WARNING: major overault of ROMS kernel to complete nesting Phase III

[arango]

The ROMS algorithms were updated to include the final Phase III of nesting. The nesting algorithms are now working for the idealized DOGBONE test case in composite and refinement grid configurations.

We get identical nesting solutions in serial with partitions, shared-memory (OpenMP), and distributed-memory (MPI) for tile partitions of 1x1, 2x2, and 3x3. I have been debugging these algorithms since early January. It was quite tedious to get this working in all these computer configurations.

The development of Phase III was very difficult and I rewrote the main interface (nesting.F) several times. I was looking for an algorithm that it is easy to adjoint. We plan to support adjoint-based algorithms (4D-Var) with nested grids in the future...

The ROMS nesting algorithms are very elegant but tricky. I assume that it will take sometime to the Users to master it. We expect fine tuning of these algorithms in the future as we build more complex nesting applications.

The information for this release will be available in the ​ROMS forum shortly.

Good luck