Amara-Adam/crop-health-monitoring
Domaine:
agriculturegeospatialType de record:
software
Créateur:
Ama
Hôte:
# Crop Health Monitoring (Sentinel-2)
Tools to turn Sentinel-2 L2A imagery into vegetation index maps (NDVI, NDRE, SAVI) for precision agriculture. Designed for small-to-medium fields with simple, reproducible steps on Windows, macOS, or Linux.
## ✨ What it does
- Downloads or ingests Sentinel-2 **Level-2A** products (surface reflectance).
- Resamples/aligns key bands to **10 m**.
- Stacks Red, Red-edge, NIR bands into a single raster.
- Computes **NDVI**, **NDRE**, **SAVI** and exports GeoTIFFs.
- (Optional) Clips outputs to your **AOI** (GeoJSON/Shape) and produces basic stats.
## 🧠 Background (short)
- Sentinel-2 MSI has 13 spectral bands (10–60 m). For crop vigor & early stress:
- **B04 (Red, 10 m)**, **B08 (NIR, 10 m)** → NDVI
- **B05 (Red-edge, 20 m)**, **B8A (Narrow NIR, 20 m)** → NDRE
- We resample B05 and B8A to **10 m** so all math is pixel-aligned.
_(For a concise domain overview used to shape this pipeline, see the attached internship report.)_ :contentReference[oaicite:0]{index=0}
## 📂 Project structure
├─ data/
│ ├─ S2/
│ │ └─ L2A/ /
│ └─ indices/
│ ├─ R10m_tif/ # resampled 10 m bands live here
│ └─ outputs/ # final indices + clipped rasters
├─ aoi/
│ └─ aoi.geojson
├─ scripts/
│ ├─ compute_indices.py
│ └─ utils.py
├─ environment.yml
└─ README.md
## 🛠️ Requirements
- Python 3.10+ (Conda recommended)
- GDAL, Rasterio, Numpy, GeoPandas, Shapely
- (Optional) QGIS for visualization
Create the environment:
```bash
conda env create -f environment.yml
conda activate s2_agro