Investigate discretisation error presumably due to steep bedslopes combined with coarse triangles

[ole]

In the vicinity of steep slopes and shallow depths we observe triangles 'sticking' out of the water. This is due to the combined constraints

• conservation of volume
• gradient limiters
• h >= 0

In addition it may be that storing the output with unique vertex values averaged exacerbates this problem by affecting neighbouring triangles.

The remedies so far have been to refine areas affected.

We need a test that will reproduce the problem and then compare the two storage schemes. Alternatively, a comprehensive solution at the level of the numerical solver, but that has not yet been discovered by the team.

[sexton]

• harbour.py seems to be a good test for this ticket due to the nature of the bathymetry.
• harbour.py stored in ANUGA\documentation\user_manual\examples
• see harbour.sww in that directory for example
• harbour.py is a potential validation exercise