Scheduled Downtime
On Tuesday 24 October 2023 @ 5pm MT the forums will be in read only mode in preparation for the downtime. On Wednesday 25 October 2023 @ 5am MT, this website will be down for maintenance and expected to return online later in the morning.
Normal Operations
The forums are back online with normal operations. If you notice any issues or errors related to the forums, please reach out to help@ucar.edu

How does CESM handle "parameterizations" to incorporate inputs from unresolved physics at the subgrid scale

krishnab

krishna bhogaonker
New Member
I was watching some of the NCAR CESM 2022 tutorials/introduction videos, and they are really helpful. One issue that Peter H. Lauritzen raised
when discussing the dynamical core of the CESM 2.2 model as well as the upcoming spectral elements core was this idea of "parameterizations" for
unresolved physics. Basically the grid or mesh for the CESM 2.2 model is based upon the Finite Volume scheme (correct me if I am wrong) and the
mesh basically has squares that are about 100 x 100 km in size.

With the 100 x 100 km grid size, these models cannot resolve physical processes that happen at a spatial scale smaller than 100 km in size. I think
Lauritzen actually says that the effective grid size for resolving processes may actually be more like 400km--based on the energy method for
stability analysis.

Hence, I was hoping someone could explain or point me to an article that explains how the fluxes from these unresolved processes are incorporated into
CESM? The language here is a bit tricky, so I am trying to be precise. Unresolved means that the phenomena cannot be discretized at a
certain spatial scale without significant error--but this does not mean that those processes are unmodeled.

I can guess at a few ways to approach this problems, but hopefully someone knows the answer. One simple method would be to take a random draw from a
statistical distribution for the unresolved phenomena and then use that in the forcing term for the larger spatial scale processes. Alternatively, someone
could iterate over the 100 x 100km mesh cells and run a smaller inner mesh to resolve the fluxes of the smaller scale physics, and then aggregate those fluxes
back to the full 100 x 100 km mesh scale. Of course the choice of best method will depend upon rigorous test and an evaluation of the accuracy versus
computational efficiency tradeoffs.

Thanks for any input and assistance. Like I said, if someone know the answer or if someone can just point me to the correct article, that would be excellent.
 

katec

CSEG and Liaisons
Staff member
The CESM2 paper by Danabasoglu et al, (2020) gives a good overview of the physical parameterizations used in each of the component model, plus references to papers explaining them each in detail. https://agupubs.onlinelibrary.wiley.com/doi/abs/10.1029/2019MS001916

There are hundreds of different parameterizations in CESM2 for an incredibly wide variety of spacial scales and physical phenomena. Each parameterization uses different methods to simulate the appropriate response of sub-grid scale physical processes to the large-scale forcing. The atmosphere, ocean, land, sea ice and land ice models all use multiple parameterizations for smaller scale processes.
 

krishnab

krishna bhogaonker
New Member
@katec Thanks so much for the link. That is very helpful. I just did not know where to look for the information. Haha, I understand that there are hundreds of parameterizations, but did not realize that so many different methods were used. I will read through the paper.
 
Top