Running a new WRF simulation with wrfout

[maR_Kus]

Hi,
I would like to use WRF output as input for a new WRF simulation. As I deleted some variables from the output and my new domain has a resolution with a parent_grid_ratio different than 3 and 5, ndown shouldn't be suited to produce the forcings. So I guess the only solution is to use UPP, right?

I was looking at the UPP-Vtables and the wrf_cntrl.parm provided with UPP v4, but there are several inconsistencies and I do not understand how to create a proper wrf_cntrl.parm which works with any of the Vtables provided with ungrib. So, first I wonder if anyone here already tried to solve this issue and has a solution to shared?

If not, I will try to play with UPP, but I am not sure on which Vtable I have to use: in particular, Vtable.ARW.UPP is for WRF-ARW output that has been processed on eta/sigma levels, while Vtable.ARWp.UPP is suited for output that has been processed by UPP and is on pressure levels. As my outputs are on hybrid levels, I assume Vtable.ARWp.UPP is not the correct one and I should convert wrfout to pressure levels with UPP. Is it the right method?

Is anyone able to help me with this?

 

[Ming Chen]

I suppose ndown should work for this case unless you delete some variables that are required for running WRF.
If you prefer to run UPP, that is also fine. Hope someone can answer the questions related to UPP.

 

[mesego]

I suppose ndown should work for this case unless you delete some variables that are required for running WRF.
If you prefer to run UPP, that is also fine. Hope someone can answer the questions related to UPP.

Hi I am meeting a similar problem. May I know what are the necessary variables for running ndown.exe? Do I need to keep all the variables in wrfout or just those variables commonly requested for boundary condition (e.g., T, U,V etc.) are enough? Thank you very much!

 

[maR_Kus]

I suppose ndown should work for this case unless you delete some variables that are required for running WRF.
If you prefer to run UPP, that is also fine. Hope someone can answer the questions related to UPP.

Hi Ming,
thanks for the suggestion, I worked a bit on UPP and finally I have been able to complete two simulations using the wrf_cntrl and Vtable provided with UPP and ungrib, respectively.
However, comparing the WRF output generated using the Vtable.ARW.UPP (i.e. I used in ungrib wrfout extracted in hybrid levels by UPP) with the same simulation obtained using the Vtable.ARWp.UPP (i.e. ungrib with wrfout extracted to some pressure levels by UPP) I found patches differences that, in case of 2m temperature, are larger than 3K in several locations.
I wonder why this happens (I do not expect the two simulations are bit-to-bit identical but such differences are too large) and which of the two methods should be preferred (i.e. forcing created by UPP in pressure or hybrid levels)?

Thanks again for support,
Markus

 

[Ming Chen]

Markus,

Thanks for the detailed description of this issue. Unfortunately I never use UPP before to process wrfout and then create initial and boundary conditions for finer-grid WRF run. Without direct experiences, it is hard for me to answer your question.
Just a preliminary thinking, I suppose we should have better results if using wrfout data in hybrid levels. This is because we can avoid errors introduced by interpolation from hybrid levels to pressure levels. Is this true in your results?
I will send your post to my colleague, who is an ex[pert in UPP. Hopefully she can get back to you with more helpful information.

 

[maR_Kus]

Hi Ming,
thanks for forwarding the post to UPP expert.
Anyway, I agree with you, wrfout data in hybrid levels should be better as it avoids any vertical interpolation.

 

[fossell]

Hi Markus,
The wrf_cntrl.parm and UPP Vtables were provided by a community user who found a way to make it work for their experiment, it wasn't ever a fully supported feature of UPP. Unfortunately we haven't had support for WRF and grib1 formatted output for quite some time so we're unable to dig into this any further. I'm not sure what would be causing the large deltas you're seeing, and I'm sorry we don't have a answer for you.