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

Suspiciously low CAPE values

James King

James King
Member
Hi all,

I have a run of CAM6-Chem from CESM2.1.3 with coupled land (CLM5-BGC-Crop) and oceans (POP2) forced by SSP1-2.6 GHGs and land use. I wanted to look at some convection parameters, however the output from the model for CIN was 0 everywhere and the output for CAPE was an order of magnitude less that I would have expected (max values over land were ~300 J/kg, whereas ERA5 gives you ~3000 J/kg). The model was set to output these variables at monthly mean temporal resolution but you would still expect more CAPE than this. My questions are:

1 - where's the CIN gone? Is this perhaps a bug?
2 - why is CAPE so low? Are the units of CAPE output by CAM definitely J/kg and not something else?
3 - are there any handy utilities for manually calculating these quantities given that I'm not sure I trust the output diagnostics?

Relevant case docs are attached. This run was done on ARCHER2, an HPC in the UK, but I've placed an example of the CAPE and CIN output in Glade for reference:

/glade/work/jamesking/CAPE_q/ transient_SSP1-2.6_base_02_CAPE_CIN.2015-08.nc

Thanks,

James
 

Attachments

  • atm_in.txt
    42.3 KB · Views: 0
  • README.case.txt
    5.3 KB · Views: 0

peverley

Courtney Peverley
Moderator
CAM-Chem folks correctly pointed out that the problem is not CAM-Chem specific. Moving back to CAM board and I'll try to track down some information for you.
 

aherring

Adam
Member
1. The CIN calculation is from CAM5's shallow convection scheme. It is not used in CAM6 physics, hence it's an empty array
2. CAPE is in J/kg and there should have values on the order of 1000 over Tropical regions. However, it's not CAPE proper, but a cloud work function or "dilute CAPE" which assumes some entrainment rate, and so values should be substantially less than conventional CAPE calculations.
3. No, nothing handy.
 

James King

James King
Member
Thanks for this Adam, that's v useful info. No worries re a method for CAPE calculations, I'll work one out myself and see what numbers it gives me with conventional assumptions.
 

aherring

Adam
Member
To provide some more context. The dilute CAPE values should be on the order of 1000 J/kg on a time-step by time-step basis, but the time-average may be on the order of 100 J/kg as your seeing. Some evidence for this is in my QJRMS paper (See FIgure's 8 and 9), where I did code up the dilute CAPE routine used in CAM6 in matlab, and fed it the climo average T and q profiles in the deep Tropics, and got order 100 J/kg.
 

James King

James King
Member
Hi Adam,

Thanks for this also - that's a very neat piece of analysis. It would make sense if the time-averaging results in lower values, since even in convectively 'hot' places there will be plenty of time when conditions are stable over the course of a month (and as you show, the convection scheme is sensitive to horizontal resolution which in this case is 0.9x1.25 deg). Unfortunately the experiment in question here needed to be run over the entire 21st century so we didn't output 3D fields at higher temporal resolution for reasons of storage space. I've got similar numbers to your fig. 8 using monthly mean T and q from my model output, taking spatial means over tropical rainforests, and running CAPE/CIN calcs using the MetPy package in Python. I think the conclusion here is that CAPE isn't the most useful diagnostic for convection in monthly mean model output...

Cheers,

James
 

aherring

Adam
Member
I doubt you have the FREQZM variable, but that's what I like to track -- it is the fraction of time the ZM scheme triggered at a particular grid cell (dilute CAPE exceed 70 J/kg trigger values). I use it various ways in that study, e.g., Figure 7. You could also divide the climo CAPE values by FREQZM to get the average CAPE when ZM is active.
 
Top