LaunchPAD is the first phase of the Climate Model Evaluation Hub for Africa. The aim of this project is to improve understanding of how climate models represent African regions. During LaunchPAD we developed diagnostics to evaluate key African climate processes. 10 early career researchers, LaunchPAD fellows, created the diagnostics.
We ran the diagnositics on CMIP6 models, with the goal that they could also be run on other climate models, including future generations of CMIP models.
In this repository, you can find the diagnostics, which you are welcome to use. The diagnostics are also being added to ESMValTool. You can also find the description of the diagnostics, including scientific rationale, as well as plots showing the results for CMIP6.
To learn more about LaunchPAD, the fellows, and their diagnostics take a look at our project website: https://launchpad6.home.blog
Each of the diagnostics have an folder within DIAGNOSTICS.
If you would like to find out more about the diagnostics, and the results for CMIP6, you can click through to the README and the "plots" directories.
If you would like to run the diagnostics, you can find information on how to run them below.
The README for each diagnostic details the rationale for the diagnostic, a step by step method, and some sample plots. These also contain the contact information of the fellow responsible for each diagnostic and their contact information. Not all of these diagnostic methods have been published and permission to cite requires permission from the authors of the diagnostic.
This repository of diagnostics has been designed to run on the JASMIN data analysis facility (https://www.jasmin.ac.uk) and with the CEDA archive of CMIP6 model output (https://help.ceda.ac.uk/article/4801-cmip6-data). However, if you download the required data from the CEDA archive you can modify configuration files in the files/CONFIG directory to run diagnostics over a different system.
All of the diagnostics use the same set of basic configuration files in files/CONFIG:
- config.py contains some hardcoded items like a list of models to search for, month names and file extensions
- config_find_files.py contains markers for model locations and specific information such as modelling group/institution
- find_files.py defines absolute paths for model and observational files on JASMIN and in the LaunchPAD group work space
- config_functions.py contains definitions that are used in diagnostic calculations
Diagnostics are found in the DIAGNOSTICS directory. In each diagnostic directory there is a diagnostic.py file which is where the diagnostic is calculated and plotted. This is the only python file you need to execute. There is also a diagnostic_config.py where some diagnostic specific details are defined such as variable names and the list of observational datasets to use. This is also where you need to specify where the LaunchPAD repository will be located on JASMIN or any computer you are using. Do this by altering the home_add field
Once you have set up github on your computer (https://help.github.com/en/github/authenticating-to-github/connecting-to-github-with-ssh) you can clone the whole LaunchPAD repository. The directory you are in when you clone the repository needs to be added to the diagnostic_config.py in the home_add field as described above. This needs to be done in the config file of each diagnostic you want to run.
> git clone git@github.com:Priority-on-African-Diagnostics/LaunchPAD.git
If you would rather only clone one or a selection of diagnostics you can. If you take this approach you must always clone the files/CONFIG area with your diagnostics of choice. Follow the steps below that show how to pull only the African Easterly Jet diagnostic:
> git clone --filter=blob:none --no-checkout https://github.com/Priority-on-African-Diagnostics/LaunchPAD.git
> cd LaunchPAD/
> git sparse-checkout set DIAGNOSTICS/Central_African_Easterlies/ files/CONFIG
> git checkout master
To run diagnostics on JASMIN move to the DIAGNOSTICS/diagnostic of your choice and alter home_add in diagnostic_config.py is described above. Load the jaspy module (> load jaspy) and run diagnostic.py. Figure will appear in the plots directory and any intermediary netcdf files will appear in intermediary_files.
If you are not running on JASMIN you can load any conda environment that contains modules in files/CONFIG and diagnostic.py before running the diagnostic.
If you have any questions about the repository please get in touch with Rachel James (rachel.james@bristol.ac.uk) or Ellen Dyer (ellen.dyer@ouce.ox.ac.uk). We would be grateful to hear about any work or projects using the diagnostics in this repository.