problem in applying another PFT vegetation type dataset in CCSM4

[fenghe@wisc_edu]

Dear all,

I got the following error message when I apply another PFT vegetation type dataset in CCSM4:
c2g_1d error: sumwt is greater than 1.0 at g= 137

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
I am working on a CCSM4 (ccsm4_0_beta32) slab ocean run with another PFT dataset.
I made the surface data using mksrfdata from CCSM4 land tools in ccsm4_0_rel08:
mksurfdata < mksurfdata.namelist >&! mksurfdat.out

In mksurfdata.namelist, I replaced the default mksurfdata.namelist with the intended grid and pft file, and I attached the modified mksurfdata.namelist file in the end of this message.

The surface data was created without any error messages, and is stored on bluefire

/ptmp/fenghe/EAH/surfdata/surfdata_0192x0288.KK10_08ka.nc

I have compared this surface data with the following surface data:

/fis/cgd/cseg/csm/inputdata/lnd/clm2/surfdata/surfdata_0.9x1.25_urb3den_simyr1850_c090702

and found the only difference is PCT_PFT.

I was satisfied with this result, and built a slab ocean startup case with CLM_FORCE_COLDSTART being turned on in env_conf.xml.

This run was not successful with the following error:
/ptmp/fenghe/e40.08kakk10cs/run/lnd.log.110313-140809
c2g_1d error: sumwt is greater than 1.0 at g= 137

My impression is that there might be some problem with my input PFT raw data, but I do not know how to modify it.

It seems that the land mask in CCSM4 is larger with the amount of land it prescribes relative to ocean compared to my input PFT dataset. As the result, there are lots of barren coastal cells in pct_pft in the surface data I created (see attached plot,: left, original ncar surface data; right, surface data with modified raw pft input).

View attachment 57

View attachment 58

 

[slevis]

I think you'll have to track this down by looking at the gridcell that triggers the error message and then looking at the corresponding gridcells in the raw data. Presumably the mksrfdata code interpolates the raw data to the surface dataset. I'm not sure what else to recommend at this time.

Sam

 

[slevis]

One more idea:
Sounds as though you changed the pft data and the grid. Maybe the problem is with the modified grid.

And another suggestion:
You may wish to try re-making the default data (i.e., the surface data provided with the model) to see that you can get something that works when you don't change the grid and raw data. Then go back and make your modifications one at a time (e.g., pft modification first or grid modification first).

Sam

I think you'll have to track this down by looking at the gridcell that triggers the error message and then looking at the corresponding gridcells in the raw data. Presumably the mksrfdata code interpolates the raw data to the surface dataset. I'm not sure what else to recommend at this time.

Sam

 

[fenghe@wisc_edu]

Hi Sam,

Thank you very much for your advice.

I have identified the problem in my input PFT raw data. There is one negative value (-0.100006) in the "bear ground" PFT in northwestern Australia.

After I corrected this bug, the slab ocean model successfully ran for one month.
View attachment 59