cmip7-prep

cmip7-prep turns raw NorESM and CESM history output into CMOR-compliant NetCDF for submission to CMIP7, and tells you what it produced.

This guide covers running it, reading what it produced, and changing it.

User's guide

• Introduction
  • The driver
  • The branded variable name
  • Mapping CMIP7 variables to model output variables
  • Mapping CMIP7 realm, frequency and sampling to output time series files
  • The grid specification file
  • The CMOR tables
  • What a run needs
• Installation
  • NIRD
  • Derecho
  • Your own machine
  • Docker
  • Checking it worked
  • The CMOR tables
• Tutorial
  • Look before you run
  • Run it
  • The arguments
  • Doing less than everything
• What a run produces
  • The output directory
  • Where the logs go
  • The CMOR output structure
  • The end-of-run table
  • Reading a validation report
• Recipes
  • Rerun after an interruption
  • Run everything again anyway
  • Redo only validation
  • One realm, one frequency
  • Check the output files join up
  • Run one stage by hand
  • Publish the validation reports
  • Limit the memory a run uses
• When something goes wrong
  • A time series step fails on a single history file
  • A step is labelled warning
  • Validation reports errors you cannot account for
  • Workers left running after a killed run
  • A run was killed partway through writing a variable
  • ModuleNotFoundError: No module named 'cmip7_prep'
  • The HTML index shows no plots for a realm
  • Memory on a shared node
• Internals
  • The mapping file format
  • Maintaining the mapping
  • The modules
  • The files in data/
  • Running the tests