4 posts
Estimate rooftop solar potential in Python for 2,148 buildings in Manchester: take a 1 m Environment Agency LiDAR DSM and Overture Maps footprints, compute roof slope and aspect, trace horizon angles for shading from neighbours and trees, build a sun map from a PVGIS typical year with pvlib, check it against pvlib's Perez model and PVGIS, then estimate kWp and kWh per year per roof and rank the best roofs.
Build a labelled deep learning dataset from free data: stream a Sentinel-2 L2A composite from the Planetary Computer, rasterise Overture building footprints over ESA WorldCover, cut georeferenced image chips, split them by spatial block so the test set does not leak, check the class balance and load it all as a TorchGeo dataset.
Turn classified LiDAR into building footprints with heights in Python: stream USGS 3DEP COPC tiles, build a DTM and a building nDSM from ASPRS class 6 points, polygonise and regularise the mask, split roofs at height steps, compute height statistics per building, score the result against OpenStreetMap with IoU, and export a GeoPackage and a 3D view.
Run Meta's Segment Anything Model on high-resolution aerial imagery with segment-geospatial (samgeo), on a CPU. Generate masks automatically, vectorise them to a GeoPackage, filter them by area, shape and NDVI to find buildings, and prompt SAM with points and boxes, checked against OpenStreetMap.