This code involves a variance decomposition analysis of climate change mitigation scenarios, in order to understand the spread in these scenarios. The scenario spread of variables such as primary energy carriers, energy use in end-used sectors, emissions and more is attributed to three drivers: (1) climate targets, (2) model differences, and (3) other scenario assumptions.
The code reads in the IPCC AR6 scenario database that is publicly available here (version Nov 2022). Both the full data as well as the metadata are required. These files can be stored in any file directory, but make sure that this directory is specified in the input.yml file. Other required datafiles are already inside this Github repository. For the calculations, it only requires standard Python packages like pandas, tqdm and yaml. For plotting, packages such as plotly are used.
As mentioned, the first step is to save the IPCC AR6 scenario data into a location, that you subsequently put into the input.yml file under the entry location_ipcc. In this file, you can also set other parameters: threshold_dataremoval dictates how many scenarios a model should at least have to be part of this analysis (default 10), sample_size_per_ms indicates the number of model-scenario pairs that is drawn for each model-climate category label when creating the samples (default 3000), resampling is the number of times the analysis is redone (including sampling), in order to average out potential stochasticity (default 100), and removal_c8 (either yes or no) indicates whether the C8 climate category should be included or not (default yes).
The calculations are done in the Main.ipynb, which reads in a Class objects from the files class_datahandling.py and class_decomposition.py and step-wise reads in data, performs the decomposition, and saves the data into the Data folder. Of particular interest is the subfolder Data/Output_files, where a netcdf file called Variances.nc is stored. This file includes the relative importance of the three drivers - the outcomes of this research.
Plotting scripts are provided in the folder Plotting Scripts, for each individual figure separately, annotated by the name of the file. At the top of each plotting script, which are jupyter notebooks, you can find a number of parameters to tune the plotting.
For more information, please have a look at our preprint: https://www.researchsquare.com/article/rs-2073170/v1. Also, please cite this reference if you use this code or data in your own research.