VIs_to_LAI: Simulate Leaf Area Index from Vegetation Indices

Authors: Jonghan Ko at Chonnam National University and Chi Tim Ng at Hang Seng University of Hong Kong

Collaborator: Jong-oh Ban at Hallym Polytechnic University

GitHub Repository: https://github.com/RS-iCM/VIs_to_LAI

HuggingFace Dataset: https://huggingface.co/datasets/jonghanko/VIs_to_LAI/tree/main

---

Overview

VIsToLAI is a Python-based, open-source software framework designed to estimate leaf area index (LAI) from time series of satellite-derived vegetation indices (NDVI, RDVI, OSAVI, and MTVI₁). By integrating empirical regression, Log–log, and machine learning modules, VIsToLAI offers a flexible, scalable workflow that bypasses destructive sampling and intensive calibration. Pretrained models, an extensible API, and interactive Jupyter notebooks streamline data ingestion, model execution, and visualization. Demonstrated on staple crops under varied conditions, VIsToLAI accurately reconstructs LAI dynamics and integrates seamlessly into remote sensing workflows for precision agriculture, crop monitoring, and ecological modeling.

---

Features

• Multiple Vegetation Indices: Support for NDVI, RDVI, OSAVI, and MTVI₁
• Three Modeling Approaches:
  • Empirical exponential regression
  • Log–log regression
  • Machine learning (Extra Trees, Gradient Boosting, DNN)
• Flexible Workflows:
  • 1D time-series simulation for point/field data
  • 2D geospatial simulation for regional mapping
• Pretrained Models: Ready-to-use models for rice, barley, wheat, and maize
• Interactive Notebooks: Jupyter notebooks for reproducible workflows
• Extensible API: Modular design for custom indices and algorithms
• Built-in Visualization: Time-series plots, scatter diagnostics, and geospatial maps
• Ensemble Methods: Combine multiple models for improved accuracy

---

Installation

Prerequisites

• Python ≥ 3.8 (recommended: Python 3.10+)
• pip package manager

Quick Install

Install the package with all dependencies:

pip install -r requirements.txt

Or install as an editable package:

pip install -e .

Full Installation (with 2D/spatial features and Jupyter)

For 2D mapping and geospatial analysis:

pip install -e ".[all]"

This includes:

• Core dependencies
• Cartopy (for 2D mapping and shapefile support)
• tqdm (progress bars)
• Jupyter notebooks

Optional Extras

# For 2D/spatial analysis only
pip install -e ".[2d]"

# For Jupyter development only
pip install -e ".[dev]"

Docker Installation

Build and run with Docker:

# Build the image
docker build -t vis-to-lai-crops .

# Run with Docker Compose (recommended)
docker-compose up --build

Access Jupyter Lab at http://localhost:8888 (check container logs for token).

---

Quick Start

1D Time-Series Simulation

Run a notebook for 1D LAI simulation:

jupyter notebook RUN_Python_Rice.ipynb

Or use Python directly:

from codes.sim_VIs_to_LAI_crops import main
import os

# Set paths
path = os.path.abspath(os.getcwd())
para_FN = path + '/data/empirical_reg_parameters_rice.txt'
wobs_FN2 = path + '/data/Rice_LAI_n_VIs.csv'
data_FN = path + '/data/Rice_FN_NICS_2021.csv'
output_FN = path + '/outputs/SLAI_rice.out'

# Model files
DNN_FN = path + '/models/rice_NN.h5'
pkl_FN = path + '/models/pickle_extra_trees_Rice.pkl'
pkl_seq_FN = path + '/models/pickle_extra_trees_Rice_seq.pkl'

# Run simulation
# reg_opt: 0=DNN, 1=ML, 3=NDVI-based, 4=four VIs-based, 5=log-log, 7=Ensemble
main(DNN_FN, pkl_FN, pkl_seq_FN, 
     reg_opt=7,    # Ensemble method
     plot_opt=1,   # Show plot
     file_opt=1,   # Save output
     flag=5.5,     # Max LAI value
     para_FN=para_FN,
     wobs_FN2=wobs_FN2,
     data_FN=data_FN,
     output_FN=output_FN)

2D Geospatial Simulation

For 2D regional mapping:

jupyter notebook RUN_Python_LAI_2D_Rice.ipynb

---

Available Notebooks

1D Time-Series Notebooks

• RUN_Python_Rice.ipynb - Rice LAI simulation
• RUN_Python_Barley.ipynb - Barley LAI simulation
• RUN_Python_Wheat.ipynb - Wheat LAI simulation
• RUN_Python_Maize.ipynb - Maize LAI simulation

2D Geospatial Notebooks

• RUN_Python_LAI_2D_Rice.ipynb - Regional rice LAI mapping
• RUN_Python_LAI_2D_Maize.ipynb - Regional maize LAI mapping

---

Model Options

The framework supports multiple regression options (reg_opt parameter):

• 0: Deep Neural Network (DNN)
• 1: Machine Learning - Extra Trees Regressor
• 2: Machine Learning - Sequential (with temporal features)
• 3: NDVI-based empirical regression
• 4: Four VIs-based empirical regression (ensemble of all VIs)
• 5: Log-log regression
• 6: Ensemble 1 (DNN + ML + VIs + Log-log)
• 7: Ensemble 2 (ML + VIs + Log-log) - Recommended

---

Project Structure

VIs_to_LAI_crops/
├── codes/                      # Core Python modules
│   ├── sim_VIs_to_LAI_crops.py      # Main 1D simulation module
│   ├── empirical_VIs_to_LAI_2D_*.py # 2D empirical modules
│   └── each_crop_model/        # Crop-specific models
├── data/                       # Input data (CSV, OBS, TXT)
│   ├── *_LAI_n_VIs.csv        # Training data
│   └── empirical_reg_parameters_*.txt  # Regression parameters
├── models/                     # Pretrained models
│   ├── *_NN.h5                # DNN models
│   └── pickle_*.pkl           # ML models
├── outputs/                    # Simulation outputs
│   └── SLAI_*.out             # Simulated LAI files
├── class_map_*/                # 2D class maps
├── vis_*/                      # 2D vegetation indices
├── Shape_*/                    # Shapefile boundaries (2D)
├── RUN_Python_*.ipynb          # Jupyter notebooks
├── setup.py                    # Package setup
├── requirements.txt            # Python dependencies
└── Dockerfile                  # Docker configuration

---

Requirements

Core Dependencies

• numpy ≥ 1.20.0
• pandas ≥ 1.3.0
• scipy ≥ 1.7.0
• scikit-learn ≥ 1.0.0
• matplotlib ≥ 3.4.0
• tensorflow ≥ 2.8.0
• keras ≥ 2.8.0
• h5py ≥ 3.0.0
• pyyaml ≥ 5.4.0

Optional Dependencies (for 2D features)

• cartopy ≥ 0.20.0 (geospatial mapping)
• tqdm ≥ 4.64.0 (progress bars)

Development Dependencies

• jupyter ≥ 1.0.0
• ipykernel ≥ 6.0.0
• notebook ≥ 6.4.0

See requirements.txt for a complete list.

---

Usage Examples

Example 1: Rice LAI Simulation with Ensemble Method

from codes.sim_VIs_to_LAI_crops import main
import os

path = os.path.abspath(os.getcwd())
main(
    DNN_FN=path + '/models/rice_NN.h5',
    pkl_FN=path + '/models/pickle_extra_trees_Rice.pkl',
    pkl_seq_FN=path + '/models/pickle_extra_trees_Rice_seq.pkl',
    reg_opt=7,  # Ensemble method
    plot_opt=1,
    file_opt=1,
    flag=5.5,
    para_FN=path + '/data/empirical_reg_parameters_rice.txt',
    wobs_FN2=path + '/data/Rice_LAI_n_VIs.csv',
    data_FN=path + '/data/Rice_FN_NICS_2021.csv',
    output_FN=path + '/outputs/SLAI_rice.out'
)

---

Citation

If you use this software in your research, please cite:

@software{vistolai2024,
  author = {Ko, Jonghan and Ng, Chi Tim},
  title = {VIs_to_LAI: Simulate Leaf Area Index from Vegetation Indices},
  year = {2024},
  url = {https://github.com/RS-iCM/VIs_to_LAI}
}

---

License

[Specify your license here - e.g., MIT, Apache 2.0, etc.]

---

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

---

Support

For questions, issues, or contributions, please visit:

• GitHub Issues: https://github.com/RS-iCM/VIs_to_LAI/issues
• HuggingFace: https://huggingface.co/datasets/jonghanko/VIs_to_LAI

---

Acknowledgments

• Chonnam National University
• Hang Seng University of Hong Kong
• Hallym Polytechnic University

---

Last Updated: August 2025