but I think we can all assume the current official numbers are lowballing it. They will rise when people get the chance to report them in later.