Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 20 Next »

If you install the spice with the addon EvaSuite, then you can do two things:

  1. Compare your simulation experiment with the well-known EObs dataset (Version 23, other versions or datasets needs your adaptation).
  2. Compare your simulation experiment and a reference simulation with EObs.

The hereon-evasuite is a stand-alone evaluation program (Evasuite-gitlab-dkrz). In the spice-package we tried to integrate automatically the process of installation, configuration and execution. This is a hard task and might be somewhat flawed in some places.

General notes for evaluation with the eva-suite addon

The period for comparison must be within the total period of the available EOBS data set (1950 onwards). Please, keep in mind that the quality of the reference data is not the same over the whole domain and over all decades. The chosen evaluation period requires to cover at least one year, otherwise the evaluation crashes when computing the seasonal cycle. If you want to consider additional observational datasets or more model simulations, you should consider to work with a stand-alone environment of the hereon-evasuite. 

The use of eva-suite in the runtime-environment of spice is always optional and can be switched off by setting ITYPE_EVAL=0 in the experimental settings ${EXPDIR}/scripts_eva-suite/eva-suite_settings file. Mean sea level pressure, 2-meter temperature, maximum and minimum temperature, total precipitation and global radiation are analyzed for BIAS and PDF. The results are stored under the directory ${SPDIR}/chain/work/${EXPID}/eva-suite/graphics.

Depending on the resolution and time period, the evaluation can be very time-consuming and memory-consuming. Therefore, the job-script of the evasuite (${EXPDIR}/scripts_eva-suite/eva-suite.job.sh) should be checked for the amount of time and memory allocated. A 1-year evaluation for a simulation with 0.44° consumes about 30 minutes computational time.

  • No labels