This project presents an AI/ML-based system to identify and classify Tropical Cloud Clusters (TCCs) using half-hourly satellite data. The workflow combines unsupervised clustering with Random Forest classification to detect convective cloud systems and assess their severity using meteorologically interpretable features.
- Preprocess satellite-based cloud observation data
- Identify potential Tropical Cloud Clusters (TCCs)
- Apply clustering algorithms to group similar convective systems
- Use Random Forest classification to label and assess TCC severity
- Extract and interpret key meteorological features influencing classification
- Clustering:
- KMeans used to discover natural groups in the cloud system data
- Classification:
- Random Forest model trained on derived features to categorize TCCs into severity levels