Questions on Albedo Modification in WRF

[lff]

Dear WRF Support Team,

I am conducting a study that requires modifying the albedo of part of the land surface to a specific, constant value. To implement this, I conducted two experiments. The first is a control experiment in which no changes were made to the albedo values. The second is a sensitivity experiment, in which I created a new monthly albedo dataset, modified only the target regions, and replaced the ALBEDO12M variable in the geo_em.d01.nc file. I also set usemonalb = .true. in the namelist.input.

&physics
sf_surface_physics = 2,
num_soil_layers = 4,
bl_pbl_physics = 1,
sf_sfclay_physics = 1,
ra_lw_physics = 4,
ra_sw_physics = 4,
cu_physics = 1,
mp_physics = 8,
radt = 25,
bldt = 0,
cudt = 5,
num_land_cat = 21,
sf_urban_physics =0,
usemonalb = .true.,


I have the following questions:

1. When usemonalb = .true. is set, does WRF read the monthly albedo values from geo_em.d01.nc (i.e., ALBEDO12M) instead of using the land surface parameter table values (e.g., VEGPARM.TBL)? Does WRF interpolate these monthly values to each simulation timestep internally? Should I also set rdlai2d = true?
2. However, if WRF uses the modified ALBEDO12M values from geo_em.d01.nc. I am puzzled by the discrepancies I observe between the values in geo_em.d01.nc and those written to the resulting wrfinput_d01 file as ALBBCK. When I compare the ALBBCK fields between the two experiments (i.e., the two wrfinput files), the difference clearly reflects the albedo modifications I applied. This suggests that the changes are recognized by WRF, but ALBBCK do not exactly match those in the original geo_em.d01.nc.
3. In the model outputs (wrfout), albedo is clearly affected by snow cover. However, when I compare albedo from control and modified simulations, I observe large differences in snow-covered regions—particularly over the Tibetan Plateau—even though no albedo modifications were made there. This made me feel confused, as I expected changes to be localized, and not to trigger such significant feedbacks in remote regions.
4. I came across a suggestion of modifying albedo in another source that states:

"Since the model can generate snowfall during the simulation, the resulting snow cover can alter surface albedo. Therefore, we need to remove the snow from the model. This can be done by restarting the simulation at regular intervals and modifying the snow variables in the wrfrst (restart) files."

Whether the unexpected changes in albedo due to snow accumulation in the model? If so, would it be reasonable to follow this method to eliminate snow effects?

Thank you in advance for your help. Any clarification on these issues would be very valuable for me.