- Step 1: Prepare an AOI file and modify the GT data (QGIS)
- Make a geojson file of an AOI in EPSG:4326.
- Clip and reproject (e.g. EPSG:28350 in Albany) the GT data and rasterize it.
- Load the GT data
Raster > Extraction > Clip by mask layer
Raster> Projections > Warp (reproject)
(the output resolution should be matched to that of S2)- Export as geotiff
- Step 2: Data downloading and preprocessing (
S2_download_and_preprocessing.ipynb
)- Download an ECARR and NDVI image via GEE
- Preprocess the GT data
- Preprocess the index data (reprojection, shape adjustment, etc)
- Step 3: Main analysis (
eucalyptus_classification.ipynb
)- Train and validate the model for eucalyptus classification (logistic regression)
- Export the result as a vector map
