The SCO-Cartovege project developed a decision support tool for the conservation of flora and habitats of subantarctic islands, by combining vegetation surveys, satellite imagery and artificial intelligence modeling. The developed remote sensing R pipeline can be transferred to others territories.
Automated pipeline for very high-resolution habitat mapping in remote environments
Cartovege is an end-to-end, fully scripted pipeline developed in R to produce high-resolution habitat and vegetation maps from remote sensing data, with minimal human intervention.
Originally developed to address the challenges of mapping in isolated, cloud-prone insular regions (e.g., French subantarctic islands), the pipeline integrates:
- Orthorectified multispectral imagery
- Digital elevation model
- Vector mask of region of interest
- Field habitat data (in situ and/or photo-interpreted)
It uses a Random Forest supervised classification framework and supports hierarchical habitat typologies (up to N levels) and the calculation of landscape metrics and diversity indices for ecological monitoring.
🛠️ Key features
- Fully automated and reproducible workflow
- Adaptable to various datasets, spatial resolutions, and ecological contexts
- Transparent script-based implementation (no GUI required)
- RMarkdown tutorials
📄 Licence & Citation
This pipeline is open source but protected by a MIT Licence.
Please cite this pipeline using the following DOI: https://doi.org/10.5281/zenodo.15827741
⚙️ Prerequisites
Make sure you have R 3.2 (or later) installed.
Clone the repository:
git clone https://github.com/dianeespel/Cartovege.git
cd Cartovege
Set up your environment: The Requirements/ folder contains the .yml files for the different environments used in the project. Choose the one that fits your needs and create the environment using conda:
conda env create -f Requirements/environment_name.yml
conda activate environment_name
🚀 Usage
-
Data preparation: Place your satellite image(s) (GeoTIFF) and reference polygons (Shapefile) in the respective folders.
-
Configuration: Before each script edit the file to specify input and output paths
-
Run the pipeline: Execute the main script from the terminal:
Rscript R_scripts/Cartovege_main_script.R
📂 Project structure
Cartovege/
│
├── data/ # Input satellite images and reference data
├── Requirements/ # Conda environments (.yml files)
├── R_scripts/ # Processing and analysis scripts
├── Rmarkdown_tutorials/ # R markdown files
├── LICENCE.txt # MIT Licence file
├── CITATION.cff # CITATION file
└── README.md # Project documentation
👤 Authors / contact
Diane ESPEL, PhD