AngeloSorte/nasa-ndvi-vegetation-stress-classification

Machine learning project using NASA MODIS NDVI data to detect vegetation stress vs stability across Europe, Africa, and Asia using Earth Engine. # 🌱 NASA NDVI Vegetation Stress Classification This project analyzes global vegetation health using NASA MODIS NDVI satellite data and applies machine learning to detect environmental stress trends. --- ## 🚀 Objective The goal of this project is to: - Analyze vegetation trends using NDVI (Normalized Difference Vegetation Index) - Compare regions (Europe, Africa, Asia) - Detect vegetation stress (declining trends) - Apply a machine learning model to classify environmental conditions --- ## 🛰️ Data Source - NASA MODIS Satellite Dataset - NDVI Product: MOD13A2 - Provided via Google Earth Engine --- ## 🧠 Machine Learning Approach We compute: - Mean NDVI per region - Temporal NDVI trend (linear slope) Then we apply: - Random Forest Classifier ### Labels: - `0` → Stable / improving vegetation - `1` → Vegetation stress (declining NDVI) --- ## 🌍 Regions Analyzed - Europe - Africa - Asia (Defined using Earth Engine bounding boxes for reproducibility) --- ## ⚙️ Tech Stack - Python - Google Earth Engine API - NumPy / Pandas - Scikit-learn - Matplotlib --- ## 📊 Output Example The model outputs: - NDVI trend per region - Classification report - Confusion matrix - Stress prediction for unseen values --- ## 📈 Example Insight Regions with negative NDVI slope indicate potential: - Deforestation - Climate stress - Land degradation --- ## 🧪 How to Run 1. Install dependencies: ```bash pip install -r requirements.txt --- ## Authenticate Earth Engine: ee.Authenticate() Run the Colab notebook or Python script --- ## 📦 Requirements See requirements.txt --- ## 👨‍💻 Author Angelo Sorte --- ## 🌱 Future Improvements Pixel-level NDVI heatmaps Time-series deep learning (LSTM) Global vegetation anomaly detection Interactive Earth Engine dashboard