Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,88 @@
+---
+license: cc-by-4.0
+task_categories:
+- other
+- feature-extraction
+tags:
+- geoscience
+- geology
+- tectonics
+- plate-tectonics
+- geospatial
+- earth-observation
+- remote-sensing
+language:
+- en
+pretty_name: GPlates Tectonic Intelligence Dataset
+size_categories:
+- 10K<n<100K
+---
+
+![GPlates Tectonic Intelligence Dataset](https://i.ibb.co/GvgsCy4d/Chat-GPT-Image-Aug-18-2026-12-48-59-AM.png)
+
+# GPlates Tectonic Intelligence Dataset
+
+Global, AI-ready tectonic context for every point on Earth — built from GPlates/EarthByte plate reconstruction data.
+
+## Dataset Summary
+
+This dataset converts the Müller et al. (2019) global plate reconstruction model into a standardized, gridded feature set. Every cell in a 1.0° global grid (360 × 180 = 64,800 points, present day / 0 Ma) is tagged with its reconstruction plate ID, seafloor age, and distance to the nearest mid-ocean ridge, subduction trench, transform fault, and coastline — turning raw tectonic geometry into tabular features ready for geological, geophysical, exploration, and environmental ML applications.
+
+## Dataset Structure
+
+**Format:** GeoParquet (WGS84 / EPSG:4326), one row per grid cell.
+
+| Field | Type | Description |
+|---|---|---|
+| `lon` | float | Cell center longitude (degrees) |
+| `lat` | float | Cell center latitude (degrees) |
+| `plate_id` | int | Reconstruction plate ID (Müller et al. 2019 model) |
+| `age_ma` | float | Seafloor age in millions of years; null over continental crust |
+| `dist_to_ridge_km` | float | Distance to nearest mid-ocean ridge segment |
+| `dist_to_trench_km` | float | Distance to nearest subduction trench |
+| `dist_to_transform_km` | float | Distance to nearest transform fault |
+| `dist_to_coastline_km` | float | Distance to nearest coastline |
+
+## Source Data
+
+- **Plate model:** Müller, R.D. et al. (2019), global full-plate motion and plate boundary model.
+- **Seafloor age grid:** Seton, M. et al. (2020), present-day oceanic crustal age and spreading parameters.
+- **Access layer:** [`gplately`](https://github.com/GPlates/gplately) (Mather et al.), built on `pygplates`.
+
+Original data distributed by EarthByte / GPlates under their respective open licenses; this derived grid is released under CC-BY-4.0 — verify against upstream terms before redistribution.
+
+## Processing Pipeline
+
+1. Downloaded core reconstruction files (rotation model, topologies, static polygons) and supporting layers (coastlines, continents, COBs, age grid) via `gplately.DataServer`.
+2. Converted vector layers to standardized GeoJSON and GeoParquet.
+3. Extracted labeled plate-boundary geometries (ridges, trenches, transforms) and coastlines as spatial features.
+4. Built a global 1.0° grid, assigned plate IDs via point-in-polygon partitioning, sampled seafloor age, and computed nearest-boundary distances (reprojected to an equidistant CRS for real-world km).
+5. Published the standardized dataset here.
+
+## Usage
+
+```python
+import pandas as pd
+
+df = pd.read_parquet("global_tectonic_context.parquet")
+df.head()
+```
+
+## Known Limitations
+
+- 1.0° resolution is coarse; finer grids (e.g. 0.25°) trade runtime for detail.
+- Bilinear interpolation on the age grid can return null values near coastlines even over oceanic crust, slightly undercounting true oceanic coverage.
+- Distances are computed on an equidistant cylindrical projection — accurate at global scale, with mild distortion near the poles.
+- Present-day (0 Ma) snapshot only; no time-series reconstruction yet.
+
+## Citation
+
+If you use this dataset, please cite the underlying sources:
+
+- Müller, R.D. et al. (2019), *Tectonics*, 38(6).
+- Seton, M. et al. (2020), *Geochemistry, Geophysics, Geosystems*, 21(10).
+- Mather, B. et al., *GPlately* (GitHub: GPlates/gplately).
+
+## License
+
+CC-BY-4.0