[mpas-developers] call for testing: del2 and del4 diffusions for shallow water model

[Qingshan Chen]

Dear all,

I have been working to implement del2 and del4 diffusions for the shallow water core of MPAS. The coding is finished now. I have done some initial testings to check the sanity of the code. I also compared the output to that of a older version of shallow water code that I was using. The results are qualitatively similar. 

We are going to push the branch swmodel_del4 containing these changes back to the trunk. Todd Ringler has agreed to be one of the code evaluators. I would appreciate it if an additional person from the MPAS development team could also take a look at the code and maybe run a few test simulations. Below I will detail the changes that I have made on this project.

CHANGES
1. In core_sw/module_time_integration.F, in the subroutine compute_tend, in the LANL_FORMULATION only, routines were added to implement the del2 and del4 diffusions. The original code inside the first loop for del2 diffusion has been removed. 

2. In core_sw/module_time_integration.F, in the subroutine compute_scalar_tend, routines were added to implement the del2 and del4 diffusions for the tracers. 

3. In support for the previous two changes, the following new variables were added to core_sw/Registry and namelist.input.sw:
config_h_mom_eddy_visc2
config_h_mom_eddy_visc4
config_h_tracer_eddy_diff2
config_h_tracer_eddy_diff4

4. Also in support for Changes 1 & 2, we enforced conditional halo update on divergence and vorticity, at the beginning of the rk4 subroutine in core_sw/module_time_integration.F. We are not absolutely sure whether these updates are necessary. My testings show that they do make a difference.

5. I added the gfortran-serial entry in the main Makefile

6. On line 277 of core_sw/module_global_dynamics.F, '1i' was changed to '1i0' so that gfortran can compile. It seems that my version of gfortran (v4.4) insists on a non-negative digit as the field width after the descriptor 'i'. 

7. Changes from the trunk up to approximately v580 has been merged to the branch.

With best regards,
Qingshan