hssiddiqui/Irrigation-Detection-SSA

Siddiqui H, Small C and Modi V (2025) Operationalizing remote sensing methods for smallholder dry season irrigation detection in sub-Saharan Africa. Front. Remote Sens. # Smallholder Irrigation Detection ## Repository Structure The repository contains scripts and files for small holder irrigation detection across sub-Saharan Africa ``` Irrigation-Detection/ ├── data-collection-polygons/ # Folder for input files │ └── ...... ├── trained-models/ # Trained transformer models │ └── ...... ├── GEE_scripts/ │ └── Interactive_EVI_Extraction.js # View Sentinel 2 EVI time series for polygons ├── Google_colab_scripts/ ├── s2_imagery_upload.ipynb # Download S2 EVI stack ├── clean_labelled_data.ipynb # Cluster cleaning script ├── smooth_inference_imagery.ipynb # Svatisky Golay Filter for smoothing ├── irrigation_detection_inference.ipynb # Run irrigation predictions on EVI stack └── utils.ipynb # Utility functions └── Sahel_labeling_v4.pdf ``` ## How to implement the methodology > [!NOTE] > Users can skip steps 1-4 and use the trained models to detect dry season irrigation in Nigeria or Burkina Faso 1. **Collect labeled data for training:** ``` GEE_scripts/Interactive_EVI_Extraction.js ``` - Open script in Google Earth Engine - Draw polygons by looking at high resolution Google Earth imagery and false color composites of dry season S2 imagery - Examine S2 EVI time series - Save 2. **Create Sentinel 2 EVI stack:** ``` Google_colab_scripts/s2_imagery_upload.ipynb ``` 3. **Clean labelled data:** ``` Google_colab_scripts/clean_labelled_data.ipynb ``` 4. **Train models:** - Detecting Irrigation in the Ethiopian Highlands 5. **Download and smooth inference imagery:** ``` Google_colab_scripts/s2_imagery_upload.ipynb Google_colab_scripts/smooth_inference_imagery.ipynb ``` 6. **Irrigation Detection:** ``` Google_colab_scripts/irrigation_detection_inference.ipynb ```