regarding Cumulus parameterization schemes

[deva_wrf]

hello everyone,
i have to ask you a question regarding the cu_physics for innermost domain having resolution 3km, i want to know is this important to set the cu physics for inner most domain for such fine resolution domain.

 

[kwerner]

Hi,
No, we actually don't recommend using a cumulus scheme for grid resolutions <= 3km. We do recommend using one for anything >= 10km, and between 3km and 10km is somewhat of a gray zone, meaning it's not clear whether a cumulus scheme should be used - users should either refrain from using domains with that grid resolution, or do some simple testing to determine if using a cumulus scheme is ideal for their particular run.

 

[Jialiwang]

Hi,

I have a follow-up question on this, I compared a 12km run with KF cumulus parameterization (let's call it CP), a 12km without any CP, and a 4km without any CP. As expected, the 12km with KF did show larger bias than the 4km without CP; however, I saw the 12km without CP did better than 12km with CP, in fact the 12km without CP does an equally good job as the 4km without CP. I am talking about precipitation, compared to PRISM 4km, one-month total precip amount.

My question is, the CP is not perfect anyways, why do we need it, even for the 12km? Is there anyway that I can approve that 12km without CP is bad and one SHOULD use a CP when running 12km?

Attached please see 3 figures for bias of these 3 model setup. again the bias is for 2017 Jun monthly total precip. unit: mm/month

 

[kwerner]

Can you let me know which version of the model you are using? Thanks!