Despite what the naysayers indicate it is possible to get great and reliable results using real water. I have seen it done at a museum I worked at in my misspent youth. The question is whether that effort would be worth it. This is something you won't be able to gauge until after you have already devoted the effort to build it and run it for a while. With real water models you are also signing up for ongoing maintenance as well as for the initial building.
For waterfalls, as the above pictures show, it is possible to achieve good results using slightly modified versions of techniques you are probably already using. If that isn't good enough you might want to consider adding a recording of the actual falls would add a nice dimension.