A potential bug for calculating variable Z3-geopotential height

[linlin@email_tamu_edu]

When I used variable Z3 to help interpolate variables in hybrid coordinate to height coordinate in CESM2_1 CAM6 in the single column mode with TWP06 case, I found the lowest level (largest index) of Z3 was at about 2.7km , which is a radical departure from the Z3 in CAM5.3.In CAM5.3, the lowest level of Z3 is about 200m. The "lifted up" Z3 values will lead to the corresponding "lifted up" other simulated variables, such as the cloud water etc., which is unreasonable. I checked the calculation of Z3. It is calculated and output in cam_diagnostics.F90, subroutine diag_phys_writeout_dry.

do k = 1, pver
      z3(:ncol,k) = state%zm(:ncol,k) + state%phis(:ncol)*rga
    end do

I printed out the state%zm and state%phis and found that the state%phis was one order of magnitude larger in CESM2_1 CAM6 than that in CAM5.3, which is the primary issue for the "lifted up" Z3. I am thinking most people will use variable Z3, that is the reason why I share my finding about Z3 in CESM2_1 CAM6.Hope some one could improve/fix the state%phis as soon as possible. Thanks,Lin

 

[eaton]

The code to compute Z3 is identical in cesm2_1_0 and cesm1_2_2.  This would indicate that perhaps the problem you observe is in the topo datasets.  I checked the 1-deg FV topo datasets for both model versions and found the values of PHIS to be very similar.  What grid are you using? 

 

[linlin@email_tamu_edu]

Yes, I agree with you since I myself checked the code to compute Z3 in CESM2_1, CESM2_0, CESM1_2 and found they were indentical. Z3 is compsed of state%zm and state%phis. This motivates me furhter track down state%zm and state%phis separately and finds the issue may be related to state%phis.I am running single column model using CESM2_1 CAM6. The dycore used might be Eulerian. Could you please check with the vertical interpolation in the Eulerain dycore?This is what I suspect becasue one very similar bug was detected for DOE E3SM model. Less  people use single column model and Eulerian dycore. Maybe the Eulerian dycore is not updated in the new CESM version model, namely CESM2_1 CAM6.For your convenience, my case is located at /gpfs/fs1/scratch/linlin/archive/scm_cesm2_1_cam6_twp06_snowDu_expZ3Please feel free to look at the raw Z3 from the .nc file. /gpfs/fs1/scratch/linlin/archive/scm_cesm2_1_cam6_twp06_snowDu_expZ3Thanks very much!Lin 

 

[eaton]

I looked at your file

scm_cesm2_1_cam6_twp06_snowDu_expZ3.cam.h0.2006-01-17-10800.nc

and see that PHIS=25897.07, or about 2643m above sea level.  This
doesn't seem right for a location of 131E, 12.4S.  PHIS is input data
so I'm guessing something is wrong there.  Where does the value of
PHIS for your run come from?

 

[linlin@email_tamu_edu]

Thank you for looking into this for me. As you know, I am running single column model in CESM2_1, CAM6. The only input file I can tell is the forcing file, which is a default file under /components/cam/cime_config/usermods_dirs/In addition to this forcing file, I myself did not add any other non-default files. I used pretty much the default settings.Also, I myself are not quite familar with the PHIS data. I cannot tell exactly this PHIS is provided by which file. Sorry about this. If so, probably the forcing file should be checked (the forcing file provide information of PHIS??). Or the other files used specifically for single column model simulations should be paid more attention to.My guessing is the CESM2_1 CAM6 is now 32 vertical levels. There should be something missed in the forcing file or interpolation method to comply with 32 vertical levels. Thanks a lot for your time,Lin