Input data

Input data is originally stored in its unedited, raw form. It is then processed with the mrmfa package specifically developed for data preprocessing for the REMIND MFA (Dürrwächter et al., 2025)¹. The package consists of R scripts which bring the raw data into the form that is used by the MFA. This includes, for example, combining different data sources into one set, filling regional gaps, interpolating in time if necessary, and formatting. The mrmfa package builds on the madrat package (Dietrich et al., 2025)², a package for transparent and reproducible input data processing developed by PIK. It takes the regional resolution as an input parameter, such that different resolutions and scopes can be used in the future.

The processed input data in the form of dimension files and parameter datasets is available open source on Zenodo, with the exception of proprietary datasets. Currently, all datasets necessary to run the plastics MFA are available, while parts of the steel and cement datasets can not be published. For each model, the input data consists of one folder ("dimensions") containing dimension files and another folder ("input_data") containing parameter datasets, the regional resolution of the input data and a text file with the input data version ("version.txt"). In order to run the model, these folders have to be stored according to the input_data_path specified in the config (e.g. "data/material/input"), and the input_data_version in the config needs to be set to the string defined in the "version.txt" file.