 So now I'd like to talk a little bit about the difference of flow across a ripple, particularly at the lower flow speed end, versus what's more common in a dune. So I've been drawing the flow separation with a line like this, but as soon as the flow goes over the crest of a ripple, the flow can actually expand, and that creates some extra turbulence. And so the flow isn't actually a particular line like this, and you can get some back flow coming along when the flow intersects the bed again. So and then when the flow is deflected across the back of the next ripple that's downstream, the boundary layer re-establishes itself again. In a dune there's so much, there's a really strong flow going across from the separation point and you end up with this zone of shear because your flow speed is very low at the bottom and it's very high at the top. There's this shear zone that creates a significant amount of turbulence. And in some cases if that shear is large enough it starts dragging this low speed water into the flow and that creates a low pressure at this zone right on the downstream side of the dune and that pulls water actually in the upstream direction to fill in that low pressure zone. So this is the example of a roller vortex where at high flow speeds in front of dunes you get flow that actually goes upstream. In some cases you can even develop, if that flow is fast enough, you can actually even develop ripples between the dunes, migrating upstream. That difference in behavior of the turbulence that causes the difference between ripples and dunes. So one of the reasons that the change from ripples to other bed forms is abrupt is that ripples in general depend on the laminar part of the flow layer whereas the other parts depend on the turbulence, the overall turbulence structure. So when we talked about the Bernoulli effect we often had a small flow change and when that's happening within the laminar sublayer you tend to get ripples. Usually when it's slow enough to form ripples and find grain sizes there's no movement in the larger grain sizes. So ripples depend on the laminar sublayer whereas other bed forms depend on the structure of the turbulence. So in general when you go from ripples into dunes you're going from an effect of the laminar sublayer into the more of the turbulence of the boundary layer and the same thing again going from ripples into upper planar and you tend not to get dunes in this transition because the grains aren't large enough, the grains aren't heavy enough to pile up into dunes. But they tend to have very very long trajectories down the flow which tends to give you planar ripples.