WRF changing HGT from wrfinput* to wrfout* ?

[Arty]

Hello,

I recently launched a new set of simulations with two domains at 7 km and 2.333 km resolutions. Upon checking the output thoroughly, I noticed that the maximum height in the coarser domain's output is not as high as it is in the input file. Here are the details :

wrfinput*

maxhgt wrfinput_d01_1991_02
1085
maxhgt wrfinput_d02_1991_02
1540.5

wrfout*

maxhgt wrfout_d01_1991-02-01_00\:00\:00 
871.9203
maxhgt wrfout_d02_1991-02-01_00\:00\:00 
1540.5

As seen above, the maximum height for d01 is lower in the wrfout* files compared to the wrfinput file, while there is no change for d02. Same observations can be made in all previous runs (same spatial configuration / various topography testing).

According to the User's Guide, it appears there's a smoothing option by default (my bad, I skipped over it apparently). Nevertheless, I'm quite eager to know more about that. If some of you have some experience having tested the option on/off, please let me know

 

[wallis]

Is smooth_cg_topo on?

 

[Arty]

Is smooth_cg_topo on?

As I'm modeling central south Pacific ocean French Polynesian islands, I did not activate this option ; outer domain elevation is 0 meter everywhere.

I'm curious however how the smoothing option applied on domain d01 only affects the results (especially rainfalls) if I de-activate it. I launched another test run to compare results because it can have an effect on the steep topography islands I'm studying.

Thanks for your time.

 

[Ming Chen]

Arty,
I run a single domain case, and I found that HGT in wrfout and wrfinput are identical. Can you confirm this is true if you run a single domain case? If so, we can determine what you found is caused by nesting.
Please let me know which version of WRF you used. Please also upload your namelist.wps and namelist.input for me to take a look.

 

[Arty]

Arty,
I run a single domain case, and I found that HGT in wrfout and wrfinput are identical. Can you confirm this is true if you run a single domain case? If so, we can determine what you found is caused by nesting.
Please let me know which version of WRF you used. Please also upload your namelist.wps and namelist.input for me to take a look.

Hi Ms. Chen,

The WRF User's Guide (WRF3.6 - which I'm using) states :

Chap. 5-13 :
smooth_option: this a smoothing option for the parent domain in the area of the nest
if feedback is on. Three options are available: 0 = no smoothing; 1 = 1-2-1 smoothing; 2
= smoothing-desmoothing.

Chap. 5-63 :
smooth_option = 2 : (default) smoothing-desmoothing option for parent domain; used only with feedback=1

So it was a surprise when I looked in the wrfout* files, but : my bad, I had to read the User's Guide more carefully. As stated above, the smoothing only applies to the parent domain, hence the fact the finer domain topography is not modified. Now I just wonder how this smoothing in d01 affects d02 outputs. I did launch a test run (1 year) setting smooth_option = 0 to compare against a previous similar run and find out.

Please note I'm most interested in the finer domain d02 results - but I nevertheless would like to know if the difference is huge (or not) between both options (d01 topography smoothed or not).

I'll post my conclusion here when it's done. In the meantime, if you or anyone else has any insight about that, I'd be glad to hear you out.

 

[Arty]

After conducting a one-year simulation to compare the activation of smoothing on d01 versus its deactivation, the results indicate that only precipitation is significantly influenced. Other variables appear to be quite insensitive to this option.

In the attached figure, I deliberately adjusted the color ranges (and quivers) to highlight relative discrepancies (%). However, please note that domain-average anomalies computed from absolute values are indicated in each subplot's title as Mean Bias Error (Mean Absolute Error) [units]. Overall, there is slightly more rainfall [a] in d02 when topography smoothing is activated in d01 and, unsurprisingly, solar irradiation [c] is consequently slightly lower.

 

[Ming Chen]

Arty,
Thanks for the test and update of your results. Note that the option smooth_cg_topo only affects outer rows and columns of topography in D01. It has slight impacts over relatively flat terrain area.
Please let us know if you have more questions regarding the nest and topography smooth.
By the way, WRFV3.6 is way too old. Would you please update to newer version of WRF?

 

[Arty]

Arty,
Thanks for the test and update of your results. Note that the option smooth_cg_topo only affects outer rows and columns of topography in D01. It has slight impacts over relatively flat terrain area.
Please let us know if you have more questions regarding the nest and topography smooth.
By the way, WRFV3.6 is way too old. Would you please update to newer version of WRF?

Unfortunately I did have a lot of troubles working with WRF4.2 at the beginning of my thesis, so I switched back using the version my predecessor - whom I use the output files via NDOWN - worked with, i.e. WRF3.6. I finally get it to work and I don't feel switching back to a newer version as my PhD is more than half-advanced ; even though I do know (now) the problem I encountered was not due to WRF version. Of course, for my future work I intend to use the latest version.

About the smoothing option, I didn't activate 'smooth_cg_option' but 'smooth_option' which apply over all domain as I understand it; and as it shows : my d01 peaks at 1085 meters when smooth_option is deactivated (= 0) whereas it peaks at 875 meters when activated (= 2). But as show above, it does not change much except for rain. Moreover, no "logical" pattern seem to emerge (in rain or other variables) from the anomalies.