Add alternative formulation of scavenging in convective clouds.

[oyvindseland]

What is the feature/what would you like to discuss?

An alternative formulation of convective cloud scavenging developed by has been developed by Pengfei Yu at the University of Colorado, Geophysical Research Letters, 46,1061–1069. The original code has been made available bu must be updated the the recent ZM files and connected to oslo-aero. Potential candidate for NorESM2.5 but should be tested in 2.3 due to project obligations. Files zm_conv_intr.F90 and zm_conv.F90

Is there anyone in particular you want to be part of this conversation?

No response

Will this change (regression test) answers?

Yes

Will you be implementing this enhancement yourself?

Yes, but I will need some help

[oyvindseland]

I have looked at the code from Pengfei Yu, published in Geophysical Research Letters, 46,1061–1069.
What you should be aware of is that the the code does not change the scavenging, only the transport. I notice from old notes that I have seen this before although I did not fully understood the idea behind the parameterisation.

The deep convective transport in NorESM (and CESM) is expressed as the fraction of aerosols which is not being scavenged:
Transport fraction=1-stratiform scavenging - convective in-cloud scavenging
Transport mass = Transport fraction * interstitial concentration mass of entrainment into updraft
Convective in-coud scavenging: Convective cloud volumeassumed activated fraction of aerosols*(precipitation rate/in-cloud cloud-water)

The problem is that this assumes that the entrained mass = cloud volume independent of updraft speed. This creates a problem because the entrained mass-flux within one timestep is often higher than the mass within the convective cloud. The solution to this should have been to calculate the scavenging within the mass flux, but the mass flux is calculated after the scavenging. Also there has been attempts to replace the convective parameterisation for the last 15 years or so, so it is possible that no one has wanted to touch the problem.

The added parameterisation I still do not understand at least from a physical point of view. It looks like it is added a flux that is dependent on the thickness of the layer, so the deeper the layer the more added the mass is reduced, i.e. a thick layer is suppoosed to have much higher vertical velocity than a thinner layer