CLM RUN_STARTDATE question ?

[xxyii]

Dear all,
I want to know if the RUN_STARTDATE in env_run.xml in CLM refered to the real world time? I have tried to use compset ICRUCLM45BGC of CLM4.5 to simulate the globa Gross primary production (GPP). The climate data was used the default set: 1991-2010 year of CRUNCEP.V4 dataset. When I set the RUN_STARTDATE=2016-01-01, STOP_OPTION=nyears, STOP_N=1, I got month-by-month output from January to December 2016. Obviously. the 2016 year is clearly out of date for my atmospheric forcing data(1991-2010), how can I still get results? I feel very confused and hope to get your answers

Sincerely,
Xiuyi Wu

 

[oleson]

It can be a real date in the sense that it reflects the year of the atmospheric forcing, but only if the date chosen is aligned with the date of the atmospheric forcing.
For example, an I2000 compset comes out of the box with these relevant settings:

RUN_STARTDATE: 0001-01-01
DATM_CLMNCEP_YR_START: 1991
DATM_CLMNCEP_YR_ALIGN: $DATM_CLMNCEP_YR_START (so it will be set to 1991)
DATM_CLMNCEP_YR_END: 2010

These settings are aligning model year 1991 with year 1991 of the atmospheric forcing.
The datm will calculate what year of forcing it needs to start with to eventually align model year 1991 with year 1991 of the atmospheric forcing (this ends up being year 2001 of the atmospheric forcing). In this case, the default is to cycle back over the atmospheric forcing data if the model runs longer than 2010-2001+1 (10 years).
For example, model year 0011 will use atmospheric forcing from 1991.

If you change RUN_STARTDATE to 2016-01-01 and don't change any other settings, model year 2016 will use year 1996 of the atmospheric forcing (this should be evident in the atm log).

If you change RUN_STARTDATE to 1991-01-01 (the date of the first year of the atmospheric forcing), then model year 1991 will use year 1991 of the atmospheric forcing.

 

[xxyii]

Thank you for your reply, now I understood it. And I have another question. I used the I2000 compset and set the RUN_STARTDATE: 1996-01-01 to build two case: (1) one is used the default climate forcing:1991-2010; (2) another was set for using the climate forcing of only 2006 by setiing the datm.streams.txt.CLMCRUNCEP.* I found that the results of two cases have a little difference, so I wonder know the reason for it. What is the operating mechanism of the CLM atmospheric forcing data?
I am looking forward to your reply.

Sincerely,
Xiuyi

 

[oleson]

I would need to know your settings for RUN_STARTDATE, DATM_CLMNCEP_YR_START, DATM_CLMNCEP_YR_ALIGN, and DATM_CLMNCEP_YR_END to understand what the differences might be caused by. I would also suggest comparing your atm log files for the two runs and look for differences. As far as how the datm operates there is some information here:

6. Data Atmosphere (DATM) — CIME cime5.6 documentation

esmci.github.io

and:

Redirecting...

 

[xxyii]

I would need to know your settings for RUN_STARTDATE, DATM_CLMNCEP_YR_START, DATM_CLMNCEP_YR_ALIGN, and DATM_CLMNCEP_YR_END to understand what the differences might be caused by. I would also suggest comparing your atm log files for the two runs and look for differences. As far as how the datm operates there is some information here:

6. Data Atmosphere (DATM) — CIME cime5.6 documentation

esmci.github.io

and:

Redirecting...

Dear Oleson,

I want to simulate the the year of 2006 model by setting two case.
the setting of the first case is:
RUN_STARTDATE=2006-01-01
DATM_CLMNCEP_YR_START=1991
DATM_CLMNCEP_YR_ALIGN=1991
DATM_CLMNCEP_YR_END=2010

the setting of the second case is:
RUN_STARTDATE=2006-01-01
DATM_CLMNCEP_YR_START=2006
DATM_CLMNCEP_YR_ALIGN=2006
DATM_CLMNCEP_YR_END=2006

I successfully runned two cases and checked the atm log files for the two runs and found the difference. The first screenshot is related to the first case, and the second screenshot is corresponding the second case. It looks like that the first case used the climate forcing data of 2006 and 2010, and the second one used only 2006 year climate forcing data. The results of two case are different and I wonder know which way is right to simulate the model of 2006.
I am looking forward to your reply.

Xiuyi

 

[oleson]

In the first case, the output from the datm should look something like this:

(datm_comp_run) atm: model date 20060103 54000s
(shr_dmodel_readstrm) file ub: /glade/p/cgd/tss/CTSM_datm_forcing_data/atm_forcing.datm7.cruncep_qianFill.0.5d.v7.c160715/Precip6Hrly/clmforc.cruncep.V7.c2016.0.5d.Prec.2006-01.nc 12
(shr_dmodel_readstrm) file ub: /glade/p/cgd/tss/CTSM_datm_forcing_data/atm_forcing.datm7.cruncep_qianFill.0.5d.v7.c160715/TPHWL6Hrly/clmforc.cruncep.V7.c2016.0.5d.TPQWL.2006-01.nc 12
(datm_comp_run) atm: model date 20060103 55800s
(datm_comp_run) atm: model date 20060103 57600s

So the precipitation should be read from 2006-01, not 2010-01.
I suggest looking at your datm.streams.txt.CLMCRUNCEPv7.Precip file.
It should contain a list of data from 1991 through 2010.
Also, look at your datm_in file. The line for the precip streams should look like this:

"datm.streams.txt.CLMCRUNCEPv7.Precip 1991 1991 2010",

 

[xxyii]

In the first case, the output from the datm should look something like this:

(datm_comp_run) atm: model date 20060103 54000s
(shr_dmodel_readstrm) file ub: /glade/p/cgd/tss/CTSM_datm_forcing_data/atm_forcing.datm7.cruncep_qianFill.0.5d.v7.c160715/Precip6Hrly/clmforc.cruncep.V7.c2016.0.5d.Prec.2006-01.nc 12
(shr_dmodel_readstrm) file ub: /glade/p/cgd/tss/CTSM_datm_forcing_data/atm_forcing.datm7.cruncep_qianFill.0.5d.v7.c160715/TPHWL6Hrly/clmforc.cruncep.V7.c2016.0.5d.TPQWL.2006-01.nc 12
(datm_comp_run) atm: model date 20060103 55800s
(datm_comp_run) atm: model date 20060103 57600s

So the precipitation should be read from 2006-01, not 2010-01.
I suggest looking at your datm.streams.txt.CLMCRUNCEPv7.Precip file.
It should contain a list of data from 1991 thrugh 2010.
Also, look at your datm_in file. The line for the precip streams should look like this:

"datm.streams.txt.CLMCRUNCEPv7.Precip 1991 1991 2010",

Dear Oleson,
I checked my datm_in file and found that I set the "datm.streams.txt.CLMCRUNCEPv7.Precip 1991 2010 2010" by mistake before and now I modified it. However, I still found a little difference between atm log files of two cases in the first part. The following figure is a screenshot of first case atm log. Only this place appeared the 2005-12.nc, which made me a little overwhelmed.

 

[oleson]

That's normal behavior. The datm needs the last time slice of the previous month (2005-12) to do interpolation to the model time step.