I am hoping to run a perpetual year experiment at 2010 conditions on UKESM AMIP and was hoping you could point me towards some documentation as to the simplest way to achieve this? I have found this page Using the UMUI - UKCA but not one which applies to Rose.
but these are not UKESM-AMIP configurations and are for 2000 not 2010. There are instructions on how to convert these timeslice runs to transient ones, and you could follow these in reverse, although the tricky bits would be making up the data files.
you can see directories containing things like the emissions and land fraction files etc., CFC surface concentrations, and volcanic AOD. These would need to be replaced within Rose for the current time-varying ones, as well as editing the radiation panels to make these only have 2014 concentrations for the trace gases. If you want year 2010 instead these numbers will need to be extracted from the historical files and made-up into new files.
For the SSTs/sea-ice you may need to think about how you make these up, i.e. do you take a mean of e.g. 2009-2019 to give a mean 2014, or just take 2014 and repeat that year-on-year. I would suggest taking a mean value to remove particular ENSO features that may exist in a single year (unless these features are perhaps what you want to include!). You could also take data from the UKESM ensembles and make up a climatology from the NEMO output if you wanted to use modelled data.
You should find that if there are any problems in not providing input data the model will exit with an error as it would run out of data as the model passes 2014 (from 2015 onwards for timeseries simulations you would follow a scenario instead).
you can see that there are both pre-Industrial and historial AMIP suites. Doing a diff (e.g. xxdiff -r suite1/ suite2/) will highlight what has been changed between these and essentially those are the files that you need to replace.
Thank you- this is all very helpful! I definitely need to have 2010 conditions for the anthropogenic emissions but I have the option of using 2010 meteorology or running with changing meteorology so will begin with creating 2010 emissions files.
I was also wondering if the input4mip original files which were used for the greenhouses gases in this wiki: https://code.metoffice.gov.uk/trac/UKESM/wiki/CmipGhg were available. I’m looking specifically for the latitudinal varying methane file which is mentioned as the path to the Jasmin workspace does not seem to exist!
The input4mip files are probably available online, but I’m not sure where they would be on JASMIN. I’m not sure who processed these files either but you could contact the Met Office and ask the UKESM team who might know more about this.
I was just going back to a few points in this and had a quick question: the radiation panel you mentioned, I had dealt with this by switching off each of the UKCA radiation panels as I wanted to turn off the feed-backs. I have found this changes the methane quite significantly in the mesosphere so am considering changing the radiation panel as you say. I was wondering if you could point me towards the right section of rose to look at ( ie, is it the radiation logicals in Section 34 of UM Science Settings or is it Section 01 02 Radiation?).
You need to be careful doing this. Did you just turn off switches in the UKCA Section 34 panel, or also in the Section 01 02 Rose panel?
You might have a situation where the base model was set-up to use UKCA CH4, O3, N2O etc., but turning these off without setting appropriate values could mean that the base climate state is now quite different. Additionally, if you are running with GLOMAP-mode aerosols then turning off all the aerosol feedbacks will also affect the base climate state as now it’s a clean atmosphere.
It’s possible to run with aerosol climatologies and then to turn off the UKCA chemical feedbacks to ensure that the dynamics are the same across multiple runs. This exists in a standard GA7/GA8 suite from UMvn11.2 (see roses-u:#183) using the aeroclim-chem option, although you probably want to make changes to the emissions etc. in the override file, or copy the settings into the suite proper to enable you to tweak the settings.
I had initially turned off the switches in Section 34 (and noticed the strange output) so have been looking at the 01 02 panel also - is it correct to turn off the equivalent switches in this panel (for sw and lw) but keep on the ones which don’t correspond (eg. oxygen)? And with these switched off to set a sensible base climate state, is this done using the Gas MMR section within Section 01 02 or by editing the sw and the lw files?
I have GLOMAP aerosols on so will need to run with the aerosol climatologies as you say, my job is a copy of u-be303 at vn11.1. Looking through that ticket, it seems that the change is including the line UM_OPT_KEYS=‘aeroclim-chem’, and then I’m guessing that within the rose-app-aeroclim-chem.conf file I need to replace the em_files listed with the ones I created for 2010?
Thank you for taking the time to look at this - I haven’t looked too closely at the panels other than UKCA before this so this is all new!
I would avoid changing anything in the S01-02 panels except to add concentrations if the values are set to missing data (which would then get picked-up when turning feedbacks off). Changing things there will affect the radiation scheme and I’m not sure of the impacts from there. To turn off feedbacks you just need to make changes to the UKCA panel and switch to aerosol climatologies.
The concentrations in the Gas MMR section? It seems the time varying Gas MMR switch is on already but I can also change fixed concentrations. Both of these are different to the ones shown in the ticket!
Also, with the aerosol climatologies on, do I keep the radaer switch on?
Yes, the gas concentrations. When you turn the UKCA feedbacks off the radiation scheme will then use the values from these panels & the ozone climatology supplied in the reconfiguration panel.
When you use aerosol climatologies these are actually from CLASSIC not GLOMAP - it isn’t possible to use the GLOMAP_CLIM options with UKCA chemistry yet, so it uses the CLASSIC settings. This should be controlled with the override file, but note here that it will override settings in the Rose panels. Look in the .cfg file held in the app/um/opt directory to see what it changes.
Within the override file in opt, many of the terms in run_ukca need changing, and currently I have copied over the run_ukca section from the rose-app.conf standard file, but I am unsure which bits I need to keep the same to allow the aeroclim-chem to work. For example, the ukca_scenario in the opt file is zero, but my code needs to use the rcp scenario etc. Simply copying over my run_ukca section leads to a fail as there are additional terms in the standard run_ukca so I’m guessing that these terms are not being overwritten and are still being read from the standard file? (ie are there any bits that must be kept for the opt file to work correctly).
It will over-ride all the settings and take the ones from the opt file. You may want to manually apply the settings needs to turn off GLOMAP & RADAER, turn on the CLASSIC climatologies etc., but leave everything else untouched. You can’t copy the whole of your run_ukca section as the bits relating to GLOMAP/RADAER and the feedback settings may be different.
Most of the changes in the opt file outside of run_ukca are the files for the CLASSIC climatologies and also turning on and initialising the chemistry - the initialisations could go as you will already have these, similarly the STASH output is already in.
Having l_ukca_chem_aero=.false. is key, and making this change will actually turn off a load of other things by setting !! at the start of a line (meaning that it’s ignored when making up the namelist). File locations don’t matter, and the scenario options aren’t important for this change (they just had to be set to something in the file).
This config file is designed to be set on a GA configuration which didn’t have either chemistry or aerosols on originally so it’s got everything turned on as required.
Ok makes sense, I think it is easier in that case to use the opt file and copy some of my settings over (scenario etc.). There are a few terms which are different in the aeroclim-chem vs the standard rose-app.conf:
dir_strat_aer, file_strat_aer, l_ukca_asad_columns=.false.,l_ukca_ibvoc=.true.,l_ukca_use_background_aerosol=.false.
Are any of these settings required by the aeroclim-chem or can these be changed to what I had initially? (I’m guessing the background_aerosol one might be key!)
I had assumed the CLASSIC climatologies were the qrclim files but these file names are not correct (eg ROSE_DATA/etc/ancil/qrclim.blck) and produce an error. I have found there are files in this directory I think are ok to use? $UMDIR/ancil/atmos/n96e/aerosol_clims/ga6_antie/v1/qrclim.blck85
I’m also receiving an error related to the stashcode 132 is a DMS file that I can see being read in in the standard rose-app.conf. I can see a similar error being corrected here https://code.metoffice.gov.uk/trac/roses-u/changeset/146263 so have commented this out in rose-app.conf. This is also meaning that I can no longer read in emissions for some species and had to add C5H8. I also receive errors now regarding some reactions eg. SO2 OH → HO2 SO3, so I am concerned that something has gone wrong.
Also have had to comment out all of the stash and the initialisations in the aeroclim-chem module.
The l_ukca_use_background_aerosol logical isn’t actually associated with this - this is an offline file read-in for the UKCA chemistry and if T means that a climatological file is expected (12-months) whereas if F it means that a time-series is applied. The l_ukca_asad_columns should be left as you currently have it.
The key thing is the qrclim files specififed. If these don’t exist then you will need to use alternatives. They will be installed via the install_ancil task I think. In the optional file, does the file have locations like $UM_ANCIL_ARCLBLCK_DIR/$UM_ANCIL_ARCLBLCK_FILE? These will be set by the install_ancil task, but could be hard-coded to something else. Looking at how I originally set-up this version those files are probably fine.
It might be better for you to make the necessary changes from the opt file to within rose. This configuration may not have been updated correctly for the UM version you have, and so could have errors, completely unrelated to what you are actually trying to do. The main thing is to specify the CLASSIC climatologies and turn off GLOMAP, then turn off the other feedbacks. The opt file is designed to be run on top of a job that may not have any aerosols or chemistry at all, so has more options changed than you would need.
Great thank you - I have a version which runs now with the optional file using the qrclim.____85 files I found (having had to comment out the SO2+OH reaction and the DMS input file and many of the stash requests being unavailable). The optional file doesn’t have those directories, it reads in from ‘$ROSE_DATA/etc/ancil/qrclim.blck’ and my install_ancil also doesn’t have these directories set.
I’ve run a few tests with this working version and can see that changing the concentrations in the radiation panel (Gas MMRs) appears to have no effect my output. I am confused by this as you mentioned before that when feed-backs were off in UKCA the radiation scheme would be using these panels? (potentially this is caused by l_clmchfcg=.true.?)
Thank you for your help on this.
I have been comparing the output from runs with ukca rad off and aeroclim-chem on to the standard UKESM and can see there are higher temperatures and differences to the methane especially in the mesosphere and upper stratosphere. I was just wondering if this is a known difference between the schemes/ something that has been found before or if something in my upper atmosphere is incorrect.
I haven’t done a detailed analysis. It’s quite possible that there will be differences, potentially large ones regionally depending on what’s been changed. UKESM will have been configured and tuned with the processes turned on, so turning bits off will lead to differences.
You may need to take a set back and think about what you are wanting to achieve from the runs. If you want to keep dynamics constrained more, perhaps nudging is better. Mohit has written instructions about a nudged UKESM-AMIP job here