multi-pass mesh generation using a fort.64 seed

[HamishB]

Hi, an idea for your consideration--

I'm wondering if it could be useful to do multi-pass mesh generation by:

• using the existing M2 wavelength mesh size function with your input DEM in the usual manner,
• run ADCIRC with the resulting mesh and output a fort.64 file (Depth-averaged Velocity Time Series at All Nodes in the Model Grid),
• then read that fort.64.nc file back into OM2D,
• and finally run a new mesh size function based on observed velocities.

The idea being to better focus the mesh effort into areas of known high flow that the M2 shallow water wave eq. might not specifically capture, e.g. tidal straits or the inlet of an estuary which includes a large river flow as a boundary condition.

i.e. use a 'real' value for |u| in the CFL calculation instead of just estimating it.

At minimum it could be used to simply plot a map view highlighting areas where your mesh is either too coarse or needlessly fine, based on initial fort.64 results.

[krober10nd]

Hey Hamish,

I don't think it's a bad idea. We've considered this in the past, but we've always stuck to a more a priori workflow based on linear waves. What I think would make more sense is to use the max vel file to get the modeled |u| for the Courant number calculation. That really isn't difficult as it just requires msh.BoundCr takes a varargin with the modeled velocity.

Then you could pass the point set of the mesh you just simulated with and a new sizing function based on the updates to the Courant number and run the generation for a couple iterations.

[HamishB]

What I think would make more sense is to use the max vel file to get the modeled |u| for the Courant number calculation.

Yes, that would be much simpler.