Territorial Transformation Index (TTI): Brazil 2017–2024 dataset and reproducible pipeline

This deposit contains the complete research package supporting the Territorial Transformation Index (TTI), a scalar, label-agnostic and spatially scale-invariant indicator of territorial change derived from satellite semantic embeddings (AlphaEarth Foundations, Google + DeepMind). The dataset represents the national application of TTI to Brazil for the period 2017–2024. The deposit includes: • National raster of territorial transformation (2000 m resolution) • State-level aggregated statistics (27 federative units) • Municipal dataset (5,506 municipalities) • Random pixel sample dataset (N = 10,000) • 25 km regular grid aggregation (five regional tiles) • Reproducible Google Earth Engine / Colab pipeline • Preprint manuscript (English and Portuguese) TTI is defined as the complement of cosine similarity between embedding vectors of two temporal states: TTI(x,t1,t2) = 1 − E(x,t1) · E(x,t2) where E represents the 64-dimensional L2-normalized embedding vectors. This release provides the first national-scale empirical characterization of embedding-based territorial transformation detection using AlphaEarth embeddings. Version 1.1 adds the national GeoTIFF raster (GW_STT_2017_2024.tif) that was missing in the initial release.