chrisroadmap / climateforcing Goto Github PK

• update docstrings to be fully PEP-compliant
• SECPERYEAR needs updating to mean tropical year length (365.24219 days): two repositories use 365 days so this will need to be in a later release (better still, make days_per_year an input with a default)
• Refactoring: pylint has graffitied all over it
• increase coverage

Simple examples of how to use

• aprp
• atmos
• utci
• solar
• add documentation via sphinx/readthedocs

• pick a style for docstrings
• apply across existing functions
• add to contributing guide
• add to CI/CD? (pydocstyle maybe)

should be updated to remove py3.6 and add py3.10 and py3.11

Currently have two different humidity calculations, one in the UTCI function and one in atmos.humidity, and they are not the same. Changing one of them will fail tests - down to me to understand the most appropriate calculation to use.

• understand differences (https://github.com/mzelinka/aprp/blob/master/code/python/demonstrate_APRP.ipynb)
• correct my code if buggy

currently fails with PermissionError

From testing, it looks like a module conflict is causing loads of junk output to be printed from the APRP code

chrisroadmap/cmip6-aerosol-forcing#1

• linting
• docs
• contributing guide
• CI
• setup.py
• pypi release
• changelog
• versioneer
• deploy tests

Currently, some take dicts of CMIP6 diagnostics (APRP) and some take parameters (mean radiant temperatures). Should move to dicts for all to keep number of function arguments down and overall module consistency

• issue request template
• pull request template
• conda-forge
  • sort out conda versioning issue (stray "v" prepended; currently you have to conda install climateforcing==v0.2.1 rather than conda install climateforcing==0.2.1)
• continous integration for deployment
  • codecov upload
  • pypi
  • conda

1. the e_s relationship is out of Broede, and I don't have a literature reference for it; convert to something that is in the literature (e.g. Alduchov & Eskridge 1996)
2. utci calculation currently has no dependence on surface air pressure and it probably should through its input into saturation specific humidity

The current climateforcing.solar module calculates instantaneous and time-mean solar zenith angle, but of course we can be more generally useful:

• allow for array input for time in cos_mean_solar_zenith_angle
• calculate direct and diffuse (time-averaged) irradiance on any non-horizontal plane
  • calculate (time-averaged) solar azimuth angle
  • diffuse irradiance on a tilted plane*

*this problem needs to be kept tractable and within the limitations of what can be provided by climate models - many sophisticated tilted irradiance relationships exist in the literature but much beyond the isotropic diffuse model might be overkill given that most climate models use plane-parallel and two-stream radiation codes.

• implement
• comparison to e.g. Met Office code
• tests

See https://github.com/chrisroadmap/ar6/blob/main/src/ar6/utils/h5.py