replacement of monthly albedo

[lff]

Hello, all!,

I am new to WRF and am currently trying to perform two experiments by replacing the surface albedo. The first set is a control run without changing albedo, and the second set modifies the albedo for a specific land surface category.

I have conducted a 10-day simulation and found that in the wrfout output files, the albedo differs between the two experiments only at the first time step, while from the second time step onwards, the albedo values are almost identical between the two runs.

My procedure for replacing albedo is as follows:

• I modified the ALBEDO12M variable inside the geo_em.d01.nc file.
• Then I ran metgrid.exe.
• In the namelist.input, I set usemonalb = true.
• Finally, I ran real.exe and wrf.exe to perform the simulations.

Below is the relevant part of the &physics section in my namelist.input:

&physics
sf_surface_physics = 4, ! Noah-MP
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,
/

&noah_mp
dveg = 2, !On LAI predicted; FVEG calculated
/


My questions are:

1. Is this approach of replacing albedo by modifying ALBEDO12M in geo_em.d01.nc and setting usemonalb = true correct?
2. Why does the albedo difference between the two experiments only appear at the first time step in the output?
3. For consistency, should I set usemonalb = true in both the control and modified experiments?
4. After switching to the NOAH scheme, I was able to see the albedo changes I made. I suspect that in NOAH-MP, albedo is only read during initialization and later replaced by dynamic calculations. Is that correct? If I want to keep albedo fixed (e.g., use prescribed monthly values) throughout the simulation, what is the proper way to achieve this with NOAH-MP?

Thank you very much for your time and help!

 

[kwerner]

Hi,
Many apologies for the delay in response and thank you for your patience. I'll try my best to answer your questions.

Is this approach of replacing albedo by modifying ALBEDO12M in geo_em.d01.nc and setting usemonalb = true correct?

Yes, I believe this is the correct approach

Why does the albedo difference between the two experiments only appear at the first time step in the output?

I believe the issue is that you may need to have the albedo updated throughout the simulation. Otherwise, albedo is just an initial condition that remains steady through the run. See the process I discuss in this previous post. Some of that doesn't apply to you because you've already replaced albedo in WPS, but the part about updating it through the sst_update process should be applicable, and may be something you can do.

For consistency, should I set usemonalb = true in both the control and modified experiments?

Yes, I believe so.

After switching to the NOAH scheme, I was able to see the albedo changes I made. I suspect that in NOAH-MP, albedo is only read during initialization and later replaced by dynamic calculations. Is that correct? If I want to keep albedo fixed (e.g., use prescribed monthly values) throughout the simulation, what is the proper way to achieve this with NOAH-MP?

Yes, it seems that it may be written that way, which is why you will likely need to update the values through the sst_update option I mention above.

 

[lff]

Thank you for your reply!

I would like to further confirm the use of the sst_update option. I have currently downloaded SST and sea ice data from ERA5, along with other data (e.g., 10 m u-component of wind, 10 m v-component of wind, and 2 m dewpoint temperature). If I want ALBBCK not to remain constant, I need to set sst_update = 1 in the namelist.input file, will this configuration correctly enable SST updates during the model simulation? Or do I need to provide an additional SST input file for the updates to take effect?

In addition, based on my understanding, if I set usemonalb = false, WRF will initialize the ALBBCK using values defined in VEGPARM.TBL for each land cover type. In this case, if I also set sst_update = 1, will ALBBCK change over time during the simulation? Or is it necessary to set usemonalb = true to allow WRF to read the ALBEDO12M data from geo_em.nc to enable temporal updates of ALBBCK?

When conducting sensitivity experiments involving albedo modification, I encountered the following two issues:
1. When I compare albedo between the control run (without albedo modification) and the modified simulation using wrfout, I observe albedo changes over the Tibetan Plateau—even though no albedo modifications were made there. These changes correspond to differences in SNOWC, so I suspect that my albedo perturbation triggered snowfall. The perturbation I made to albedo was relatively small, but the changes in albedo over the plateau are much larger than my imposed perturbation. This confuses me—are such significant remote feedbacks reasonable?

2. I came across a suggestion in another source that says: “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.” Would it be reasonable to follow this method to eliminate snow effects? If snowfall occurs in the region where I modified albedo, how exactly should this be implemented? How should I determine appropriate restart intervals? Are there other recommended ways to reduce the influence of snow?

Currently, I am using the Noah land surface scheme, not Noah-MP.
Thank you in advance for your help—any clarification on these issues would be very valuable to me.

 

[kwerner]

If I want ALBBCK not to remain constant, I need to set sst_update = 1 in the namelist.input file, will this configuration correctly enable SST updates during the model simulation? Or do I need to provide an additional SST input file for the updates to take effect?

No, you don't need to provide additional SST data for this. Setting sst_update=1 should ensure that it's updated correctly.

In addition, based on my understanding, if I set usemonalb = false, WRF will initialize the ALBBCK using values defined in VEGPARM.TBL for each land cover type. In this case, if I also set sst_update = 1, will ALBBCK change over time during the simulation? Or is it necessary to set usemonalb = true to allow WRF to read the ALBEDO12M data from geo_em.nc to enable temporal updates of ALBBCK?

I'm actually not positive whether ALBBCK will change over time with usemonalb turned on, but yes, to allow the model to use data from geogrid, you have to have it turned on. You could run a short test to determine whether it's changed.

1. When I compare albedo between the control run (without albedo modification) and the modified simulation using wrfout, I observe albedo changes over the Tibetan Plateau—even though no albedo modifications were made there. These changes correspond to differences in SNOWC, so I suspect that my albedo perturbation triggered snowfall. The perturbation I made to albedo was relatively small, but the changes in albedo over the plateau are much larger than my imposed perturbation. This confuses me—are such significant remote feedbacks reasonable?

I think this is probably reasonable behavior, assuming the updated albedo values aren't unreasonable. Again, this is something i'd recommend running a short simulation to verify that everything looks okay, or just checking on the output files as they're generated.

2. I came across a suggestion in another source that says: “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.” Would it be reasonable to follow this method to eliminate snow effects? If snowfall occurs in the region where I modified albedo, how exactly should this be implemented? How should I determine appropriate restart intervals? Are there other recommended ways to reduce the influence of snow?

This is something I'm not sure about. I'm going to ask a colleague who may know and get back to you with an answer. I'm sorry I don't know!