Cloud Classification through Machine Learning and Global Horizontal Irradiance Data Analysis

This repository contains the pre-trained model and code to prepare GHI data and perform cloud classification as described in our paper "Cloud Classification through Machine Learning and Global Horizontal Irradiance Data Analysis".

Table of Contents

• Introduction
• Installation
• Usage
  • Data Preparation
  • Running the Model
  • Detailed Steps
• Contributing
• License
• Citation

Introduction

This project provides a pre-trained machine learning model to classify cloud types using Global Horizontal Irradiance (GHI) data. Users can prepare their GHI data, run the pre-trained model, and get predictions.

Installation

To get started with this project, clone the repository and install the necessary dependencies.

git clone https://github.com/anabelalusi/cloud-classification.git
cd cloud-classification
pip install -r requirements.txt

Usage

Data Preparation

1. Organize your data: Ensure your data file (for example "data.csv") is placed in the project directory
2. Data format: Your data.csv file should contain the following columns: ghi (ghi data at minute resolution), ghi_cs (a clear sky model), kt_modificado (the value of the modified clearness index), delta_kt_modificado (variation between consectuvie kt* values)

Running the Model

1. Ensure correct input data: Your input data files should be in the correct format and located in the project directory.
2. Run the provided script: Execute the script in a Python environment to prepare the data, compute features, and make predictions using the pre-trained model.

python run_model.py --model_path https://github.com/anabelalusi/cloud-classification/blob/main/cloud-classification-XGBoost.pkl --input_data path/to/processed/data --output_predictions path/to/save/predictions

Detailed steps

1. Insert Site Data: Provide the solar total irradiance (Gs), cosine of the zenith angle (cz), and the orbital correction factor (Fn).
2. Load data: Load the GHI and clear sky GHI data from the provided CSV files.
3. Calculate Clearness Index: Compute the clearness index and the modified clearness index from the loaded data.
4. Create DataFrame: Organize the data into a DataFrame and clean any rows with NaN values.
5. Feature Calculation: Define a function to calculate the necessary features from 33-minute windows of data.
6. Run the Pre-trained Model: Load the pre-trained model, make predictions based on the calculated features, and map numerical predictions to cloud class names.
7. Display Results: Output the predictions and their corresponding cloud class names.

License

This project is licensed under the MIT License. See the LICENSE file for details.

Citation

If you use this code or our models in your research, please cite our paper:

@article{Lusi2024cloudclassification,
  title={Cloud Classification through Machine Learning and Global Horizontal Irradiance Data Analysis},
  author={Anabela Rocío Lusi, Pablo Facundo Orte, Elian Wolfram, José Ignacio Orlando},
  journal={Quarterly Journal of the Royal Meteorological Society},
  year={2024},
  volume={150},
  issue={765},
  pages={5435-5451},
  doi={https://doi.org/10.1002/qj.4880}
}

For more information, please contact us at [email protected]