CESM Models

To use this grid please make the following changes in scripts/ccsm_utils/Case.template/config_grids.xml

--- config_grid.xml    (revision 47680)
+++ config_grid.xml    (working copy)
@@ -93,7 +93,7 @@
 <GRID sname="ne30np4_ne30np4"       alias="ne30_ne30"   compset="(DOCN|XOCN|SOCN|AQUAP)">a%ne30np4_l%ne30np4_oi%ne30np4_r%r05_m%gx1v6_g%null_w%null</GRID>

 <GRID sname="ne60np4_ne60np4"       alias="ne60_ne60"   compset="(DOCN|XOCN|SOCN|AQUAP)">a%ne60np4_l%ne60np4_oi%ne60np4_r%r05_m%gx1v6_g%null_w%null</GRID>

 <GRID sname="ne120np4_ne120np4"     alias="ne120_ne120" compset="(DOCN|XOCN|SOCN|AQUAP)">a%ne120np4_l%ne120np4_oi%ne120np4_r%r05_m%gx1v6_g%null_w%null</GRID>

-<GRID sname="ne240np4_ne240np4"     alias="ne240_ne240" compset="(DOCN|XOCN|SOCN|AQUAP)">a%ne240np4_l%ne240np4_oi%ne240np4_r%r01_m%gx1v6_g%null_w%null</GRID>

+<GRID sname="ne240np4_ne240np4"     alias="ne240_ne240" compset="(DOCN|XOCN|SOCN|AQUAP)">a%ne240np4_l%ne240np4_oi%ne240np4_r%null_m%gx1v6_g%null_w%null</GRID>

 <!--- cism grids --> 

@@ -863,6 +863,8 @@
 <griddom grid="ne240np4" mask="gx1v6">
   <ATM_DOMAIN_FILE>domain.lnd.ne240np4_gx1v6.111226.nc</ATM_DOMAIN_FILE>
   <LND_DOMAIN_FILE>domain.lnd.ne240np4_gx1v6.111226.nc</LND_DOMAIN_FILE>
+  <OCN_DOMAIN_FILE>domain.ocn.ne240np4_gx1v6.111226.nc</OCN_DOMAIN_FILE>
+  <ICE_DOMAIN_FILE>domain.ocn.ne240np4_gx1v6.111226.nc</ICE_DOMAIN_FILE>
 </griddom>

 <gridmap atm_grid="ne240np4" ocn_grid="gx1v6">

A bug in the CESM 1.2.0 release code causes RRTMG to give wrong results on big-endian platforms (e.g. on Mira). This is caused by a bias in the output of a random number generator.

A solution is to change the RRTMG external to this fixed tag:

https://svn-ccsm-models.cgd.ucar.edu/rrtmg/trunk_tags/rrtmg_130827/src

This fix should be present in CESM 1.2.1.

A work around is to add the following to user_nl_cam in your case directory.

  prescribed_aero_datapath= '/your/path/to_inputdata/atm/cam/chem/trop_mam/aero'
  aerodep_flx_datapath= '/your/path/to_inputdata/atm/cam/chem/trop_mam/aero'

Or, you can edit these datapaths in cesm1_2_0/models/atm/cam/bld/namelist_files/use_cases/1850_cam5_pm.xml before you create a new case.

do_circulation_diags MUST be .true. in all cases with WACCM's QBO forcing enabled. Cases with qbo_use_forcing = .true. and do_circulation_diags = .false. will produce invalid output.

b40.rcp4_5.2deg.wcm.carma.bc0tg.002

This sets up a control run by default. The following must be added to user_nl_cam to get an actual nuclear winter:

ncdata = "b40.rcp4_5.2deg.wcm.carma.bc5tg.IndPak.002.cam2.i.2013-01-01-00000.nc"

The SE version of gravity_waves_sources in CESM 1.2 has a frontogenesis function ~14x larger than it should be due to a unit conversion error. The main effect is to trigger frontal waves more often and in the tropics. This only affects WACCM-SE.

This will be fixed for CESM 1.3 in the next CAM tag but a workaround patch for gravity_waves_sources is listed below.

@@ -39,7 +39,7 @@
     call initEdgeBuffer(edge3,3*nlev)
     allocate(deriv(0:Nthreads-1))

-    psurf_ref = hypi(plev+1)
+    psurf_ref = hypi(plev+1)/100._r8

   end subroutine gws_init

In WACCM-SE's gravity_waves_sources module, a race condition is present that may cause problems for some runs. For instance, it may cause the model to crash if using the Intel compiler with DEBUG set to TRUE.

To work around this you may either change gravity_waves_sources to make the variables "ie" and "ncol" thread-private, or simply run with NTHRDS_ATM set to 1.

The problem occurs when the solar_parms_file input file contains data for leap days (Feb 29) while the calendar is 'NOLEAP'. A work around to this problem is to use the following alternative datasets for solar_parms_file which do not contain leap days:

B_1850-2000_WACCM_CN
 solar_parms_file = '$DIN_LOC_ROOT/atm/waccm/solar/proxy_solar_Solomon_Richmond_1845-2008_daily_noleap_c110526.nc'

B_1955-2005_WACCM_CN
F_1955-2005_WACCM_CN
B_RCP2.6_WACCM_CN
B_RCP4.5_WACCM_CN
B_RCP8.5_WACCM_CN
 solar_parms_file = '$DIN_LOC_ROOT/atm/waccm/solar/wasolar_1948-2140_daily_noleap_c110525.nc'

In the models/drv/bld/namelist_files/namelist_definition_drv.xml variable orb_velp should be orb_mvelp this is only a problem if you attempt to set the variable. The fix is to correct the name in this file and it is fixed in https://svn-ccsm-models.cgd.ucar.edu/drv/seq_mct/trunk_tags/drvseq4_2_34

CESM allows different coupling frequencies for different models. However, there are some limitations. The coupling frequencies are set via the env_run.xml variables *_NCPL. Generally, the atmosphere, land, and ice model coupling frequencies should match. The ocean model coupling frequency should be equal to or less than the atm coupling frequency. And the runoff model coupling frequency should be greater or equal to the atm coupling frequency and less than or equal to the ocean coupling frequency. The known problem is that there are no checks at setup or run-time that these constraints are met. If they are not met, there can be problems with the simulation. In summary, users should ensure that the following is true,

  ATM_NCPL = LND_NCPL = ICE_NCPL >= ROF_NCPL >= OCN_NCPL

especially for fully active configurations.

In modifying the clm to introduce tropical crop pfts, I have come across the following problem:

The if-statements in biogeochem/CNAllocationMod.F90 and biogeochem/CNNDynamicsMod.F90 that check for soybean do not include the option for irrigated soybean. I'm correcting this in my version of clm4.5 that will run with 78 pfts.

The Bugzilla report contains detailed information regarding this bug and the fix.

The Bugzilla report contains detailed information regarding this bug and the fix.

The Bugzilla report contains detailed information regarding this bug and the fix.

When running CLM45CNDV model (compset: ICLM45CNDV) and after running 400 years, there are no trees in the Amazon. ICNDV works all fine.

The behavior has been attributed to Gordon's physiology changes to the model that improve the GPP simulation. This means that the dgvm needs retuning, and we decided not to proceed with such retuning at this time because we expect this dgvm to be replaced with Rosie Fisher's work in the not too distant future.

It will map to other high resolutions like: ne120np4, 0.23x0.31, 0.47x0.61 however. The problem is that it does the mapping, but puts garbage data values (like e317) in some fields that the program does NOT touch. They happen to be very important indexing fields that can NOT be messed with.

This is fixed when you use the new netCDF library: netCDF-4.3.0-rc3.

The file clmi.ICRUCLM45BGCCROPmp24.0241-01-01.1.9x2.5_g1v6_simyr2000_c130515.nc was interpolated using interpinic for CROP simulations. But, because of Bug 1725, interpinic gave a file that has CROP NOT spunup, but starting up from bare soil. So we plan to delete this file, and get a version that will work. Users should either wait for that file, or spin up CROP on their own using the standard procedure.

See the bugzilla report for detailed information and a suggested work around.

In mid-October, Tony Craig pointed out an issue in how the POP restarted a multi-instance run that had created restart / rpointer files: once the instance-specific files existed (due to successfully completing a run), you could not start from the beginning of the run again because the initial conditions would not be read in correctly from the restart files in the reference directory.

This bug was fixed by editing pop2.buildnml.csh in the cesm_pop_2_1_20131028 and cesm_pop_2_1_20131101 POP trunk tags, made as part of the 1.3.0 development cycle (and the fix will be incorporated in the CESM 1.2.1 minor release)

See the bugzilla report for detailed information and a suggested work around.

The ecosystem in CESM 1.2 includes climatological riverine inputs of various tracers. There is a bug for the Fe input, in that it overwrites the atmospheric Fe flux, instead of being added to it.

Detailed information and a code fix is available in the Bugzilla report.

See the bugzilla report for detailed information and a suggested work around.

See the bugzilla report for detailed information and a suggested work around.

Fixed code will be included in CESM1.3.0.

The variable scalelength was introduced into ecosys in CESM 1.2, in order to have a depth dependent remineralization length scale. It is used as a multiplier to modfiy the %diss as a function of depth. The bug is that it is not applied consistently to all uses of %diss.

The global impact of this is small. However, it can be non-negligible when mixing is deep.

Fixed code will be included in CESM1.3.0.

A fix to the CFC initialization process that went in to 1.1.1 and 1.2.0 only worked for hybrid runs, a second patch was needed to handle branch runs correctly and is included in CESM1.2.1.

The Bugzilla report contains detailed information regarding this bug and the fix.

Here are the details associated with creating the case:

create_newcase -case /glade/u/home/shields/cesm1/scripts/tests/b.day2.7.again
-res T31_gx3v7 -mach yellowstone -compset B_1850_CN

And here are the custom XML settings:

env_run.xml: RUN_TYPE = branch
env_run.xml: RUN_REFCASE = b.day2.6
env_run.xml: RUN_REFDATE = 0001-02-01

The log output with error statements and the associated fix are detailed in the Bugzilla report.

There is a similar issue described in this DiscussCESM forum post:
https://bb.cgd.ucar.edu/bug-cesm120-when-using-branch-brnchretaincasenametrue

If you try to set up an ne120_f09_g16 or ne120_f02_t12 case, the grids are not set correctly because of a missing underscore in config_grid.xml. There is no separator between the ocean grid [gx1v6 or tx0.1v2] and the r% header indicating the name of the runoff.

More details and a patch are available from the Bugzilla report.

For cesm using the pgi compiler on yellowstone add the following to the env_mach_specific for your case:

 
if( $?LSB_JOBID) then
   setenv LD_PRELOAD /opt/ibmhpc/pe1304/ppe.pami/gnu/lib64/pami64/libpami.so
endif

A port to the new edison architecture at NERSC has been completed with limited testing by adding a scripts and a Machines tag. These tags are compatible with the cesm1_2_0 release and the release in the collections directory on edison has been updated with this change. Note that when using multiple threads we exploit the SMT capability of the system and therefore the pe count for cost estimate and the Model Cost values in the timing file are a factor of 2x too large. We will correct this in a future tag.

A partial port to the machine stampede was included in the 1.2.0 release. We apologize - this port is incomplete and should not have been included in a release version of the code.

When running create_clone to duplicate a case from another source tree, I noticed that the variable CCSM_MACHDIR is not updated to the new tree. Click on the Bugzilla link for fix.

Currently in env_build.xml we have

<entry id="EXEROOT" value="/blah/$CCSMUSER/$CASE/bld" />

and in env_run.xml

<entry id="RUNDIR" value="/blah/$CCSMUSER/$CASE/run" />

In many cases you may want to change both of these variables, but you cannot currently reference one variable in the other file - that is you cannot set:

<entry id="RUNDIR" value="$EXEROOT/../run" />

The workaround is not to use $EXEROOT in env_run.xml but rather to set it individually

There is an xlf compiler problem using ne30_ne30, compsets FSTRATMAM and FAMIPC5 that requires updating the PE/threading layout. Please post a question to the DiscussCESM Forums if you need help determining an optimal layout.

Janus is being upgraded and is switching from dotkits to modules. The machine was unavailable for porting and testing at the time of this CESM release. Please post a question to the DiscussCESM Forums if you need help running on Janus after the machine becomes available again.