Climate Disaster Warning System

This repository contains the code, models, and supporting materials for detecting and analyzing climate-related disasters—specifically fire, flood, sea-level rise, and earthquake events. The project integrates deep learning and geospatial data analysis to support early warning systems and climate research.

📁 Project Structure

model/
  Fire/
    fire_detection_resnet50_V1.h5
    Fire_Detection.ipynb
    D-Fire/
  Flood/
    flood_detection.ipynb
    optimizer_vit.pth
    resnet_confusion_matrix.csv
    resnet_hard_predictions.csv
    resnet_metrics.pkl
    resnet_model_checkpoint.pth
    resnet_probability_predictions.csv
    resnet_test_metrics_summary.csv
    resnet_test_summary_metrics.csv
    vit_model.pth
  Sea-Level Rise/
    CSR_GRACE_GRACE-FO_RL06_Mascons_all-corrections_v02.nc
    SLR_GRACE.ipynb
    Data/
  Earthquake/
    input/
      test/...
      sample_submission.csv
      train.csv
    earthquake_detection.ipynb
    lgbm_flood_4.pkl
    lgbm_importances.png
    submission.csv
.gitignore
LICENSE
README.md
requirements.txt

🧠 Models Overview

🔥 Fire Detection

• Model: ResNet50 (Keras-based)
• Approach: Binary image classification (fire vs. non-fire) with transfer learning
• Justification: ResNet50's deep architecture and residual connections help mitigate vanishing gradients and boost accuracy on image tasks.

🌊 Flood Detection

• Models: ResNet and Vision Transformer (ViT)
• Approach: Image-based flood classification and evaluation
• Justification: ResNet is a proven CNN model, while ViT captures global context via self-attention, enhancing performance in complex flood imagery.

🌐 Sea-Level Rise Analysis

• Data Source: GRACE satellite NetCDF files
• Tools: Data processing and visualization in Jupyter Notebooks
• Justification: GRACE data offers precise Earth gravity measurements, enabling accurate inferences about sea-level and mass redistribution trends.

🌎 Earthquake Detection

• Model: LightGBM Regressor, CatBoostRegressor, SVR, NuSVR, KernelRidge
• Approach: Time-series or seismic data analysis for earthquake event detection and prediction
• Justification: Deep learning models can capture temporal and spatial patterns in seismic data, improving the accuracy of earthquake detection and early warning.

📥 Datasets & Pretrained Models

• Fire & Flood Datasets: Download from:

  • Kaggle Fire Dataset
  • Kaggle Flood Dataset Place files under:
  • model/Fire/D-Fire/
  • model/Flood/

• Sea-Level Data: Download from NASA’s GRACE portal: NASA GRACE Data

• Earthquake Data: LANL Earthquake Prediction Place files under:

  • Earthquake Detection/...

• Trained Models: Pretrained models can be downloaded from this Google Drive folder. Place them in the appropriate directories as shown in the project structure above.

📚 Components

Fire Detection

• Fire_Detection.ipynb: Full pipeline for training and evaluating the ResNet50 model.
• fire_detection_resnet50_V1.h5: Trained model weights.
• D-Fire/: Dataset directory for training/testing.

Flood Detection

• flood_detection.ipynb: Includes training and evaluation of both ResNet and ViT models.
• Evaluation metrics: CSV and PKL files track performance, predictions, and confusion matrices.

Sea-Level Rise

• [SLR_GRACE.ipynb](model/Sea-Level Rise/SLR_GRACE.ipynb): Notebook for visualizing and analyzing NetCDF-formatted satellite data.
• CSR_GRACE_GRACE-FO_RL06_Mascons_all-corrections_v02.nc: Satellite data file.
• Data/: Additional supporting data.

Earthquake Detection

• earthquake_detection.ipynb: Notebook for training and evaluating the earthquake detection model.

🚀 Getting Started

1. Clone the repository

   git clone https://github.com/md-hameem/Climate-Disasters-Warning-Systems.git
   cd Climate-Disasters-Warning-Systems

2. Install Dependencies Ensure Python 3.x is installed. Then run:

   pip install -r requirements.txt

3. Run Notebooks Launch Jupyter and open the relevant .ipynb files in each subdirectory.

🗂️ Notes

• Large model files are excluded via .gitignore.
• Ensure the appropriate models and datasets are placed in their respective folders before running the notebooks.

📄 License

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

---

📬 Contact

For questions, suggestions, or contributions, feel free to:

• Open an issue or submit a pull request
• Email: