Runaway SOM when adding heat flux or prescribe SST

[bowen_zhao@yale_edu]

Hi,I'm running the component set  E_1850_CAM5 with a resolution of 1.9x2.5_gx1v6 while trying increase the SST over region [10N,15N;160W,180W]. I tried both adding a constant heat flux (ranging from 0.1W/m2 to 10 W/m2) and forcing SST to restore to my reference temperature with restoring timescale of 10-30days. Nonetheless the run never succeeded and always got terminated around two or three months after running. Sometimes I got a message in cesm.log file while other times:  zisocl: Error: Tried to extend CL to the model top(shr_sys_abort) ERROR: zisocl: Error: Tried to extend CL to the model top (shr_sys_abort) WARNING: calling shr_mpi_abort() and stopping
while other times cesm.log file simply says"Primary job  terminated normally, but 1 process returneda non-zero exit code.. Per user-direction, the job has been aborted." I checked the history files. Tropical SST alrady reaches 306-310K at two or three months. The PRECC or PRECL maximum are way too large too in the atm.log file.I checked back and forth and really have no idea what I could do wrong unless the system is just too sensitive to my interested region?Below is how I modify the docn_comp_mod.F90 with changes are all labeled after Bowen 02/21/2017:!!Bowen 02/21/2017         lon_sens=(/160.0 , 180.0/)         lat_sens=(/10.0  , 15.0/)         hf_sens=0.1_r8!!Bowen 02/21/2017         do n =1,lsize         if (imask(n) /=0) then             !--- pull out h from av for resuse below ---            hn = avstrm%rAttr(kh,n)            !--- compute new temp ---            o2x%rAttr(kt,n) = somtp(n) +&               (x2o%rAttr(kswnet,n) +&  ! shortwave                 x2o%rAttr(klwup ,n) +&  ! longwave                x2o%rAttr(klwdn ,n) +&  ! longwave                x2o%rAttr(ksen  ,n) +&  ! sensible                x2o%rAttr(klat  ,n) +&  ! latent                x2o%rAttr(kmelth,n) -&  ! ice melt                avstrm%rAttr(kqbot ,n) -&  ! flux at bottom                (x2o%rAttr(ksnow,n)+x2o%rAttr(kioff,n))*latice) *&  ! latent by prec and roff                dt/(cpsw*rhosw*hn) !!Bowen 02/21/2017          if ((lon(n) >= lon_sens(1)) .and. (lon(n) = lat_sens(1)) .and. (lat(n) 0            o2x%rAttr(kt,n) =max(TkFrzSw,o2x%rAttr(kt,n))                    ! reset temp            somtp(n) = o2x%rAttr(kt,n)                                        ! save temp         endif         enddo       endif   ! firstcall
Any suggestions are highly appreciated!

 

[dbailey]

The SOM cannot be a net source or sink of heat. If you wish to add heat in the tropics, you need to remove it in the polar regions. Prescribing SST is not supported in SOM mode.Dave

 

[bowen_zhao@yale_edu]

Thank you so much! It's my first time to run a model so please forgive my ignorance.If I do want to continue with adding heat in slab ocean models, then I suspect, as you suggested, I should remove exact amont heat from the polar regions? Is it correct that it has to be exactly equal amount?

 

[bowen_zhao@yale_edu]

Thanks for your comments.But I don't think I really understand why slab ocean mdoel works in this way. Could you kingdly explain more?If I add some small heat fluxes, wouldn't the atmosphere just adjust itself by emitting more OLR?Also, I'm aware of some successful runs where they remove the radiative cooling aerosols, which is also a positive energy perturbation to the system. Then what's difference between modifying aerosol and imposing heat flux?I'm looking forward to hear from you.

 

[dajianzheng@163_com]

Hi Bowen,I am working on similar issue. Could you tell me how do you get the lon and lat variables? Thanks,Jian

 

[bowen_zhao@yale_edu]

Hi Jian, I attached a file below. This approach does not work but you could find how to get lon and lat. Search for "Bowen" to see modifications. Bowen

 

[dajianzheng@163_com]

Hi Bowen,I didn't see the attachment. Could you send it again? Thanks

 

[bowen_zhao@yale_edu]

Sorry abou that. Should be visible now.

 

[dajianzheng@163_com]

I see it now. Thank you Bowen.