CCSM3 Model Setup andRuntime Issues for a

Permian (250Ma)SimulationChristine Shields

Jeff Kiehl

NCAR/CGD/CCR shields@ucar.edu

Thanks to Steve Yeager for creating theocean grids and coupler mapping files.

Thanks to CSEG for making CCSM3code and scripts Paleo-inclusive.

Thanks to everyone who helped withmodel troubleshooting….

Setup Procedures for CCSM2apply to CCSM3…

For details, see “Procedures for GeneratingInput Files for CCSM2”, Brady, Yeager,Shields, Levis, Otto-Bliesner, 2003.

See Paleo WG liason, Zav Kothalava zav@ucar.edu

Model Setup AtmosphereInitial Data Set: Use available tools to create initial dataset. (See Setup documentation). You will need a Topography/Bathymetry dataset (preferably

netcdf) for your time period.

Physics Changes: Namelist parameters specify trace gases,

solar constant, aerosol/optical depth values*

Dynamics/Time Step: May need to change atmosphere/land

timestep **

Model Setup Atmosphere *Aerosol mass mixing ratio calculations are tuned to present day and require

setting a scaling factor specified in the namelist. For Paleo,set these to “0”unless you know this information for your time period. You will also need toexplicitly set an optical visible depth, but don’t use the old “tauvis”way , use anew parameter called “tauback”.

carscl (carbon scaling factor) sulscl (sulfur scaling factor) ssltscl (sea salt scaling factor dustscl (dust scaling factor) tauback (if you want a “tauvis” of .14, this parameter needs to be set to .28).

(see CAM3 documentation for details).

Model Setup Land

Surface Data Set: Use available tools to create surface

data sets. (See Setup Documentation). You will need a (2x2) vegetation map based on LSM surface

types.

Runoff Data Set: Use available tools to create runoff

dataset. You will need your topography mapped onto the (2x2) grid.

For LSM types: The NCAR Land Surface model (LSM version 1.0) Coupled To TheNCAR Community Climate Model, NCAR/TN-429+STR, pp 7.

Model Setup Land

Surface Data Specification: The raw data sets used to create the surface_data file are

specified in the namelist. mksrf_fvegtype mksrf_fsoitex mksrf_fsoicol mksrf_flanwat mksrf_fglacier mksrf_furban mksrf_flai

Once the surface_data file is created at runtime, all subsequentruns only need to specify the location of this file with theBuildscript variable,

fsurdat

Model Setup Land

Dynamics/Time Step **

If it is necessary to change the timestep, thiscan be done in the namelist. Be sure theatmosphere and the land have the sametimestep. Also be sure to adjust the runoffmodel (rtm) timestep if you wish to beconsistent with the CCSM3 control runs. Thistimestep is 3 hours. This parameter is called

rtm_nsteps.

Model Setup Land

Source code Mods: You will need to modify two subroutines for deep past

simulations. 1. clm_varpar.F90 (rtmlon and rtmlat) runoff input file lat and lon dimensions if other than .5x.5 2. mksrfdatMod.F90 You will need to COMMENT OUT the hardwired code that inserts the Ross Ice Shelf and creates glacier points over all land points south of -79. degrees latitude.

Model Setup Ocean

Grids: Use Scrip and kmtED to create POP grid and

kmt files. (See Setup Documentation for details). You will need the topography/bathymetry (netcdf) file used to

create the atmosphere initial dataset.

Use available tools to create your regionmask file.

You will need to modify the ascii “region_ids” file by hand toreflect your new region mask. This file is automatically includedwith POP input files as part of the released code.

Example View ofPermian Gridgenerated bySteve Yeagerwith kmtED…

This is a view ofthe south poleand theequatorialislands.

Model Setup Ocean

Other Initial Datasets:- Initial Temperature/Salinity* You may use one of several POP namelist options to

specify your T/S initialization.

- transport_contents You may modify this input dataset to specify where

you want the ocean model to automatically computemass transports. This file can be modified by handfrom the default file available with the released code.(You supply the “i” and “j” grid point locations).

*

User generated profiles for temperature andsalinity can be used to initialize the model.

In pop namelist, set INIT_TS_OPTION to mean,and specify TS_INIT_FILE to the ascii datasetwhich contains the profile information fortemperature and salinity.

The profile used here was taken from aCretaceous run (courtesy of Bette Otto-Bliesnerand Esther Brady), at year 1100 in hopes ofinitializing with an ocean state close(r) toequilibrium.

Model Setup Ocean

Other POP namelist options: POP now automatically computes the MOC and NHT, but this is

hardwired for present day. The simplest solution is to turn thesefeatures OFF and compute these variables off-line.

moc = .false. n_heat_trans = .false. n_salt_trans = .false.

Source code Mods: In your source code mod directory (src.pop), you will place the

modified pop_in (above), transport_contents, and region_idsfiles.

Model Setup Ice

Grids: CSIM = POP grids. By setting up the

POP grid, you have done your work forthe CSIM grid!

Initialization: For the Permian, we initialized with a no-

ice state. There is now a “no ice”initialization option in the ice Buildscript.

Model Setup Coupler

Physics Changes: Namelist parameters can be specified to control the

orbital parameters for your time period.

If Time period < 1Ma, use

orb_year (default) If Time period > 1Ma, use orb_eccen (eccentricity) orb_obliq (obliquity) orb_mvlep (moving vernal equinox)

Model Setup Coupler

Mapping Files: Use Scrip tools to create mapping files

to/from the atmosphere/ocean grids, aswell as to create a runoff mapping file.*

Again, refer to the CCSM2 SetupProcedures document.

*note: Using Scrip for deep past time periods is non-trivial… you may experience some difficulty with thistool.

Model Runtime GOTCHAS

COUPLER/OCEAN

Potential problem with POP area weights at poles.

This can be identified through the coupler history variable called“areafact_o_cpl2comp”, which is the ratio between the grid areathat the coupler “sees” versus the grid area POP “sees”. Valuesshould range from .99 – 1.02; ratios larger than this may causeinstabilities in your simulation if you have no ice at the poles. Youmay need to hand edit your pop mapping files to force POP areaweights to “1” at large ratio points.

Another solution to this problem is to lower the ocean time step fora short time to enable the model to run past the instability.

Model Runtime GOTCHAS

Watch out for shallowocean inlets…salinities mayincrease tounreasonablevalues.

The solution is to:

a) Make your inletsmarginal seas

b) “borrow” freshwater fromsomewhere elsein your ocean tofreshen youinlets.

Model Runtime GOTCHAS

**ATMOSPHERE/LAND

The time step in the atmosphere and land may need to be reduced ifthe model crashes with instabilities generated due to strong initial andboundary forcings. Note that the coupling timestep between theatmosphere and the land in the COUPLED model is once per hour,(compared to the stand-alone CAM/CLM where the coupling is EVERYtime step). The default time step for the low resolution model (T31), is30 minutes. This, in combination with the CAM/CLM coupling of everyhour, may lead to large growth in instabilities, and therefore modelcrashes. Reducing the model time step to 20 minutes has been shownto eliminate the problem.

Model Runtime GOTCHAS

LAND Surface Datasets

At runtime, the model will create a surface_data. [atmgrid]_[ocngrid].nc filebased on the mksrf_xxx_paleo.nc files created with the land setup tools, andthe atmosphere/ocean mapping files. If there are locations where the ocean“thinks” there is a land point, and the land has “thinks” that point is ocean(and has no surface information for that point), then the land model willautomatically insert wetlands at those grid points. You may want to adjustthese erroneous wetland points in the newly createdsurface_data.[atmgrid]_[ocngrid].nc file. Use the correctedsurface_data_[atmgrid]_[ocngrid].nc file for your actual model runs. This canbe easily done offline. I wrote an “ncl” script to correct the wetland points forthe Permian.

Model Runtime GOTCHAS

Run Scripts…

CCSM3 uses a new set of scripts to create the necessary build, run, andarchiving scripts used by the model. Be sure to familiarize yourself withthe new procedures, which are different from previous versions of themodel. These scripts create your namelists and buildscripts based onthe type of run, i.e. initial, branch, hybrid, etc. It is advisable to create anew set of scripts for each type of run, (for example if you want toBranch off of your initial paleo run), and make the necessarymodifications for your physics and boundary forcings, RATHER thanjust copying over the initial scripts and making the the “type”modifications yourself. These new scripts are complex and havesubtleties that are easy to miss.

Model DiagnosticsAtmosphere:Mark Steven’s AMWG packagehttp://www.cgd.ucar.edu/cms/diagnostics

Land:Keith Oleson’s LMWG Packagehttp://www.cgd.ucar.edu/tss/clm/diagnostics

Ocean:Scott Weese’s (NCAR/OCE) Spatial and Timeseries Ocean Package Contact Scott for the base code, note that this is an UNSUPPORTED

package. I have made a few changes unique to the Paleo world, so feelfree to contact me for those changes.

Ice:Julie Schramm’s (NCAR/OCE) Ice Diagnostic Package Contact Julie for the most recent code. I have not used this for the

Permian, as the Permian is basically a no-ice world! However, the codeis available but as in the ocean package, it is UNSUPPORTED.