Controlling outputs in ACCESS-CM2

Earth System Coupled Model
1

Hi,

I want to reduce the output of an ACCESS-CM2 suite as much as I can.
In the UM, you change this by turning off or on the stash requests.
I am just trying to workout what the different frequencies mean for the time profile as it is not clear and I am not sure where to find documentation. Is anyone able to tell me what each time profile here means?

  • T3HMN = 3hr mean?
  • T3HR = 3hr frequency?
  • T6H = 6hr?
  • T6HDMPM = 6hr?
  • T6HR = 6hr?
  • TALLTS
  • TCOUP = coupling timestep?
  • TCOUPMN
  • TDAYM = day?
  • TDAYMAX = daily max?
  • TDAYMIN
  • TDMPMN

With editing the ocean diag table, is it simply a matter of commenting entries?

Thanks in advance
Sebastian

2

Not sure on the specifics for CM2, For a general UM run you can check the documentation at https://code.metoffice.gov.uk/doc/um/latest/papers/umdp_C04.pdf section A.1 to help decode the STASHC file.

E.g.

&umstash_time
    iend=-1,
    ifre=1,
    intv=-1,
    ioff=0,
    iopt=1,
    isam=1,
    istr=1,
    ityp=2,
    tim_name='VTACC1HR',
    unt1=2,
    unt2=1,
    unt3=2,
/

is for a time profile named VTACC1HR, accumulated from samples every 1 timestep, output every 1 hour.

3

Does that answer the question re UM?

In MOM6 (I’m not sure what was used in CM2) there are two sections, the file list and the fields list.

The files list is configured based on what grid the data is saved on (rho2 for pressure levels, native, z, sfc for surface, static for time-invariant). So you might turn off one of those vertical files if its not needed. Otherwise you can also control at the variable level by deleting / commenting out / modifying each line.

(See Diagnostics — MOM6 0.2a3 documentation)

4

Thanks, it sort of helps but not really. For some of the UM time profiles it is simple, but others are less clear.

For example the unt1/2/3 input can be labelled dump periods or timesteps but it is not clear what that means, making it hard to decipher what the time units are.

ACCESS-CM2 ocean is mom5, here is a sample of the ocean diag table:

# WGOMD/CMIP5 Table 2.3
# all of these fields are time dependent 
#
# Rosetta stone: u=sea_water_x_velocity
 "ocean_model","u","u"                       ,"HISTORY/ocean_month","all",.true.,"none",2
#
# Rosetta stone: v=sea_water_y_velocity
 "ocean_model","v","v"                       ,"HISTORY/ocean_month","all",.true.,"none",2

If I wanted to suppress one of these variables it is just commented or deleted?

5

Are you looking at these in rose? @sonyafiddes says to run rose edit which should launch a gui, then look in the index for ‘Time Profiles’ and then there are textual descriptions for them.

I pretty sure that is correct.

6

Yeah I am looking at this in the rose gui but it doesn’t really help. See below the entry for TDMPMN. The gui doesn’t say what a dump period (unt1) or timestep (unt2) is, so can’t work out if this is a monthly mean output or somehting else

image
image763×825 59.6 KB

7

Chermelle will solve this :slight_smile:

8

Hi Sebastian,

The dump frequency is a configurable setting, that is set in another part in the UM.

If you look under um > namelist > Model Input and Output > Dumping and Meaning you can find the frequency choices for dumping.

My speciality is high-resolution modelling, not running climate models, but in a tutorial I downloaded the dumping frequency has been set to Gregorian months.

Hope that helps.

1 Like
9

The biggest clue would be to look at the data itself. If there are multiple entries per Gregorian month then it would be per timestep.
Just FYI, the timestep is defined in um → namelist → Top Level Model Control → Model Domain and Timestep.

Let me know if you want me to have a look at the data.

10

https://accessdev.nci.org.au/trac/wiki/ticket/139/StashTable#TDMPMN

This might be easier ??

Manually adding/removing or introducing of the STASH item has been recorded in ACCESS-CM2 tutoring by Chloe: https://youtu.be/tw05r9-o_SI?t=3610

Basically, use rose/cylc to open the STASH code list, then unclick those you don’t want, then click TidyStashTransformPruneDuplicated

Give it a try, and let me know if that works for you.

1 Like
11

Thanks @ars599

FYI the accessdev trac wiki will disappear shortly. @paulleopardi has done some great work to transfer the wiki to GitHub. It is available here

The ticket you referenced is here:

1 Like
12

Thanks all for the help
The reason I was asking is that I only want to archive monthly outputs and was trying to work out which time profiles to keep on.
I think that TDMPMN will correspond to monthly but was checking what the other profiles are.
I assume that T6HDMPM means that the value is averaged every 6 hours over the dumping period, based on that table? so it is also a monthly output?

13

The UM climate mean system is tied into the restart dumps in a confusing way. TDMPMN and T6HDMPM are both used for monthly means. Fields using them will have the UPMEAN use_name In CM2 we don’t save seasonal or annual means as some Met Office suites do.

T6HDMPM is used mainly for pressure level diagnostics. I’m not sure whether this is because they’re more expensive to calculate or to give a better match to observations which might be based on 6 hourly analyses.

Don’t turn off anything that has the UPUKCA or UPCOUP use_name. These fields are used internally for coupling with the chemistry scheme and ocean/ice. They’re not written to output but the model will break without them.

Depending on which suite you have it might have high frequency output (3 and 6 hourly) set. You can definitely turn these off.

If you click on the dom_name label it’ll sort by the output spatial domain. You’ll get the most effect by turning off fields saved on all model levels, DALLRH and DALLTH. Perhaps also tiled CABLE output (stash code 800 and onwards in section 3).

Screenshot 2024-02-15 151627
Screenshot 2024-02-15 1516271060×629 86.8 KB

Here you could turn off 0,253 with the UPMEAN use_name but not 0,253 with UPUKCA.

The single level DIAG fields aren’t worth worrying about

1 Like
14

Thanks @MartinDix thats really good to know - I had turned off almost all UPUKCA variables :open_mouth:

Are there any other components of the model namelists (ocean sea ice or atmosphere) that can’t be turned off?

15

I think only the UM has this interaction between the diagnostics and coupling. You can turn off any of the MOM or CICE diagnostics without causing a problem.

1 Like