3: Starting a Historical Simulation from a Spun-Up State

3: Starting a Historical Simulation from a Spun-Up State#

By default the FHISTC_LTso starts from a default initial condition in 1979, but you can start from another year and use initial conditions coming from a previous run.

Why Start from a Spun-Up State?

In the control exercise, you ran FHISTC_LTso using its default setup, which starts around 1979.

In this exercise, you will run the same compset, FHISTC_LTso, but start the simulation in 1850 using initial conditions from a spun-up pre-industrial simulation.

This allows you to compare two historical atmosphere-only simulations:

  • fhist.control: FHISTC_LTso starting from the default initial conditions, around 1979.

  • fhist.1850: FHISTC_LTso starting in 1850 from spun-up pre-industrial initial conditions.

Both cases use FHISTC_LTso. The main differences are the model start date and the initial condition source.

By comparing these two cases, you will learn how to distinguish between:

  • The compset

  • The forcing datasets

  • The model start date

  • The source of initial conditions

Exercise: Customize your CAM history files

Create, configure, build, and run a case called fhist.1850 using:

Compset: FHISTC_LTso Resolution: ne16pg3_ne16pg3_mg17

Use the same CAM history output settings as in the control case.

Initialize the model using restart files from: b.e21.B1850.f19_g17.CMIP6-piControl-2deg.001 at model year: 0321-01-01

Start the simulation on: 1850-01-01

Run for 5 days.

Check your solution:

  • How can you check that there is a difference between the set up of this and your control?

  • How can you check that it is running the way you intended: using ssts & ghg forcings from time-varying, historical files?

  • What year is the model running?

Questions to Think About

Before looking at the solution, consider:

  • What is the default start date of fhist.control?

  • What is the start date of fhist.1850?

  • Are both cases using the same compset?

  • Are both cases using historical forcing datasets?

  • Which settings control the initial conditions?

  • Which settings control the calendar date of the simulation?

Click here for hints

How do I start my run using initialization datasets from a previous run?

  • Look at chapter Modifify the run type. In that chapter, you learned different ways to start the model, and how to use initial conditions/restart files from a different case to start up a hybrid case. Use that method for the current exercise.

For this exercise, you will create a hybrid run, which allows the model to:

Use restart files from another simulation. Start from a different calendar date. Use a different forcing period than the source simulation.

Where are restart files located?

Many spun-up simulations are available through the CESM inputdata repository (DIN_LOC_ROOT).

For this exercise, the restart files are referenced through:

b.e21.B1850.f19_g17.CMIP6-piControl-2deg.001

  • Data from spun-up model runs can be found in inputdata (DIN_LOC_ROOT) .

How do I start my run in 1850?

  • Look at the description of the xml variable RUN_STARTDATE

How do I output 3 hourly instantaneous variables?

  • Use namelist variables: nhtfrq, mfilt, fincl.

  • For more information, look at the chapter:
    NAMELIST MODIFICATIONS -> Customize CAM output

I am getting an error: This compset and grid combination is untested in CESM

  • overide this error by adding --run-unsupported to the create_newcase command

Click here for the solution

# Set environment variables

Set environment variables with the commands:

tcsh

set CASENAME=fhist.1850
set CASEDIR=/glade/u/home/$USER/cases/$CASENAME
set RUNDIR=/glade/derecho/scratch/$USER/$CASENAME/run
set COMPSET=FHISTC_LTso
set RESOLUTION=ne16pg3_ne16pg3_mg17

bash

export CASENAME=fhist.1850
export CASEDIR=/glade/u/home/$USER/cases/$CASENAME
export RUNDIR=/glade/derecho/scratch/$USER/$CASENAME/run
export COMPSET=FHISTC_LTso
export RESOLUTION=ne16pg3_ne16pg3_mg17

# Create a new case

Create a new case with the command create_newcase:

cd /glade/u/home/$USER/code/my_cesm_code/cime/scripts/
./create_newcase --case $CASEDIR --res $RESOLUTION --compset $COMPSET --run-unsupported

# Change the job queue and account number

If needed, change job queue and account number.
For instance, to run in the queue tutorial and the project number UESM0014

cd $CASEDIR
./xmlchange JOB_QUEUE=tutorial,PROJECT=UESM0014 --force

This step can be redone at anytime in the process.

# Setup

Invoke case.setup with the command:

cd $CASEDIR
./case.setup

# Customize namelists

Edit the file user_nl_cam and add the lines:

nhtfrq(2) = -3
mfilt(2) = 240
fincl2 = 'TS:I','PS:I', 'U850:I','T850:I','PRECT:I','LHFLX:I','SHFLX:I','FLNT:I','FLNS:I'
interpolate_output(2) = .true.
interpolate_nlat(2)   =  91
interpolate_nlon(2)   = 180

This creates a second CAM history stream, h1, with 3-hourly instantaneous output.

You can edit the file with a text editor. Alternatively, you can add the lines using echo:

echo "nhtfrq(2) = -3">> user_nl_cam    
echo "mfilt(2) = 240">> user_nl_cam
echo "fincl2 = 'TS:I','PS:I', 'U850:I','T850:I','PRECT:I','LHFLX:I','SHFLX:I','FLNT:I','FLNS:I'">> user_nl_cam
echo "interpolate_output(2) = .true.">> user_nl_cam    
echo "interpolate_nlat(2)   = 91">> user_nl_cam    
echo "interpolate_nlon(2)   = 180">> user_nl_cam    
echo "">> user_nl_cam

You build the namelists with the command:

./preview_namelists

This step is optional as the script preview_namelists is automatically called by case.build and case.submit. But it is nice to inspect the namelists to make sure that your changes made their way into:

$CASEDIR/CaseDocs/atm_in

# change starting date and ref case

Set the simulation start date:

./xmlchange RUN_STARTDATE=1850-01-01

Specify the reference case:

./xmlchange RUN_REFCASE=b.e21.B1850.f19_g17.CMIP6-piControl-2deg.001

Specify the restart date:

./xmlchange RUN_REFDATE=0321-01-01

Configure the run type:

./xmlchange RUN_TYPE=hybrid

Verify the settings:

./xmlquery RUN_TYPE,RUN_STARTDATE,RUN_REFCASE,RUN_REFDATE


**# Set run length**

If needed, change the ``run length``. If you want to run 5 days, you don't have to do this, as 5 days is the default.

./xmlchange STOP_N=5,STOP_OPTION=ndays

------------

Verify the Configuration
Check the Run Type

./xmlquery RUN_TYPE

Expected:

RUN_TYPE: hybrid

Check the Start Date

./xmlquery RUN_STARTDATE

Expected:

1850-01-01

Check the Reference Case

./xmlquery RUN_REFCASE,RUN_REFDATE

Expected:

RUN_REFCASE: b.e21.B1850.f19_g17.CMIP6-piControl-2deg.001 RUN_REFDATE: 0321-01-01

**# Build and submit**:

qcmd – ./case.build ./case.submit


Compare with the Control Case

Compare the settings of fhist.control and fhist.1850:

Setting	fhist.control	fhist.1850
Compset	FHISTC_LTso	FHISTC_LTso
Resolution	ne16pg3_ne16pg3_mg17	ne16pg3_ne16pg3_mg17
Forcing Protocol	Historical	Historical
Initial Conditions	Default	Restart files from spun-up simulation
Start Date	~1979	1850

The key difference is the initialization strategy and model start date.

**# Check your solution**

When the run is completed, look at the history files into the archive directory.  
    
(1) Check that your archive directory on derecho (The path will be different on other machines):

cd /glade/derecho/scratch/\(USER/archive/\)CASENAME/atm/hist ls


As your run is only 5-day, there should be no monthly file (``h0``)

Notice that the start year is `1850`.

(2) Look at the contents of the ``h1`` files using ``ncdump``.

- The file should contain the instantaneous output in the file ``h1`` for the variables:

    float FLNS(time, lat, lon) ;
            FLNS:Sampling_Sequence = "rad_lwsw" ;
            FLNS:units = "W/m2" ;
            FLNS:long_name = "Net longwave flux at surface" ;
    float FLNT(time, lat, lon) ;
            FLNT:Sampling_Sequence = "rad_lwsw" ;
            FLNT:units = "W/m2" ;
            FLNT:long_name = "Net longwave flux at top of model" ;
    float LHFLX(time, lat, lon) ;
            LHFLX:units = "W/m2" ;
            LHFLX:long_name = "Surface latent heat flux" ;
    float PRECT(time, lat, lon) ;
            PRECT:units = "m/s" ;
            PRECT:long_name = "Total (convective and large-scale) precipitation rate (liq + ice)" ;
    float PS(time, lat, lon) ;
            PS:units = "Pa" ;
            PS:long_name = "Surface pressure" ;
    float SHFLX(time, lat, lon) ;
            SHFLX:units = "W/m2" ;
            SHFLX:long_name = "Surface sensible heat flux" ;
    float T850(time, lat, lon) ;
            T850:units = "K" ;
            T850:long_name = "Temperature at 850 mbar pressure surface" ;
    float TS(time, lat, lon) ;
            TS:units = "K" ;
            TS:long_name = "Surface temperature (radiative)" ;
    float U850(time, lat, lon) ;
            U850:units = "m/s" ;
            U850:long_name = "Zonal wind at 850 mbar pressure surface" ;

Note that these variables have no ``cell_methods`` attribute becasue the output is instantaneous. 


</details>
</div>