Add ocean stochastic physics

[pjpegion]

A first cut of ocean stochastic physics necessary for coupled data assimilation is needed. Development of ocean stochastically perturbed physics tendencies (SPPT) and energetic planetary boundary layer perturbations will be included in the initial implantation.

[DeniseWorthen]

Phil--could you give an outline of how this works? I'd like to understand what exactly MOM is receiving from the stochastic physics, on what frequency, how MOM is using that information etc. If you could just flesh out the algorithm that would be helpful for my understanding of the changes you are proposing.

[pjpegion]

run_stochastic_physics_ocn creates random patterns that are correlated in space and time. These patterns are used as multipliers in the physics subroutines. SPPT randomly perturbs the physics tendencies with a number between 0 and 2. The energetic PBL stochastic scheme perturbs the KE production and dissipation with two different random numbers between 0 and 2.

For these schemes random numbers are updated on the thermodynamic time-step, and I don't see the need at this time to have the random patterns update at any of the short time-steps.

So, in short. init_stochastic_physics_ocn takes inputs from the ocean (grid layout and mpi decomposition and time-step information) to set up the random patterns, and returns the logicals do_sppt and do_epbl that control what to perturb.

run_stochastic_physics_ocn just returns the random pattern(s), and these random patterns need to be available in MOM_diabatic_driver.F90 and MOM_energetic_PBL.F90 along with either the do_sppt and do_epbl logicals, or I suppose the allocation of these arrays can depend on the logicals in the initialization step, and the deep down subroutines just check to see if the arrays are allocated.