Model Analysis Repository
The Latin word "mar" translates to "sea". This repository will contain a collection of (mainly) ocean-focused analyses to inform next-generation ocean and climate model development.
- Interactively (clone the repository, edit the notebooks, and run)
- Execute the batch script
run_mar.sh - Visit https://dora.gfdl.noaa.gov/analysis/mar
Jupyter notebooks are the encapsulation of a particular analysis. There are relatively few constraints on how an analysis built, but there are a few interfaces to be aware of:
The batch and web engines for MAR (items 2 and 3 above) will set two runtime environment variables. Use one or both of these fields to determine the top-level path to a model experiment to analyze:
MAR_DORA_ID: The experiment ID in the dora databaseMAR_PATHPP: The top-level path to the post-processing experiment directory of the experiment (e.g./some/path/pp/)
Each notebook should have a default set of model years to analyze (e.g. 1981-2010). The MAR engines will also provide two optional, additional variables,
STARTYR and ENDYR, that can be used to override the defaults in the notebook.
If your notebook produces scalar metrics, it should write those results to a YAML file. See the SST_bias_NOAA_OISSTv2.ipynb notebook for an example of
how to construct a YAML file. Some examples of scalar fields might be RMSE and bias of a field, or the average depth of the Mediterranean outflow plume.