Hello,
I am trying to use WRF with the BEP-BEM urban physics schemes for domains centered on prominent urban regions. I have been attempting to utilize the NUDAPT-44 data instead of the conventional urban morphology based on urban land cover class. I have followed the appropriate steps as outlined in the NUDAPT44 documentation (https://ral.ucar.edu/sites/default/files/public/product-tool/NUDAPT_44_Documentation.pdf). When attempting to utilize both BEP and BEP-BEM I come across separate problems:
BEP (sf_urban_physics = 2):
geogrid.exe and metgrid.exe run successfully, fully extracting the urban fraction and URB_PARAM data from NLCD 2011 and NUDAPT, respectively. I have checked that each of the variables is present in geo_d0x and met_d0x files. After running real.exe, however, the URB_PARAM variables (BUILD_AREA_FRACTION, HEIGHT_HISTOGRAMS, BUILD_SURF_RATIO, BUILD_HEIGHT) in wrfinputd0x are all set to zero for all pixels. As such, when wrf.exe is run it does not have any NUDAPT parameter values to use, and much of the initial output lines in rsl.out files say “USING DEFAULT URBAN MORPHOLOGY”.
BEM (sf_urban_physics = 3):
When running real.exe, the program does not complete successfully and ends with a fatal call…
d02 2013-06-02_03:00:00 alloc_space_field: domain 3 , 35210496 bytes allocated
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 2066
frame/module_domain.f: Failed to allocate grid%tglev_urb3d(sm31:em31,1:model_config_rec%num_urban_layers,sm33:em33).
I gather that this has something to do with not enough memory available? I have been attempting this on a personal machine. Would the only answer be to find a high-power cluster to run real.exe on?
The WRF version I am using is 3.8.1. I have attached namelist files for reference. Thank you for your help!
Regards,
Markus
I am trying to use WRF with the BEP-BEM urban physics schemes for domains centered on prominent urban regions. I have been attempting to utilize the NUDAPT-44 data instead of the conventional urban morphology based on urban land cover class. I have followed the appropriate steps as outlined in the NUDAPT44 documentation (https://ral.ucar.edu/sites/default/files/public/product-tool/NUDAPT_44_Documentation.pdf). When attempting to utilize both BEP and BEP-BEM I come across separate problems:
BEP (sf_urban_physics = 2):
geogrid.exe and metgrid.exe run successfully, fully extracting the urban fraction and URB_PARAM data from NLCD 2011 and NUDAPT, respectively. I have checked that each of the variables is present in geo_d0x and met_d0x files. After running real.exe, however, the URB_PARAM variables (BUILD_AREA_FRACTION, HEIGHT_HISTOGRAMS, BUILD_SURF_RATIO, BUILD_HEIGHT) in wrfinputd0x are all set to zero for all pixels. As such, when wrf.exe is run it does not have any NUDAPT parameter values to use, and much of the initial output lines in rsl.out files say “USING DEFAULT URBAN MORPHOLOGY”.
BEM (sf_urban_physics = 3):
When running real.exe, the program does not complete successfully and ends with a fatal call…
d02 2013-06-02_03:00:00 alloc_space_field: domain 3 , 35210496 bytes allocated
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 2066
frame/module_domain.f: Failed to allocate grid%tglev_urb3d(sm31:em31,1:model_config_rec%num_urban_layers,sm33:em33).
I gather that this has something to do with not enough memory available? I have been attempting this on a personal machine. Would the only answer be to find a high-power cluster to run real.exe on?
The WRF version I am using is 3.8.1. I have attached namelist files for reference. Thank you for your help!
Regards,
Markus