Known Problems and Solutions for WRF Versions 3¶
WRFV3.9.1.1¶
Recurring issues from previous version(s)¶
ERA-Interim Data Problem
When using ERA-Interim data, along with the default namelist.input option ‘surface_input_source = 3’, you may get the following error. See the solution posted in the WRFV3.8 section.-------------- FATAL CALLED --------------- FATAL CALLED FROM FILE: LINE: 2963 mismatch_landmask_ivgtyp -------------------------------------------
“random seed” Compilation Error for G3
Some compilers will complain with the following error. See the solution posted in the WRFV3.8 section.call random_seed (PUT=seed) 1 Error: Size of 'put' argument of 'random_seed' intrinsic at (1) too small (12/33) module_cu_g3.f90:3195:41:
Restart Problem
Beginning with the V3.9 release, the restart capability is broken, unless one restarts from a boundary condition time. This means if one has 6 hourly lateral boundary conditions, and if one restarts at 6 hourly interval, restart works fine. Restarting from at times between the boundary times may fail. We are still looking for a solution for this problem.Half vs. Full 1D Coefficients for Hybrid Vertical Coordinate in small_step Routine
In the original implementation of the hybrid vertical coordinate (v3.9), the cpp directives utilized capitalization to differentiate between the use of full-level (c1f, for example) and half-level (c1h, for example) 1d coeffieicent arrays. For several locations in the dyn_em/module_small_step.F routine, the use of the “mu” array with the “t_2” field incorrectly was assigned to full levels. In four statements where the half-level t_2 array is used, the half-level c1h and c2h 1d arrays replace the incorrect full-level 1d arrays c1f and c2f. See the solution posted in the WRFV3.9 section.“Ran out of valid boundary conditions in file wrfbdy_d01” infinite loop
If a user tried to start a simulation after the last LBC valid period, the WRF model would get into an infinite loop and print out repeated statements. See the solution posted in the WRFV3.9 section.THIS TIME 2000-01-24_18:00:00, NEXT TIME 2000-01-25_00:00:00 d01 2000-01-25_06:00:00 Input data is acceptable to use: wrfbdy_d01 2 input_wrf: wrf_get_next_time current_date: 2000-01-24_18:00:00 Status = -4 d01 2000-01-25_06:00:00 ---- ERROR: Ran out of valid boundary conditions in file wrfbdy_d01
Compiling Problem with GNU 7.2.0
When compiling the WRFV3 code with GNU Version 7.2.0, it fails with an internal compiler error: module_ra_rrtmg_lw.f90:5981:0. See the solution posted in the WRFV3.9.1 section.use rrlw_kg06, only : fracrefa, absa, ka, ka_mco2, & internal compiler error: in gfc_trans_use_stmts, at fortran/trans-decl.c:4920 module_ra_rrtmg_lw.f90:5981:0: internal compiler error: Abort trap: 6 gfortran: internal compiler error: Abort trap: 6 (program f951) Please submit a full bug report, with preprocessed source if appropriate.
WRFV3.9.1¶
Recurring issues from previous version(s)¶
ERA-Interim Data Problem
When using ERA-Interim data, along with the default namelist.input option ‘surface_input_source = 3’, you may get the following error. See the solution posted in the WRFV3.8 section.-------------- FATAL CALLED --------------- FATAL CALLED FROM FILE: LINE: 2963 mismatch_landmask_ivgtyp -------------------------------------------
“random seed” Compilation Error for G3
Some compilers will complain with the following error. See the solution posted in the WRFV3.8 section.call random_seed (PUT=seed) 1 Error: Size of 'put' argument of 'random_seed' intrinsic at (1) too small (12/33) module_cu_g3.f90:3195:41:
Restart Problem
Beginning with the V3.9 release, the restart capability is broken, unless one restarts from a boundary condition time. This means if one has 6 hourly lateral boundary conditions, and if one restarts at 6 hourly interval, restart works fine. Restarting from at times between the boundary times may fail. We are still looking for a solution for this problem.Half vs. Full 1D Coefficients for Hybrid Vertical Coordinate in small_step Routine
In the original implementation of the hybrid vertical coordinate (v3.9), the cpp directives utilized capitalization to differentiate between the use of full-level (c1f, for example) and half-level (c1h, for example) 1d coeffieicent arrays. For several locations in the dyn_em/module_small_step.F routine, the use of the “mu” array with the “t_2” field incorrectly was assigned to full levels. In four statements where the half-level t_2 array is used, the half-level c1h and c2h 1d arrays replace the incorrect full-level 1d arrays c1f and c2f. See the solution posted in the WRFV3.9 section.“Ran out of valid boundary conditions in file wrfbdy_d01” infinite loop
If a user tried to start a simulation after the last LBC valid period, the WRF model would get into an infinite loop and print out repeated statements. See the solution posted in the WRFV3.9 section.THIS TIME 2000-01-24_18:00:00, NEXT TIME 2000-01-25_00:00:00 d01 2000-01-25_06:00:00 Input data is acceptable to use: wrfbdy_d01 2 input_wrf: wrf_get_next_time current_date: 2000-01-24_18:00:00 Status = -4 d01 2000-01-25_06:00:00 ---- ERROR: Ran out of valid boundary conditions in file wrfbdy_d01
diff_opt=2 Problem¶
Problem: In the file dyn_em/module_diffusion_em.F, some undefined k indices were used outside k loops instead of kts. This happened in one line of the recently changed (since V3.9.1) code to fix the slope terms in diff_opt=2. This affects the v-wind component computation in the y direction at the surface over sloped terrain. This problem only occurs when using the diff_opt = 2 option, along with the use of terrain data, in the y direction (e.g., this would apply to all real data cases over land. This would not apply to any of the idealized cases, unless terrain was added. This does not apply to the hill_2d_x case, as it is x-direction specific).
Solution: Download the fix for diff_opt=2, which is a modified module_diffusion_em.F file. Put the modified file in the WRF/dyn_em directory, then go back to WRFV3/ and recompile. It is not necessary to issue a ‘clean -a’ or to reconfigure. Simply recompile, and the compilation should be fairly quick.
Out of Bounds Error in module_diffusion_em.F¶
Problem: In the dyn_em/module_diffusion_em.F file for v3.9.1, there is an out-of-bounds error in the horizontal_diffusion_v_2 subroutine. The computation of the field zy_at_v in the i,j loop (not the i,k,j loop). The “k” index for the array zy should be “kts”. This bug was introduced with v3.9.1 and fixed in v3.9.1.1. This is the original (WRONG) line (line 2907) zy_at_v(i, kts, j) = 0.5 * (zy(i, k, j) + zy(i, k + 1 , j)) Here is the correct line in v4: zy_at_v(i, kts, j) = 0.5 * (zy(i, kts, j) + zy(i, kts + 1 , j))
Solution: Modify the module_diffusion_em.F file to reflect the correct line, as indicated above. Save the file and then go back to WRFV3/ and recompile. There is no need to clean the code, or to reconfigure. A simple recompilation should suffice.
WRFV3.9¶
Recurring issues from previous version(s)¶
ERA-Interim Data Problem
When using ERA-Interim data, along with the default namelist.input option ‘surface_input_source = 3’, you may get the following error. See the solution posted in the WRFV3.8 section.-------------- FATAL CALLED --------------- FATAL CALLED FROM FILE: LINE: 2963 mismatch_landmask_ivgtyp -------------------------------------------
“random seed” Compilation Error for G3
Some compilers will complain with the following error. See the solution posted in the WRFV3.8 section.call random_seed (PUT=seed) 1 Error: Size of 'put' argument of 'random_seed' intrinsic at (1) too small (12/33) module_cu_g3.f90:3195:41:
Restart Problem¶
Problem: Beginning with this release (V3.9), the restart capability is broken, unless one restarts from a boundary condition time. This means if one has 6 hourly lateral boundary conditions, and if one restarts at 6 hourly interval, restart works fine. Restarting from at times between the boundary times may fail.
Solution: We are looking for a fix to the problem.
Ndown Program Problem¶
Problem: When using the new (since V3.9) physics suites capability, along with the ndown.exe program, a fatal error occurs (the actual numbers listed below are dependent on the physics suite chosen):
------------------ ERROR ---------------
namelist : sf_surface_physics = 2
input_files : SF_SURFACE_PHYSICS = -1 (from wrfinput files).
---- ERROR: Mismatch between namelist and global attribute SF_SURFACE_PHYSICS
NOTE: 1 namelist vs input data inconsistencies found.
--------------- FATAL CALLED -------------
Solution: Download the fix for the ndown problem, which is a modified ndown_em.F file. Place it in your WRFV3/main/ directory, and then unpack the file. You will then need to recompile the code; however, since this is not a configuration or registry change, it will not be necessary to issue a ‘clean -a’ or to reconfigure.
Using a Regional lat-lon Projection¶
Problem: Beginning with this release (V3.9), when using a lat-lon projection that is regional (NOT global), there was no check implemented to assign map factors (from what they were, to what they need to be, for backward compatibility).
Solution: Download the fix for the regional lat-lon projection, which is a modified module_initialize_real.F file. Place it in your WRFV3/dyn_em/ directory, and then unpack the file. You will then need to recompile the code; however, since this is not a configuration or registry change, it will not be necessary to issue a ‘clean -a’ or to reconfigure.
Groundwater Option with Noah-MP¶
Problem: Beginning with this release (V3.9), when using the Noah-MP ‘groundwater’ option, there were 2 “sign” errors in the code, as well as a problem with the FANGLE parameter calculation.
Solution: There are modifications to several files. The options are to update to the most recent version of the code, or to download the following tar files, unpack them, and place them in their correct corresponding directories:
Download Registry.EM_COMMON to fix the groundwater Noah-MP issue - place in WRFV3/Registry/
Download start_em.F to fix the groundwater Noah-MP issue - place in WRFV3/dyn_em/
Download module_physics_init.F to fix the groundwater Noah-MP issue - place in WRFV3/phys/
Download module_sf_noahmp_groundwater.F to fix the groundwater Noah-MP issue - place in WRFV3/phys/
Download module_sf_noahmpdrv.F to fix the groundwater Noah-MP issue - place in WRFV3/phys/
You will then need to recompile the code. Because there are changes to the registry, it will be mandatory to issue a ‘clean -a’, and then to reconfigure, prior to recompiling the code.
Half vs. Full 1D Coefficients for Hybrid Vertical Coordinate in small_step Routine¶
Problem: In the original implementation of the hybrid vertical coordinate (v3.9), the cpp directives utilized capitalization to differentiate between the use of full-level (c1f, for example) and half-level (c1h, for example) 1d coeffieicent arrays. For several locations in the dyn_em/module_small_step.F routine, the use of the “mu” array with the “t_2” field incorrectly was assigned to full levels. In four statements where the half-level t_2 array is used, the half-level c1h and c2h 1d arrays replace the incorrect full-level 1d arrays c1f and c2f.
Solution: Download the fix for the half vs. full HVC issue, which is a modified module_small_step_em.F file. Place it in your WRFV3/dyn_em/ directory, and then unpack the file. You will then need to recompile the code; however, since this is not a configuration or registry change, it will not be necessary to issue a ‘clean -a’ or to reconfigure.
“Ran out of valid boundary conditions in file wrfbdy_d01” infinite loop¶
Problem: If a user tried to start a simulation after the last LBC valid period, the WRF model would get into an infinite loop and print out repeated statements:
THIS TIME 2000-01-24_18:00:00, NEXT TIME 2000-01-25_00:00:00
d01 2000-01-25_06:00:00 Input data is acceptable to use: wrfbdy_d01
2 input_wrf: wrf_get_next_time current_date: 2000-01-24_18:00:00 Status = -4
d01 2000-01-25_06:00:00 ---- ERROR: Ran out of valid boundary conditions in file wrfbdy_d01
Solution: Download the fix for the wrfbdy_d01 infinite loop issue, which is a modified input_wrf.F file. Place the file in the share directory, and then recompile the code. There is no need to issue a ‘clean -a’ or to reconfigure prior to recompiling. See the GitHub PR for additional information.
WRFV3.8.1¶
Recurring issues from previous version(s)¶
Problem when using tmn_update
An error was introduced beginning with WRF Version 3.7 in a restart run when the tmn_update option is used. This error is present in WRF Versions 3.7, 3.7.1, 3.8 and 3.8.1. When tmn_update is activated, a moving-averaged skin temperature (TLAG) is used as a lower boundary condition for the land model. Since TLAG is left out of the restart file, any restart using the tmn_update option will start with 0 K as the lower boundary condition for the land model. This will cause the soil temperature to cool gradually, eventually affecting the skin and atmospheric temperature. The tmn_update option can be used for the following land models: SLAB, Noah, NoahMP, RUC and CLM. See the solution posted in the WRFV3.7 section.Problem with Cloud Fraction
Beginning in V3.8, a bug was introduced to cloud fraction. This is seen by most radiation schemes, when a microphysics scheme is used that does not have cloud ice (e.g. WSM3, Kessler). When WSM3 or Kessler is used, cloud fraction is not computed, and hence a radiation scheme will not see any cloud. This results in unrealistic surface temperature and incorrect upwelling long-wave radiation at the top of the atmosphere. See the solution posted in the WRFV3.8 section.ERA-Interim Data Problem
When using ERA-Interim data, along with the default namelist.input option ‘surface_input_source = 3’, you may get the following error. See the solution posted in the WRFV3.8 section.-------------- FATAL CALLED --------------- FATAL CALLED FROM FILE: LINE: 2963 mismatch_landmask_ivgtyp -------------------------------------------
“random seed” Compilation Error for G3
Some compilers will complain with the following error. See the solution posted in the WRFV3.8 section.call random_seed (PUT=seed) 1 Error: Size of 'put' argument of 'random_seed' intrinsic at (1) too small (12/33) module_cu_g3.f90:3195:41:
Restart issues¶
Problem: Tests revealed that the following options do not produce correct restart results: TEMF and global runs.
Solution: Work will be underway to fix these.
WRFV3.8¶
Recurring issues from previous version(s)¶
Problem when using tmn_update
An error was introduced beginning with WRF Version 3.7 in a restart run when the tmn_update option is used. This error is present in WRF Versions 3.7, 3.7.1, 3.8 and 3.8.1. When tmn_update is activated, a moving-averaged skin temperature (TLAG) is used as a lower boundary condition for the land model. Since TLAG is left out of the restart file, any restart using the tmn_update option will start with 0 K as the lower boundary condition for the land model. This will cause the soil temperature to cool gradually, eventually affecting the skin and atmospheric temperature. The tmn_update option can be used for the following land models: SLAB, Noah, NoahMP, RUC and CLM. See the solution posted in the WRFV3.7 section.
New Option bl_mynn_edmf for MYNN¶
Problem: This new sub-option when using MYNN (bl_pbl_physics = 5) is still experimental. Some stability issues still exist with bl_mynn_edmf = 2.
Solution: Use with caution, or wait until later.
New Tiedtke¶
Problem: Water budget issue has been improved in this version, but further improvement could still happen.
Solution: Use with caution and feedback is welcome.
Problem with Cloud Fraction¶
Problem: Beginning in V3.8, a bug was introduced to cloud fraction. This is seen by most radiation schemes, when a microphysics scheme is used that does not have cloud ice (e.g. WSM3, Kessler). When WSM3 or Kessler is used, cloud fraction is not computed, and hence a radiation scheme will not see any cloud. This results in unrealistic surface temperature and incorrect upwelling long-wave radiation at the top of the atmosphere.
Solution: Download the fix for the cloud fraction problem, which is a modified module_radiation_driver.F file. You will need to unpack the file and place it in the WRFV3/phys/ directory. Rename the file to “module_radiation_driver.F” and recompile the code. It will not be necessary to issue a ‘clean -a’ or to reconfigure. Simply recompile, and the compilation should be rather quick.
ERA-Interim Data Problem¶
Problem: When using ERA-Interim data, along with the default namelist.input option ‘surface_input_source = 3’, you may get the following error:
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: LINE: 2963
mismatch_landmask_ivgtyp
-------------------------------------------
Solution: There currently is no formal fix for this problem, but until we track down the exact cause and solution, a work-around is to simply change “surface_input_source” to 1 (instead of 3) in the namelist.input file, and then re-run real, which should correct the problem, and not cause any changes in your output.
“random seed” Compilation Error for G3¶
Problem: Some compilers will complain with the following error:
call random_seed (PUT=seed) 1 Error: Size of 'put' argument of 'random_seed' intrinsic at (1) too small (12/33) module_cu_g3.f90:3195:41:
Solution: Download this fix for the “random seed” compiling error, which is a modified module_cu_g3.F, place it in your WRFV3/phys/ directory and recompile the code.
WRFV3.7.1¶
Recurring issues from previous version(s)¶
Restart Issues
Tests revealed that the following options do not produce correct restart results: TEMF, Thompson aerosol-aware (mp_physics=28), Kain-Fritsch and Grell-type schemes with cu_rad_feedback on and global runs. Work is still underway to fix this issue.MAC Mavericks OS
Current WRF/WPS code doesn’t support CLANG, the default c compiler on this Mac OS. See the solution posted in the WRFV3.6 section.Problem with Morrison Scheme
When using the Morrison scheme without any cumulus turned on in any of the domains (for example, running a single domain with Morrison scheme only), problems exist because the Morrison scheme uses some tendency arrays from output with a cumulus scheme. However these arrays were not allocated when no cumulus scheme is used. See the solution posted in the WRFV3.7 section.sf_urban_physics options 2 and 3
An error was introduced in WRFV3.7 and remained an issue in V3.7.1, that causes the urban physics options 2 (BEP) and 3 (BEM) to fail occasionally. This was caused because memory access to some arrays were not available for these options. See the solution posted in the WRFV3.7 section.Problem when using tmn_update
An error was introduced beginning with WRF Version 3.7 in a restart run when the tmn_update option is used. This error is present in WRF Versions 3.7, 3.7.1, 3.8 and 3.8.1. When tmn_update is activated, a moving-averaged skin temperature (TLAG) is used as a lower boundary condition for the land model. Since TLAG is left out of the restart file, any restart using the tmn_update option will start with 0 K as the lower boundary condition for the land model. This will cause the soil temperature to cool gradually, eventually affecting the skin and atmospheric temperature. The tmn_update option can be used for the following land models: SLAB, Noah, NoahMP, RUC and CLM. See the solution posted in the WRFV3.7 section.Problem when using KF with NSSL schemes
Model instability can occur when using the combination of cu_physics=1 (Kain-Fritsch cumulus scheme) with mp_physics=17 or 18 (NSSL 2-moment) at typical time steps. The model may be stable at sufficiently small time step. See the solution posted in the WRFV3.7 section.
Missing Quotation in Registry.EM_COMMON File¶
Problem: In the Registry.EM_COMMON file, there is a missing quotation mark following the description for PTROPNN. Some compilers will issue a fatal error and the compiliation will not be successful.
Solution: Edit the Registry/Registry.EM_COMMON file to add the missing quotation, and then save the file. It will be necessary to issue a ‘clean -a’ and then to reconfigure before attempting to recompile the code.
sf_urban_physics¶
Problem: If using this physics option with multiple domains, and it is not set for all domains, problems are seen near the boundaries with several variables.
Solution: This must be turned on for all domains.
Problem with Morrison Scheme¶
Problem: When using the Morrison scheme without any cumulus turned on in any of the domains (for example, running a single domain with Morrison scheme only), problems exist because the Morrison scheme uses some tendency arrays from output with a cumulus scheme. However these arrays were not allocated when no cumulus scheme is used.
Solution: If you wish to use this scheme without any cumulus, you must edit Registry.EM_COMMON, and update this line, from:
package morr_two_moment mp_physics==10 - moist:qv,qc,qr,qi,qs,qg;scalar:qni,qns,qnr,qng
to
package morr_two_moment mp_physics==10 - moist:qv,qc,qr,qi,qs,qg;scalar:qni,qns,qnr,qng;state:rqrcuten,rqscuten,rqicuten
Once you update the file, save the file, and then you will need to go back to the WRFV3/ directory, issue a ‘clean -a’, reconfigure, and recompile the code.
Problem with icloud = 3¶
Problem: The new option, icloud=3, introduced in Version 3.7 has an error that causes a large cold bias in the lower atmosphere. When the option is used, the temporary values of cloud water and ice are changed to include estimates from the RH-based partial cloud scheme in the radiation driver. After the radiation is called, these fields need to be reset to the original, microphysics-only values, but they are not reset in the original implementation.
Solution: Download the fix for the problem with icloud=3, which is a modified module_radiation_driver.F file. Place it in the WRFV3/phys/ directory. You will need to rename the file to module_radiation_driver.F and then you will need to recompile the code. It is not necessary to issue a “clean -a”, or to reconfigure. Just simply recompile.
WRFV3.7¶
Recurring issues from previous version(s)¶
Restart Issues
Tests revealed that the following options do not produce correct restart results: TEMF, Thompson aerosol-aware (mp_physics=28), Kain-Fritsch and Grell-type schemes with cu_rad_feedback on and global runs. See the solution posted in the WRFV3.6 section.MAC Mavericks OS
Current WRF/WPS code doesn’t support CLANG, the default c compiler on this Mac OS. See the solution posted in the WRFV3.6 section.
RRTMG with o3input¶
Problem: An error was introduced in 3.7 code that did not allow a user to change the o3input option - it would always be 0, which means to use the default ozone profile in RRTMG, which is only a function of height. o3input = 2 uses the CAM ozone profile that has a latitudinal, height, and monthly variation. This is the recommended option with the use of the RRTMG radiation.
Solution: Download the fix for the RRTMG with o3input issue, which is a modified module_check_a_mundo.F file, place it in the WRFV3/share/ directory, and rename it to “module_check_a_mundo.F”. You will need to recompile the code, but it is not necessary to issue a ‘clean -a’ or to reconfigure.
RRTMG Fast Longwave Option¶
Problem: There are errors in this newly introduced option (ra_lw_physics=24).
Solution: Do not use this option until next release.
Input/Boundary data¶
Problem: When using GFS data that contain max wind and tropapause level data, discontinuity was introduced in wrfinput and wrfbdy files.
Solution: Edit Vtable.GFS and remove the last 10 entries (UMAXW and below).
sf_urban_physics with multiple domains¶
Problem: If using this physics option with multiple domains, and it is not set for all domains, problems are seen near the boundaries with several variables.
Solution: This must be turned on for all domains.
Problem with Morrison Scheme¶
Problem: When using the Morrison scheme without any cumulus turned on in any of the domains (for example, running a single domain with Morrison scheme only), problems exist because the Morrison scheme uses some tendency arrays from output with a cumulus scheme. However these arrays were not allocated when no cumulus scheme is used.
Solution: If you wish to use this scheme without any cumulus, you must edit Registry.EM_COMMON, and update this line, from:
package morr_two_moment mp_physics==10 - moist:qv,qc,qr,qi,qs,qg;scalar:qni,qns,qnr,qng
to
package morr_two_moment mp_physics==10 - moist:qv,qc,qr,qi,qs,qg;scalar:qni,qns,qnr,qng;state:rqrcuten,rqscuten,rqicuten
Once you update the file, save the file, and then you will need to go back to the WRFV3/ directory, issue a ‘clean -a’, reconfigure, and recompile the code.
sf_urban_physics options 2 and 3¶
Problem: An error was introduced in WRFV3.7 and remained an issue in V3.7.1, that causes the urban physics options 2 (BEP) and 3 (BEM) to fail occasionally. This was caused because memory access to some arrays were not available for these options.
Solution: Download the fix for the issue with urban physics 2 and 3, which is a modified module_sf_urban.F file, place it in your WRFV3/phys/ directory (rename it to module_sf_urban.F). You will then need to go back to WRFV3/ and recompile the code. It is not necessary to issue a “clean -a”, or to reconfigure. Just simply recompile.
Problem when using KF with NSSL schemes¶
Problem: Model instability can occur when using the combination of cu_physics=1 (Kain-Fritsch cumulus scheme) with mp_physics=17 or 18 (NSSL 2-moment) at typical time steps. The model may be stable at sufficiently small time step.
Solution: Download the fix for using KF with NSSL schemes, which is a modified module_mp_nssl_2mom.F file. Place the file into your WRFV3/phys/ directory, and then recompile WRFV3. There is no need to issue a clean -a or to reconfigure. Simply recompile the code.
Problem when using tmn_update¶
Problem: An error was introduced beginning with WRF Version 3.7 in a restart run when the tmn_update option is used. This error is present in WRF Versions 3.7, 3.7.1, 3.8 and 3.8.1. When tmn_update is activated, a moving-averaged skin temperature (TLAG) is used as a lower boundary condition for the land model. Since TLAG is left out of the restart file, any restart using the tmn_update option will start with 0 K as the lower boundary condition for the land model. This will cause the soil temperature to cool gradually, eventually affecting the skin and atmospheric temperature. The tmn_update option can be used for the following land models: SLAB, Noah, NoahMP, RUC and CLM.
Solution: The fix is to add the letter ‘r’ (for restart) in the 8th column of the Registry.EM_COMMON file, for variables TLAG, so that the original line:
state real TLAG i&j misc 1 - d=(interp_mask_field:lu_index,iswater)u=(copy_fcnm) "TLAG" "DAILY MEAN SFC TEMPERATURE OF PRIOR DAYS" "K"
becomes
state real TLAG i&j misc 1 - rd=(interp_mask_field:lu_index,iswater)u=(copy_fcnm) "TLAG" "DAILY MEAN SFC TEMPERATURE OF PRIOR DAYS" "K"
One must do ‘clean -a’, reconfigure, and recompile the code after making this registry change.
If you have not previously made any changes to your Registry.EM_COMMON file, and you are okay with replacing it, instead of making the above change, you can simply replace your current file with the corrected Registry.EM_COMMON.v37 file. Once you download the file, you will need to unpack it, change the name to Registry.EM_COMMON, and place it in your WRFV3/Registry/ directory. After this, go back to WRFV3/, issue a ‘clean -a’, reconfigure, and recompile.
(Thanks to Heimo Truhetz of University of Graz, Austria, for reporting and finding the fix for the problem.)
WRFV3.6.1¶
Recurring issues from previous version(s)¶
Restart Issues
Tests revealed that the following options do not produce correct restart results: TEMF, Thompson aerosol-aware (mp_physics=28), Kain-Fritsch and Grell-type schemes with cu_rad_feedback on and global runs. See the solution posted in the WRFV3.6 section.MAC Mavericks OS
Current WRF/WPS code doesn’t support CLANG, the default c compiler on this Mac OS. See the solution posted in the WRFV3.6 section.
Vertical Interpolation in Real (ARW Only)¶
Problem: The vertical interpolation in Real does not correctly increase the number of levels of data that it is interpolating from. The Real program will stop and report an inability to find the trapping level. This only occurs with metgrid data manufactured with the max wind or tropopause included in the GFS data, and when the WRF model top is above 20 mb.
Solution: Download the fix for the real program vertical interpolation issue, which is a modified module_initialize_real.F file, place it in the dyn_em directory, rename it to “module_initialize_real.F,” and then recompile the code (there is no need to do a ‘clean -a’ or to reconfigure).
Trouble Building WRFV3 for Various Compilers, and testing on IBM Platforms¶
Problem: Compilers that do not support the 2003 FLUSH intrinsic are unable to build WRF. Some platforms do allow the use of the CALL FFLUSH option.
Solution: Download the fix for the issue building WRFV3 for various compilers and platforms, which is a packaged file that includes modifications to the following files. Unpack the downloaded file and then place the individual files in their directories indicated below.
Makefile - in the top WRFV3/ directory
fortran_2003_fflush_test.F - in the WRFV3/tools/ directory
fortran_2003_flush_test.F - in the WRFV3/tools/ directory
configure -in the top WRFV3/ directory
module_wrf_error.F - in the WRFV3/frame/ directory
After that, do a ‘clean -a’, then reconfigure and recompile WRFV3.
Building WRFV3 with Older Compilers¶
Problem: Some older compilers complain about the 2003 standard allocate statement with the third argument in the ‘start_em.F’ file.
Solution: Download the fix for building WRFV3 with older compilers, which is a modified start_em.F file that allows the first two arguments in the allocate statement. Place the file in the WRFV3/dyn_em directory, rename the file to “start_em.F,” then recompile the code (no need to do a ‘clean -a’ or to reconfigure).
Compilation Troubles¶
Problem: There is an additional space between mpi_x_comm_size and the opening parentheses in one of the if-def lines in the module_io_quilt.F file. This causes the compile to break for some compilations, since the substitution pattern is not found.
Solution: Download the fix for the additional space compilation issue, which is a modified module_io_quilt.F file. Place it in the WRFV3/frame directory and change the name to “module_io_quilt.F”, then recompile code (no need to do a ‘clean -a’ or to reconfigure).
Building WRF Serially¶
Problem: WRF is not capable of building serially with no nest.
Solution: Download the fix for the problem with building WRF serially, which is a modified mediation_feedback_domain.F file. This has some additional code that will now allow it to build serially with no nest. Place in the WRFV3/share/ directory and rename to “mediation_feedback_domain.F”, then recompile the code (there is no need to do a ‘clean -a’ or to reconfigure).
IBM Compilers¶
Problem: Some IBM compilers (particularly older ones) complain that the code is too complex to build (referring specifically to ‘module_first_rk_step_part1.o).
Solution: Download the fix for the IBM compiler issue, which is a modified module_first_rk_step_part1.F that removes unnecessary code. Place it in the WRFV3/dyn_em/ directory and rename it to “module_first_rk_step_part1.F”, then recompile the code (there is no need to do a ‘clean -a’ or to reconfigure).
Trajectories¶
Problem: The dimensions of the trajectory variables is set to the vertical index, which causes problems when the number of trajectories is more than the number of vertical levels (default is set to 25).
Solution: Download the fix for the trajectory issue, which is a modified module_em.F file. Place it in the WRFV3/dyn_em/ directory and rename to “module_em.F”, then recompile the code (there is no need to do a ‘clean -a’ or to reconfigure).
WRFV3.6¶
Calculation for Climate Extremes¶
Problem: In V3.6, the statistics for climate extremes (option output_diagnostics = 1) will not be correct for a nested run.
Solution: Fix will be in V3.6.1 release.
Lake Physics¶
Problem: In V3.6, the lake physics option only works for a single domain.
Solution: Nesting is currently being worked on, and will be available in 3.6.1.
Bug Fix for Option fractional_seaice with sf_sfclay_physics = 1¶
Problem: In V3.6, the surface layer physics option 1 is renamed from option 11 (revised scheme) from previous versions of WRF code, where fractional_seaice option wasn’t available. When this option was renamed to 1 in V3.6, and fractional_seaice option was added, but an error is introduced. When fractional_seaice option is turned on, the surface layer physics used underneath is the old version. Thanks to Charles Seman of NOAA/OAR for pointing this out.
Solution: Download the fix for the issue with fractional_seaice and sf_sfclay_physics=1, which is a modified module_surface_driver.F file, to the phys/ directory, then recompile.
Option cu_rad_feedback¶
Problem: In V3.6 this option (which lets the radiation see the sub-grid cloud produced by some cumulus schemes) is not activated. This affects the Grell schemes (Grell 3, Grell-Freitas, and Grell-Devenyi) and the Kain-Fristch (KF) scheme (if the KF scheme is used with the “cu_rad_feedback” option activated).
Solution: Download the fix for the cu_rad_feedback issue, which is a modified module_physics_init.F file, place it in the WRFV3/phys directory, and then recompile.
Bug Fix for SBM Scheme¶
Problem: Three variables were left out in the SBM registry file in the released code. The full-version of the SBM scheme will not work with this bug.
Solution: Download fix #1 for the SBM problem, which is a modified registry.sbm file, place it in the WRFV3/Registry/ directory, and rename it to registry.sbm. Then if using the fast version of this scheme, download fix #2 for the SBM problem, which is a modified module_mp_fast_sbm.F file, place it in the WRFV3/phys directory. Do a ‘clean -a’, and recompile.
Bug Fix for Option swint_opt¶
Problem: An error is introduced in 3.6 for this option (which interpolates shortwave solar radiation to each time step at the non-radiation steps). When this option is turned on, both SWDOWN and GSW (net) should be affected, but GSW was left out. This affects SLAB, RUC, PX and CLM as well as topo_shade, and sst_skin options.
Solution: Download the fix for the swint_opt issue, which is a modified module_radiation_driver.F file, place it in the phys/ directory, module_radiation_driver.F. Recompile.
Configure File for Some OS Using PGI Compiler¶
Problem: With some mpich installations, using the PGI compiler still requires mpif90 -f90=pgf90. We have added additional options to accommodate it.
Solution: Download the fix for the issue with the configure file with some OSs and PGI, which is a modified configure_new.defaults file, place it in the arch/ directory.
Restart Issues¶
Problem: Tests revealed that the following options do not produce correct restart results: TEMF, Thompson aerosol-aware (mp_physics=28), Kain-Fritsch and Grell-type schemes with cu_rad_feedback on and global runs.
Solution: Work will be underway to fix these.
MAC Mavericks OS¶
Problem: Current WRF/WPS code doesn’t support CLANG, the default c compiler on this Mac OS.
Solution: Work is underway to resolve this issue. In the meantime, one can install gcc instead.
WRFV3.5.1¶
Nesting with 1:2 ratio¶
Problem: There is a problem with nest ratio of 1:2.
Solution: We are working on resolving the problem.
Grell-Freitas Cumulus Scheme (new in V3.5)¶
Problem: There are 2 bugs found in one of the physics files. One bug is related to erroneous use of a computed tendency that will subsequently impact the computed rainfall rates, and the other bug is related to the forcing calculation.
Solution: Download the fix for the GF cumulus scheme issue, which is a modified module_cu_gf.F file, place it in the WRFV3/phys directory, then recompile the code. There is no need to do a ‘clean -a’ or to reconfigure.
Compilation Problems¶
Problem: There are certain compilers that have a problem with the way a couple of the calculations were written in the code for files module_bl_mynn.F and module_radiation_driver.F, causing WRF to encounter errors during compilation.
Solution: Download fix #1 for the V5 compilation problem, which is a modified module_bl_mynn.F file. Then download fix #2 for the V5 compilation problem, which is a modified module_radiation_driver.F file. Place both files in the WRFV3/phys directory, then recompile the code.
Restart Failure when Using Sea-Ice Options¶
Problem: When using certain sea-ice options (namelist options: SEAICE_ALBEDO_OPT=2; SEAICE_THICKNESS_OPT=1; SEAICE_SNOWDEPTH_OPT=1) restarts were failing.
Solution: Download the fix for the restart failure when using sea-ice options, which is a modified WRFV3/Registry/Registry.EM_COMMON file. You will have to do a ‘clean -a’ and then reconfigure and recompile the code for this correction to work.
WRFV3.5¶
Bug fix for the Problem with the Run-time Control of IO stream variable sets¶
Problem: An error (seg-fault) occasionally occurs due to uninitialized data in the head of the linked-list, which should not be part of the traversal. This will affect any users who use the ‘run time control of IO stream variable sets.’
Solution: Download the fix for the run-time control problem, which is a modified WRFV3/frame/module_domain.F file, then recompile. There is no need to do a ‘clean -a’ or to reconfigure.
Bug fix for running Ideal cases em_les and em_seabreeze2d_x¶
Problem: Due to a code modification prior to the release of V3.5, landmask does not remain constant during these 2 cases, and causes inaccurate values for several variables. This was corrected prior to the release of V3.5.1, and therefore only applies to V3.5.
Solution: Download the fix for the issue with em_les and em_seabreeze2d_x, which is a modified WRFV3/dyn_em/module_big_step_utilities_em.F file, then recompile. There is no need to do a ‘clean -a’ or to reconfigure.
WRFV3.4.1¶
Recurring issues from previous version(s)¶
OpenMP And MPI For Some Physics in V3.2
Certain physics options do not work correctly with OpenMP or MPI compile. These aremp_physics = 9 (Milbrandt-Yau scheme) (OpenMP)
sf_urban_physics = 2 (BEM) (OpenMP)
ishallow = 1 (MPI)
There is no known problem with these physics options if OpenMP or MPI is not used. See the solution posted in the WRFV3.2 section.
Compile V3.3 with ifort
With the use of ifort 10.1, the model may seg fault with CAM and RRTMG radiation options with default compiler option.See the solution posted in the WRFV3.3 section.
Bug Fix For Computing Radar Reflectivity for the Thompson Scheme¶
Problem: There is a bug in computing radar reflectivity for Thompson scheme. The bug is about not utilizing spherical snow assumption used in the scheme.
Solution: Download the bug fix for the problem with computing radar reflectivity for the Thompson scheme, which is a modified phys/module_mp_radar.F file. Then recompile the code without cleaning or reconfiguring.
Problem with the ‘override_restart_timers’ namelist option¶
Problem: When using this namelist option for running restarts, in combination with using the ‘sst_update’ or ‘fdda’ options, this was not working as it should, since these use input files at set intervals.
Solution: Download the fix for the override_restart_timers option, which is a modified input_wrf.F file, into the shared/ directory, then recompile WRF, without needing to clean or reconfigure the code.
WRFV3.4¶
Recurring issues from previous version(s)¶
OpenMP And MPI For Some Physics in V3.2
Certain physics options do not work correctly with OpenMP or MPI compile. These aremp_physics = 9 (Milbrandt-Yau scheme) (OpenMP)
sf_urban_physics = 2 (BEM) (OpenMP)
ishallow = 1 (MPI)
There is no known problem with these physics options if OpenMP or MPI is not used. See the solution posted in the WRFV3.2 section.
Compile V3.3 with ifort
With the use of ifort 10.1, the model may seg fault with CAM and RRTMG radiation options with default compiler option.See the solution posted in the WRFV3.3 section.
Bug Fix For Program ndown¶
Problem: With the use of default hypsometric_opt = 2 option (introduced in 3.4), program ndown broke down. Because of this change in the default option, do not use 3.4 ndown for model data created prior to 3.4. However one may explicitly set hypsometric_opt = 1 in namelist section &domains to allow 3.4 ndown to work with old data.
Solution: Download the bug fix for the problem with ndown, which is a corrected module_initialize_real.F file, to the WRFV3/dyn_em/ directory and recompile.
Incorrect Sounding in Fire Example¶
Problem: The input_sounding file provided in test/em_fire/two_fires directory in 3.4 is incorrect, which will cause running ideal.exe to fail.
Solution: Please downlownd the correct < file here.
WRFV3.3.1¶
Recurring issues from previous version(s)¶
OpenMP And MPI For Some Physics in V3.2
Certain physics options do not work correctly with OpenMP or MPI compile. These aremp_physics = 9 (Milbrandt-Yau scheme) (OpenMP)
sf_urban_physics = 2 (BEM) (OpenMP)
ishallow = 1 (MPI)
There is no known problem with these physics options if OpenMP or MPI is not used. See the solution posted in the WRFV3.2 section.
Compile V3.3 with ifort
With the use of ifort 10.1, the model may seg fault with CAM and RRTMG radiation options with default compiler option.See the solution posted in the WRFV3.3 section.
Bug Fix for Morrison Microphysics Option¶
Problem: A bug was introduced in V3.3.1 in Morrison microphysics scheme. The bug affects cloud ice fallspeed, and hence cirrus lifetime and radiative forcing. The effect of the bug is less in shorter simulations, but significant in terms of high cloud cover and LW radiative forcing in longer-term climate runs. This bug is not present in V3.3 and earlier.
Solution: Edit phys/module_mp_morr_two_moment.F, and change line 1289 to
AIN(K) = (RHOSU/RHO(K))**0.35*AI
Or download the modified module_mp_morr_two_moment.F file. Either way, the code will need to be recompiled (but not cleaned or reconfigured).
“Hanging” Problem for Some Machines¶
Problem: For machines with relatively small per-processor memory, such as blue gene, this fix removes the problem of just “hanging” when trying to open netcdf files. For users who have no troubles with the current netcdf IO, this fix is not required.
Solution: Reduce the MaxTimes variable to ten thousand from nine hundred thousand. Please look at code external/io_netcdf/wrf_io.F90 (or external/io_pnetcdf/wrf_io.F90 if you run parallel netCDF). Change line 44 from:
integer , parameter :: MaxTimes = 900000
to:
integer , parameter :: MaxTimes = 10000
The code needs to be recompiled (no ./clean -a is necessary).
PGI dmpar compile not giving bit-for-bit results between two identical runs¶
Problem: When compiling with PGI in dmpar, the results will not be identical from one run to the next, using the exact same namelist configuration.
Solution: In the arch/configure_new.defaults file, you will see the following line:
FCOPTIM = -fastsse -Mvect=noaltcode -Msmartalloc -Mprefetch=distance:8 -Mfprelaxed # -Minfo=all =Mneginfo=all
Comment out ‘-fastsse’ so that it reads:
FCOPTIM = -O3 #-fastsse -Mvect=noaltcode -Msmartalloc -Mprefetch=distance:8 -Mfprelaxed # -Minfo=all =Mneginfo=all
You will then need to do a ‘./clean -a’ and then reconfigure and recompile.
WRFV3.3¶
Recurring issues from previous version(s)¶
OpenMP And MPI For Some Physics in V3.2
Certain physics options do not work correctly with OpenMP or MPI compile. These aremp_physics = 9 (Milbrandt-Yau scheme) (OpenMP)
sf_urban_physics = 2 (BEM) (OpenMP)
ishallow = 1 (MPI)
There is no known problem with these physics options if OpenMP or MPI is not used. See the solution posted in the WRFV3.2 section.
Problem with Program real.exe¶
Problem: There is a bug in program real which was introduced in V3.3. If you have getting a seg fault running program real, or having trouble running wrf.exe, this might be the cause.
Solution: Download the fix for the problem with real.exe, which is a modified module_initialize_real.F file, to the WRFV3/dyn_em/ directory. Recompile.
Problem with New Configure Script¶
Problem: There is a problem with the new configure script on Opensuse and Ubuntu Linux. It will fail on creating configure.wrf file, and may result in file removal in the top WRFV3/ directory. If configure.wrf is successfully created, then this fix doesn’t affect you.
Solution: If files are removed from WRFV3 directory, you will have to redownload the V3.3 tar file, or re-untar the file. Then download the fix for the problem with the new configure script, which contains a modified configure script and arch/configure_new.default file, and unpack it in the WRFV3 directory.
Bug in SKEB code¶
Problem: A bug is reported to use SKEB (stochastic kinetic-energy backscatter scheme) with a variable (idiv) used but not defined.
Solution: The change is to add
idiv = 7
after line 889 of dyn_em/module_stoch.F. Or download the fix for the bug in the SKEB code, which is the modified version of the aforementioned file.
Compile V3.3 with ifort¶
Problem: With the use of ifort 10.1, the model may seg fault with CAM and RRTMG radiation options with default compiler option.
Solution: Reduce the optimization level from -O3 to -O2 appears to resolve the problem.
WRFV3.2.1¶
Recurring issues from previous version(s)¶
Restart Issue
There are problem with restart produing idential results as continous run for physics options level 3 MYNN PBL and PX LSM. Use of BEP urban option + nesting also has a restart issue. See the solution posted in the WRFV3.1 section.Thompson Scheme
WRF Versions 3.2 and 3.2.1 contain a MAJOR bug in the Thompson et al. microphysics option number 8. The bug applies for ALL simulations using this option but is worst for high resolution convective cases because it involves the production of graupel from rain and snow collisions. Any users of this scheme should not trust the results of simulations using these code versions because, instead of creating graupel due to rain/snow collisions, the bug was causing the removal of graupel. This bug is NOT found in earlier versions of WRF.
Besides the bug fix for graupel, other changes have been made to the scheme including a more subtle bug related to rain evaporation that has existed in ALL WRF code versions containing the Thompson et al (2008) scheme since late 2006. This bug was minor in comparison but is fixed in the new code. Any users wishing to fix the rain evaporation bug (involves 2 lines of code) in previous WRF versions may contact the author for details to include this change only.
Also, the intended bug fix replacement file contains a new method to determine the graupel Y-intercept parameter as a design change as well as changes to handle mass/number imbalance checks, especially related to sedimentation. The latter changes are better mathematically and should be hardly visible in real simulations, but came about due to testing the sedimentation scheme in isolation from the rest of the code. The bug-fix code should also be released as part of the next standard WRF code release, though further changes are possible between now and Spring 2011.
See the solution posted in the WRFV3.2 section.OpenMP And MPI For Some Physics in V3.2
Certain physics options do not work correctly with OpenMP or MPI compile. These aremp_physics = 9 (Milbrandt-Yau scheme) (OpenMP)
sf_urban_physics = 2 (BEM) (OpenMP)
ishallow = 1 (MPI)
There is no known problem with these physics options if OpenMP or MPI is not used. See the solution posted in the WRFV3.2 section.
WRFV3.2¶
Recurring issues from previous version(s)¶
Restart Issue
There are problem with restart produing idential results as continous run for physics options level 3 MYNN PBL and PX LSM. Use of BEP urban option + nesting also has a restart issue. See the solution posted in the WRFV3.1 section.
Thompson Scheme¶
Problem: WRF Versions 3.2 and 3.2.1 contain a MAJOR bug in the Thompson et al. microphysics option number 8. The bug applies for ALL simulations using this option but is worst for high resolution convective cases because it involves the production of graupel from rain and snow collisions. Any users of this scheme should not trust the results of simulations using these code versions because, instead of creating graupel due to rain/snow collisions, the bug was causing the removal of graupel. This bug is NOT found in earlier versions of WRF.
Besides the bug fix for graupel, other changes have been made to the scheme including a more subtle bug related to rain evaporation that has existed in ALL WRF code versions containing the Thompson et al (2008) scheme since late 2006. This bug was minor in comparison but is fixed in the new code. Any users wishing to fix the rain evaporation bug (involves 2 lines of code) in previous WRF versions may contact the author for details to include this change only.
Also, the intended bug fix replacement file contains a new method to determine the graupel Y-intercept parameter as a design change as well as changes to handle mass/number imbalance checks, especially related to sedimentation. The latter changes are better mathematically and should be hardly visible in real simulations, but came about due to testing the sedimentation scheme in isolation from the rest of the code. The bug-fix code should also be released as part of the next standard WRF code release, though further changes are possible between now and Spring 2011.
Solution: Download the fix for the Thompson scheme problem, which is a modified module_mp_thompson.F file, and replace the one in phys/ directory. Recompile.
Uninitialized Variable Problem¶
Problem: A problem with potentially uninitialized ProcOrient field in linked lists of state variables is uncovered in V3.2. This could lead to sporatic hanging of MPI jobs.
Solution: Download the fix for the uninitialized variable problem, which is a modified gen_allocs.c file, and replace the same file in WRFV3/tools/ directory, and do a full recompile that begins with ‘clean -a’.
Grid-nudging File Generation Error¶
Problem: There is an error introduced in V3.2 that causes some fields being incorrectly zeroed out in the real program when writing out the wrffdda files.
Solution: Download the fix for the grid-nudging file generation error, which is a modified reg_parse.c file, and replace the one in WRFV3/tools/ directory. Do ‘clean -a’ (note, one must use ‘-a’), and recompile.
If you are compiling with WRF-Chem, there is another change needed in Registry.EM_CHEM: search for string ‘igr’, and change it to ‘i{10}r’. Do ‘clean -a’, and then recompile.
Additional namelist Required To Specify IO Format Option¶
Due to expanded IO capability in V3.2 (runtime IO and increased IO streams), several runtime options require additional namelist.
sst_update: it requires io_form_auxinput4 = 2 in &time_control
find_input_stream: it requires io_form_auxinput2 = 2 in &time_control
Other namelists like io_form_history, io_form_input, io_form_boundary and io_form_restart are also required in the namelist.input.
Possible Performance Hit with V3.2 code¶
Problem 1: Two fixes introduced in V3.2 may result in performance hit. One of them is the change for RRTM longwave radiation option. Since the fix involves adding model layers above model top, the lower the model top (p_top), the more likely to see a slow down.
Solution: One solution is to increase the time interval to call the scheme hence maintain the same wallclock time. The other solution is to modify phys/module_ra_rrtm.F, line 31, and change layer depth deltap from 4 mb to a larger value. However with larger layer depth, one may not obtain optimal results.
Problem 2: The other performance hit is related to adding semi-lagrangian fall terms in all WSM/WDM schemes. This could affect finer grid model run more.
Compile V3.2 with ifort¶
With the new code, one must ust at least ifort 10 to compile
OpenMP And MPI For Some Physics in V3.2¶
Problem: Certain physics options do not work correctly with OpenMP or MPI compile. These are
mp_physics = 9 (Milbrandt-Yau scheme) (OpenMP)
sf_urban_physics = 2 (BEM) (OpenMP)
ishallow = 1 (MPI)
There is no known problem with these physics options if OpenMP or MPI is not used.
Solution: Do not attempt to use the above listed options with OpenMP or MPI compile.
WRFV3.1.1¶
Recurring issues from previous version(s)¶
Occasional Spurious High PBL Height In MYJ Scheme
MYJ PBL scheme occasionally produces unrealistically high PBL and high TKE in sporadic columns. This is reported by PSU researchers. See the solution posted in the WRFV3.0 section.OpenMP compile
Certain physics options do not work correctly with OpenMP compile. These are cu_physics = 3 (Grell-Devenyi) and 5 (new Grell 3D scheme); mp_physics = 7 (Goddard GCE) and 10 (Morrison); and ra_lw_physics = 3, ra_sw_physics = 3 (both CAM radiation). There is no known problem with these physics options if OpenMP is not used. Likewise, there is no known OpenMP problems with other physics options. See the solution posted in the WRFV3.0 section.Problem with the Use of GWD Option with MYJ PBL
The model may abort if gravity wave drag option (gwd_opt = 1) is used with MYJ PBL. See the solution posted in the WRFV3.1 section.Problem with DFI Option
The model may abort when DFI option is used. See the solution posted in the WRFV3.1 section.Restart Issue
There are problem with restart produing idential results as continous run for physics options level 3 MYNN PBL and PX LSM. Use of BEP urban option + nesting also has a restart issue. See the solution posted in the WRFV3.1 section.
Running WRF with OpenMP¶
Problem: If you are running WRF model with OpenMP (smpar compile) and you have set something like ‘unlimit’ or ‘ulimit -s unlimited’, or set MP_STACK_SIZE to 64000000 or larger, the model still crashes, this may be a problem of over specifying stack size.
Solution: Set stack size sufficiently large, but not unlimited value. A few GB is usually more than enough. If your computer or compiler does not recognize MP_STACK_SIZE, try KMP_STACKSIZE and/or OMP_STACKSIZE.
WRFV3.1¶
Recurring issues from previous version(s)¶
Occasional Spurious High PBL Height In MYJ Scheme
MYJ PBL scheme occasionally produces unrealistically high PBL and high TKE in sporadic columns. This is reported by PSU researchers. See the solution posted in the WRFV3.0 section.OpenMP compile
Certain physics options do not work correctly with OpenMP compile. These are cu_physics = 3 (Grell-Devenyi) and 5 (new Grell 3D scheme); mp_physics = 7 (Goddard GCE) and 10 (Morrison); and ra_lw_physics = 3, ra_sw_physics = 3 (both CAM radiation). There is no known problem with these physics options if OpenMP is not used. Likewise, there is no known OpenMP problems with other physics options. See the solution posted in the WRFV3.0 section.
Problem with the Use of GWD Option with MYJ PBL¶
Problem: The model may abort if gravity wave drag option (gwd_opt = 1) is used with MYJ PBL.
Solution: Download the fix for the problem with the use of the GWD option with MYJ PBL, which contains a modified module_pbl_driver.F and module_bl_gwdo.F file, into WRFV3 directory. Unzip and untar the file there, and it will replace the two modules in phys/ directory. Recompile. This fix should work for V3.1.1 code too.
Problem with DFI Option¶
Problem: The model may abort when DFI option is used.
Solution: Download the fix for the problem with DFI, which is a modified dfi.F file, to the WRFV3/share/ directory, and rename it to dfi.F. Recompile. This fix works for V3.1.1 code too.
Bug In Using Urban Physics Options¶
Fixed in V3.1.1.
Problem: If you experience problems with using urban physics option with nesting using V3.1 code (segamentation fault sometimes), please consider update to V3.1.1 release.
Bug In Spectral Nudging¶
Fixed in V3.1.1
Problem: A major bug is uncovered in V3.1 code, and it effectively disabled the function.
Solution: Edit phys/module_fddagd_driver.F, and change line 335 to
IF( config_flags%grid_fdda /= 0 ) THEN
Recompile.
Compilation Problem on IBM Using XLF Compiler¶
Fixed in V3.1.1
Problem: Several users have reported errors compiling WRFV3.1 under version 12.1.0.0 and 12.1.0.3 of the IBM XLF Fortran compilers. Errors are of the form:
"filename.f90", 1517-023 (S) The aliasing table size is insufficient. Recompile specifying the ALIAS_SIZE option with a value of 1073741824."
The errors have been reported for share/wrf_timeseries.F and phys/module_radiation_driver.F.
Solution: Disregard the instruction in the error message concerning ALIAS_SIZE. Instead, modify the two files to move the USE module_domain statement so that it appears before rather than after the USE module_dm like this (delete the lines with “-” and add the lines with “+”):
Index: share/wrf_timeseries.F
===================================================================
--- share/wrf_timeseries.F (revision 3829)
+++ share/wrf_timeseries.F (working copy)
@@ -367,8 +367,8 @@
SUBROUTINE write_ts( grid )
+ USE module_domain
USE module_dm
- USE module_domain
IMPLICIT NONE
Index: phys/module_radiation_driver.F
===================================================================
--- phys/module_radiation_driver.F (revision 3829)
+++ phys/module_radiation_driver.F (working copy)
@@ -1246,8 +1246,8 @@
,kts, kte &
,num_tiles )
+ USE module_domain
USE module_dm
- USE module_domain
USE module_bc
USE module_model_constants
Observation Nudging for Nests¶
Fixed in V3.1.1
Problem: The user specified horizontal radius of influence for nests (obs_rinxy) is not applied correctly. This applies to all versions since WRFV2.2. The code incorrectly divides the user-input value obs_rinxy by the ratio of the nest gridsize to coarse-mesh gridsize. In addition, errors have been found with nest observation intialization that would cause observation not to be recognized on a nest.
Solution: Download the fix for the problem with observation nudging for nests, which is a modified module_fddaobs_rtfdda.F file, to WRFV3/phys/ for V3.1 and rename it and recompile. For V2.2 code, modify the line below in phys/module_fddaobs_rtdda.Fi:
schnes=1/float(3**(inest-1))
to
schnes=1.0
Compiling tc_em.F on Some Platforms¶
Fixed in V3.1.1*
Problem: If you encounter a compile problem with tc_em.F (e.g. when using ifort), it might be due to duplicate declaration of variables itmp and icnt. Note also this does not affect compiling other executables. Also note that tc.exe needs to be compiled serially. If you use this program, compile it separately from other WRF programs.
Solution: Edit main/tc_em.F, and remove one set of declaration of these variables. Still requires serial and no-nesting compile for this utility.
Restart Issue¶
Problem: There are problem with restart produing idential results as continous run for physics options level 3 MYNN PBL and PX LSM. Use of BEP urban option + nesting also has a restart issue.
Solution: A fix for PX LSM has been provided by EPA. Edit Registry.EM and change the following two lines to (a letter ‘r’ is added in the 8th column) (included in V3.1.1 bug fix release):
+state real soilctop isj misc 1 Z i012r "SOILCTOP" "SOIL CAT FRACTION (TOP)" ""
+state real soilcbot isj misc 1 Z i012r "SOILCBOT" "SOIL CAT FRACTION (BOTTOM)" ""
No solutions yet for Level 3 MYNN PBL and BEP + nesting. Working on the problem for future release.
New Thompson Scheme with g95¶
Probably Fixed in V3.1.1
Problem: We have encountered some problem with the new Thompson scheme with the use of g95 compiler: the model cannot pass the initialization phase.
Solution: Use the old Thompson scheme (98) for the time being until it is fixed.
Nest Ratio 1:2 Problem¶
Fixed in v3.1.1
Problem: Nest ratio 1:2 is not working correctly in current and 3.0* releases.
Solution: Working on a solution. In the mean time, use another nest ratio.
WRFV3.0.1.1¶
Recurring issues from previous version(s)¶
Occasional Spurious High PBL Height In MYJ Scheme
MYJ PBL scheme occasionally produces unrealistically high PBL and high TKE in sporadic columns. This is reported by PSU researchers. See the solution posted in the WRFV3.0 section.OpenMP compile
Certain physics options do not work correctly with OpenMP compile. These are cu_physics = 3 (Grell-Devenyi) and 5 (new Grell 3D scheme); mp_physics = 7 (Goddard GCE) and 10 (Morrison); and ra_lw_physics = 3, ra_sw_physics = 3 (both CAM radiation). There is no known problem with these physics options if OpenMP is not used. Likewise, there is no known OpenMP problems with other physics options. See the solution posted in the WRFV3.0 section.Using Shared Libraries with NETCDF 3.6.2 and Later
If one encounters the following problem at runtime, it probably means that you are using netcdf 3.6.2 with shared library build enabled. See the solution posted in the WRFV3.0 section.>wrf.exe wrf.exe: error while loading shared libraries: libnetcdf.so.4: cannot open shared object file: No such file or directory
netCDF 4 WRF does not yet support netcdf 4. Please use netcdf 3 instead, and 3.6+ if you have file sizes greater than 2 Gb.
Bug Fix for Moving Nest Runs¶
Problem: Nest vertical motion at the time when the nest moves was reset to initial values (mostly zeros). It should be from previous nest and interpolated from the parent domain at the leading edge. This was an error introduced in V3 code.
Solution: Download the bug fix for moving nest runs, which is a corrected start_em.F file, to WRFV3/dyn_em/, rename the file, and recompile. Note that this error will change results.
Bug Fix for Time Series Output¶
Problem: An error was reported in computing earth-relative wind component for time series output - they were rotated by the negative of the correct angle.
Solution: Download the bug fix for time series output, which is a corrected wrf_timeseries.F file, to WRFV3/share/, rename the file, and then recompile WRF.
Bug Fix for Restart When Adaptive Time Stepping Is Used¶
Problem: When adaptive time stepping is activated, the model would not restart properly.
Solution: Download the bug fix for restart with adaptive time stepping, which is a corrected adapt_timestep_em.F file, to WRFV3/dyn_em/, rename the file, and then recompile WRF.
Bug Fix for GRIB 1 Model Output¶
Problem: An error in one of the Grib flags was found by the AFWA operational group, causing the winds to be interpreted as “earth relative”.
Solution: Download the bug fix for GRIB1 model output, which is a corrected grib1_routines.c file, to WRFV3/external/io_grib1/*, rename and recompile.
Bug Fix for RRTM Longwave Radiation Scheme¶
Problem: Two bugs were discovered by Mike Iacono of AER for the RRTM longwave radiation scheme (ra_lw_physics = 1). The first is the ozone profile which is upside down, and the second is a missing factor of 0.6 to get the ozone to volumn mixing ratio. These would affect long simulations but can be up to 0.5 K per day less radiative cooling in the upper and lower levels, little effect at the mid-levels, and about 8W/m-2 at the surface.
Solution: Download the bug fix for RRTM longwave radiation, which is a modified module_ra_rrtm.F file, to directory phys/, rename and recompile.
Minor Update for 2-Moment Morrison Scheme¶
Problem: The update provides bug fixes for calculation of evaporation of melted snow and graupel. It also includes changes to the threshold mixing ratios required for rime-splintering ice multiplication.
Solution: Download the update for the 2-moment Morrison Scheme, which is a modified module_mp_morr_two_moment.F file, to directory phys/, rename and recompile.
Bug Fix for Restart Run Using Periodic BC, TKE Diffusion Option and DM Parallelization¶
Problem: There is a bug in restart runs when TKE diffusion option is used with periodic boundary condition and DM parallelization.
Solution: Download the bug fix for a restart run using periodic BC, TKE diffusion, and DM parallelization, which contains a modified Registry.EM file, and start_em.F file, into the WRFV3/ directory, unzip and untar it. Do ‘clean -a’, and recompile.
Bug Report with Positive-Definite, Periodic BC and DM Parallelization¶
Problem: When positive-definite advection option is used with periodic boundary condition and DM parallelization, minor differences exist compared to run made with single processor, initially at the boundaries.
Solution: Working on a solution, or run without positive-definite advection option.
Compile Error When Using ESMF for Coupling¶
Problem: If one uses ESMF for coupling purpose, one might encounter an compile error for module external/io_esmf/module_esmf_extensions.F90:
module_esmf_extensions.f, line 450: This name has already been assigned a data type. [SN] INTEGER(ESMF_KIND_I4) :: year, seconds, Sn, Sd
Solution: Edit this routine and change the following line from
INTEGER(ESMF_KIND_I4) :: year, seconds, Sn, Sd
to
INTEGER(ESMF_KIND_I4) :: year, seconds
Do ‘clean -a’ and recompile.
WRFV3.0.1¶
Recurring issues from previous version¶
See the solutions posted in the WRFV3.0 section.
Occasional Spurious High PBL Height In MYJ Scheme
MYJ PBL scheme occasionally produces unrealistically high PBL and high TKE in sporadic columns. This is reported by PSU researchers.OpenMP compile
Certain physics options do not work correctly with OpenMP compile. These are cu_physics = 3 (Grell-Devenyi) and 5 (new Grell 3D scheme); mp_physics = 7 (Goddard GCE) and 10 (Morrison); and ra_lw_physics = 3, ra_sw_physics = 3 (both CAM radiation). There is no known problem with these physics options if OpenMP is not used. Likewise, there is no known OpenMP problems with other physics options.Using Shared Libraries with NETCDF 3.6.2 and Later
If one encounters the following problem at runtime, it probably means that you are using netcdf 3.6.2 with shared library build enabled.>wrf.exe wrf.exe: error while loading shared libraries: libnetcdf.so.4: cannot open shared object file: No such file or directory
Compile Optimization Issue on IBM Power Series¶
Posted 8/05/08 - fixed in version 3.0.1.1
Problem: The default optimization level for the WRF compilation on IBM Power Series is -O3 -qhot. This level is considered an aggressive optimization. Compared with -O2 (which was used in WRF V2), the -O3 -qhot optimization level gives an approximate 30% performance speed-up. It is known that the cu_physics=99 option is especially impacted with the -O3 -qhot optimization level.
Solution: If you have a concerns about the WRF default optimization level, you may lower this in the configure.wrf file to either -O3 (without -qhot) or further reduce it to -O2.
Maximum number of processors for given domain sizes¶
Posted 8/05/08 - fixed in version 3.0.1.1
Problem: The ARW core requires that no patch (subdomain allocated to an MPI task) be no smaller in its horizontal dimensions than 6x6 for coarse domain and no smaller than 9x9 for a nest. For each dimension of each of your domains, divide the domain width by the number of tasks over which that domain will be decomposed. The result should not be less than 6 for the coarse domain and not less than 9 for any nest. For example, if you are running a simulation with a nested domain that is 424 in X and 325 in Y, then the largest number of tasks in X is 47 and in Y is 36 (for a total of 1692). The smallest number in X and Y over all domains is the most tasks you can run for the simulation.
The correct number of processors in X and Y might not be the number in X and Y that the model computes automatically. For example, running with 16 MPI tasks, the model will come up with 4 processors in X and 4 in Y on its own. If you need 4 in X and 2 in Y, you must override the model’s choice by specifying nproc_x and nproc_y in the domains section of the namelist.input file.
Error behavior:
Beginning with version 3.0.1, the model will generate a message:
tfp_message: module_dm.b <line number>
and abort. Prior to 3.0.1, the behavior is more dangerous: the code may still run, but incorrectly and with the possibility of crashing unexpectedly later in the run.
Also beginning with 3.0.1, there is an option that will cause the model to detect this “overdecomposition” condition and then automatically reduce the number of processors decomposing a problem domain. To enable this option, add -DALLOW_OVERDECOMP to the ARCH_LOCAL line of your configure.wrf file, clean, and recompile. Note that the -DALLOW_OVERDECOMP option is new and not yet fully tested. Use at your own risk. For the time being, if you still run into trouble, work out the maximum number of processors by hand using the method described above and only specify configurations that will not cause an overdecomposition condition.
netcdf 4¶
Posted 8/05/08
Problem: WRF does not yet support netcdf 4. Please use netcdf 3 instead, and 3.6+ if you have file sizes greater than 2 Gb.
WRFV3.0¶
Observation Nudging for Nests (posted 4/27/09)¶
Problem: The user specified horizontal radius of influence for nests (obs_rinxy) is not applied correctly. This applies to all versions since WRFV2.2. The code incorrectly divides the user-input value obs_rinxy by the ratio of the nest gridsize to coarse-mesh gridsize.
Solution: The following patch applied to file phys/module_fddaobs_rtfdda.F fixes the problem: Change this line:
schnes=1/float(3**(inest-1))
to
schnes=1.0
KF scheme issue (Posted 4/10/09)¶
Problem: An error was introduced to V3.0 (also in V3.0.1 and V3.0.1.1) KF schemes (cu_physics=1 and cu_physics=99). The error may cause long lasting cloud in rare areas.
Solution: The fix is to replace line line 2189 (NCA(I,J) = TADVEC) with NCA(I,J) = REAL(NIC)*DT in phys/module_cu_kfeta.F, and replace liine 1925 with the same change in module_cu_kf.F.
Nest Ratio 1:2 Problem (Posted 4/9/09)¶
Problem: Nest ratio 1:2 is not working correctly in current release.
Solution: Working on a solution. In the mean time, use another nest ratio.
Bug In Lin Microphysics Scheme (posted 10/16/08)¶
Problem: A bug was introduced to the Lin microphysics scheme (mp_physics=2) in V3.0 code. It can cause over-prediction of cloud ice and graupel. It may also cause over-prediction of surface rainfall. This bug has been fixed in V3.0.1 code.
Solution: Download the modified module_mp_lin.F file to directory phys/ if you are still using V3.0 code. Rename and recompile.
Occasional Spurious High PBL Height In MYJ Scheme (posted 8/1/08)¶
Problem: MYJ PBL scheme occasionally produces unrealistically high PBL and high TKE in sporadic columns. This is reported by PSU researchers.
Solution: Download the modified module_bl_myjpbl.F (from PSU). The scheme computes the PBL depth by using the TKE profile criteria from Penn State, instead of the current MYJ method that searches for the first level of background TKE. This fix reduces the PBL height in general.
Using Gridded nudging and IO Quilting Together (posted 7/28/08, Fixed in V3.0.1)¶
Problem: There is an error to correctly read gridded fdda data when gridded nudging (grid_fdda = 1) and io quilting (nio_tasks_per_group > 0 and nio_group >= 1) are used together.
Solution: Download the modified module_io.F.fix into frame/ directory, rename the file to module_io.F. Recompile. Note that the error may exist in the prior versions of the model, however, this modified file may not be compatible with earlier version of the code.
OpenMP compile (posted 5/16/08, updated 5/23/08)¶
Problem 1: WRF build mechanism does not actually incorporate OpenMP when smpar and dm+sm are selected. This is fixed in V3.0.1.
Solution 1: After running configure and selecting either (smpar) or (dm+sm) option, edit the resulting configure.wrf file and add $(OMP) to the FCFLAGS line.
Problem 2: Certain physics options do not work correctly with OpenMP compile. These are cu_physics = 3 (Grell-Devenyi) and 5 (new Grell 3D scheme); mp_physics = 7 (Goddard GCE) and 10 (Morrison); and ra_lw_physics = 3, ra_sw_physics = 3 (both CAM radiation). There is no known problem with these physics options if OpenMP is not used. Likewise, there is no known OpenMP problems with other physics options.
Solution 2: Do not attempt to use the above listed options with OpenMP compile.
Using Adaptive Time Stepping with Nest (posted 6/20/08, Fixed In V3.0.1)¶
Problem: Some users encountered difficulties when using adaptive time stepping with nesting.
Solution: A suggestion has been made by the developer to set namelist variable max_step_increase_pct = 5, 51, 51, i.e. to make this control a domain-dependent variable, and use large value for the nests. To enable this, one will need to modify Registry/Registry.EM and change the following line to:
rconfig integer max_step_increase_pct namelist,domains max_doms 5 h \
"max_step_increase_pct"
Do ‘clean -a’, ‘configure’, and compile again so that the domain-dependent namelist can be used.