wiki:CDB_Optim

Version 2 (modified by lalibert, 8 years ago) (diff)

--

Finding which models are avalaible for your diagnostic

The CDB offers a simple script, find_optimset.sh, to find the optimal set of models available for a given diagnostic. As explained in the tutorial?, the script first looks up for the following header informations:

  • Experiments and year range requested (e.g. historical:1965,1970, for tas_zonal)
  • Variables and time frequencies required (e.g. tas:day,atmos,day, for tas_zonal)

For example, find_optimset.sh would go through this sequence for the tas_zonal script:

  1. Find all models with 'historical' ouptuts
  2. For each model, find all the rNiMpL numbers that contain tas at the daily frequency
  3. For each of the model's rNiMpL, loop through years 1965-1970.
  4. Once the loop is over, if some years are missing, echo the list of missing years.
  5. Repeat 2-4 for all other models.
  6. Once this loop is over, the script echos the list of available models for this experiment.

The tas_zonal diagnostic requests by default the experiments historical:1965,1970, historical:1965,2004 and rcp45:2060,2099. In such a situation, the script runs the previously described algorithm on these three experiments and then echos the models found in some but not all experiments as well as the models that are found for all experiments. This way, it is possible to know the maximal set of models to use and also to see which models have missing data. This makes it easier for the user to fix the archive, as described here?.