Set of R scrpts for data processing

License

AGPLv3, see file `LICENSE`.

Purpose

Data processing scripts to analyze a large ensemble of simulated crop yield data from the GGCMI[1] Phase2 experiment[2] and for pixel level evaluation as conducted by Müller et al.[3] in comparison to the Ray et al.[4] data. The analysis based on these scripts is published by Müller et al.[5] and this archive of scripts is meant to increase transparency and reproducibility. 

 

Usage

The scripts can be run as is, but require

  • a version of R, compatible with the scripts (I used version 4.1.2)
  • installation of all R libraries used
  • access to the raw GGCMI output data, as referenced in the paper[3]
  • adjusting of paths

There are some interdependencies, also indicated in the comments in the scripts.

Before you will be able to run

  • `ggcmi_read_data_for_pixel_irs.R` or
  • `ggcmi_irs_clustering.R`


you will have to run

  • `ggcmi_create_data_for_pixel_irs.R`

which generates the pre-processed data needed.

The evaluation script `ggcmi_evaluation_boxplot.R` is independent of the other scripts, but needs pre-computed reference data supplied as `[maize|soy|rice|wheat]_agmerra_hist_[default|harmnon|fullharm]_yield_boxdata_geoshare.Rdata` from [3].

The file `runR_slurm_largemem.jcf` is a job description file for submitting a script to the HPC2015 cluster at PIK.

References

[1]: https://agmip.org/ag-grid-2/
[2]: http://dx.doi.org/10.5194/gmd-13-2315-2020
[3]: http://dx.doi.org/10.5194/gmd-10-1403-2017
[4]: https://dx.doi.org/10.1038/ncomms6989
[5]: https://dx.doi.org/10.22541/essoar.168394775.56087254/v1