Pre-urban (urban-to-cropland) experiment in WRF: best practice for removing urban surface effects?

[UV Rohith]

Hello Sir/Mam,

I am running WRF with the Noah LSM using the modified 21-category IGBP-MODIS land use (urban = category 13), with all urban canopy models turned OFF. My objective is a pre-urban (urban-to-cropland) sensitivity experiment to quantify UHI impacts over Indian Cities.

From previous discussions, I understand that changing LU_INDEX alone is not sufficient, since urban influence in WRF–Noah also enters through static surface fields such as LANDUSEF, Z0, ALBEDO12M, LAI12M, GREENFRAC, and FRC_URB2D, as well as through land-use lookup tables.

I would appreciate guidance on best practice for this setup:

1. Is it sufficient to modify the geogrid output (geo_em*.nc) by replacing all urban (13) pixels with cropland values for LU_INDEX, LANDUSEF, roughness, albedo, LAI, and setting FRC_URB2D = 0 etc..., followed by a full rerun of real.exe?
2. When using Noah without UCM, are changes to tables such as VEGPARM.TBL or LANDUSE.TBL recommended, or should default cropland parameters be used?
3. Are there known pitfalls related to soil or surface initialization when converting urban to non-urban land cover in this configuration?

Any references or practical advice from similar Noah-based pre-urban experiments would be very helpful. THANKS IN ADVANCE.

 

[Ming Chen]

Please see my answers below:

Hello Sir/Mam,

I am running WRF with the Noah LSM using the modified 21-category IGBP-MODIS land use (urban = category 13), with all urban canopy models turned OFF. My objective is a pre-urban (urban-to-cropland) sensitivity experiment to quantify UHI impacts over Indian Cities.

From previous discussions, I understand that changing LU_INDEX alone is not sufficient, since urban influence in WRF–Noah also enters through static surface fields such as LANDUSEF, Z0, ALBEDO12M, LAI12M, GREENFRAC, and FRC_URB2D, as well as through land-use lookup tables.

I would appreciate guidance on best practice for this setup:

1. Is it sufficient to modify the geogrid output (geo_em*.nc) by replacing all urban (13) pixels with cropland values for LU_INDEX, LANDUSEF, roughness, albedo, LAI, and setting FRC_URB2D = 0 etc..., followed by a full rerun of real.exe?

It is sufficient to modify just LU_INDEX and set FRC_URB2D - 0 at urban points. LANDUSEF is used to determine LU_INDEX. Other variables are determined based on landuse type. Therefore, once you change LU_INDEX, these variables will change accordingly.

1. When using Noah without UCM, are changes to tables such as VEGPARM.TBL or LANDUSE.TBL recommended, or should default cropland parameters be used?

No you shouldn't change these TBL files.

1. Are there known pitfalls related to soil or surface initialization when converting urban to non-urban land cover in this configuration?

This is possible. The input data describes 'true' condition of surface, and soilm/soilt represent the condition of 'urban' points.

Any references or practical advice from similar Noah-based pre-urban experiments would be very helpful. THANKS IN ADVANCE.