RECOVER: Ecosystem Condition and Recovery Modeling

This project aims to extract ecological indices from Google Earth Engine (GEE) and train regression models to predict ecosystem conditions.

Project Structure

.
├── data/                        # Dataset files (CSV, Shapefiles, GeoJSON)
├── notebooks/                   # Jupyter notebooks for exploration and analysis
├── scripts/
│   ├── extraction/              # GEE data extraction scripts
│   ├── processing/              # Data transformation and feature engineering
│   ├── ml/                      # Model training, evaluation, and prediction
│   ├── analysis/                # Trajectory analysis, clustering, benchmarking
│   ├── visualization/           # Plots, maps, and cluster visualizations
│   ├── validation/              # Clustering validation suite
│   └── utils/                   # Diagnostics and utility scripts
├── models/                      # PyTorch model checkpoints
├── plots/                       # Generated visualizations and performance plots
├── pyproject.toml               # Project metadata and dependencies
└── README.md                    # Project overview and documentation

Getting Started

Prerequisites

• Python 3.8+
• Google Earth Engine account and project access

Installation

1. Clone the repository.
2. Install dependencies using your preferred package manager (e.g., pip):

   pip install .
   

Usage

1. Data Extraction

Use scripts/extraction/gee_extraction.py to extract indices from GEE.

python scripts/extraction/gee_extraction.py

2. Model Training

Train the multi-output regression model using scripts/ml/train_regression_model.py.

python scripts/ml/train_regression_model.py

3. Evaluation

Evaluate the trained model using scripts/ml/evaluate_model.py.

python scripts/ml/evaluate_model.py

4. Reference Condition Prediction

Predict reference states and uncertainty intervals using scripts/ml/predict_reference_conditions.py.

python scripts/ml/predict_reference_conditions.py

5. Benchmarking

Select similar natural sites (benchmarks) and calculate the Benchmarked Condition Index (BCI) using scripts/analysis/select_benchmark_sites.py.

python scripts/analysis/select_benchmark_sites.py

6. Visualization

Generate benchmarking distribution plots using scripts/visualization/plot_benchmarking_results.py.

python scripts/visualization/plot_benchmarking_results.py

License

MIT