CMOR MIP tables for use with CMOR v3.13.1 and newer versions in preparation for CMIP7.
Note that versions of CMOR after v3.10 will be able to use these MIP tables, but minimum version 3.13 is required to correctly output the realization, initialization, physics and forcing indices.
To support overriding of long names, in the very small number of cases where this is required to match the Data Request v1.2.2.3, CMOR v3.13.2 is needed and the examples here have been updated to use it.
The CVs JSON file is now hosted in the tables-cvs directory. This file will be updated around 00Z each day via a github action using the esgvoc tools. Initially only two source_id entries exist for demonstration purposes, but these will be extended as the Essential Model Documentation is collected and processed. An update to modelling groups will be sent in the second half of February.
A large number of changes will be made to the CVs as new models, grids and eventually experiments are registered making this unwieldly. To reference a particular version of the CVs JSON file we recommend using the commit hash (e.g. 58efe39) from here and the date.
With the introduction of branded variable names and an updated set of global attributes the tables here look a little different to those for CMIP6.
- Variables are arranged in MIP tables by realm and indexed by branded variable name.
- Frequency is no longer defined for a specific variable and any valid frequency can be set via the input JSON file (the same is true for region).
Notable changes to the input JSON file used by CMOR
drs_specsshould be set to the drs_specs version this will beMIP-DRS7.0.0.0initially, but will be updated as changes to the data definitions are produced, e.g. new version of the Data Request.regionis required (usually"glb"for global variables, note the change in case as of Data Request v1.2.2.3)archive_idis"WCRP"frequencymust be specified and does not contain any suffixes as in CMIP6 (e.g.6hrPtand similar have been removed as has1hrCMused for the diurnal cycle diagnostics. The time sampling is now described in the Branded Variable Name.*_indexfields are now strings and must have the appropriate prefix, e.g.realization_indexshould be"r1"rather than1tracking_prefixhas been updated with the value required for CMIP7long_nameandcell_measuresboth have separate files keyed by the CMIP7 compound names. Modelling groups are asked to use these files as shown in the example notebooks
The tables and examples presented here are derived directly from Data Request version v1.2.2.3 with the following changes;
long_nameandmodeling_realmfields have been mostly "homogenised". Data Request variables sharing the same branded name also share the samelong_name(16 Data Request variables in v1.2.2.3 are exceptions, all sea ice variables).commentfields have been left blank as we have not yet "homogenised" this data.cell_measuresare currently blank, with a separate JSON file containing them indexed by the CMIP7 Compound name from the Data Request -- the examples show how to re-introduce this metadata, and updated guidance will be added here.
- There are some branded variable names that appear in both atmos and landIce MIP tables. We are expecting to change this in the next version of the Data Request to avoid duplication.
Each of these use the tables and the testing CVs JSON file
- Simple CMOR demo notebook (python script equivalent)
- Get the right environment either using conda with
cmor_environment.ymlor pixi using thepixi.lockfile
- Get the right environment either using conda with
- Example of "re-cmorising" CMIP6 data
Note in particular the lines used to add cell_measures metadata to variables.
Testing of these tables has been limited, so please report problems / suggestions via the issues.
The construction script uses the Data Request API and a set of reference files (adapted from CMIP6Plus) to construct the MIP tables and associated files.