Versions follow Semantic Versioning (<major>.<minor>.<patch>).
Backward incompatible (breaking) changes will only be introduced in major versions with advance notice in the Deprecations section of releases.
This version of the software was used to produce v0.4.0 of the GHG concentrations. As far as we are aware, this version fixes all major bugs in v0.3.0 of the GHG concentrations. The difference in ERF between the CMIP6 data and v0.4.0 is less than 0.02 W / m^2. In most cases, it is much less than this. For the full list of updates since v0.3.0 (and before), see below.
For further analysis of the changes, see https://github.com/climate-resource/CMIP6-vs-CMIP7-GHG-Concentrations.
- Updated to produce the v0.3.0 concentrations. Re-wrote a bunch of stuff in the process, including no longer producing the 0.5 degree resolution data. This can be added in future again if there is demand. (#57)
- Reset the CI. As a result, all sorts of results could have changed. (#64)
- Re-wrote the mean-preserving interpolation module. The behaviour is now completely different, but also much easier to control and more performant. (#70)
-
Added reading of data from the NOAA network
Supports reading both flask and in-situ data for CO2, CH4, N2O and SF6 (#7)
-
Added auto-generation of a Zenodo DOI into the production config via
scripts/write-run-config.py. (#65) -
The new
local.mean_preserving_interpolationmodule. This is tested in isolation by the tests intestsand could be re-used in other applications. (#70) -
Added a first list of references to the output file. Future work (#62) will make these even more refined as needed, but this already provides better visibility for observations people than we had previously. (#78)
-
Use data from Droste et al., 2020 for the remaining PFCs (and c-C4F8, which previously used AGAGE), except c8f18, for which we still use CMIP6 (due to a lack of papers since Ivy et al., 2012). (#81)
-
Added a DOI into the generated files. (#65)
-
Added a check to ensure that no negative values appear in the outputs. (#71)
-
- Switched to writing the output data as float, rather than double, to save space and avoid the illusion of precision.
- Switched to writing files in time chunks that better suit the access pattern of CMIP (files are now written separately for 1-999, 1000-1749 and 1750- rather than all in one).
(#74)
-
- Used data from Chapter 7 of the WMO 2022 ozone assessment, Velders et al., 2022 and Western et al., 2024 for global-means of ODSs and HFCs where available.
- Improved the interpolation for SF6-like gases to be smoother and more reliable
- Used AR6 values for radiative efficiencies and equivalent calculations
(#76)
-
Updated to a newer version of Velders et al. (2022) (#77)
-
Added CMIP IPO/ESA funding acknowledgement. (#78)