Digital targeting · 2023 to present
Bedrock outcrop detection
LiDAR, radar and optical data fused with machine learning to map likely outcrop: 96% of known outcrops recovered while flagging under 2% of the ground.
- Client
- Exploration client (confidential)
- Commodity
- Digital targeting
- Years
- 2023 to present
A remote sensing and machine learning study on a 293 km² glaciated Canadian Shield property: Ontario's open single photon LiDAR, Sentinel-1 radar, Sentinel-2 and ASTER optical archives fused with scikit-learn, PyTorch and weights-of-evidence methods to map likely outcrop. The model recovered 96% of known outcrops while flagging under 2% of the ground for traverse, delivered in about three days, and the client found close agreement with 1976 Ontario Geological Survey bedrock mapping.