bhima_processing: analysis code for GEB-PlantFATE v1.0 afforestation scenarios

Analysis and figure code for a coupled GEB-PlantFATE study of afforestation in the Ghod subbasin, Maharashtra, India.

Stefaniak, E. Z., Joshi, J., Artuso, S., Maxwell, T. L., Smilovic, M., Hofhansl, F., and de Bruijn, J. A.: GEB-PlantFATE v1.0: Coupling of forest demographic and soil hydrological models for simulating the hydrological impacts of afforestation and biodiversity. [JOURNAL / STATUS — ADD DOI WHEN AVAILABLE]

Processes daily model output for twelve afforestation and biodiversity scenarios into catchment-scale aggregates, spatial diagnostics and publication figures. Model output is archived separately at doi:10.5281/zenodo.18457345.

What this is

This repository processes raw GEB-PlantFATE model output into the catchment-scale aggregates, spatial diagnostics and publication figures reported in the paper. It contains code and one small spatial file; the model output itself is archived separately at 10.5281/zenodo.18457345 and must be downloaded to run the pipeline.

The underlying study applies the coupled GEB-PlantFATE model to a 628 km² section of the Ghod subbasin in Maharashtra, western India, across twelve scenarios combining six afforestation levels with high-biodiversity (10 species), low-biodiversity (1 species) and no-PlantFATE model setups, forced with ISIMIP3b GFDL-ESM4 output under SSP3-7.0 over 2020–2049.

Contents

Both R and Python are used. R handles most time-series processing and plotting; Python handles Zarr access, spatial operations and map production. Two Google Earth Engine JavaScript files export biomass reference layers and are run in the GEE code editor rather than locally.

- code/ — preprocessing, aggregation and PlantFATE diagnostics
- code/figures_publication/ — one script per publication figure
- code/google_earth_engine/ — GEE export scripts
- code/util/ — cell-list generation and Zarr packing helpers
- data/maps/region.gpkg — study area boundary

The pipeline runs in four broad stages: configuration and spatial setup; aggregation of daily gridded output to basin-wide and yearly summaries; extraction of per-cell environmental forcing and single-cell PlantFATE re-runs; and figure production.

## Requirements

Python ≥ 3.13, managed with uv (uv sync); dependencies pinned in uv.lock. R packages are not pinned — see the README for the list. All scripts are run from the repository root. Full setup instructions, the data download and extraction steps, the figure-to-script mapping, and known issues are in README.md.

Data not redistributed

Figure A2 uses relative density and dominance tables from appendices 3 and 4 of Jezeera (2016), which remain the copyright of the author and are not included here. The README explains how to obtain them.

Related resources

- Model output: doi:10.5281/zenodo.18457345
- GEB: v0.4, doi:10.5281/zenodo.22094305 — https://github.com/GEB-model/GEB
- PlantFATE: [DOI PENDING] — https://github.com/jaideep777/Plant-FATE
- GEB documentation: https://docs.geb.sh/
- Regional context: Kanade et al. (2023), doi:10.5281/zenodo.8255959

Funding

International Institute for Applied Systems Analysis, Strategic Initiatives Program — projects RESIST and FairStream.

Licence

GNU General Public License v3.0.

---

This documentation was drafted with the assistance of an AI language model and reviewed and corrected by the authors.