A lightweight satellite analysis pipeline for monitoring vegetation change using Landsat (5/7/8/9) and Sentinel-2 imagery through the Microsoft Planetary Computer API.
The system computes yearly NDVI (Normalized Difference Vegetation Index) composites and visualizes deforestation trends in an interactive Streamlit map.
- Processes Landsat 5–9 and Sentinel-2 scenes from 1985–2024
- Computes NDVI using dataset-specific scale and offset values
- Masks clouds, water, snow, and shadows using QA and SCL bands
- Outputs Cloud-Optimized GeoTIFF (COG) composites per year
- Visualizes NDVI and ΔNDVI (change) in a FastAPI + React Dashboard
- Streams imagery efficiently via the Planetary Computer — minimal local storage required
Language: Python
Core Libraries: Dask, StackSTAC, Xarray, RioXarray, Rasterio, GeoPandas, NumPy
Visualization: FastAPI, React, Leafmap, Matplotlib
Data Source: Microsoft Planetary Computer STAC API
| Year | Observation |
|---|---|
| 1995 | Dense vegetation with minimal disturbance |
| 2002 | Visible clearing in northern region |
| 2021 | Significant NDVI decline due to deforestation |
NDVI change progression — 1995 → 2002 → 2021