Change CLUBB flag (clubb_l_min_wp2_from_corr_wx)

[bstephens82]

Issue Type

Other (please describe below)

Issue Description

@vlarson has requested that NCAR change the CLUBB flag clubb_l_min_wp2_from_corr_wx from true to false. UWM has noticed that this change improves some results using CLUBB single-column, and would prefer this flag be set to false in CAM by default. I have tested this change and diagnostics are here. It does not seem there are large differences in general, but it does brighten the clouds a bit on average (by about 0.45 W/m2) and reduce LWCF by about 0.15 W/m2 (these are results from a 2-year test).

Will this change answers?

Yes

Will you be implementing this yourself?

Any CAM SE can do this

[vlarson]

Here's a brief explanation of what the CLUBB option l_min_wp2_from_corr_wx=T does and why I think it's flawed. When l_min_wp2_from_corr_wx=T, CLUBB boosts the values of wp2 in order to ensure that the correlation between vertical velocity w and total moisture rt lies between -1 and 1. While of course we don't want correlations to stray outside [-1,1], boosting wp2 when the simulation is producing unphysical results can be dangerous. In two of our deep convective single-column cases at UWM (arm97 and twp_ice), this correlation clipping can produce large values of wp2 for several hours. An example of this, from arm97, is shown by the red line, which has l_min_wp2_from_corr_wx=T. When l_min_wp2_from_corr_wx=F (green line), the extreme values of wp2 are diminished:

(Ignore the blue and black lines.)

We haven't tested l_min_wp2_from_corr_wx=F much in global CAM runs. The runs that Ben has done show that setting l_min_wp2_from_corr_wx=F changes the results a little, but not much. Maybe that means that the correlation clipping isn't triggered much in global runs. But it seems safer to me to shut it off.

[bstephens82]

@adamrher @vlarson here are the 10-year results. It does look like they are much more similar now after 10 years.