Changes to enable coupling with WW3 using 3d vortex formulation

[uturuncoglu]

This are changes in the model side that needs to be merged to complete the 3d vortex coupling. Since, schism-dev/schism-esmf#32 is already merged. this could be merged as well.

[uturuncoglu]

@josephzhang8 JFYI.

[josephzhang8]

Looks good; thx @uturuncoglu.
Shall I go ahead and merge?

[uturuncoglu]

@josephzhang8 Thanks for asking, let is wait for the discussion to solve the build issue under UFS coastal. This could include some extra work about it. More info: schism-dev/schism-esmf#32 (comment)

[uturuncoglu]

@josephzhang8 Are we expecting answer change in this version of SCHSIM? This branch is sync with the master and I am running DATM+schism configuration (no WW3). I am seeing some difference in the output when it is compared with the RT baseline. Both air_pressure and wind_speed are different. Do you think that switching from nws = 2 to nws = 4 will create issue. I'll check CDEPS side too to be sure that the data coming from there is same with the baseline.

[uturuncoglu]

@josephzhang8 I have just check that mediator history file is fine. So, the difference is caused by SCHISM.

[uturuncoglu]

@josephzhang8 Here is the details from first time step,

 air_pressure   (nSCHISM_hgrid_node,time)  t_index =      1     1
       1066     1066  (     1,     1) (     1,     1) (     1,     1) (     1,     1)
                1066   1.000000000000000E+05   1.000000000000000E+05 1.3E+03  1.000000000000000E+05 1.3E-02  1.000000000000000E+05
                1066   1.013250000000000E+05   1.013250000000000E+05          1.013250000000000E+05          1.013250000000000E+05
                1066  (     1,     1) (     1,     1)
          avg abs field values:    1.000000000000000E+05    rms diff: 1.3E+03   avg rel diff(npos):  1.3E-02
                                   1.013256718750000E+05                        avg decimal digits(ndif):  1.9 worst:  1.9
 RMS air_pressure                     1.3250E+03            NORMALIZED  1.3163E-02

 wind_speed   (two,nSCHISM_hgrid_node,time)  t_index =      1     1
                2132  (     1,     1,     1) (     1,     1,     1)
                2132   0.000000000000000E+00   0.000000000000000E+00
                2132   0.000000000000000E+00   0.000000000000000E+00
                2132  (     1,     1,     1) (     1,     1,     1)
          avg abs field values:    0.000000000000000E+00
                                   0.000000000000000E+00

We have only difference in air pressure but not with wind. Second coupling time step is also same but in third we have also difference in wind.

[josephzhang8]

Yes, differences are expected due to changes from nws=2 to 4. USE_ATMOS assumes atmos model will fill in those arrays on SCHISM mesh. Also, I added air-sea exchange for this option (which was missing before).

[uturuncoglu]

@josephzhang8 Okay. Thanks for clarification. It seems that the model is modifying the given forcing as input or through the couplign interface. Is this a correct assumption? If so maybe it would be nice to add unmodified forcing to the output too since this could confuse the user if they compare the input with the output.

[josephzhang8]

@uturuncoglu: no the code does not modify inputs like wind, pressure etc. I think this is related to reading inputs in 2 time records '1' and '2'; in _step, time interpolation is done from these 2 to get the values at the 'current' time. What are the var names you used for outputs?

[uturuncoglu]

@josephzhang8 So, we have difference due to the temporal interpolation. Is it because model internal time step is different than the coupling time step? I am just checking air_pressure and also wind_speed. Anyway, if you think that that is fine. I'll update the baselines and check other configurations. If they are fine then we could merge this PR. I'll update you when it is ready.

[josephzhang8]

Below is the code in _step that save vars read in from ATMOS model:

  do i=1,np_global
    if(ipgl(i)%rank==myrank) then
      nd=ipgl(i)%id
      windx2(nd)=rwild6(1,i)
      windy2(nd)=rwild6(2,i)
      pr2(nd)=rwild6(3,i)
      airt2(nd)=rwild6(4,i)
      shum2(nd)=rwild6(5,i)

      if(ihconsv/=0) then
        hradd(nd)=rwild6(6,i)
        srad(nd)=rwild6(7,i)
      endif !ihconsv/
      if(isconsv/=0) then
        fluxprc(nd)=rwild6(8,i)
        prec_snow(nd)=rwild6(9,i)
      endif !isconsv/
    endif !ipgl
  enddo !i

So basically, first 5 vars are saved into '2' and the others are saved as the values at this (SCHISM) time step (no time interp). Time interp is done for the first 5 vars. I think older versions only have first 3 vars.

So I think you can use the new results as RT, with the understanding that results might evolve further based on the specific requirements of the coupling. Thx.

[uturuncoglu]

@josephzhang8 Tests are good and I created baselines. This PR can be merged now. Once it is merged, I'll update UFS coastal to point recent version. Thanks for your help.