WRF 4.3 output the value of AFWA_VIS was too unreasonable.

[wyw]

Hello,
I want to use &AFWA to calculate the visibility. In the model output file, the values of the AFWA_VIS was 10 times approximately as large as the GFS 0.25° dataset and observation data in my study area, especially on land. I used different combinations of physical schemes, however the results did not improve significantly. So I want to know why the results are so big and what else can we do to improve the results.

The best physical schemes （the wind, temperature and rain were closed to the observation）in my study area as follow:

mp_physics	WSM6
cu_physics	TDK
ra_lw_physics	RRTM
ra_sw_physics	Dudhia
bl_pbl_physics	YSU
sf_sfclay_physics	Monin-Obukhov
sf_surface_physics	Noah

the &afwa setting in the namelist.input :
&afwa
afwa_diag_opt = 1
afwa_vis_opt = 1
afwa_cloud_opt = 1
afwa_ptype_opt = 1
/

The results of visibility at the same time of wrf results and gfs0.25° as follow:
The coloabr tick number of the first two picture are same, and showing the maximum value oeach.
The third and fourth images set the same maximum value of colorbar.

View attachment 7037

 

[Ming Chen]

Hi,
The visibility diagnosed in AFWA package is dependent on multiple factors. Specifically, if your lowest model level is above 20m, then the model will use 2m T and RH in the interpolation to obtain data at 20m height. This will to a certain degree affect the result.
Please take a look at the code phys/module_diag_afwa.F, which performs the calculation of visibility. I am suspicious that the input data for its calculation might be an issue in the overestimation of your case. I don't think the method itself is wrong. To confirm this, you may extract the code, provide GFS data as input and see whether you can get similar results to GFS.

 

[wyw]

ok. I have refined the vertical level, and set the lowest model level about 5m, the result of visibility was normal.
thank you very much.

 

[Ming Chen]

Would you please clarify the relative importance of changes in vertical levels and the lowest model level height? Thanks!

 

[wyw]

Previously, I checked the lowest model level which was 50m, as you said, the lowest model level is above 20m, then the model will use 2m T and RH in the interpolation to obtain data at 20m height. And that will to a certain degree affect the result. The vertical layer was set at 40(e_vert=40), and the thickness of the layer was shown in image:


Now, the vertical layer was set at 50 (e_vert=50), and I also add some parameters about the vertical levels:
e_vert = 50, 50
auto_levels_opt = 2
dzbot = 5
max_dz = 1500
dzstretch_s =1.3
dzstretch_u =1.1

the thickness of the layer was shown in image:


After resetting the vertical layer, the lowest model level is below 20m, then the model will use T and RH close to 20m in the interpolation to obtain data at 20m height.

These are some of the settings I've made to improve the results of the model. could you please give me some advice?

 

[Ming Chen]

Many thanks for the update and kind information. I guess these settings are what we can do at the user's end. Other important issues are dependent on physics schemes involved.

 

[janac4]

Hello, I'm interested in computing visibility variables with WRF AFWA diagnostic options. I managed to run a simulation with this activated and got AFWA_VIS results different of 0. However I don't know for sure how to interpret these results as visibility depends strongly on the direction and AFWA_VIS gives a single value in each cell. Is it taken into account? How does the model compute these type of variables.
Thank you.

 

[Ming Chen]

Please clarify what you mean by saying that "visibility depends strongly on the direction"?

 

[janac4]

Please clarify what you mean by saying that "visibility depends strongly on the direction"?

The clear visibility distance for a user depends on which way they're facing. In each part of the model, it could be that visibility is good to the north but not to the south. Can the model consider this? How should I understand the "AFWA_VIS" variable's result?

 

[Ming Chen]

I don't think WRF considers 'direction' in its calculation of AFWA_VIS.

 

[mirabelle]

Previously, I checked the lowest model level which was 50m, as you said, the lowest model level is above 20m, then the model will use 2m T and RH in the interpolation to obtain data at 20m height. And that will to a certain degree affect the result. The vertical layer was set at 40(e_vert=40), and the thickness of the layer was shown in image:
View attachment 7079

Now, the vertical layer was set at 50 (e_vert=50), and I also add some parameters about the vertical levels:
e_vert = 50, 50
auto_levels_opt = 2
dzbot = 5
max_dz = 1500
dzstretch_s =1.3
dzstretch_u =1.1

the thickness of the layer was shown in image:
View attachment 7078

After resetting the vertical layer, the lowest model level is below 20m, then the model will use T and RH close to 20m in the interpolation to obtain data at 20m height.

These are some of the settings I've made to improve the results of the model. could you please give me some advice?

Hi，

How was the content of this image generated? Such as "Full level index" and "Height"

 

[haiqingsong]

Previously, I checked the lowest model level which was 50m, as you said, the lowest model level is above 20m, then the model will use 2m T and RH in the interpolation to obtain data at 20m height. And that will to a certain degree affect the result. The vertical layer was set at 40(e_vert=40), and the thickness of the layer was shown in image:
View attachment 7079

Now, the vertical layer was set at 50 (e_vert=50), and I also add some parameters about the vertical levels:
e_vert = 50, 50
auto_levels_opt = 2
dzbot = 5
max_dz = 1500
dzstretch_s =1.3
dzstretch_u =1.1

the thickness of the layer was shown in image:
View attachment 7078

After resetting the vertical layer, the lowest model level is below 20m, then the model will use T and RH close to 20m in the interpolation to obtain data at 20m height.

These are some of the settings I've made to improve the results of the model. could you please give me some advice?

What is your p_top_requested ？

 

[ankitpatte0698]

Previously, I checked the lowest model level which was 50m, as you said, the lowest model level is above 20m, then the model will use 2m T and RH in the interpolation to obtain data at 20m height. And that will to a certain degree affect the result. The vertical layer was set at 40(e_vert=40), and the thickness of the layer was shown in image:
View attachment 7079

Now, the vertical layer was set at 50 (e_vert=50), and I also add some parameters about the vertical levels:
e_vert = 50, 50
auto_levels_opt = 2
dzbot = 5
max_dz = 1500
dzstretch_s =1.3
dzstretch_u =1.1

the thickness of the layer was shown in image:
View attachment 7078

After resetting the vertical layer, the lowest model level is below 20m, then the model will use T and RH close to 20m in the interpolation to obtain data at 20m height.

These are some of the settings I've made to improve the results of the model. could you please give me some advice?

hi, i want to know how did you get this type of file where you get height and thickness of each respective levels.

 

[Ming Chen]

Please clarify what is "this type of file " you mentioned in your post. Also, please start a new thread.

 

[ankitpatte0698]

Please clarify what is "this type of file " you mentioned in your post. Also, please start a new thread.

I meant how can we get these type of information i.e. height and thickness for each model level; for our respective simulations. I check my wrfinput file but I couldn't find/or didn't understand which variables i should look into.
Thanks in advance

 

[janac4]

Previously, I checked the lowest model level which was 50m, as you said, the lowest model level is above 20m, then the model will use 2m T and RH in the interpolation to obtain data at 20m height. And that will to a certain degree affect the result. The vertical layer was set at 40(e_vert=40), and the thickness of the layer was shown in image:
View attachment 7079

Now, the vertical layer was set at 50 (e_vert=50), and I also add some parameters about the vertical levels:
e_vert = 50, 50
auto_levels_opt = 2
dzbot = 5
max_dz = 1500
dzstretch_s =1.3
dzstretch_u =1.1

the thickness of the layer was shown in image:
View attachment 7078

After resetting the vertical layer, the lowest model level is below 20m, then the model will use T and RH close to 20m in the interpolation to obtain data at 20m height.

These are some of the settings I've made to improve the results of the model. could you please give me some advice?

Hello, I would like to understand how this namelist options affect model levels and why with them we can achieve better model visibility values. What about the other variables?

 

[janac4]

I meant how can we get these type of information i.e. height and thickness for each model level; for our respective simulations. I check my wrfinput file but I couldn't find/or didn't understand which variables i should look into.
Thanks in advance

Hello,
This information can be found in the rsl.out.0000 file after running real.exe.

 

[ankitpatte0698]

Hello,
This information can be found in the rsl.out.0000 file after running real.exe.

Thanks a lot