---
license: cc-by-4.0
pretty_name: SolarMap.PH rooftop solar detections
tags:
  - remote-sensing
  - solar-energy
  - rooftop-solar
  - philippines
  - geospatial
  - sentinel-2
---

# SolarMap.PH data products

Open rooftop-solar detections for the Philippines, built from satellite imagery by [SolarMap.PH](https://solarmap.ph). Rooftops are detected with CLIP image embeddings plus a gradient-boosted classifier, then segmented and snapped to OpenStreetMap buildings. All geometry is EPSG:4326 (WGS84). CC-BY-4.0.

The trained model itself lives in a separate repo: [`xmpuspus/solar-map-ph-clf-v4`](https://huggingface.co/xmpuspus/solar-map-ph-clf-v4).

## Coverage

Per-building and per-tile detections across NCR plus Cebu, Davao, Cagayan de Oro, Iloilo, Calabarzon, Bacolod, and Legazpi, with city- and barangay-level aggregates.

## What's here

| File | Granularity |
|---|---|
| `per_building_solar_ncr.geojson` | Per OSM building (commercial/industrial/public; residential roofs suppressed) |
| `rooftop_solar_<region>.geojson` | Per 240 m tile center, by region |
| `solar_map_ph_2026Q2.geojson` | Per city |
| `city_detection_counts*.json`, `city_solar_saturation.json` | City and 240 m-cell aggregates |
| `solar_saturation_ncr.geojson`, `hot_spots_2026Q2.geojson` | Saturation and hotspot layers |
| `residential_solar_aggregate.json` | Residential counts only, no geometry (privacy) |
| `franchise_cities_polygons.geojson` | Distribution-utility franchise boundaries |
| `SCHEMA.md` | Full field-by-field schema for every file |
| `BENCHMARKS.md`, `MODEL_CARD.md` | Evaluation tables and model card |
| `dataset_v4.npz` | The labeled training tiles |

See `SCHEMA.md` for the exact property schema of each file (building id, footprint area, panel area, kWp estimate, calibrated confidence, and so on).

## Notes

- Residential rooftops are intentionally suppressed in the per-building file; only aggregate counts are published, for privacy.
- `kwp_estimate` is a rule-of-thumb (panel area / 6.0), not a metered figure.
- Detections are model output, not a verified installation registry.

## Links and citation

- Code and pipeline: https://github.com/xmpuspus/solar-map-ph (MIT)
- Trained model: https://huggingface.co/xmpuspus/solar-map-ph-clf-v4
- Site and methodology: https://solarmap.ph
- Cite: SolarMap.PH (2026, v1.0.0). https://doi.org/10.5281/zenodo.20178050