Hi there,
I wish to continue a closed topic. Sorry, I had a busy time.
Summary: I am trying to shorten the running time of an isimip3a suite, which runs in two period: a spinup from 1801 and then from 1901-2022, the historic run. That information is in <./model_scenario_info_isimip3a.dat>. It requires some input data, which now are in the folders </home/users/eempvg/MODELS/Obsclim/drive_files.dat> listed on. Since I want to run a shorter period we remove some files from the list, which had 24 rows at the beging an now has 6. We changed the “drive_nfiles_hist” from 23 to 5, “styr_drive_hist” from 1801 to 1981 and “styr_post_hist” from 1901 to 1981. The year (1981) was selected to match the dumpfiles’ availability.
I am presenting the problem and how it has progressed including the recommendation for the closed topic. Everythingh seemed to be fine, but the suit u-dh532 failled just after the spinups with the following error:
{MPI Task 5} [FATAL ERROR] INIT_DRIVE: Error reading file name/time pair from file (IOSTAT=-1 IOMSG=end-of-file during read, unit 1, file /home/users/eempvg/MODELS/Obsclim/drive_files.dat)
Any help is really apreciated, remember my new suite is u-dh532.
I’ll have a look, but you’ll need to enable read permission for other users to your home directory first so that I can see /home/users/eempvg/MODELS/Obsclim/drive_files.dat. The command
The error appears to be occurring in subroutine init_drive because the variable nfiles in namelist jules_drive is causing the program to try to read beyond the end of drive_files.dat. Tracing the source of nfiles back through app/jules/rose-app.conf, suite.rc, and model_scenario_info_isimip3a.dat, it appears that it could come from either drive_nfiles_hist under drive_info or drive_nfiles_hist_isimip3a under drive_nfiles_hist_isimip3a_info. I’m not sure how all the suite machinery works, but I suspect that it’s the latter value (equal to 24) that is being used.
I had a look in the suite’s README file to see whether there was any information on what the different drive_nfiles_hist variables mean. There wasn’t, but I did notice that the FAQs at the end suggest a different way of changing the start date. Have you tried that?
Hi David,
I didn’t try; thank you for pointing that out. I defined RESTART_HIST=true and RESTART_HIST_YEAR=1981, as suggested in the README, and now the model is running, but I need to wait to see if it works and if the model goes further than before.