Reversal of UHI Drivers in a Sahelian City - processed raster and pre-fit ML models

Processed 10-band raster stack and pre-fit machine-learning model artifacts (XGBoost, Random Forest, SVM) supporting the analysis in "Reversal of UHI Drivers in a Sahelian City: Low Built-Up Density Increases Heat in Ouagadougou" (Lindner, Adamson, Ajadi, Christa, Hagan; 2026). Files: ouaga_aligned_stack.tif - 10-band GeoTIFF at 30 m resolution covering the Ouagadougou administrative boundary, March-May 2022-2024 hot-season composite. Bands: NDVI, NDBI, BSI, DEM, distance_to_water, distance_to_roads, built_density, green_density, LST, hotspot. CRS: UTM Zone 30N. Hotspotters_Models.zip - Pre-fit binary classifiers (xgb_model.pkl, rf_model.pkl, svm_model.pkl) trained on the raster above using a 70/30 random train/test split with random_state=42. These artifacts reproduce the exact published F1, Cohen's κ, and SHAP results. The raster is fully regenerable from public satellite sources (Landsat 8/9, Sentinel-2, Copernicus DEM GLO-30, JRC Global Surface Water, ESA WorldCover, OpenStreetMap) using the source code at github.com/helyne/ouaga-urban-heat-drivers.