Come up with Surface-Inundation Plot Strategy for Two-Layers

[mandli]

Usually when we plot in GeoClaw we plot the sea-surface difference from "sea-level" in originally wet areas (b < sea_level) and then the depth of the water where it was originally dry. This works well in this case as we often set sea_level = 0 making the two plots look consistent.

Unfortunately for the multi-layer code for at least one layer this is no longer the case. Say we have the top surface at sea_level = 0 and the internal surface set to -20 meters. Then if this internal layer "inundates" and we use the same strategy as above we would go from something around -20 meters to something positive blowing out the single colorbar.

Possible options:

1. Use two colorbars, one for the sea-surface and one for the inundation
2. Extend the colorbar to always contain the appropriate ranges
3. Find some other rule for plotting surface/inundation

[mandli]

For the record I think (1) will be the easiest to implement but perhaps not the most elegant, (3) would be great but I have not come up with anything more elegant than (1), and I think (2) is a bad idea for multiple reasons.

[mandli]

@aks2203 can you copy your comments you made in #271 to this issue to continue the discussion?

[aks2203]

I have some plots I made with multilayer inundation. These first set have the inundated area color bar set from white to red ({1.0:'r',0.0:'w'}), and the cross section plots only showing eta, with different colors for each surface. The cross sectional plots help depict where the overhead views have trouble, as one can see where the cells go from deep to shallow and the surface looks more elevated.

Here the inundated area has color bar {1.0:'r',0.0:'c'}, but with both of these color bars there is discontinuity where the initially wet surface meets the inundated cells.

This third set is an attempt to fix the discontinuity in color between where the surface is being plotted and the depths of the inundated cells are being plotted, but it doesn't quite look the way I'd hoped.

So I don't have a good solution, but perhaps looking at these will be helpful.