Model and Code for publication 'Inferring the Controlling Factors of Ice Aggregation from Targeted Cloud Seeding Experiments'

# Inferring the Controlling Factors of Ice Aggregation from Targeted Cloud Seeding Experiments > **Authors:** Huiying Zhang , Fabiola Ramelli, Christopher Fuchs , Nadja Omanovic , Anna J. Miller , Robert Spirig, Zhaolong Wu , Yunpei Chu , Xia Li, Ulrike Lohmann , and Jan Henneberger > **Affiliation:** ETH Zurich > **Contact:** huiying.zhang@env.ethz.ch ## 📖 Introduction This repository contains the source code, machine learning models, and datasets associated with the research project/paper: **"Inferring the Controlling Factors of Ice Aggregation from Targeted Cloud Seeding Experiments"**. It includes: 1. **Machine Learning**: aggregation rate prediction models (XGBoost, AdaBoost, etc.) 2. **Visualization**: Correlation plots and Probability Density Function (PDF) analysis etc. --- ## 📂 Repository Structure The project is organized as follows: ```text . ├── README.md ├── casual_analysis │ └── casual.ipynb ├── correlation_plots │ ├── EDR_temp.ipynb │ ├── ICNCt0_IWC.ipynb │ ├── ICNCt0_Temp.ipynb │ └── residenceTime_Temp.ipynb ├── correlation_subplots │ └── agg_rate │ └── ICNCt0_AggRate.ipynb ├── environment.yml ├── ml_models │ ├── AdaBoost.pkl │ ├── BayesianRidge.pkl │ ├── CatBoost.pkl │ ├── DecisionTree.pkl │ ├── ExtraTrees.pkl │ ├── GradientBoosting.pkl │ ├── KNN.pkl │ ├── LightGBM.pkl │ ├── LinearRegression.pkl │ ├── RandomForest.pkl │ ├── Ridge.pkl │ ├── XGBoost.pkl │ └── models_performance.csv ├── models_training │ ├── model_training.ipynb │ └── physical_ml_models_2.0_withoutEDR_groupBased.ipynb └── pdf ├── icnc_pdf.ipynb ├── irre_rim_pdf.ipynb └── majsiz_pdf.ipynb ``` --- ## 🚀 Getting Started ### 1. Prerequisites The code relies on Python and several data science libraries. We recommend using **Conda** to manage the environment. ### 2. Installation To set up the environment and reproduce the results, please follow these steps: ```bash # 1.Create the environment from the provided file conda env create -f environment.yml # 2. Activate the environment conda activate ice ``` --- ## 📊 Data Usage The data is located in `casual_analysis/data/`. * **Main Dataset:** `merged_with_edr_postprocessed.csv` * **Metadata:** Please refer to `data_description.md` in the data folder for a detailed explanation of column names, physical units, and measurement techniques. --- ## 🤖 Model Usage Models are stored in `ml_models/`. You can load them using Python's `joblib` or `pickle` library to make predictions on new data. **Example (Python):** ```python import joblib import pandas as pd # 1. Load a trained model (e.g., XGBoost) model_path = 'ml_models/XGBoost.pkl' model = joblib.load(model_path) # 2. Load your data (ensure feature columns match the training data) # data = pd.read_csv('your_new_data.csv') # 3. Predict # predictions = model.predict(data) ``` To retrain the models or view the training process, run the notebook: `models_training/model_training.ipynb` --- ## 📈 Visualizations (Notebooks) To view the analysis plots (PDFs and Correlations), start the Jupyter Notebook server: ```bash jupyter notebook ``` * **Correlation Analysis:** Go to `correlation_plots/` (e.g., `EDR_temp.ipynb`). * **PDF Analysis:** Go to `pdf/` (e.g., `icnc_pdf.ipynb`). --- ## 📝 Citation If you use this code or data in your research, please cite our paper: ```bibtex @article{zhang2025inferring, title={Inferring the Controlling Factors of Ice Aggregation from Targeted Cloud Seeding Experiments}, author={Zhang, Huiying and Ramelli, Fabiola and Fuchs, Christopher and Omanovic, Nadja and Miller, Anna J and Spirig, Robert and Wu, Zhaolong and Chu, Yunpei and Li, Xia and Lohmann, Ulrike and others}, journal={EGUsphere}, volume={2025}, pages={1--31}, year={2025}, publisher={Copernicus Publications G{\"o}ttingen, Germany} } ``` ## 📄 License This project is licensed under the [CC-BY](https://www.google.com/search?q=LICENSE) - see the https://www.google.com/search?q=LICENSE file for details. ``` ```