Transient forcing component sets

[cnrwil004@myuct_ac_za]

Hi all

I have done ensemble runs (from spun-up ICs) of CESM1.2 with a B_2000 (fixed year 2000 forcing, if I understand correctly, without CN) compset. If it would be possible, I would now like to do matching ensemble runs with transient forcing conditions. However, it would appear that all compsets using RCP forcings include CN, whereas, for example, the B20TR transient forcing compset does not require this (I'm guessing this has something to do with a more implicit specification of atmospheric constituent concentrations in the RCP set-up, requiring more complete simulation of their evolution?). Having already completed the 2000-forcing runs without CN-cycling and given that my spun-up ICs were also produced by a run not including dynamic CN, time constraints do not allow me to now rerun these simulations with a different compset for comparison purposes. I assume that comparing runs with and without CN included is not particularly meaningful (I've done test runs to compare output and the variability in tropical forest locations, particularly, appear to be significantly different). Also, it would seem that I would have to do some separate spin-up for the CN runs (although I have not been able to find thorough documentation on this subject, so maybe I don't understand correctly at all).

The comparisons that need to be done, however, are for primarily qualitative comparison of variable distributions, so the exact forcing "pathway" is not vitally important. I am wondering whether there is any way to set up a run with transient forcing conditions, starting from year 2000 conditions (or nearby), which can be meaningfully compared to output with fixed year 2000 conditions (the investigation, in essence, is about interpretation of climate distributions under non-stationary conditions and how this compares with distributions produced under fixed "external" forcings)? One other possibility would be to use spun-up B_1850 ICs (which I also have), to do ensemble runs under these conditions, to then be compared with B20TR runs.

I realise that this is something I should have investigated more completely previously, but I am still a student trying to find my feet in an operational climate modelling context and I would like to now try to salvage whatever comparisons I can do with the runs I have done, but understand that this might not be possible. Any advice would be greatly appreciated.

Thank you in advance
Stefaan

 

[bundy]

I am by no means an expert on compsets, coupled runs or forcings, but as you haven't received an answer here, I will share what I know. First of all, the compsets are short-cuts to set up the model to run a very specific way.  Once it is set up, lots of things can be changed via the env*xml files or the component namelist files. My advice to you is to run create_newcase & cesm_setup for a few different compsets and then compare the run/namelist files to see what is different.  I find the README.case file that is generate to be especialy helpful as it lists all the options you ended up with, but also gives the 'longname' description of the compset. If you can interpret the longname, you can pick out what you want and create your own compset by giving it a custom longname.Keep in mind, of course, that you should change thing cautiously and not be surprised if something doesn't work, or, even if it does work, you need to do extensive testing to make sure it is doing what you think it should. 

 

[cnrwil004@myuct_ac_za]

Thank you, Bundy, for the answer, I have tried to change some settings and am now trying to resolve some resulting, apparently resolution-dependent, input file errors. I will keep your advice in mind as I proceed.