I hope everybody is enjoying the sunshine; as I have mentioned in other questions, I am running a JULES suite belonging to ISIMIP3a of the fire scheme. I was able to run the complete experiment, which ranged from 1801 to 2021; now, I want to run another experiment with some parameter variations, but I only need the 21st-century. So, I planned to skip running for around 2000 years using the output dump files from my first successful experiment. So, I changed the start year to 1993 to give it some time before the 21st century. However, I have not persuaded the model not to run the spinclim (spin up + transition period) of ISIMIP3a, which ranges from 1801 to 1900. So, the model is asking for a spinclim file and failed:
{MPI Task 5} [FATAL ERROR] file_ts_open: /gws/nopw/j04/mohc_shared/isimip-jules/isimip3a/InputData/climate/atmosphere/spinclim/20CRv3-ERA5/20crv3-era5_spinclimfill_rsds_global_daily_1801_1810.nc: Start time for the first file must match data_start
This is probably a very specific question, but if somebody knows how I can turn off the sipinclim from being able to run my experiment from 1993, I would appreciate the help.
Hi Maria Paula
You can look at the jules_spin and jules_initial namelists in your file ~/roses/u-dh119/app/jules/rose-app.conf.
Maybe there is something there that needs to be changed?
Patrick
Sorry for disappearing; I had a busy week. I gave a loof to rose-app.conf, I understand the model is reading initial conditions files and dump files, but I am unsure if that is OK. Turn False initial conditions so it does not need to look for the file, but It the model failed with an error I don’t understand. I considered using “file=${DUMP_FILE}” so the model looks for initial conditions in the dump file and turns True again using the file.
I checked out a copy of u-dh119 to see whether I could understand what was going on. One thing that mystifies me is where the suite is getting the name of the file in the FATAL ERROR message from. Nowhere in the suite do the strings InputData or spinclim occur. Have you any idea where the suite is picking up the file name?
I don’t know where sorry. I was looking where to change that filename as well because the folder where the function is looking for the initial conditions had 10-year files until 1900, so I was trying at least to be able to make it take the last 10-year file only, but I never found where to change it.
Note: I do know that some of the model specificities are on ./model_scenario_info_isimip3a.dat
I think I understand what is going on. The problem is with the specification of the driving data in namelist jules_drive in ~/roses/u-dh119/app/jules/rose-app.conf. The namelist variable file points to /gws/nopw/j04/mohc_shared/isimip-jules/isimip3a/obsclim/20CRv3-ERA5/drive_files.dat. This file contains a list of files of driving data together with the times of the first data in each file. The FATAL ERROR is occurring because the start time of the first file (in 1801) disagrees with the time specified by the namelist variable data_start (in 1993).
One solution would be to create your own cut-down version of the drive_files.dat file starting in 1991 and to modify the file namelist variable to point to it. You would have to start your experiment in 1991 instead of 1993 and you might also have to alter drive_nfiles_hist.
Hi David,
Thank you for identifying all that information; I will try what you suggested. I need to figure out how to create my 10-year input file for 1901 to 2000; that is what I understand is necessary to replace the current one. I would appreciate it if you had any idea how to do it.
Ignore the last comment. I just saw the file you mentioned and noticed it lists files for every decade. I hadn’t seen the files for years after 1900 but realised they were on a different route. I made a copy of drive_files.dat, modified that, modified it and where the model specified the route of the file. I reduced the number of files needed in drive_nfiles_hist and changed the data_start to match the first file year.
