Running experiment with evaluation
The only difference to the explanations given for the experiments without evaluation ( Install | Configure and run test simulations) is that you have to specify the settings in ${SPDIR}/chain/gcm2icon/sp001_eva-suite/scripts_eva-suite/eva-suite_settings
before you run spice. Please read here eva-suite_settings to check the settings carefully and be aware of the format for the settings of EVAL_START
, EVAL_END
, SUBREGIONS
and SUBREGIONS_ID
.
If you have already done another experiment with SPICE in advance, then you can use this old experiment as a reference simulation and you can setup the Eva-Suite to compare your new test simulation against the reference simulation by specifying REFID
, REFDATADIR
, and REFITYPE_TS
. Otherwise, leave the three settings blank.
If you are ready with checking the settings, you can start your test simulation with
cd ${SPDIR}/chain/gcm2icon/sp001_eva-suite
./subchain start
Be aware that you need to have at least a one-year experiment to achieve a convenient evaluation.
If you also want to evaluate the icon2icon experiments (downscalings of ICON), you need also to adapt the file eva-suite_settings
there (icon2icon/sp002_eva-suite
). Then, you can again start your simulation.
If you want to trigger the evaluation manually after your simulation has finished, then go to the subchain directory of your experiment, edit scripts_eva-suite/eva-suite_settings
, and type
./subchain eva-suite
Â