Re: [Anuga-user] Problem with water surface elevations artificially high at wet-dry transitions

[Gareth D. <gar...@gm...>]

Hi Scott,

I suspect this is just a file-storage issue -- and would suggest you 
don't modify the wet/dry thresholds.

The ANUGA sww format stores results at vertices. In practice the 
numerical method will give multiple values at each vertex (one per 
triangle that touches it), because we are using a cell-centered 
finite-volume method. By default the sww file only stores one value per 
vertex (I think using an average of the values from each contributing 
point). This will lead to "apparent" wet-dry artefacts like what you 
describe, near discontinuities in the stage.

If that's the cause, then the underlying numerical model will not have 
such artefacts. To work around it, you just need to tell ANUGA to store 
multiple values per vertex.

To do this, you can add a flag to project.py:
set_store_vertices_uniquely = True
Then call the following when setting up the domain
domain.set_store_vertices_uniquely(project.store_vertices_uniquely)

Cheers,
Gareth


On 29/5/19 1:18 am, Scott David wrote:
>
> Hi all,
>
> I am working on using anuga to model large flooding events in canyons. 
> One problem I encountered is that I have water creeping up the valley 
> walls (the resulting water stage is much higher along the walls than 
> in the center of the canyon). My assumption is that this arises from a 
> cellular roughness along the wet-dry interface of steep canyon walls. 
> I have tried increasing the model resolution and using different flow 
> algorithms, which has helped but not solved this issue. My guess here 
> is I need to specify a threshold for a depth needed for a cell to be 
> wet or dry. I have tried implementing this in my code using 
> domain.set_minimum_allowed_height(project.minD2store) and 
> domain.set_minimum_storable_height(project.minD2store) , where 
> project.minD2store is some value. However, the code seems not to 
> recognize this threshold given my results have depth values less than 
> the threshold. If anyone has suggestions as to why my script isn’t 
> using these lines of codes or any other ideas on how to minimize the 
> water surface elevation rise along canyon walls, I would appreciate it.
>
> I have tested this on anuga version 2.0.3 in Ubuntu on Windows and 
> version 2.0.2 in Ubuntu. As well, I have uploaded an example of one my 
> scripts on github, https://github.com/scottdavid2/ANUGA_modeling. The 
> file project.py sets all the variables and boundary conditions and the 
> file run_anuga_klickitat_parallel.py calls in project.py and sets up 
> the domain, initial conditions, and runs the model.
>
> Thanks,
>
> Scott David
>
>
>
> _______________________________________________
> Anuga-user mailing list
> Anu...@li...
> https://lists.sourceforge.net/lists/listinfo/anuga-user