Landuse type mismatch in urban - real.exe and met_em nc files

[lmm_ch]

Hi,
I am using a combination of NLCD and MODIS data for a domain covering part of U.S. and Canada. In my met_em and geo_em files, the land use types are fine; I have 4 urban classes in U.S. cities and only 1 urban class (13) for MODIS. But after I run real.exe, some U.S. cities have type 13 in wrfinput file. It is the same for wrfout files as well. Seems that some NLCD urban classes are converted back to MODIS urban class. Do I need to modify anything before running real.exe?
Thanks!
Chris

 

[Ming Chen]

Please send me your name list.wps and namelyst.input to take a look. Thanks.

 

[Ming Chen]

Chris,
Please also tell me which version of WPS and WRF you are running. Thanks.
Ming Chen

 

[lmm_ch]

Hi Ming,

Attached please find my namelist.input and namelist.wps, and I am using WPS 4.0 and WRF 4.0.

In addition, when I turn off sf_urban_physics, the lu_index would be fine.

Thanks and I am looking forward to your reply.

Chris

 

[Ming Chen]

I looked at your case. You are right that at some grid points the land use type is reset to 13 after running REAL. This is because the urban fraction in these points are larger than 0.5 but the land use type is not urban, --- this is apparently inconsistent. To address this issue, we force the model to reset the land use type as urban (=13). Since the urban fraction is an important parameter in urban physics module, this reset is only conducted when urban physics is turned on.
Please let me know you have more questions about this issue.

Ming Chen

 

[lmm_ch]

Hi,

Thanks a lot for your help! But I observed large areas of urban = 13 even in the centers of cities, such as Los Angeles and Phoenix. I checked LU_INDEX of met_em files, and the original LU_INDEX values over these replaced areas are 24, 25, or 26. Why are these grids being changed to 13?

Thanks again!

Chris

 

[Ming Chen]

Please use the code I attach here to replace the original one, then rebuild WRF and run REAL.
Please let me know how the result looks. Thanks.

Ming Chen

 

[lmm_ch]

The bug has been perfectly fixed. Thank you!

Chris