Error in transient urban simulation using ctsm5.1.dev115

[liliyao]

Hi, I was running offline CLMBGC simulation with 1x1_smallvilleIA resolution and with dynamic LULC, using clm5.0_ctsm5.1.dev115. The surface file and land-use timeseries are attached.

I got an error when I set “do_transient_urban = .true.” and “run_zero_weight_urban = .true” in user_nl_clm. The log files are attached. (The case ran successfully without these two lines.) At the bottom of the lnd log file, it shows it has an error when reading the initial conditions.

I tested this case with transient urban turned on using some other dev versions. I got the same error when using dev111, dev112. But it ran successfully in some previous versions like dev083, dev090.

So, I was wondering how to run transient urban simulation in more recent dev versions.

Thank you very much.

 

[oleson]

What is finidat set to in your lnd_in?
What is use_init_interp set to in your lnd_in?

 

[liliyao]

Hi Dr. Oleson, in my lnd_in, the finidat is the /inputdata/lnd/clm2/initdata_map/clmi.I1850Clm50BgcCrop-ciso.1366-01-01.0.9x1.25_gx1v7_simyr1850_c200428.nc, and "use_init_interp = .true." Thanks and Happy New Year!

 

[oleson]

Thanks for that information. I'm not sure why you are getting that error. See if you can run this test in cime/scripts:

./create_test ERS_Lm25.1x1_smallvilleIA.IHistClm50BgcCropQianRs.cheyenne_gnu.clm-smallville_dynurban_monthly

That is a dynamic urban test that passes in ctsm5.1.dev115. If it runs successfully for you, then I suggest comparing your case with that case to see what might be different. It does look like you are using the surface dataset and the landuse timeseries files that the test uses, at least the names are the same.

Fyi, the run_zero_weight_urban = .true setting should not be necessary in this version of model.

 

[liliyao]

Hi Dr. Oleson. I don't have an account on cheyenne. I run models on NERSC, and I failed running this test on NERSC. It looks like this test is built for cheyenne. Do you have any other suggestions? Thank you very much.

 

[oleson]

Of course, sorry.
If you provide me with all of the steps you took to get to the point you got the error, I can try to replicate the failure on our machine.
I assume you haven't made any changes to the model code or the surface/landuse datasets?

 

[liliyao]

Thanks, Dr. Oleson. Yes, I did not make any changes to the model code or the surface/landuse datasets. I have attached my steps. Please let me know if there is anything unclear.

 

[oleson]

Thanks. Those steps work on our machine for a 5 day run.
Two things you could try when setting up a new case, although I doubt they have anything to do with the error:
1. Use the stub runoff model (SROF) instead of MOSART in your compset (however, it looks like MOSART_MODE gets set to NULL regardless).
2. Remove run_zero_weight_urban = .true. from user_nl_clm (however, it runs fine here even with it set to true).

Can you do an ncdump -h on the finidat_interp_dest.nc file in run/init_generated_files and attach the results?

 

[oleson]

Looking back at your lnd log file you posted previously, it looks like your run might be using a previously generated finidat_interp_dest.nc file that doesn't have the right number of landunits on it, instead of doing a new interpolation.
Either try creating/running your case again from the beginning, or remove the run/init_generated_files directory and run again.

 

[liliyao]

Thank you very much Dr. Oleson. Yes, after removing the run/init_generated_files directory, removing "run_zero_weight_urban = .true." from user_nl_clm, and ran again. The case ran successfully.

However, it failed again when I added "run_zero_weight_urban = .true." to user_nl_clm. The error message is same as previously described.

I need to consider the transient urban area in my future simulation. To support transient urban area, do I need to set “run_zero_weight_urban = .true." in user_nl_clm?

 

[oleson]

No, you don't need that in this version of the code. There is a variable called PCT_URBAN_MAX on the dataset that takes care of that.

 

[liliyao]

Got it. Thanks for your kind help!