Urban Physics in Ndown

[Tanmoy]

Hello WRF community,

I hope everyone's doing well. Currently, I'm trying to run WRF simulations (WRF v 4.5.2) using ndown.exe for 3 domains. where my innermost domain d03 will simulate urban physics. So, accordingly, I followed the steps mentioned for ndown in the WRF user guide (Welcome to WRF User’s Guide! — WRF Users Guide documentation), for the d01 to d02 and similarly, I'm doing it for d02 to d03.However, when I was trying to execute the 3rd WRF simulation using d03 as single domain and wrfinput and wrfbdy from ndown, it started showing error:

 DYNAMICS OPTION: Eulerian Mass Coordinate
   alloc_space_field: domain            1 ,              72415352  bytes allocated
  med_initialdata_input: calling input_input
   Input data is acceptable to use: wrfinput_d01
----------------- ERROR -------------------
namelist    : sf_urban_physics   =          1
input files : SF_URBAN_PHYSICS   =          0 (from wrfinput files).
  ---- ERROR: Mismatch between namelist and global attribute SF_URBAN_PHYSICS
 CURRENT DATE          = 2050-12-25_00:00:00
 SIMULATION START DATE = 2050-12-25_00:00:00
NOTE:       1 namelist vs input data inconsistencies found.
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE:  <stdin>  LINE:    1298
NOTE:  Please check and reset these options
-------------------------------------------
Abort(1) on node 0 (rank 0 in comm 0): application called MPI_Abort(MPI_COMM_WORLD, 1) - process 0

So, I just enabled sf_urban_physics for both d02 and d03 before the second ndown.exe and followed the ndown steps again. It did not show any error afterwards while running the wrf.exe for d03 as single domain.

Is this correct or am I missing something?? It would be nice if anyone could explain why this happened?? Also, is it ok to run ndown for more than 2 domains??

Sincerely,
Tanmoy

 

[Hans]

According to WRF Users Guide documentation, the urban canopy model needs to use the same settings in all domains

 

[Tanmoy]

Hello Hans,
But, my study region is domain d03. Also, my d01 and d02 are of very low resolution (30km and 6km respectively). For urban physics to come into play the grid size should be atleast 2km or less, if I'm not wrong so enabling it for d01 and d02 would be basically useless. So, I kept

sf_urban_physics = 0,0,1;

 

[Arty]

Hello Hans,
But, my study region is domain d03. Also, my d01 and d02 are of very low resolution (30km and 6km respectively). For urban physics to come into play the grid size should be atleast 2km or less, if I'm not wrong so enabling it for d01 and d02 would be basically useless. So, I kept

sf_urban_physics = 0,0,1;

Did you try to activate SF_UBRAN_PHYSICS in all domains, even though useless at coarse resolution ? I suppose

I'm no expert here, but from my own experience, NDOWN/WRF is very sensitive to physics option. Some parameters require input data that NDOWN shall provide in wrfinput*/wrfbdy* files (or sometimes wrflowinp* if required - files not produced by NDOWN) : if they're not there, then run will fail.

I understand you already properly ran your d01 and d02 ? For furtther help, please describe what you did do yet, and how. We might be able to find a solution.

Fa'aitoito !

 

[Tanmoy]

Did you try to activate SF_UBRAN_PHYSICS in all domains, even though useless at coarse resolution ? I suppose

I'm no expert here, but from my own experience, NDOWN/WRF is very sensitive to physics option. Some parameters require input data that NDOWN shall provide in wrfinput*/wrfbdy* files (or sometimes wrflowinp* if required - files not produced by NDOWN) : if they're not there, then run will fail.

I understand you already properly ran your d01 and d02 ? For furtther help, please describe what you did do yet, and how. We might be able to find a solution.

Fa'aitoito !

Hello Arty,

I haven't tried to activate SF_URBAN_PHYSICS for all the domains.

I'm also sharing the entire step that I followed in the attached text file (One-way nested run using ndown.exe). I basically followed the steps from WRF User guide. Please also let me know if I did any mistake. I have also attached all the namelist files of the respective steps.

Best Regards,
Tanmoy

 

[htan2013]

Hi Tanmoy,

I think that you might need to re-run the ndown.exe for the D03 to create wrfbdy and wrfinput. Probably because the wrong setting in sf_urban_physics during running the ndown.

 

[Arty]

Running NDOWN can be "cumbersome" for such cases, as rightly stated in the User's Guide. What I understand from your step-by-step approach is that you ran for each domain separately the NDOWN process, in order to have your d01 inputs, then d02 inputs, then d03 inputs, right ?

Even though my own case was quite complicated, using a 2-ways double-nested simulations, I have not activated urban physics and don't know much about required data for that scheme. However, if that option is only put in action when scale/landuse requires it, I would try to apply it in all three domains (though I'm not sure how this option works).

Something is also not clear to me right now: did you run your complete simulation for d01 and d02 yet ? Or just short runs to create your input files for the finer domains ?

P.S. : note that there are alternatives, when looking to run multiple nested domains, to avoid running NDOWN several times ; though it necesitates some experience to properly use NCO/CDO to create required wrfinput* and wrflowinp* files (e.g. CDO remap function).

 

[Tanmoy]

Hi Tanmoy,

I think that you might need to re-run the ndown.exe for the D03 to create wrfbdy and wrfinput. Probably because the wrong setting in sf_urban_physics during running the ndown.

Hello htan2013,

So, should I turn on sf_urban_physics for d02 also before running ndown??

Best wishes,
Tanmoy

 

[Tanmoy]

Running NDOWN can be "cumbersome" for such cases, as rightly stated in the User's Guide. What I understand from your step-by-step approach is that you ran for each domain separately the NDOWN process, in order to have your d01 inputs, then d02 inputs, then d03 inputs, right ?

Even though my own case was quite complicated, using a 2-ways double-nested simulations, I have not activated urban physics and don't know much about required data for that scheme. However, if that option is only put in action when scale/landuse requires it, I would try to apply it in all three domains (though I'm not sure how this option works).

Something is also not clear to me right now: did you run your complete simulation for d01 and d02 yet ? Or just short runs to create your input files for the finer domains ?

P.S. : note that there are alternatives, when looking to run multiple nested domains, to avoid running NDOWN several times ; though it necesitates some experience to properly use NCO/CDO to create required wrfinput* and wrflowinp* files (e.g. CDO remap function).

Hello Arty,

I indeed ran the simulations for my complete time period for d01 and d02.

Regarding the alternative, me, my supervisor and colleagues, were actually looking for alternatives to run simulations faster rather than going for normal WRF runs by somewhat maintain the accuracy of the results. Anyway we have now decided to stick to the normal simulation steps as you rightly mentioned, NDOWN indeed is cumbersome and confusing for 3 domains and I myself would prefer it for 2 domain study. However I would still like to know why this error occurred, so, that not only me but anyone else can also benefit from this conversation in future.

Best Wishes,
Tanmoy

 

[htan2013]

Hello htan2013,

So, should I turn on sf_urban_physics for d02 also before running ndown??

Best wishes,
Tanmoy

Yes, I think you should try that.
Also, I believe there are bugs in the ndown process. After processing the ndown, I notice that the LANDMASK in the new wrfinput is empty, so are other urban-related variables. Can you also me verify this after you turn on the sf_urban_physics for ndown? Thank you.

 

[Tanmoy]

Yes, I think you should try that.
Also, I believe there are bugs in the ndown process. After processing the ndown, I notice that the LANDMASK in the new wrfinput is empty, so are other urban-related variables. Can you also me verify this after you turn on the sf_urban_physics for ndown? Thank you.

Thanks for the info, I'll look into it and let you know also.

 

[Ming Chen]

Tanmoy,

When running WRF over multiple domains, it is required that the physics (except cumulus scheme) should be same over all domains. When running ndown, initial and boundary conditions are derived from the corse grid run. Any inconsistency may cause issues in subsequent runs.

A detailed description of NDOWN program can be found at Chapter 5: WRF Model

Hope this is helpful.

 

[Tanmoy]

Tanmoy,

When running WRF over multiple domains, it is required that the physics (except cumulus scheme) should be same over all domains. When running ndown, initial and boundary conditions are derived from the corse grid run. Any inconsistency may cause issues in subsequent runs.

A detailed description of NDOWN program can be found at Chapter 5: WRF Model

Hope this is helpful.

Hello @Ming Chen

Thank you very much for the clarification.

Sincerely,
Tanmoy